OverUnity Research

So you want to measure the power output of your "Joule Thief" or other proposed OU device but lack a DSO or a true thermal RMS measurement device. Your waveform is rather crusty and you don't trust your metering or the level is too low to make meaningful calculations. You are tired of guessing at the power by the brightness of LED's

Not to fret. I have devised a simple method to determine the power output of your flea power device. This method, however can be scaled up for higher power measurement.

Items needed: A cheap indoor / outdoor thermometer such as a Radio Shack unit, or if you can get your hands on one, an expensive Fluke dual thermocouple thermometer, a 1/4 watt resistor, some styrofoam insulation and some tape. I found the Radio Shack unit to work well enough compared to my Fluke so I used it for the charts that follow.

Any 1/4 watt (250 milliwatt) resistor will develop a certain temperature versus power input when insulated from ambient, so you will need to tape your measuring probe to the resistor and surround it with some styrofoam insulation.

Now you need to subtract off ambient so a dual thermometer is handy. I have presented a chart of data painstakingly acquired over several days time. You can make your own chart or use this one.

You don't need to use 100 Ohms, any value resistor with a relatively small body such as a 1/4 watt unit  will develop the charted temperatures when the specified power levels for that resistor are present.

So you need only to hook the 1/4 Watt resistor of your choice (value hopefully impedance matched to your device) and measure it's insulated body temperature.

Then you can read the power developed in the resistor according to the temperature rise over ambient developed by the resistor.

Best to use a metal film precision resistor if you plan on making your own chart, as it makes calculations more precise (these types introduce less drift with temperature).

Any questions?

How do I account for all that free energy I'm pumping back into the battery powering the circuit?

(No, I'm not serious but your excellent and valiant effort will need an answer for this question.  :( )

Quote from: WaveWatcher on 2013.03.24, 15:10:57
How do I account for all that free energy I'm pumping back into the battery powering the circuit?

(No, I'm not serious but your excellent and valiant effort will need an answer for this question.  :( )

Here's my answer to that:

The battery voltage should start to  overcharge and terminal voltage soar to well above nominal, in that case I would load the battery with the appropriate resistor value (and power rating) to get the voltage to level off to nominal float values. Now you can read the power returned by the temperature rise of that resistor or just measure the voltage across the resistor and use E^2 / R to see how much free power is generated.

In another method, I would replace the battery with a large capacitor, give it a kickstart and load the capacitor with the appropriate resistor to get the voltage to level off and stabilize. Then do the above.

I don't think we have to worry about that scenario yet !

Excellent suggestion ION.

Your work is very much appreciated! I would encourage everyone to become familiar with this method. Me too.

:)

Yes, excellent work ION.

When I read it earlier I just found myself nodding at the simplicity and the use of the scientific method. The line of best fit will identify replicable anomalies easily when compared to a control.

I am pretty sure that all of the current OU claimants in this area of science will not produce any results using this method, but for serious scientists it will be a big help!

RIM.

ION
Geeze we owe you for this one big time !!
I can not imagine the countless hours this will save the community ...


And yes Measurements are fun,especially when you know how to do them  :)

Where do we sign up ?

Maybe we could learn on Larry's circuit??



Thx
Chet

LT would be wise to at least check his DSO methods with an alternative. To make great claims and rely on only one test method can surely be a possible cause of error, especially if you are unskilled in DSO good practice,  skill in using the instrument and some of the fine methods POYNT has written about.

Thanks for all your support. I urge you to run a calibration test for yourself, and use this as a backup to other good instrumentation methods.

I have also run the resistor in free air, without insulation, but the temperature rise is not so high and flea power signals can get lost in the noise, so I prefer the insulated version with it's higher sensitivity.

This would not be necessary if we were talking real power such as watts or tens of watts. I started with this method to address the "Joule Thief" proponents with outputs in the milliwatt range, but the method is scalable to any power level.

In the future I'll be posting info on how to build a simple, direct readout power meter.......stay tuned.

Works great ION, i used an insulated 1/2 Watt 330Kohm resistor (need it for the Tesla/Blocking replication from Steve which works up till 1000V i guess).

I made a calibration test from 25V till 396V (max. PS).
As my 330K Ohm resistor went from 333K till 323K when heated up i Incorporated this value in the calculations

(Be aware of the decimal point seperator which is a comma in Europe and i needed to use it in my excel spreatsheet)

Lets see what this Tesla/Blocking thingy can put out.

Regards Itsu

Good work, ION.

QuoteAny 1/4 watt (250 milliwatt) resistor will develop a certain temperature versus power input when insulated from ambient, so you will need to tape your measuring probe to the resistor and surround it with some styrofoam insulation.

From your plot, we see (for example) that 500mW results in a temperature-rise of approx 83F.  How much time does it take for your insulated resistor to reach this equilibrium temperature?

Steve
I am so glad you asked the time question,I'm trying to get my head around the "Time" part
in this  well insulated protocol [as opposed to "fixed loss to ambient" Protocol ].

Thx
Chet

Quote from: PhysicsProf on 2013.03.25, 06:03:36
Good work, ION.

From your plot, we see (for example) that 500mW results in a temperature-rise of approx 83F.  How much time does it take for your insulated resistor to reach this equilibrium temperature?

I let everything soak until there was no change in the fractional least significant digit, in this case 0.1F. This would take about 15 to 20 minutes per sample.

Of course, the insulation represents a thermal resistance to ambient, and the more heavily insulated, the higher the end equilibrium temperature. I used about one square inch of styrofoam wrapped with black electrical tape. More insulation will result in more sensitivity and higher temperature points along the curve of input power, so with the lack of a "standard"  insulation technique, each person should create a calibration curve for their unique insulation set.

The whole thing can be modeled as a resistor divider network with the thermal resistance to ambient being the lower part of the divider. We can refine this and get into more detail as there is more interest generated in this approach. There are refinements in lead dress, which will allow less leakage of our internal heat buildup to the ambient surroundings.

Itsu: Good work. You now have a calibrated curve for your load resistor. Next time perhaps use a precision resistor so you don't have so much delta R vs temp.

Bear in mind that once your curve is worked out for a particular value, any resistor value of similar body type will generate the equivalent temperature profile so you can use any preferred value that will impedance match to your device under test.

Quote from: ramset on 2013.03.25, 11:01:21
Steve
I am so glad you asked the time question,I'm trying to get my head around the "Time" part
in this  well insulated protocol [as opposed to "fixed loss to ambient" Protocol ].

Thx
Chet

Chet: This is actually a fixed thermal resistance (loss) to ambient method, similar to the box or tent method, but with much better insulation. The tent or box method is for much higher output devices. We are talking flea power here.

This calibrated resistor method is for hooking up to small devices like Joule Thief circuits, where people are guessing at power out by eyeballing LED's. There are many ways to measure power, this is but one.

Hi ION
I have an idea i would like to throw your way,in reguards to P/in and P/out measurements on a pulse motor.Could we use two incandescent bulb's insted of resistor's-as these are just resistor's that show the heat by way of light.If we have our supply battery hooked in parallel with a large value cap,and between the cap and battery we have a small incandescent bulb.The pulse motor would be drawing from the large cap(path of lowest resistance)and the current would then have to pass through the incandescent bulb to keep the cap charged to run the pulse motor.

On the output of the pulse motor,we would have a large cap with the same size incandescent bulb across that cap.We can then use a temp gun to measure the heat in both bulb's,and we would also have some sort of visual reference aswell.This way we would have a smooth DC input current flowing through the input bulb,and a smooth DC current flowing through the output bulb.We could also measure the voltages across both bulbs to use as another measurement of P/in and P/out.For this to work,the cap would have to be of a fairly big value-around 20,000uf,depending on the system ofcourse.

Your thought's?

Quote from: tinman on 2013.03.25, 14:29:42
Hi ION
I have an idea i would like to throw your way,in reguards to P/in and P/out measurements on a pulse motor.Could we use two incandescent bulb's insted of resistor's-as these are just resistor's that show the heat by way of light.If we have our supply battery hooked in parallel with a large value cap,and between the cap and battery we have a small incandescent bulb.The pulse motor would be drawing from the large cap(path of lowest resistance)and the current would then have to pass through the incandescent bulb to keep the cap charged to run the pulse motor.

On the output of the pulse motor,we would have a large cap with the same size incandescent bulb across that cap.We can then use a temp gun to measure the heat in both bulb's,and we would also have some sort of visual reference aswell.This way we would have a smooth DC input current flowing through the input bulb,and a smooth DC current flowing through the output bulb.We could also measure the voltages across both bulbs to use as another measurement of P/in and P/out.For this to work,the cap would have to be of a fairly big value-around 20,000uf,depending on the system ofcourse.
Your thought's?

On the output side it would work, assuming you have a calibration curve of surface temp of the bulb versus power input for comparison.

On the input side, it is a bit trickier, since power on this side is a product of current through the bulb and source voltage. In other words to find input power, you need to multiply current times source voltage. Though we may know the current through the bulb, we cannot integrate into this the source voltage to obtain power product by a heat level alone.

So on the input side, you could use brightness of the bulb to obtain a value of current  passing through the bulb (assuming you had priorly created a curve of amps versus temperature). Once you get the relative amps from your curve, you can multiply by the source voltage to obtain power. We are also wasting a lot power to light the bulb, which is not delivered to our load, and must be accounted for.

A more exact way on the input side would be to use a one ohm precision shunt resistor in series with your voltage source. Now you can read voltage across the one ohm, which correlates directly to amps, and multiply this by the source voltage to obtain power.

I have covered the means to deal with the case of large pulses or spikes  on the shunt resistor by using a simple RC filter in front of your meter.

Checkout the posts on my bench under blocking oscillator efficiency tests.

Because many folk may not want to get into a full computer controlled thermal bridge as outlined in my other thread, here is a simpler method that can determine the power output of your DUT.

I fabricated my build with very thin strands from type "J" ( Iron and Constantan ) thermocouple wire, wrapped and glued to 1/4 watt 100 Ohm resistors.

You could use very fine Copper and Iron, or Copper and Constantan for your thermo elements.

Each side was insulated with styrofoam to reduce the effects of air drafts, and raise the sensitivity.

The output from the self  powered unit is low but is easily read on my Fluke using the uA range setting.

The first method, (direct readout) requires a calibration curve to be generated.

With the null balance method, this is not necessary. You tune your power supply voltage for null and read power output on your power supply E X I. (I prefer this method).

You could use a dual thermometer. You could also use a powered bridge with thermistors, but I wanted to keep it simple and self powered. Also it is hard to sense the heat generated by those tiny resistors.

Although designed for power under a watt, it is easily scaleable to any power level with the appropriate sized resistors.

For greater output, additional thermocouples can be wired in series, but tips must be insulated electrically.

If you really want to build a quality thermal RMS power meter, see the attached datasheet. Linear Technology made a very nice chip to do the job, it is no longer available but you can use some of the ideas presented by the late master Jim Williams. See also app note 61 by Linear Tech.

@Ion

Really interesting circuit! The LT1088 or equivalents is exactly what we need for power measurement of signals of any shape, I see no flaw. For big powers we can add external resistances, so there is no restriction. I was not aware of the existence of such circuits. They drastically simplify the RMS measurements. Thanks  O0

exnihiloest


I hope this makes it through the language barrier
Some things by there very nature are self evident ,......I hit my thumb with a hammer it will hurt.

It has been my life experience that this Model works in other areas ....certain observations become "profound"!!

When I man with Skills and experience Like ION is interested in something such as the SM device ,To me it becomes "Profoundly obvious"there is much more than "faith" at work..............

A casual dismissal is inappropriate !
thx
Chet  
PS
Sorry for the off topic ,I will remove this post by tomorrow.....

ION,

Your willingness to share such simple but excellent solutions is inspiring  O0

I think folks should use the comparative method with the DC circuit & power supply. That should convince anyone.

To think I bought a Bird 4421 for performing such measurements. Of course, at the time the customer wouldn't have accepted such a simple arrangement even though the accuracy of your method can be just as accurate as the Bird with proper application and use.

I have a couple of full reels of LM50C's I'm willing to share with fellow forum members for the price of return postage if anyone develops a method to use them accurately. I can build a POC before anyone commits to the cost of postage  ;)

Thermocouple wires, when cut from the same spool, can be very well matched , far better than any other temperature sensor except perhaps a quality platinum RTD, These other sensors are fine but must be hand selected for match.

In the self powered balance bridge method, matching is the important issue, not absolute accuracy. I have some AD590's, thermistors and some lower quality platinum RTD's on hand. For ultimate simplicity I chose the thermocouples.

I chose a self powered regime using the inherent matching that comes from thermocouples cut from the same length of wire.

Granted the temperature IC sensors and other types have higher gain, but the need for a power supply and hand matching ruled those out.

+/- 3 C initial accuracy of the LM50C would mean matching is necessary if you want accuracy in your bridge. The resistors should also be matched and capable of the expected temperatures without excessive drift.

Of course the LM50C's (or any IC sensor or thermistor) will work just fine without matching if a zero adjustment is inserted in the bridge. You could also use the neg tempco of diodes or transistor junctions in a bridge with zero adj.

Thanks for the offer on the LM50C's they are handy little devices, I might take you up on a few.

I have a reel of LMC6482's which were used in the same bridged LM50C design. I'm sure I have the calibrated voltage sources, as well. I just need to find them.

Matching shouldn't be off by much they were ordered as matched but I don't remember the specs.

These were used in a MV motor starter pole sensing circuit design I did about 15 years ago which included MLX series Hall-effects. I have a ton of those but they are useless unless programmed.

When finished, that circuit had a very long life of reliability and repeatability with an accuracy of better than 1%. At least UL & CE liked it  :)

The voltage references were key. I'll continue to look for them after all of the eggs are found and the grandkids have passed out  ;D

Quote from: ramset on 2013.03.31, 14:09:30
..
I hope this makes it through the language barrier
...

I'm afraid that this didn't. I just thanked Ion for his proposition of using the LT1088 which is a bright solution. No irony from me, the matter that I regularly see from Ion is always relevant. So I'm surprised that there is a comment about my reply and I don't understand what you mean with "dismissal"  ???.

Quote from: ramset on 2013.03.31, 14:09:30
exnihiloest


I hope this makes it through the language barrier
Some things by there very nature are self evident ,......I hit my thumb with a hammer it will hurt.

It has been my life experience that this Model works in other areas ....certain observations become "profound"!!

When I man with Skills and experience Like ION is interested in something such as the SM device ,To me it becomes "Profoundly obvious"there is much more than "faith" at work..............

A casual dismissal is inappropriate !
thx
Chet  
PS
Sorry for the off topic ,I will remove this post by tomorrow.....

?????????

HHMMMm
Since you copied the post I'll respond where this particular topic is more relevant
the TPU thread

thx
Chet

Hi Folks,

I would like to show two circuits for measuring low power (uW-mW) waveforms in a wide frequency range, besides ION's contributions. I dig the circuits from old Ham Radio Magazin articles.  This magazin was finished existing in 1990, first issued in 1968, so I hope no copyright issues involved to show some pages from it.
From the first article I show first the schematic and if there is interest to build it I have the full article to upload, alltogether 6 pages. The second article is a one page long only.
I did not build the first circuit, and I tested the second circuit many years ago with an old Selenium photo cell , wrapped it up with a grain of wheat lamp into an Alu pipe and it worked nicely. I improved it a little by including the tiny lamp in a resiistor network to make it even less frequency dependent and to have a quasi brightness independent input impedance, this way the rf current indicator role turned into approximately a 5-10% accurate rf power measuring device with reasonable bandwidth but with some ten mW and higher sensitivity.

Gyula

gyula:

Thanks for the circuit upload, and your interest in low power measurements.

Regarding the first circuit, notice that the input is  AC coupled with a 0.01 uF capacitor. This limits the low frequency passband, so the circuit as drawn is rated from 1 MHz to 500 MHz. In other words, it will quickly roll off sensitivity below 1 MHz and not work well with very low frequency signals.

It can be modified to work down to DC. The use of light bulb filaments will work, but provide an extremely non-linear input impedance and transfer function.

If you can, upload the rest of the article as I enjoy understanding the designers intent.

Thanks again.

Hi ION,

Yes I agree and I forgot to mention the 1MHz lower frequency limit, sorry.  Good quality paralleled poly capacitors of many microFarad value to replace the 100nF coupling cap C1 (together with C2)  could shift the lower frequency limit down in the kHz range, albeit still degrading sensitivity below that limit. Surely some compensation method would be in order, probably not impossible if needed.

Here are all the pages of the article.

Gyula

Thanks for the article, gyula, I learned something new.

The term "baretter" was not known to me although I had used the effect of non-linear tungsten filament in my work.

Now I have a much better understanding of why small incandescent bulbs were used in the design.

Posted here for archiveand not to clutter the bifilar thread. Thought I'd give this a try as an alternative method for testing pm's folded line.

There is a bug that needs working out, i.e. you have to know what the power factor of the DUT input is so you can compensate R1.

There are many ways to do this, but I'll leave it for now, close enough for higher COP devices.

Pre-calibration of the device would not be difficult and the chart of temperature rise vs power can be used to determine power out if needed.

Quote from: ION on 2017.05.06, 15:38:02
Posted here for archiveand not to clutter the bifilar thread. Thought I'd give this a try as an alternative method for testing pm's folded line.

There is a bug that needs working out, i.e. you have to know what the power factor of the DUT input is so you can compensate R1.

There are many ways to do this, but I'll leave it for now, close enough for higher COP devices.

Pre-calibration of the device would not be difficult and the chart of temperature rise vs power can be used to determine power out if needed.

OK,this is getting confusing  ???

Did not myself and TK carry out this very test,only where we substituted the thermocouple for the grain of wheat incandescent bulbs--see post 379 in the bifilar coil thread.

I do not know what you mean by compensating R1 in reference to the power factor?.
Are we now saying ohms law has a bug,where there are cases where the RMS value across a resistor dose not equate to the power being dissipated by that resistor?.


Brad

Quote from: TinMan on 2017.05.07, 00:47:43
OK,this is getting confusing  ???

Did not myself and TK carry out this very test,only where we substituted the thermocouple for the grain of wheat incandescent bulbs--see post 379 in the bifilar coil thread.

I do not know what you mean by compensating R1 in reference to the power factor?.
Are we now saying ohms law has a bug,where there are cases where the RMS value across a resistor dose not equate to the power being dissipated by that resistor?.
Brad

Brad

What I am saying is that the DUT input may have a power factor problem (as we have seen by the cosine) and may actually draw less real power than the R1 resistor. In effect the input of the DUT may be at a slight disadvantage in not drawing the same full power that R1 will, so if anything the bridge will read in favor of a slightly lowered COP. This is good in that we err in the right direction, if the device has a high COP, it will not be a big problem.

But this can be compensated by adjusting R1 so that it matches the real input power of the DUT. Not a big problem, I just haven't figured out a perfect way to do it yet. No, Ohms law does not have a bug. R1 will have the full real power of E^2/R, but the input to the DUT won't because of the power factor. So not the perfect test, but I have shown others ways in this thread.

e.g. R1 can be removed from it's present connection and be fed from a power supply,
and the power supply can be adjusted to null the bridge. Then E^2(from the power supply) /R will equal the actual power delivered at the output of the DUT. In this way it is like an old balance scale comparing DUT output to a reference voltage and current. See some of my earlier drawings in this thread.

Hope that helps.

Quote from: ION on 2017.05.07, 03:06:29
Brad

What I am saying is that the DUT input may have a power factor problem (as we have seen by the cosine) and may actually draw less real power than the R1 resistor. In effect the input of the DUT may be at a slight disadvantage in not drawing the same full power that R1 will, so if anything the bridge will read in favor of a slightly lowered COP. This is good in that we err in the right direction, if the device has a high COP, it will not be a big problem.

But this can be compensated by adjusting R1 so that it matches the real input power of the DUT. Not a big problem, I just haven't figured out a perfect way to do it yet. No, Ohms law does not have a bug. R1 will have the full real power of E^2/R, but the input to the DUT won't because of the power factor. So not the perfect test, but I have shown others ways in this thread.

e.g. R1 can be removed from it's present connection and be fed from a power supply,
and the power supply can be adjusted to null the bridge. Then E^2(from the power supply) /R will equal the actual power delivered at the output of the DUT. In this way it is like an old balance scale comparing DUT output to a reference voltage and current. See some of my earlier drawings in this thread.

Hope that helps.

Ah-ok,yes that helps.

Cheers

Brad

Weighing in on Low Power True RMS Measurements

Introducing (again) The Poor Man's True RMS Thermal Bridge

Well the poor mans true thermal RMS measuring technique seems not to have garnered much interest. Perhaps I did not use a catchy  enough title and did not have the code words that trigger instant views. Perhaps it is because no one likes to read technical stuff and its not a youtube video. Perhaps it is my fault for not teaching the art in a clear enough manner.

For less than five cents and some short pieces of wire, you can construct a thermal balance bridge that has many nice features for low or high power true RMS measurements of energy devices e.g

Extremely easy to build, a child could do it!

Easy to calibrate and setup

Extremely low cost for the bridge (not counting external DVM)

Able to handle different impedance on each side of the bridge by changing resistors.

Able to resolve differences of 1mW (or less depending on the care of construction.

Good resolution for signals less than 50 mW

Accurate readings not affected by crest factor, or frequency

Thermocouple temperature  meter not needed, a good quality DMM with mV,  or better yet uV resolution will work.

True thermal RMS.
-----------------------------------------------------------------------------------------------------------------
When DC or a wave of any shape is put into a resistor, it develops a heating value in the resistor.

This heating value is fairly linear, especially at lower levels of temperature.

But we need to compensate for the ambient ambient temperature so how do we do this?

Since this is a balance bridge (like a balance scale) we just use two resistors with separate thermocouples.

The thermocouples are wired to subtract and therefore the net output of the bridge is the difference between the two thermocouples,or conversely the difference in heat developed in the two resistors.

So we are measuring the difference in heat value of the two resistors.
----------------------------------------------------------------------------------------------------------------
What you will need / building it:

Two resistors, the value will depend on the impedance you want this bridge to be e.g. 50 ohms or 10,000 ohms. It is good to size the resistors to dissipate no greater mW than the expected output or range of the DUT.

e.g. I have used 1/4 Watt (250 mW) 100 ohm resistors in my bridge for low level work. You can go to 1/8 Watt or less if you want it to be really low level and fast acting, (less mass).

You will need to fashion two thermocouples, and the way I usually make it is to use thin strands pulled from stranded T/C wire. Best to use type E thermocouple wire for highest output. Why thin strands? we want a very low thermal mass and tiny thermocouples. I have also used iron and constantan, a bit lower output than type E, but choose the thermocouple wire that has the highest output over the range of 25C to 100C.

You will want to epoxy the thermocouple tips in the middle of the resistor body, using just a small amount of epoxy to keep the mass down. (improves response and settling time). 

Now there are some construction tricks we can employ to greatly improve the sensitivity and accuracy of these bridges. We will get into that later. The basic starter bridge depicted will be accurate enough for zero to 500 mW.

We will get into wire selection for the making of our thermocouples in following pages.

See the attached drawing for the construction.

Indeed, it is possible to measure a high frequency AC output power with two identical NI resistors and thermocouples or thermistors.
One resistor has the measured RF output current flowing through it and the other has a DC current generated by an op-amp in a feedback loop with the two thermocouples/thermistors.

The difficulty with it is getting identical resistors and  thermocouples or thermistors with a calibrated thermal loss to ambient and no thermal cross-talk.

Another method for measuring small RF output power is Vasik's circuit (http://www.qsl.net/sz1a/download/build%20an%20rf%20power%20meter.pdf).

Measuring input power cannot be done this way because the power dissipated in a resistor connected in series with the input terminals of the DUT is not the same as the power delivered to that DUT. 
So high frequency AC input power measurement is much more difficult. It can be done with a circuit like the one below that converts watts to volts for displaying them on a digital voltmeter.

Dear Verpies

QuoteThe problem with it is getting identical resistors and  thermocouples or thermistors with a calibrated thermal loss to ambient and no thermal cross-talk.

We do not worry about loss to ambient, as it is a null balance method, not a quantitative method.

Matching of thermocouples is not necessary if they are cut from the same roll of wire.

Cross talk can be held to a minimum with careful construction. The method is still under construction and many of your concerns have been priorly considered and simple solutions worked out, but not yet posted.

How identical would you consider the resistors need to be?...I have resistors that are 0.02% over the range of interest. There are trimming and compensating methods if need be tighter tolerance.

QuoteMeasuring input power cannot be done this way because the power dissipated in a resistor connected in series with the input terminals of the DUT is not the same as the power delivered to that DUT. 

I already explained that there are two methods to use the bridge, one method connects to the input to the DUT in which case power factor to the DUT is compensated by increasing R1

The other method (manual) is by using a power supply to balance the bridge and noting power required. Such null balance methods are tried and true for over 200 years.

I have used the method with great success over the years, and was only trying to share what I have learned from over 30 years of working in the field of thermocouple thermometry.

Lastly, I am familiar with your circuit, Vasiks circuit and other RF bridge circuits that have been published over the years, but I admit not all of them.

I find this approach works just fine, meets my needs, and solves many problems, but not all.
That's why I wanted to share it with others.

Perhaps I need to make the intent of the thread clearer as you are well respected by all on this forum including me, and your opinions carry much weight, however your caveats have cast this work in a negative light, and IMO perhaps unfairly.

If it doesn't meet your criteria or you find it objectionable, kindly disregard and by all means, don't use it.

You could also petition the administrator to have it removed with the claim that it is technically unworkable.

Regards

P.S. I think I may be wasting my time on this forum trying to be helpful, maybe time better spent elsewhere.

Your method is not objectionable.  On the contrary, it is a good method for measuring RF output power, especially when the "balancing input" is automatically fed from the output of an op-amp that has these two thermocouples at its input.

I will change the word "problem" to "difficulty" if it came across to you as too negative of a word.  My reply was meant to underline difficulties of your approach so people know what to pay special attention to when building it.  These difficulties can be solved with attention to these details so your system is not "unworkable".

Also, I was not worried about the absolute thermal loss to ambient but the differential between two such losses (for these two resistors/tc combos)

Quote from: ION on 2017.05.07, 21:20:56
P.S. I think I may be wasting my time on this forum trying to be helpful, maybe time better spent elsewhere.

You must be joking!
...or you must not realize that your posts are usually so complete that there is no need for people to add anything to them, hence you get little feedback.

Quote from: verpies on 2017.05.07, 20:08:08
...
Measuring input power cannot be done this way because the power dissipated in a resistor connected in series with the input terminals of the DUT is not the same as the power delivered to that DUT. 
...

Hi Verpies,

In that series resistor in question, suppose we learn about the RF power dissipated in it (as per ION's proposed setup),  can we use the P=I*I*R formula to get the current through the resistor? 
If yes, then ION's proposed setup would also be good for input power measurement, if we consider the RF input voltage to the DUT and we calculate input power from the thus received current and voltage values.  Would this be correct?

Regarding your schematic with the ADL5391B, it looks very good to me, is this IC still available or only the ADL5391 is available now? This latter operates to 2 GHz, I cannot find info on the B type, did it become obsolote?

Thanks,  Gyula

Quote from: gyula on 2017.05.07, 21:47:42
In that series resistor in question, suppose we learn about the RF power dissipated in it (as per ION's proposed setup),  can we use the P=I*I*R formula to get the current through the resistor? 

Yes, but you must remember, that this measurement gives you the rms average current.

Quote from: gyula on 2017.05.07, 21:47:42
If yes, then ION's proposed setup would also be good for input power measurement, if we consider the RF input voltage to the DUT and we calculate input power from the thus received current and voltage values. 
Would this be correct?

No, because in general, the product of average input current and RF input voltage does not equal average input power ^* :(

Quote from: gyula on 2017.05.07, 21:47:42
Regarding your schematic with the ADL5391B, it looks very good to me, is this IC still available or only the ADL5391 is available now?

Don't build this circuit with the ADL5391 because it will be crap!!!  Itsu did that and it did not work.
The "B" suffix means a "buffered" 1GHz chip with high input impedances and I got it in a beta sample program.
It is possible to use the unbuffered 2GHz version in this application but some changes would need to be made because the unbuffered version has low input impedances (this is on my TO DO list with Itsu).


* ^{...the product of average input current and average input voltage does not equal average input power, either (unless it is DC)}

Okay, I understand, thanks.

Gyula

Quote from: gyula on Today at 22:47:42

QuoteIf yes, then ION's proposed setup would also be good for input power measurement, if we consider the RF input voltage to the DUT and we calculate input power from the thus received current and voltage values.
    Would this be correct?

From verpies:

QuoteNo, because in general, the product of average input current and RF input voltage does not equal average input power * :(

So no one really read the part where I said it could be compensated by trimming R1 upwards to compensate for the power factor on the input to the DUT (two different places in the post) .

Also the thermal differences to ambient are not a factor when the device is constructed properly.

My intent was to offer something simple that works surprisingly well for it's simplicity and is easily built by those not advanced in the electronic art. There are numerous more sophisticated methods of accomplishing the task available, but these may require a skill level that only EE's and advanced Techs can build and properly setup and evaluate. This point was missed.

My advice, if too simple and doesn't fit your needs, don't read this thread, there are plenty of others methods available.

One can nit pick anything "ad absurdum" if one has a special need to.   ???

Think I'm wasting my time. Forget about it.

Maybe deleting the thread would be best, because elegant simplicity is sometimes not appreciated or valued in the world of complication and sophistry.

And to think it was Einstein that said: "Things should be made as simple as possible, but no simpler".

It's based on the attached but I guess the Linear Tech guys had it all wrong:  C.C

Quote from: ION on 2017.05.07, 19:46:36
Weighing in on Low Power True RMS Measurements

Introducing (again) The Poor Man's True RMS Thermal Bridge

Well the poor mans true thermal RMS measuring technique seems not to have garnered much interest. Perhaps I did not use a catchy  enough title and did not have the code words that trigger instant views. Perhaps it is because no one likes to read technical stuff and its not a youtube video. Perhaps it is my fault for not teaching the art in a clear enough manner.

For less than five cents and some short pieces of wire, you can construct a thermal balance bridge that has many nice features for low or high power true RMS measurements of energy devices e.g

Extremely easy to build, a child could do it!

Easy to calibrate and setup

Extremely low cost for the bridge (not counting external DVM)

Able to handle different impedance on each side of the bridge by changing resistors.

Able to resolve differences of 1mW (or less depending on the care of construction.

Good resolution for signals less than 50 mW

Accurate readings not affected by crest factor, or frequency

Thermocouple temperature  meter not needed, a good quality DMM with mV,  or better yet uV resolution will work.

True thermal RMS.
-----------------------------------------------------------------------------------------------------------------
When DC or a wave of any shape is put into a resistor, it develops a heating value in the resistor.

This heating value is fairly linear, especially at lower levels of temperature.

But we need to compensate for the ambient ambient temperature so how do we do this?

Since this is a balance bridge (like a balance scale) we just use two resistors with separate thermocouples.

The thermocouples are wired to subtract and therefore the net output of the bridge is the difference between the two thermocouples,or conversely the difference in heat developed in the two resistors.

So we are measuring the difference in heat value of the two resistors.
----------------------------------------------------------------------------------------------------------------
What you will need / building it:

Two resistors, the value will depend on the impedance you want this bridge to be e.g. 50 ohms or 10,000 ohms. It is good to size the resistors to dissipate no greater mW than the expected output or range of the DUT.

e.g. I have used 1/4 Watt (250 mW) 100 ohm resistors in my bridge for low level work. You can go to 1/8 Watt or less if you want it to be really low level and fast acting, (less mass).

You will need to fashion two thermocouples, and the way I usually make it is to use thin strands pulled from stranded T/C wire. You can make it with iron and copper wire. Why thin strands? we want a very low thermal mass and tiny thermocouples. I prefer to use iron and constantan, but choose the thermocouple wire that has the highest output over the range of 25C to 100C.

You will want to epoxy the thermocouple tips in the middle of the resistor body, using just a small amount of epoxy to keep the mass down. (improves response and settling time). 

Now there are some construction tricks we can employ to greatly improve the sensitivity and accuracy of these bridges. We will get into that later. The basic starter bridge depicted will be accurate enough for zero to 500 mW.

We will get into wire selection for the making of our thermocouples in following pages.

See the attached drawing for the construction.

ION

Where dose the power(voltage source) come from to put a voltage across the volt meter in your thermocouple circuit?.

OK-i am a bit lost,as it looks like your circuit shows nothing more than wire wrapped around the two resistor's,and connected to a volt meter. ???


P.S--dont take it to heart that not many people respond to your post's. \
I have asked on a number of occasions now,on the BPC thread,for help understanding my scope shot's i supplied--trying to learn.
Seems no one is interested too much in my post's either  C.C


Brad

Quote from: TinMan on 2017.05.08, 11:10:30
Where dose the power(voltage source) come from to put a voltage across the volt meter in your thermocouple circuit?.

Just two different metals put together (to form thermocouple) produce voltage themselves.
Check this link https://en.wikipedia.org/wiki/Thermocouple

Regards

PS this was used even for electric power generation http://www.douglas-self.com/MUSEUM/POWER/thermoelectric/thermoelectric.htm

Brad said:

QuoteOK-i am a bit lost,as it looks like your circuit shows nothing more than wire wrapped around the two resistor's,and connected to a volt meter. ???

Not just any wires but wires of dissimilar metals.

Two different types of wire twisted together produce a voltage when heated. The voltage of the hot junction is always in reference to the "reference junction" or cold junction. The two junctions form a differential voltage generating circuit. Thus if the two junctions are used and placed on two resistors, there will be a voltage developed if there is a difference in heating of the two resistors.

In this way you can compare e.g input power vs output power and the difference will be a voltage.

For a better understanding of thermocouples and how they work see here:

https://www.youtube.com/watch?v=JQUY_bs59a4

I was going to get into wire selection for the thermocouple in following posts, but briefly, a type "E" thermocouple (Chromel and Constantan) produces an output of 6.319 mV/100 Deg C.

You could use copper and iron wire as in the video, however in the video he is trying to generate high currents so he uses thick wire. If you just want the voltage, you can use very thin copper and iron wire which will cut down thermal cross talk between the junctions.

I'll post some pictures of the devices I have built and more info later if there is an interest.

This method produces much more resolution at levels under 50mW and is far more linear than the GOW and light meter method. See earlier posts.

Regards.

Quote from: ION on 2017.05.08, 13:09:10
Brad said:

Not just any wires but wires of dissimilar metals.

Two different types of wire twisted together produce a voltage when heated. The voltage of the ht junction is always in reference to the "reference junction". The two junctions form a differential voltage generating circuit. Thus if the two junctions are used and placed on two resistors, there will be an voltage developed if there is a difference in heating of the two resistors.

For a better understanding of thermocouples and how they work see here:

https://www.youtube.com/watch?v=JQUY_bs59a4

I was going to get into wire selection for the thermocouple in following posts, but briefly, a type "E" thermocouple (Chromel and Constantan) produces an output of 6.319 uV/Deg C.

You could use copper and iron wire as in the video, however in the video he is trying to generate high currents so he uses thick wire. If you just want the voltage, you can use very thin copper and iron wire which will cut down thermal cross talk between the junctions.



This method produces much more resolution at levels under 50mW and is far more linear than the GOW and light meter method. See earlier posts.

Regards.

QuoteI'll post some pictures of the devices I have built and more info later if there is an interest.

Yes,there is much interest--from me anyway  O0


Brad

Brad

I am still looking for info on the best wire to use and e.g. how much iron and copper will produce. I have in the past used types J, T, and E, but would like to find out more about iron and copper, since everyone will have that on hand.

Attached is a brief tutorial on thermocouples. Pay special attention to the second drawing to understand thermocouples as differential voltage generating devices.

Since we don't expect to go over 100C the wire types need not be exotic.

Also attached is a pic of my first very crude (but worked well) differential power measurement box, with top styrofoam cover removed.

Later versions were more advanced, this was only for proof of principle when I  first started the thread. The graphs in the beginning of the thread were obtained using this box. 1/4 Watt resistors carbon film were used, Iron-Constantan TC wire, and cyano glued to resistors.

More to come.

Regards

Quote from: ION on 2017.05.07, 23:10:20
So no one really read the part where I said it could be compensated by trimming R1 upwards to compensate for the power factor on the input to the DUT...

I read it but it made no sense to me for measuring arbitrary AC input power.
Perhaps you could elaborate what data you hope to obtain from the dual R/TC arrangement and how can that data be translated to input power of an arbitrary DUT.

Please make a schematic of such measurement.  Just copy/paste the LT1088 diagram and connect it up to a DUT's input terminals and an arbitrary AC source using MS-Paint or similar.  It does not have to be pretty.

Quote from: ION on 2017.05.07, 23:10:20
My advice, if too simple and doesn't fit your needs, don't read this thread,...

Simple is beautiful and it suits my needs for RF output power measurement if the thermal symmetry is well maintained.

Quote from: ION on 2017.05.07, 23:10:20
Also the thermal differences to ambient are not a factor when the device is constructed properly.

Does that mean it is easy to construct so that it maintains thermal symmetry and immunity to external temperature gradients ?

Quote from: ION on 2017.05.07, 23:10:20
It's based on the attached but I guess the Linear Tech guys had it all wrong:  C.C

They had it right for an rms-dc current converter, but how can this device measure arbitrary input power?
BTW: That's what I meant by driving the "balancing input" by an op-amp.

Quote from: verpies on 2017.05.07, 21:45:16
Your method is not objectionable.  On the contrary, it is a good method for measuring RF output power, especially when the "balancing input" is automatically fed from the output of an op-amp that has these two thermocouples at its input.

Brad: Here is a starter block diagram for how to measure e.g the power output of a device that has an arbitrary waveform. It does not address the input power measurement, that will be in the next block diagram.

Notice that this schematic requires the operator to null the system. The operator can be replaced by using an op amp that servoes the balancing resistor based on the error signal. Interesting improvements arise when this is done e.g. a slow frequency sweep of the input will give a plot of power output by either monitoring and plotting the power supply output or the servo correction signal output. This will help identify the most efficient operating points.  I have not yet drawn this diagram but will as time permits.

Regarding the question of ambient temperature effects, they are automatically nulled out by virtue of the differential measuring method e.g. as ambient temperature rises, the thermal emf of each thermocouple rises but since they are connected as a differential measurement, the ambient is always subtracted out automatically.

Thermal symmetry can be quite acceptable if relatively large styrofoam blocks e.g. 1" x 2" x6" are used to sandwich the resistors and thermocouples.

Remember, we are not trying to get the performance of a $1000 instrument, just a tool that can be built for under a dollar or less for the experimenter on a budget, not for the EE, who will surely scoff at such an attempt.

The input power measuring and balancing against output power will possibly be the next block diagram as time permits.

Quote from: ION on 2017.05.08, 20:53:33
Brad: Here is a starter block diagram for how to measure e.g the power output of a device that has an arbitrary waveform. It does not address the input power measurement, that will be in the next block diagram.

Notice that this schematic requires the operator to null the system. The operator can be replaced by using an op amp that servoes the balancing resistor based on the error signal. Interesting improvements arise when this is done e.g. a slow frequency sweep of the input will give a plot of power output by either monitoring and plotting the power supply output or the servo correction signal output. This will help identify the most efficient operating points.  I have not yet drawn this diagram but will as time permits.

Regarding the question of ambient temperature effects, they are automatically nulled out by virtue of the differential measuring method e.g. as ambient temperature rises, the thermal emf of each thermocouple rises but since they are connected as a differential measurement, the ambient is always subtracted out automatically.

Thermal symmetry can be quite acceptable if relatively large styrofoam blocks e.g. 1" x 2" x6" are used to sandwich the resistors and thermocouples.

Remember, we are not trying to get the performance of a $1000 instrument, just a tool that can be built for under a dollar or less for the experimenter on a budget, not for the EE, who will surely scoff at such an attempt.

The input power measuring and balancing against output power will possibly be the next block diagram as time permits.

Ok,looks good.

So i will need a uV volt meter , analog  is ok ?.

Brad

Quote from: TinMan on 2017.05.08, 22:35:02
Ok,looks good.
So i will need a uV volt meter , analog  is ok ?.
Brad

Some of the better DMM have a uV scale. I wouldn't buy anything just yet until we research what is available. Meanwhile you could try with your present better grade DMM. Sometimes just a simple analog uA meter will work. I prefer the center zero deviation types for that. Let me take some time to see what is out there.

More importantly you need to get  a small amount (approx 6 inches) of TC wire , fine stranded type so that we can pull separate fine strands or just single strand but around 26 to 32 AWG. I'll have to do some research where to get it.
The other types J, T, and K will also work at slightly lower output. J would be the next choice.

E is 6.319 mV / 100 Deg C or approx 63 uV / Deg C (Chromel-Constantan)
J is 5.269 mV / 100 Deg C or approx 53 uV / Deg C (Iron-Constantan)
T is 4.279 mV / 100 Deg C or approx 43 uV / Deg C (Copper-Constantan)
K is 4.096 mv / 100 Deg C or approx 41 uV / Deg C (Chromel Alumel)

I tried plain old copper and iron fine gauge wire, but the output was way too low, so that's out.

Regards

I want to add that the circuit that verpies has posted here in reply #32 and elsewhere on this forum and other forums of the wideband DC to 1GHz wattmeter using ADL5391B is an excellent choice and should be used if you have the skill and time to construct it.

I have often commented very positively in regard to the use of it. I believe Farnell and Mouser used to have an evaluation kit for the I/C. So if you want to build a lab quality instrument that is the way to go. As a matter of fact I would consider to purchase an eval board ($253)  but can't as yet find them in stock anywhere.

Regarding some of the other RF power meters that have been referenced, they are high frequency RF grade and do not go down to DC or low frequency which is one of the criteria I would want as some of the devices I hope to evaluate would go to LF or near DC.

Of course a pure DC output device would not at all need exotic measuring equipment.

Regards

Edit: found the I/C in stock at Digikey

Quote from: ION on 2017.05.08, 13:28:00
Also attached is a pic of my first very crude (but worked well) differential power measurement box, with top styrofoam cover removed.

Wrapping the resistors with the thermocouple wire is not a good idea because that thermocouple winding will pick up the magnetic flux variations from the current flowing in the resistors and the induced voltage will swamp the thermocouple signal approximately 100000 times.

See the scopeshot below for a 10Ω ¼W THT resistor wrapped with 7 turns of wire while a 1MHz 340mA_P-P square AC is flowing through it.
Green trace is the current through the resistor and the yellow trace is the voltage induced in the 7 turns of wire (4.8V_P-P).

Dear ION.

I found... wires.com seems to stock most of everything, great for experimenters this side of the pond.

With regards to Verpies's circuit I could ask my friend Richard at RM Cybernetics about building it.

If a suitable PCB layout was produced and presented to him the price shouldn't be too prohibitive.

Cheers Graham.

Dear Verpies.

Would a " K " type thermocouple bead Epoxied to the resistor body affect results?

Cheers Graham.

Quote from: Grumage on 2017.05.09, 10:58:33
Would a " K " type thermocouple bead Epoxied to the resistor body affect results?

It would make it slower to respond but it would be more immune to the magnetic induction/interference from the current flowing in the resistor.

OK,i have just had a brain wave-or brain fart  :D

Wont say now,but will throw it together tomorrow,and present a video of my idea here,and you guys can tell me what you think.

This will eliminate any inductive problem,and can be fine tuned to be exact-to show us the smallest difference in actual current on the input,and output.

Should cost all of about $5.00 to put together,but i would think we would all have the required parts lying around,and any DMM should work ok in showing us the difference,as the amplitude would be in mV,not uV


Brad.

Quote from: verpies on 2017.05.09, 10:27:46
Wrapping the resistors with the thermocouple wire is not a good idea because that thermocouple winding will pick up the magnetic flux variations from the current flowing in the resistors and the induced voltage will swamp the thermocouple signal approximately 100000 times.

See the scopeshot below for a 10Ω ¼W THT resistor wrapped with 7 turns of wire while a 1MHz 340mA_P-P square AC is flowing through it.
Green trace is the current through the resistor and the yellow trace is the voltage induced in the 7 turns of wire (4.8V_P-P).

Dear Verpies

I'm trying to figure out how you got (7 turns) from the picture I posted, since it only shows one half turn of the twisted portion of the junction. Also you do not explain what load you had on the 7 turns. As you know an unloaded coil will have quite a self resonant rise and tend to exaggerate the voltage produced across it, even by just capacitance coupling from a nearby object.

1) I never suggested anyone should use 7 turns of TC lead wire around the resistor, I don't know where that came from. I use only one half turn of the twisted portion that forms the thermocouple junction. Since it is twisted it is immune to current in the resistor except perhaps  for some common mode signal capacitively coupled. The common mode as well as series mode AC signal is rejected by most good DC uV meters, as generally it would be very high frequency, low level. The output of the thermocouple is DC voltage or current depending on how it is used.

2) The resistor you used may have been "spiral cut" to trim it's value. This creates a few turns on the resistor film which can indeed inductively couple to the 7 turns, As you know, a wire passing through the center of a coil cannot induce current into the coil by Faraday induction because it is at right angles 90 degrees to the coil. It can however excite the turns capacitively, which is what I see in your scope  shot.

3) As I said earlier, there are improvements that I would discuss at a later time so as not to overly complicate the basic device. A couple of the improvements are to add ferrite beads to the thermocouple leads both series mode and common mode to attenuate any capacitive or inductive coupling effects. I did not complete the fine points of advanced construction yet.

4) Depending on the skill of the constructor at forming  a tiny thermocouple junction, it as not at all necesary to wrap the junction or it's associated wires around the body of the resistor. A tiny bead of epoxy will thermally connect the tiny junction to the resistor.

5) There is a well known  technique  that should the constructor wish to add a turn or two of the wires around the body of the resistor, it can be done in a  in a non-inductive manner. It is the same method used to make non-inductive wirewound resistors. Why would an extra turn or two help?
The value of the extra turn is more precise temperature reading by also heating the thermocouple wires leading to the junction so that the leads do not adversely drain heat from the junction.
--------------------------------------------------------------------------------------------------------------
There seems to be an intent on killing this method before it is "off the ground". Trying to warn others that it is a bad method is probably working, having it's effect by now.  :(

I have not yet had a chance to post the finer details of the method such as the HF rejection curves.
Sorry for that, I have a lot of domestic and health related issues to deal with right now.

P.S Brad

As I mentioned in the beginning there are many ways to sense the temperature of the resistors e.g. thermistors, Thermal IC chips, diodes. You could build a bridge that uses any of those devices but it will not be passive, will need a power supply, and will have a propensity to pick up HF noise that would not be easily filtered due to the non-linearity and rectification of some of the devices.

I chose thermocouples because I am quite experienced in using them and they solve a lot of the problems that would be introduced with other temperature sensing methods and it is a passive method, not requiring a power supply, which could introduce additional capacitance coupling.

Also I have a lot of good uV meters and measuring instruments in the shop. Nowadays they are very common. Back 50   years ago they were more difficult to construct, usually requiring chopper stabilised DC amplifiers.
When I get time I will try to demonstrate using a passive analog 50-0-50 uA meter.

To each his own.

I'm gonna take a break from this jousting session, it's bad for my health.

Quote from: Grumage on 2017.05.09, 10:35:32
Dear ION.

I found... wires.com seems to stock most of everything, great for experimenters this side of the pond.

With regards to Verpies's circuit I could ask my friend Richard at RM Cybernetics about building it.

If a suitable PCB layout was produced and presented to him the price shouldn't be too prohibitive.

Cheers Graham.

Dear Graham

I would recommend that you not bother with building the thermal RMS bridge, that I have used over the years with success. Apparently, it has been deemed to have too many theoretical problems that I have yet been able to locate.  ;)
-------------------------------------------------------------------------------------------------------
Regarding the circuit that verpies posted, attached is the pdf with the circuit and PCB layout for the evaluation board right from the manufacturer of the IC. Notice the use of a few chokes and capacitors to keep HF out of the device and to provide transformation, which have not been included in the circuit verpies posted, and may not even be necessary. You may wish to read this document and evaluate their need or not and modify appropriately before boards are produced. Maybe check with verpies on that. I believe the chokes are intended for singled ended to balanced transformation for AC signals but may be removed for DC operation so they may not be needed.

Regards

Quote from: ION on 2017.05.09, 15:42:07
Maybe check with verpies on that.

Dear ION.

That's probably a good idea....

I can't seem to find the Carburettor anywhere.....  :o

Cheers Graham.

Quote from: ION on 2017.05.09, 15:42:07
I would recommend that you not bother with building the thermal RMS bridge.

Grum is probably going to throw up his hands and run away from his monitor when he reads that in my opinion the dual thermal bridge is right up his alley and in my opinion he should build it because he understands heat flow well and a sub milimeter precision of mechanical assembly is not anything new to him. He also has intuitive understanding that closely spaced parallel wires interfere with each other.

Quote from: ION on 2017.05.09, 15:42:07
Regarding the circuit that verpies posted, attached is the pdf with the circuit and PCB layout for the evaluation board right from the manufacturer of the IC. Notice the use of a few chokes and capacitors to keep HF out of the device, which have not been included in the circuit verpies posted. You may wish to read this document and evaluate their need or not and modify appropriately before boards are produced. Maybe check with verpies on that.

That evaluation board is not DC-coupled, it performs a different function and uses more pins of the multiplier chip.
Anyway, I do not recommend that Grum gets involved in  building it until  he can have a debugged version  of the Watt2Volt converter PCB sent to him for the "A" version of the chip.

Quote from: ION on 2017.05.09, 14:34:58
I'm trying to figure out how you got (7 turns) from the picture I posted, since it only shows one half turn of the twisted portion of the junction.

It  was a low resolution picture and it looked to me this way. With half a turn I would expect the induced EMF to be 14 times smaller but that is still a lot  compared to the  thermocouple signal.

Quote from: ION on 2017.05.09, 14:34:58
Also you do not explain what load you had on the 7 turns.

Just the scope probe.

Quote from: ION on 2017.05.09, 14:34:58
As you know an unloaded coil will have quite a self resonant rise and tend to exaggerate the voltage produced across it,

Not with a 1MHz signal which  is  far away with such resonance.

Quote from: ION on 2017.05.09, 14:34:58
...even by just capacitance coupling from a nearby object.
I use only one half turn of the twisted portion that forms the thermocouple junction. Since it is twisted it is immune to current in the resistor except perhaps  for some common mode signal capacitively coupled. The common mode as well as series mode AC signal is rejected by most good DC uV meters, as generally it would be very high frequency, low level. The output of the thermocouple is DC voltage or current depending on how it is used.
3) As I said earlier, there are improvements that I would discuss at a later time so as not to overly complicate the basic device. A couple of the improvements are to add ferrite beads to the thermocouple leads both series mode and common mode to attenuate any capacitive or inductive coupling effects. I did not complete the fine points of advanced construction yet.

Capactive coupling could be an issue, too. 
You  should post scopeshots of thermocouple interference tests with a 1MHz 100mA square AC current flowing through the heater/resistor.

Quote from: ION on 2017.05.09, 14:34:58
2) The resistor you used may have been "spiral cut" to trim it's value. This creates a few turns on the resistor film which can indeed inductively couple to the 7 turns,

Not in this case but it is a valid concern and you should warn potential builder about such Gotcha.

Quote from: ION on 2017.05.09, 14:34:58
Why would an extra turn or two help?

Because the EMF from the extra turns can be arranged to cancel  itself, as in a bifilar coil of the  3rd kind.
Also if half of the coil has an equal and opposite pitch then it also cancels the helix's axial current.

Quote from: ION on 2017.05.09, 14:34:58
As you know, a wire passing through the center of a coil cannot induce current into the coil by Faraday induction because it is at right angles 90 degrees to the coil.

Yes, but when the distances are comparable to the diameer of the wire, the near field  is so twisted that it can still cause induction.  Even the pitch of the helix can matter.

Dear Verpies

QuoteQuote from: ION on Today at 15:34:58

    Why would an extra turn or two help?

Because the EMF from the extra turns can be arranged to cancel  itself, as in a bifilar coil of the  3rd kind.
Also if half of the coil has an equal and opposite pitch then it also cancels the helix's axial current.

In this statement, I was asking the question rhetorically as a prelude to the follow up sentence and you did not have to answer, as I had already stated how to create the non inductive wind  in the prior sentence:

"it can be done in a  in a non-inductive manner. It is the same method used to make non-inductive wirewound resistors."

"Why would an extra turn or two help?" I asked (rhetorically)

I was posing different  reason for adding one or two turns and why you would even want to do it, namely:

"The value of the extra turn is more precise temperature reading by also heating the thermocouple wires leading to the junction so that the leads do not adversely drain heat from the junction."

Maybe we have a language barrier here.

Regarding the interference of the RF signal by injection into the thermocouple, I can assure you it is not a problem.

I tested it with both a Kiethley 155 Null Microvolt meter and a Fluke 87. I  had a steady readout of around 4 mV with the full input of 6.72 volts from the FG  into the resistor, up to 5 MHz. Quickly switching on and off the generator did not affect the readings on either of the meters. The readings only changed if there was a long interval after switchoff, as the resistor cooled down, the readings were not visibly affected by the RF.

I looked at the actual signal on the scope and at 5 MHz it was in the range of 30 mV AC, but this is not a problem, as you see, most good instruments that are designed to read low level uV DC also reject HF AC quite well, and 30mV of AC is nothing.

I have worked in the temperature and process control industry for half a lifetime as a chief engineer designing equipment that used thermocouples and the  front end amplifiers had to reject everything from noise on a 480 volt mains to HF RF from local transmitters. These were sometimes difficult problems to solve, but solve them we did, and usually very simply.

I see the problems you keep bringing and your concerns as quite  trivial by comparison and if my time were not so valuable right now,  I would flood this page with information showing that your concerns are not an issue.

This exercise very much reminds me of the numerous exchanges I had with customers that were Ph.D's that had a lot of  theoretical but very little practical experience. It is becoming tiresome to say the least. What do you gain by continuing on like this? Why don't you just  build it and see that it works just fine and is not affected  by your caveats or just stand by and watch this fail and have a good laugh (it won't, I can assure you).

Regards

Quote from: ION on 2017.05.09, 20:02:50
The value of the extra turn is more precise temperature reading by also heating the thermocouple wires leading to the junction so that the leads do not adversely drain heat from the junction.
Maybe we have a language barrier here.

We don't.  Notice, that I did not disagree with the above - I just added more advantages from my perspective.

Quote from: ION on 2017.05.09, 20:02:50
Regarding the interference of the RF signal by injection into the thermocouple, I can assure you it is not a problem.
I tested it with both a Kiethley 155 Null Microvolt meter and a Fluke 87. I  had a steady readout of around 4 mV with the full input of 6.72 volts from the FG  into the resistor, up to 5 MHz. Quickly switching on and off the generator did not affect the readings on either of the meters. The readings only changed if there was a long interval after switchoff, as the resistor cooled down, the readings were not visibly affected by the RF.
I looked at the actual signal on the scope and at 5 MHz it was in the range of 30 mV AC, but this is not a problem, as you see, most good instruments that are designed to read low level uV DC also reject HF AC quite well, and 30mV of AC is nothing.

So you are aware that there is 30mV AC HF interference superimposed on several μV/ºC thermal signal in your system.
Wouldn't an ounce of prevention (avoiding the interference) be worth more than a pound of cure (low pass filtering)?  The latter is not infallible.
 
Did you test it with asymmetrical HF waveforms, that create a DC component after low pass filtering?

Quote from: ION on 2017.05.09, 20:02:50
What do you gain by continuing on like this?

Exploration. Advancement of knowledge.
Qualitative and quantitative characterization. Perfection of the design.
List of Gotchas for future builders. 

Aren't these the very reasons for existence of forums like this ?

Quote from: ION on 2017.05.09, 20:02:50
Why don't you just  build it and see that it works just fine and is not affected by your caveats

My caveats?  I thought that caveats were not personal...
A 1 turn wrap around a 10Ω resistor produces interference like this:

From Verpies

QuoteDid you test it with asymmetrical HF waveforms, that create a DC component after low pass filtering?

Asymetrical signals do not communicate a DC signal through a capacitance nor through a transformer by induction, even after filtering. I'm sure you know this or you could rewrite the physics books by being able to pass DC through a transformer or a capacitor. After the capacitor or transformer the average of the wave is always zero DC regardless of the input waveshape.
Only when there is leakage, rectification or asymetrical saturation of a device could the result have a DC component.

QuoteWouldn't an ounce of prevention (avoiding the interference) be worth more than a pound of cure (low pass filtering)?  The latter is not infallible.

The filtering is already in the instruments I use to measure the TC signals. If additional filtering were necessary, I would certainly use it. I did mentioned using common mode and series mode ferrite beads.
I haven't even gotten to the finer points in the method of discussing additional filtering, where and why it might be needed, because I didn't want to cloud the basic understanding of the principle.

QuoteSo you are aware that there is 30mV AC HF interference superimposed on several μV/ºC thermal signal in your system.
Wouldn't an ounce of prevention (avoiding the interference) be worth more than a pound of cure (low pass filtering)?  The latter is not infallible.

If it were needed I would certainly us it, the 30mV AC is well filtered by the low pass filter of the measuring instrument. Precaution noted.

QuoteExploration. Advancement of knowledge.
Qualitative and quantitative characterization. Perfection of the design.
List of Gotchas for future builders. Aren't these the very reasons for existence of forums like this ?

*I'm beginning to believe forums serve a different, not so benign purpose, and what I wrote recently about brotherhood was indeed very naive of me.

QuoteMy caveats?  I thought that caveats were not personal...

Answered in part in prior sentence* They can be subtly personal.

QuoteA 1 turn wrap around a 10Ω resistor produces interference like this:

You call it interference when seen on a scope but it's only interference if it can register an effect on the thermocouple DC measuring instrument. Most DC measuring instruments (even common DMM's) are designed to handle much larger interfering signals and reject them quite well, allowing only the expected DC component of the TC to pass.

Regarding your scope shot, it shows the typical signal I would expect to find on the TC leads, or even just an unconnected scope probe in the vicinity of the resistor, and it is of little consequence compared to the much larger AC signals on TC leads that I have successfully coped with in the past without difficulty as most skilled in the art have.

Regards
 

Well i didnt get time to get the whole current differential device together,due to a late finish at work,and friends turning up.

But i did get to test the current input side,and the results are better than expected  O0
For every mA of current increase,i can get a 5mV increase on the DMM-so 5mV for every mA increase  ;)

HF,stray inductance,and the likes,will have no effect on this devices measurements.

Being that it works so well,and with the twist of a knob,can be calibrated with ease,i am going to spend the money,and get all new parts,and build it properly.

So,in saying that,i will hold of on the video,until such time as i have a presentable device.


Brad

Quote from: TinMan on 2017.05.10, 10:53:52
Well i didnt get time to get the whole current differential device together,due to a late finish at work,and friends turning up.

But i did get to test the current input side,and the results are better than expected  O0
For every mA of current increase,i can get a 5mV increase on the DMM-so 5mV for every mA increase  ;)

HF,stray inductance,and the likes,will have no effect on this devices measurements.

Being that it works so well,and with the twist of a knob,can be calibrated with ease,i am going to spend the money,and get all new parts,and build it properly.

So,in saying that,i will hold of on the video,until such time as i have a presentable device.
Brad

Brad
It will be very interesting to see what you have come up with. O0

Meanwhile attached is a block diagram of how to test a common JT or Blocking Oscillator using the thermal bridge. It will give a fairly accurate indication of true RMS power output. Accuracy depends on how well your bridge is constructed, and how well insulated with respect to ambient and each side of the bridge (crosstalk). There are many ways to improve the basic idea and it's performance, but what I have presented thus far is for conceptual and initial  understanding purposes, by no means a finished, perfected device.

As you know the output of a JT presents an asymmetrical current pulse into a load resistor. Use this method as a backup to the values you are getting on a scope measuring the output.
Input to the JT is measured the usual way, since it is DC.

Regards

Nice setup Ernie, with the TC bridge.   :)

Is there any calibration involved?

Quote from: ION on 2017.05.10, 12:32:08
Brad
It will be very interesting to see what you have come up with. O0

Meanwhile attached is a block diagram of how to test a common JT or Blocking Oscillator using the thermal bridge. It will give a fairly accurate indication of true RMS power output. Accuracy depends on how well your bridge is constructed, and how well insulated with respect to ambient and each side of the bridge (crosstalk). There are many ways to improve the basic idea and it's performance, but what I have presented thus far is for conceptual and initial  understanding purposes, by no means a finished, perfected device.

As you know the output of a JT presents an asymmetrical current pulse into a load resistor. Use this method as a backup to the values you are getting on a scope measuring the output.
Input to the JT is measured the usual way, since it is DC.

Regards

Looks good ION,and i believe that my setup can be used to calculate the P/out of a JT.

One thing to note though.
You cant place a resistor where the LED go's in a JT,as you have in your schematic.
You would need to add a diode in series with the resistor,and then some how account for the power dissipated by that diode as well as the resistor.

With my design,you would only need the LED,as per the standard JT.

I should be collecting my new 4 channel scope on friday,and hopefully my copper tape rolls will arrive in the mail on friday also.

So this weekend is going to be full on,as i want to build this equivalent to your thermocouple device as well.

I see a late friday night coming on  C.C


Brad

Ah..... oh to be a little younger Brad...

I love the elegance in its simplicity ION, nice one.  O0

Regarding calibration, well it's all a bit " hit n miss " without a visit to your local standards laboratory but, most modern test instruments are pretty good, aren't they?

I spent some time in the test and standards lab calibrating what were called " sub standard " instruments for field use. The " standard volt and amp " meters were kept in amazing housings!!  Oh they were all analogue back then, those were the days!

Gotta go.... back later.

Quote from: poynt99 on 2017.05.10, 12:44:02
Nice setup Ernie, with the TC bridge.   :)

Is there any calibration involved?

The really interesting thing about the bridge is that the resistors don't have to be matched in resistance, just in size. In fact, they can be widely different in resistance as long as they are the same approximate physical size and power rating e.g. for power measurements under
250 mW 1/4 or 1/8 watt resistors are ideal. For higher power measurements, the resistors can be scaled up accordingly in size.

I know this all sound counter intuitive, but once you begin thinking in the realm of heat production and EMF from heat, it will start to make sense.

I imagine it should be possible to build a device that can measure levels in the  0 to 50 mW range with acceptable accuracy.

At some point I want to build one that uses tiny surface mount resistors and see how low it can go, and maintain some decent level of accuracy.

I'm sure you know the technique of using heat and thermocouples to generate mV to drive a meter movement is actually used in some older RF ampmeters.

Because the bridge is immune to small resistance changes, we don't have to worry about slight drifts in the resistors. It may be difficult to understand this but once you begin thinking in terms of the final result of generating heat, the value on the right side R2  can be anything that matches the voltage range your power supply, and the value of R1 can be roughly impedance matched to the DUT.

If you wish to see how well the bridge balances, you can apply the same power level to each resistor, you should get a null output if the bridge is properly constructed, if not there are ways to trim and calibrate the TC mV output of the system. I will get into that later, but calibration should not be required if physical symmetry of the constructed device is maintained.

Regards

P.S. Thermocouples are used in some vacuum measuring bulbs where the filament temperature is related to the number of molecules the filament can transfer heat to,
so a higher vacuum = higher filament temperature=more TC output.

Here are a few graphs from two different builds, using different components. One is a round styrofoam chamber, the other is flat sandwich construction.

As you can see from the graphs, there is  a healthy output even under 50 mW.

Tracking was also very good, as I put voltage into both resistors in parallel to observe the deviation at several points (which was acceptable) .

I'll be moving on from this to some new ideas, but will continue to work in the background perfecting the method and going to even lower values of milliwatts.

Some of the ideas I will be working on in background:
a)  Eliminating the need for a thermocouple/heater combination by using a low slope PTC Thermistor or Platinum RTD of the non inductive thin film type as both load and sensor.

b) for very small mW measurement, using the tempco of a heating element (filament) inside a vacuum tube as both load and sensor. As you may know some of these bismuth coated filament rods are folded back and forth and are low inductance. Just a fun experiment.

c) measuring the emission vs filament voltage of e.g. a 6AL5 dual diode vacuum tube, both passively, Cathode to Anode mVolts or with a plate voltage applied. Initial tests look very interesting, very low to no emission under 3 Volts on the filament, but quite stable and high output above that approximately up to 500 mV at 6 volts on the filament, passively measured A to K.

d) using an infra red sensor to accurately read the IR energy from the load resistor in a non contact manner.

By the way, if you wish to know how small the power we are chasing really is, take a 100 Ohm 1/4 watt resistor and put 5 volts into it. This is 250 mW. You can just barely hold it to a spot slightly below the lower lip.(I found this to be an accurate spot on the body to detect temperature, there are probably others, but I won't go there, that's for your own private experiments) This is a reasonable amount of power as you almost can't keep it touching.

Now keep reducing the voltage until you can no longer sense the heat. What mW did you find it to be? For me it was around 40 to 60 mW where any warmth was first detected, but I'm almost dead anyway so maybe someone else will get better results.

Probably better to run this experiment working up from zero.

Regards.