PopularFX
Home Help Search Login Register
Welcome,Guest. Please login or register.
2019-03-24, 17:28:55
News: If you have a suggestion or need for a new board title, please PM the Admins.

Pages: 1 ... 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 [19] 20 21 22 23 24
Author Topic: Investigating "anomalies" in Bifilar coils  (Read 122038 times)

Hero Member
*****

Posts: 1644

I can confirm, COP>1 when using a 2.2nF cap.

@ 1.6MHz with and without the 1:1 toroid, both cases the calculated output is higher then the input.
Screenshot 1 is the input power (non 1:1 toroid)
Screenshot 2 ia the output power (non 1:1 toroid9.


Regards Itsu
   

Hero Member
*****

Posts: 1298
Almost, but it is like pissing in its gas tank and expecting it to run well.

In this case, the signal coming from the Current Sensing Resistor was erroneous by 215% too small.
The scope was doing it's job properly with this garbage signal. Garbage In - Garbage Out.

Is there a similar problem with the scopeshots posted in #436?




---------------------------
"The easiest person to fool is yourself" -- Richard Feynman
   

Hero Member
*****

Posts: 1298
OK, I've been fiddling around trying to get the bulb brightness system to give some intelligible numbers. I took some DC brightness data, as I said yesterday, plotting Lux vs. DC power using a power supply, the Fluke 87-iii for voltage and another DMM for current. I also calculated the filament resistance based on R=V/I (or R=V2/P, same thing) and plotted it on the same chart.  So if you read a Lux value from the lightmeter, you then can find the Filament Resistance and the equivalent DC Power on the chart.

So I just ran one preliminary test just to see what the numbers would do.

This is using the original PBT coil set with the load resistor removed and the GOW bulb in its place,  the new coupling transformer, and the F43 FG.

So.... The Lux meter reads 3.49 Lux. The scope's Math Average input power is 108 mW. The phase shift is -78.41 degrees (taken at higher resolution than the scopeshot below). The manual Input Power calculation from the scope's values gives 117 mW. Reading the Chart for 3.49 (actually 3.5) Lux gives a DC equivalent of 115 mW and a filament resistance of 73 ohms. Manual calculation of power out from the CH3 voltage and the filament resistance gives 109 mW.

So this method pretty much hangs together without major errors, I think.

Now to find some "sweet spot" where the numbers might indicate OU.


---------------------------
"The easiest person to fool is yourself" -- Richard Feynman
   

Hero Member
*****

Posts: 3038


Buy me some coffee


In this case, the signal coming from the Current Sensing Resistor was erroneous ( 215% too small ).
The scope was doing its job properly with this garbage signal ...and Garbage In = Garbage Out.

So precision non inductive resistors are no good as CVR's ?

And now confirmation by Itsu of an OU capacitor  O0  :D

Quote
Almost, but it is like pissing in its gas tank and expecting it to run well.

Well i guess that would depend on how much alcohol you had to drink the night before.
If you are pi*sing in your lawnmowers fuel tank,one would think you had plenty to drink the nigh before.


Brad


---------------------------
Never let your schooling get in the way of your education.
   

Hero Member
*****

Posts: 1298
Another data set at 238 kHz:
Lux = 1.94, and from graph this = 94 mW DC equivalent and 68 ohms DC resistance
Math average In = 88.4 mW
Manual In calculation 13.4 x 0.0409 x cos (-78.48) = 109.5 mW
Out calculation 2.532/68 = 94 mW


---------------------------
"The easiest person to fool is yourself" -- Richard Feynman
   

Hero Member
*****

Posts: 1298
So precision non inductive resistors are no good as CVR's ?
It is possible I had a bad connection, although I did wiggle stuff around. Later tests were much more stable though, and gave much lower, but still "ou", COP.

Quote

And now confirmation by Itsu of an OU capacitor  O0  :D

 :D   :o

Quote

Well i guess that would depend on how much alcohol you had to drink the night before.
If you are pi*sing in your lawnmowers fuel tank,one would think you had plenty to drink the nigh before.


Brad


---------------------------
"The easiest person to fool is yourself" -- Richard Feynman
   

Hero Member
*****

Posts: 1298
And another data set:
0.83 Lux, which gives 72 mW and 63.5 ohms from the chart.
Input Math average = 72.9 mW
Manual Input calculation 13.7 x 0.0364 x cos -81.39 = 74.7 mW
Output calculation 2.112/63.5 = 70.1 mW


---------------------------
"The easiest person to fool is yourself" -- Richard Feynman
   

Hero Member
*****

Posts: 1298
OK, last one for now. I tuned the frequency to give exactly 2.00 Lux on the lightmeter, which corresponds to 96 mW and 68.5 ohms on the chart.
238.19 kHz
Scope Math average IN is 96.3 mW
Manual IN calculation is 13.3 x 0.0415 x cos -78.86 = 106.6 mW
Output calculation is 2.572/68.5 = 96.4 mW
« Last Edit: 2017-05-07, 14:41:13 by TinselKoala »


---------------------------
"The easiest person to fool is yourself" -- Richard Feynman
   

Hero Member
*****

Posts: 704
To all the guys that fought the good fight in this, my hat goes off to you  O0

The dedication and focus by the good men in these pages and the work done by them is uplifting.

Do not be discouraged, there is yet much to learn.

Some of the best minds and hearts are here and have contributed unselfishly, it will not be forgotten, it does not go unnoticed. The universe is listening! And it will be delivered.

Lets hope this effort can indeed solidify the bond of  brotherhood in the "quest".

That was an awesome post. This has been an awesome thread to read.
   

Hero Member
*****

Posts: 1468
Is there a similar problem with the scopeshots posted in #436?
They are much better.
I wonder if getting rid of the reference clipleads altogether would improve the signals even further.

It nags at my mind that we are neglecting the inductance of the loop area formed by the entire circuit.
   

Hero Member
*****

Posts: 1468
So precision non inductive resistors are no good as CVR's ?
They should be!
The way you get the signal from CSRs to the scope matters, too.

A lot can go wrong along the way...
« Last Edit: 2017-05-07, 15:25:26 by verpies »
   

Hero Member
*****

Posts: 1298
I calibrated my test GOW bulb by using a DC power supply and monitoring the current and voltage while recording the LUX reading. SO each data point is Voltage, Current, and Lux. DC power is calculated the usual way by VxI in the spreadsheet. Filament resistance is also calculated in the spreadsheet by R=V/I at each point. The bulb is taped right up on the lightmeter sensor and the sensor and bulb are tucked away in an opaque bag (the meter's carry-case!). About 18 inches of twisted pair wire connects the bulb to the DUT. (This could be improved.)
Once the data are gathered by stepping up the voltage in small increments and recording the three variables at each step, the values are entered into the spreadsheet (I use LibreOffice Calc on Linux) and the graph is generated by using the Chart Wizard and then fine tuning grids, axes, titles etc. I found that using the smooth lines option with curves fit automatically gave the best looking lines to fit the data and to allow interpolation between data points.

Here's how to use the resulting Bulb Calibration nomograph:

EDIT: I see now that the horizontal minor grid vertical spacing is sometimes a little irregular. That's a bug in the spreadsheet, I guess, and it makes it harder to be accurate when reading the graph.



---------------------------
"The easiest person to fool is yourself" -- Richard Feynman
   

Group: Administrator
Hero Member
*****

Posts: 2778
It's not as complicated as it may seem...
Yes, the problem here is with the measurement of the input power.
There is a very simple, cheap, readily accessible (no scope required) and accurate method of measuring Pin; feed the ac source through a DC-powered amplifier, and measure the DC power feeding the amplifier.

This method has its challenges too, mainly that in order to be capable of supplying 100's of mA to the DUT, the idle current of the amplifier might swamp out the flea power Pin's and Pout's you are dealing with here.

I have illustrated the method below in case anyone cares to investigate its merit. Steps to its use:

1) Determine the idle power for the power amplifier (PA). Measure Vavg and Iavg with no signal at the input and no connection to DUT. This is Pidle.

2) Determine the total power used (Ptot) with an input signal, and connection to DUT using Vavg and Iavg (Ptot=Vavg x Iavg).

3) Compute Pin via Pin = Ptot - Pidle

Rather than building up a circuit to perform this function, one could also crack open their FG and install banana jacks for the insertion of a CVR and measurement of V+ for the output stage (similar to what is shown here). The extra wiring involved is not a concern because any resultant fluctuations due to parasitic inductance is averaged out by the meters. The only critical requirement is to place the DMM probes as close to the CVR body as possible.


---------------------------
Never let your belligerence get in the way of your brilliance!
   

Hero Member
*****

Posts: 1298
There is a very simple, cheap, readily accessible (no scope required) and accurate method of measuring Pin; feed the ac source through a DC-powered amplifier, and measure the DC power feeding the amplifier.

This method has its challenges too, mainly that in order to be capable of supplying 100's of mA to the DUT, the idle current of the amplifier might swamp out the flea power Pin's and Pout's you are dealing with here.

I have illustrated the method below in case anyone cares to investigate its merit. Steps to its use:

1) Determine the idle power for the power amplifier (PA). Measure Vavg and Iavg with no signal at the input and no connection to DUT. This is Pidle.

2) Determine the total power used (Ptot) with an input signal, and connection to DUT using Vavg and Iavg (Ptot=Vavg x Iavg).

3) Compute Pin via Pin = Ptot - Pidle

Rather than building up a circuit to perform this function, one could also crack open their FG and install banana jacks for the insertion of a CVR and measurement of V+ for the output stage (similar to what is shown here). The extra wiring involved is not a concern because any resultant fluctuations due to parasitic inductance is averaged out by the meters. The only critical requirement is to place the DMM probes as close to the CVR body as possible.

Hmm.... I have a jarful of NOS Motorola MC1590G broadband amps.... do you think one would be suitable, and if so would you care to draw up a complete schematic for use?


---------------------------
"The easiest person to fool is yourself" -- Richard Feynman
   

Hero Member
*****

Posts: 1298
Meanwhile back on the bench.....

I have a 1.00 ohm NI resistor that I mounted to a scope probe cable with the probe cut off. I used this long ago at Steve Weir's urging when dealing with the Ainslie affair. SO I dug it out and installed it on the Capacitor Test circuit, and eliminated the other probe ground clips.

(But before I did that, I verified that the improvised current probe gives the right values with DC current, as compared to an ammeter reading. It is spot on (at 1x attenuation setting of course)).

EDIT: The Load resistors are noticeably warm.


---------------------------
"The easiest person to fool is yourself" -- Richard Feynman
   

Group: Administrator
Hero Member
*****

Posts: 2778
It's not as complicated as it may seem...
Hmm.... I have a jarful of NOS Motorola MC1590G broadband amps.... do you think one would be suitable, and if so would you care to draw up a complete schematic for use?

With only 8mA output drive, it is most likely not suitable, unless coupled to an output booster. I have included an app-note on output boosters. Using an output booster can allow for a number of op-amp choices, provided they have a large enough bandwidth and slew-rate. I know you have a jar full, but I think there are better and simpler choices than the MC1590G as an op-amp driver.

Perhaps there are a number of individuals here that might be willing and eager to design something up; Verpies?


---------------------------
Never let your belligerence get in the way of your brilliance!
   

Hero Member
*****

Posts: 1298
With only 8mA output drive, it is most likely not suitable, unless coupled to an output booster. I have included an app-note on output boosters. Using an output booster can allow for a number of op-amp choices, provided they have a large enough bandwidth and slew-rate. I know you have a jar full, but I think there are better and simpler choices than the MC1590G as an op-amp driver.

Perhaps there are a number of individuals here that might be willing and eager to design something up; Verpies?

I have some TL082 units on hand which have been my general purpose go-to opamps. Suitable?


---------------------------
"The easiest person to fool is yourself" -- Richard Feynman
   

Hero Member
*****

Posts: 1298
Sanity check on the present CapTests ....

"Current" scope reading set up now gives these values:

Irms measured through the 1R00 TKprobe was 75.2 mA
Vrms across the 18R80 load was 1.38 V  and 1.38/18.8 = 73.4 mA 

Frequency 1.6889 MHz, phase1->2 -80.82 degrees

IN: Scope math average 52.7 mW
3.34 x 0.0752 x cos(-80.82) = 40.1 mW

OUT: 1.382/18.8 = 101.3 mW



---------------------------
"The easiest person to fool is yourself" -- Richard Feynman
   

Group: Administrator
Hero Member
*****

Posts: 2778
It's not as complicated as it may seem...
I might suggest the following sanity check:

1) Change the capacitor to 1uF (film if possible)
2) Decrease frequency to about 3.5kHz (1u/2.2n factor)
3) Repeat test.
4) Post results.


---------------------------
Never let your belligerence get in the way of your brilliance!
   

Group: Administrator
Hero Member
*****

Posts: 2778
It's not as complicated as it may seem...
I have some TL082 units on hand which have been my general purpose go-to opamps. Suitable?
I think we would need a device with much higher GBW product, such as with the AD8038.


---------------------------
Never let your belligerence get in the way of your brilliance!
   

Hero Member
*****

Posts: 1298
I think we would need a device with much higher GBW product, such as with the AD8038.

Unfortunately I have to work with what I have on hand.


---------------------------
"The easiest person to fool is yourself" -- Richard Feynman
   

Group: Administrator
Hero Member
*****

Posts: 2778
It's not as complicated as it may seem...
Unfortunately I have to work with what I have on hand.
I could send you the parts, then you'll have them on hand.  :)


---------------------------
Never let your belligerence get in the way of your brilliance!
   

Hero Member
*****

Posts: 1298
I might suggest the following sanity check:

1) Change the capacitor to 1uF (film if possible)
2) Decrease frequency to about 3.5kHz (1u/2.2n factor)
3) Repeat test.
4) Post results.

OK, but first I want to post these results. Comparing with, and without, the toroidal coupling transformer.

With transformer (#81):
f = 1.6882 MHz,  phase angle -82.27
IN: Math average 55.8 mW, sine computation 41.3 mW
OUT: 99.8 mW

Without transformer (and changing nothing else except scope channel v/div settings) (#82)
f = 1.6875 MHz,  phase angle -82.70
IN: Math average 229 mW, sine computation 133 mW
OUT: 432 mW
Load resistors now decidedly warm

So my transformer is also an attenuator, but it isn't responsible for the COP>1 measurements or undue phase shifts in this Cap Test system.




---------------------------
"The easiest person to fool is yourself" -- Richard Feynman
   

Hero Member
*****

Posts: 1298
I might suggest the following sanity check:

1) Change the capacitor to 1uF (film if possible)
2) Decrease frequency to about 3.5kHz (1u/2.2n factor)
3) Repeat test.
4) Post results.

OK, here you go. I found a 1uF 10 percent poly film cap in my stash and measured it to be 0.997 uF.

Without coupling transformer, at 3.531 kHz (The F43 is hard to set precisely, being analog with a big knob), the phase shift was -71.49 degrees.
IN: Math average 371 mW, sine calculation 310 mW
OUT: 385 mW


---------------------------
"The easiest person to fool is yourself" -- Richard Feynman
   

Hero Member
*****

Posts: 1298
I could send you the parts, then you'll have them on hand.  :)

If you think it is really worthwhile, I'm happy to do it.

I'll PM you my address.


---------------------------
"The easiest person to fool is yourself" -- Richard Feynman
   
Pages: 1 ... 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 [19] 20 21 22 23 24
« previous next »


 

Home Help Search Login Register
Theme © PopularFX | Based on PFX Ideas! | Scripts from iScript4u 2019-03-24, 17:28:55