PopularFX
Home Help Search Login Register
Welcome,Guest. Please login or register.
2020-02-28, 07:10:52
News: If you have a suggestion or need for a new board title, please PM the Admins.

Pages: [1]
Author Topic: LTJT - poynt99 Tests #1  (Read 12324 times)

Group: Administrator
Hero Member
*****

Posts: 2906
It's not as complicated as it may seem...
Here are a few pictures of the temporary lab I set up for testing Lawrence Tesung's Joule Thief circuit. Scope shots of the circuits to follow shortly.

.99
   

Group: Administrator
Hero Member
*****

Posts: 2906
It's not as complicated as it may seem...
This first run shows data for this particular build of the LTJT using a core I had on hand. It has a u (permeability) quite a bit higher than what Lawrence is using, so the frequency of oscillation is of course correspondingly-lower; about 6kHz or so.

For purposes of better accuracy, only the screen shots where the high number of cycles is shown should be used to obtain the MEAN power trace, and RMS voltage and current trace values. The low cycle count shots are there for wave form inspection (and frequency) only.

01.png is showing the average INPUT POWER as 24.59mW. Note that multiplying the RMS voltage and current yields 40.07mW as a result. It is not shown, but using the scope to take the "RMS" of the input power trace yields a result of about 42mW.

03.png shows that the OUTPUT POWER is at 47.49mW, but this must be divided by 10 to take into account the value of the CSR is 10 Ohms, not 1 Ohm. Therefore, the OUTPUT POWER is about 4.75mW.

.99
   
Group: Guest
Poynt!

That math trace is looking damn good!  You are really in the linear charging zone there, so much so that you could filter out the noise by eye and do the math on paper!  lol

The integral of y= kt is 1/2 k t^2 for for t = the start of the ramp to t = the end of the ramp!  That gives you the energy associated with the orange power ramp function.  Then you could average energy that over your cycle time to get the average power!

That might sound complicated to some so here is another approach.   Look at the orange power trace.  Notice that the ramp fits inside a rectangular box.  Just calculate the area of that box, and then take 1/2 of that area, and you now have the energy associated with the linear ramp of increasing power.  Then you just have to average that energy over one full cycle to get the average power.  That will give you exactly the same result as the integral formula.

The glass is more than half-full and the days are getting longer for us Northern Hemisphere types!

MileHigh
   

Jr. Member
**

Posts: 84
Open Source Experimentalist
Hi .99,

I noticed that all of Lawrence's scope shots presented show the oscilloscope channel signal tripping on a "falling" signal ..... I see that you have your oscilloscope channel trip set on a "rising" signal.

Have you found that a "rising" signal trip is a better setting or easier to tune the circuit ..... or is this a mute point in my observation of this particular setting?

Best Regards,
Glen
 :)  
   
Group: Guest
01.png is showing the average INPUT POWER as 24.59mW. Note that multiplying the RMS voltage and current yields 40.07mW as a result. It is not shown, but using the scope to take the "RMS" of the input power trace yields a result of about 42mW.

03.png shows that the OUTPUT POWER is at 47.49mW, but this must be divided by 10 to take into account the value of the CSR is 10 Ohms, not 1 Ohm. Therefore, the OUTPUT POWER is about 4.75mW.

.99

I'm a little surprised that it's that inefficient.  No surprise at all that it's under unity.  I suppose that you will be doing some more runs Poynt, and we will get data that is a variation on the theme.  With this first result, I will be surprised if you can approach even 50% efficiency.  I suppose if you go air core you might get better numbers also.

It all makes sense and the results are as expected.  There ain't no zipons floating around here!

Milehigh
   

Group: Administrator
Hero Member
*****

Posts: 2906
It's not as complicated as it may seem...
Hi .99,

I noticed that all of Lawrence's scope shots presented show the oscilloscope channel signal tripping on a "falling" signal ..... I see that you have your oscilloscope channel trip set on a "rising" signal.

Have you found that a "rising" signal trip is a better setting or easier to tune the circuit ..... or is this a mute point in my observation of this particular setting?

Best Regards,
Glen
 :)  

Hi Glen. I've never found the trigger slope to cause any critical effects. Just to check, I just switched the input scope to trigger on the falling edge, and all the numbers remained exactly the same, and well they should.

.99
   

Group: Administrator
Hero Member
*****

Posts: 2906
It's not as complicated as it may seem...
I'm a little surprised that it's that inefficient.  No surprise at all that it's under unity.  I suppose that you will be doing some more runs Poynt, and we will get data that is a variation on the theme.  With this first result, I will be surprised if you can approach even 50% efficiency.  I suppose if you go air core you might get better numbers also.

It all makes sense and the results are as expected.  There ain't no zipons floating around here!

Milehigh

I am surprised too. 20% efficiency seems quite poor, but again, there is that nasty collector LED sucking up a substantial amount of power. You can see how much brighter it is than the Red LED, not that this means an awful lot, but it is noticeably brighter.

.99
   

Group: Administrator
Hero Member
*****

Posts: 2906
It's not as complicated as it may seem...
Attached is my hand drawing of the circuit from my note book I'm keeping for these tests. It shows the probe locations (PT=probe tip, PG=probe gnd) and the toroid coil resistance and inductance of each winding, BEFORE I cut the core. I have not taken new inductance readings yet.

I used #22 AWG wire, which corresponds to about 0.64mm diameter I believe.

The core type is also noted there: SANLIN SL5 T22x14x8, P/N 2170.

Note that the Duracell battery I am currently using is probably quite dead, as it's reading just over 1V.

.99
   

Group: Administrator
Hero Member
*****

Posts: 3737


Buy me some coffee
Great stuff Darren, Dam thats the first time ive seen someone take notes   O0
   
Group: Professor
Hero Member
*****

Posts: 2688
Glad to see the RMS versus Mean issue resolved (in agreement) and that .99 is taking and showing measurements.  
But I've long since cut out the LED across the transistor -- cut out that dam LED!     As we said some time ago.


I will have other suggestions in the future as I have time (pressing family stuff right now).  The overall question is far from resolved, MH.
« Last Edit: 2011-02-06, 23:44:59 by PhysicsProf »
   
Group: Guest
PhysicsProf:

I am sure that Poynt will make additional measurements.  You can't forget that every example submitted by Lawrence included the LED connected to the transistor collector so it makes sense that Poynt start with the reference design, which is a standard Joule Thief with an additional L3 pick-up coil driving another LED and a resistance as a load.  His claims were based on this configuration so it needs to be analyzed.

Note the collector LED was already discussed and Poynt offered to measure the average power across that component also.  And indeed, as you state, removing it completely from the circuit is another option.

MileHigh
   
Group: Guest
Hi Poynt99,

I see a ~5.75 mW output there on your RMS scope shot values.

Channel 2 represents the voltage drop across a 10 ohm resistor.

Consider if Channel 1 was 2.03 VRMS and Channel 2 was 28.3 mVRMS:

Then the output voltage to your load would be 2VRMS and the output current through your load (assuming no KCL violation exists) would be 28.3 mVRMS / 10 Ω or 2.83 mARMS giving 5.7449 mW.

The calculated Power Trace shows 47.49 mV V that we can divide by 10 for the CSR to get 4.749 mW.

5.7449 · cos(θ) = 4.749 so our phase angle, θ,  should equal 34.442°

Does this sound right? Can we use cos(θ) in a nonsinusoidal waveform calculation?

 :)
   
Group: Guest
Glad to see the RMS versus Mean issue resolved (in agreement) and that .99 is taking and showing measurements.  
But I've long since cut out the LED across the transistor -- cut out that dam LED!     As we said some time ago.


I will have other suggestions in the future as I have time (pressing family stuff right now).  The overall question is far from resolved, MH.

I recall reading some posts regarding that LED, but I didn't agree with everything I read. For all intents and purposes that LED is in the input section and any activity it produces should be reflected in the input power readings, not the output power readings.

It is not a factor in the load considerations at the output.

It is however a feature of maintaining a biasing current in the primary winding when the input is driven off. This has an effect of causing an asymmetric BH curve in the core because it causes more hysteresis in one direction.

So while it has no real effect on the load, it may have a real effect on the output voltage, especially in a Pulsed DC configuration such as this.

Harvey
   

Group: Tinkerer
Hero Member
*****

Posts: 1618
I recall reading some posts regarding that LED, but I didn't agree with everything I read. For all intents and purposes that LED is in the input section and any activity it produces should be reflected in the input power readings, not the output power readings.

It is not a factor in the load considerations at the output.

It is however a feature of maintaining a biasing current in the primary winding when the input is driven off. This has an effect of causing an asymmetric BH curve in the core because it causes more hysteresis in one direction.

So while it has no real effect on the load, it may have a real effect on the output voltage, especially in a Pulsed DC configuration such as this.

Harvey


It would seem that the "collector LED" is illuminated
by inductive kickback pulses from the primary
winding;  a partial dissipation of the energy released
when the transistor goes into cutoff.

Would this dissipation deprive the secondary circuit
of that energy?
 


---------------------------
The animal mind ALWAYS reacts to what it does not understand. This is what sets dogs barking. If you are going to tell the truth, you are going to have to be okay with barking dogs, because they will harry your passage until you pass through town.
Les Visible - 27 February 2020
   
Group: Guest

Would this dissipation deprive the secondary circuit
of that energy?
 

Yes it would.  When the transistor switches off the magnetic energy stored in the core gets split between the collector LED and the discharge cycle through the L3 secondary and the LED and the two resistors.

This was mentioned to Lawrence several times and he dismissed it and stated that the device was still over unity.

MileHigh
   
Group: Guest
In the case that the battery voltage is below the LED threshold (Darren said his was as low as 1 V) then very little current will flow from the battery through the LED if any even though the LED may be conducting. In some devices, they will remain on until the current falls below a specific level much like an SCR. In other devices they may deplete strongly in the PN junction forcing the device off when the voltage drops below the threshold. It depends on how the device is manufactured.

Keep in mind that the energy stored in the magnetic field is not trying to reach the battery negative, but instead it is trying to balance the primary winding back to a neutral state. It simply finds the path through the LED and battery less restrictive than fighting its way back through the windings while more collapsing field is inducing opposing EMF. In the event that the path is completely blocked (like the high impedance of the transistor) then the field may find its way to the secondary instead.

Remember though, that any change in flux in the core will induce an EMF and resulting current in the secondary. This goes both ways, building the field and collapsing the field. The transfer from primary to secondary would work just as well with a push pull H-bridge if not better.

So the Input LED acts to keep the current flowing through the primary just a bit longer essentially taking the place of the transistor momentarily and if the battery were above the threshold would also keep the primary biased as I mentioned before.

But in the way it is being used here, I must agree that the energy stored in the field is split between the inputs and outputs as the conditions do not allow any further energy to be drawn from the battery. Of course that may be proving the Lead Out / Bring in  concept if the battery is off and the LED is conducting beyond the energy stored in the core during the inductive charge cycle in addition to that distributed to the output stage.

What do you guys think?
Should the input LED be added as a Load to the system?
What is the best way to quantify the energy stored in the field and the energy released to the Load(s)?

:)
   
Group: Guest
Harvey:

The input LED, a.k.a. the collector LED is definitely part of the load.  I think that it's important to remember that the basic JT is not really a transformer, it's really a timing circuit that charges and discharges an inductor.

Once the toroidal inductor is charged then the power output associated with the decreasing magnetic flux in the toroidal core, i.e.; the stored energy discharge, has two possible paths to follow.

First path:  The EMF source for the first path is the battery voltage plus the 13 turns in the L2 winding times the rate of change of magnetic flux with respect to time.  The load in the first path is the collector LED and the one-ohm resistor.

Second path:  The EMF source for the second path is the 17 turns in the L3 winding times the rate of change of magnetic flux with respect to time.  It's the identical rate of change of flux as in the first path.  The load in the second path is the output LED and a 10-ohm resistor and a 100-ohm resistor.

The impedance of the load in the first path is obviously much lower than the second path because the second path has 110 ohms of series resistance whereas the first path doesn't.  The EMFs for the two paths are comparable.  Therefore most of the energy stored in the toroid will discharge through the first path.

This situation could be turned around by increasing the number of windings in L3 and lowering the series resistance in the second path.  Then most of the energy would be discharged through the second path.  However, that was not the configuration that was specified by Lawrence.  I think it's important to emphasize this fact.

For those interested, the rate of change of magnetic flux with respect to time is clearly shown on the scope captures.  It's indicated by the yellow voltage trace during the discharge cycle.  Also, some readers might not realize that the yellow voltage trace and the blue current trace during the discharge cycle are directly related to each other.  The yellow voltage trace is proportional to the first derivative of the current trace with respect to time.  By the same token the blue current trace is proportional to the integral of the voltage trace with respect to time.  Once all of the stored energy in the core is sucked dry, the L3 voltage drops off suddenly and the charging cycle starts up again.

MileHigh
   
Hero Member
*****

Posts: 931
Here are a few pictures of the temporary lab I set up for testing Lawrence Tesung's Joule Thief circuit. Scope shots of the circuits to follow shortly.

.99

Dear Poynt99,

Have you received my components package via the Post Office Mail yet?  It should be in a hard paper envelop.  Please let me know as soon as you get it.  The expected time was “within 7 – 10 days”.

As I mentioned, you need to supply your own breadboard, the battery holder and the AA battery.  Otherwise, the components should be ready to be assembled.

I am not surprised that your LT-JT is way under unity.  As I stated multiple times before, FLEET is a pseudo resonance device.  It appears that your LT-JT is nothing near that condition.

Please use the components I supplied to build your second LT-JT on a breadboard.  It does not matter if the wires are too long.  TK showed the change in waveform with long wires.  In your case, you want to use the long wires as a tuning parameter to get the best COP value.

With the breadboard, you can easily change the waveforms by changing the holes used.  That is another important tuning parameter.  You can remove the JT LED for testing easily.

When you have two prototypes to compare, the fun begins. 

@PhysicsProf, glad to see that you are back.  When you have your new DSO, please post some of your photos and screen shots.  Use Mean Power Value for display.  (You can optionally display the RMS for comparison purposes).   You have an advantage over Poynt99 because you have a few prototypes with Tseung FLEET Comparison Index greater than 1 already.

@All, Poynt99 showed clearly that NOT every build can show COP > 1.  Most people who worked with JT thought that they achieved success when they see the lights on.  The basic JT is a very tolerant circuit.  It will light the LEDs with very wide range of values.  FLEET is different.  The COP > 1 results will require much tuning.   The second build by Poynt99 using the components I supplied will be much closer to the required pseudo resonance condition.  On the breadboard with long wires, some tuning can be done.  Thank you for your postings and comments.  There is much fun ahead.
   

Group: Administrator
Hero Member
*****

Posts: 2906
It's not as complicated as it may seem...
The package hasn't arrived yet Lawrence. Maybe today.  ;)

.99
   

Group: Administrator
Hero Member
*****

Posts: 2906
It's not as complicated as it may seem...
Hi Poynt99,

I see a ~5.75 mW output there on your RMS scope shot values.

Channel 2 represents the voltage drop across a 10 ohm resistor.

Consider if Channel 1 was 2.03 VRMS and Channel 2 was 28.3 mVRMS:

Then the output voltage to your load would be 2VRMS and the output current through your load (assuming no KCL violation exists) would be 28.3 mVRMS / 10 Ω or 2.83 mARMS giving 5.7449 mW.

The calculated Power Trace shows 47.49 mV V that we can divide by 10 for the CSR to get 4.749 mW.

5.7449 · cos(θ) = 4.749 so our phase angle, θ,  should equal 34.442°

Does this sound right? Can we use cos(θ) in a nonsinusoidal waveform calculation?

 :)

Pave from the RMS values does come close the the correct Pave value (according to P(t)), but this won't always be the case. Even if the phase angle compensation were a valid method to find the true Pave, how would we determine the phase angle to plug it in?

.99
   

Group: Administrator
Hero Member
*****

Posts: 2906
It's not as complicated as it may seem...
I recall reading some posts regarding that LED, but I didn't agree with everything I read. For all intents and purposes that LED is in the input section and any activity it produces should be reflected in the input power readings, not the output power readings.

It is not a factor in the load considerations at the output.

It is however a feature of maintaining a biasing current in the primary winding when the input is driven off. This has an effect of causing an asymmetric BH curve in the core because it causes more hysteresis in one direction.

So while it has no real effect on the load, it may have a real effect on the output voltage, especially in a Pulsed DC configuration such as this.

Harvey

I think that the collector LED needs to be considered as part of the output loop because although it only consumes energy during the "off" phase of the period, this is nonetheless, dissipated energy that could have been transformed to the output winding instead.

.99
   

Group: Administrator
Hero Member
*****

Posts: 2906
It's not as complicated as it may seem...

Should the input LED be added as a Load to the system?
What is the best way to quantify the energy stored in the field and the energy released to the Load(s)?

:)

I would suggest that, unless Lawrence is quite firm about leaving the collector LED installed in order for his circuit to operate at COP>1, we remove the collector LED from the circuit allowing us to get a more accurate COP for the circuit as a whole. The efficiency of power transfer to the output winding should increase a significant amount by removing that LED.

The power in the INPUT CSR really should be included also, but worst case it only represents about 1.5mW (~ 6% of the total INPUT power) of unaccounted-for power.

.99
   
Hero Member
*****

Posts: 931
I would suggest that, unless Lawrence is quite firm about leaving the collector LED installed in order for his circuit to operate at COP>1, we remove the collector LED from the circuit allowing us to get a more accurate COP for the circuit as a whole. The efficiency of power transfer to the output winding should increase a significant amount by removing that LED.

The power in the INPUT CSR really should be included also, but worst case it only represents about 1.5mW (~ 6% of the total INPUT power) of unaccounted-for power.

.99

Dear Poynt99,

My advice is to wait for the component parts to arrive and build another LTJT on a breadboard.  It is a matter of getting as close to the known COP > 1 situation as possible.  Your existing build may be too far away from that situation and is difficult to tune with the hard wiring.

It is perfectly acceptable to remove the JT LED for testing.  PhysicsProf is already doing that with good success.  He has received his ATTEN DSO and is checking it out.  It will probably take him a few days.  You should have received the components package by then.  It would be interesting to see two sources of screen shots.

There is possibility of the posting of experimental results from the Beijing Physics Society as well.

That will make things more exciting.
   

Group: Tinkerer
Hero Member
*****

Posts: 3169
tExB=qr
Attached is my hand drawing of the circuit from my note book I'm keeping for these tests. It shows the probe locations (PT=probe tip, PG=probe gnd) and the toroid coil resistance and inductance of each winding, BEFORE I cut the core. I have not taken new inductance readings yet.

I used #22 AWG wire, which corresponds to about 0.64mm diameter I believe.

The core type is also noted there: SANLIN SL5 T22x14x8, P/N 2170.

Note that the Duracell battery I am currently using is probably quite dead, as it's reading just over 1V.

.99

Add the date to your notes.
   
Pages: [1]
« previous next »


 

Home Help Search Login Register
Theme © PopularFX | Based on PFX Ideas! | Scripts from iScript4u 2020-02-28, 07:10:52