Storing Cookies (See : http://ec.europa.eu/ipg/basics/legal/cookies/index_en.htm ) help us to bring you our services at overunity.com . If you use this website and our services you declare yourself okay with using cookies .More Infos here:
https://overunity.com/5553/privacy-policy/
If you do not agree with storing cookies, please LEAVE this website now. From the 25th of May 2018, every existing user has to accept the GDPR agreement at first login. If a user is unwilling to accept the GDPR, he should email us and request to erase his account. Many thanks for your understanding

User Menu

Custom Search

Author Topic: Lidmotor's Penny circuit help needed.  (Read 52572 times)

TinselKoala

  • Hero Member
  • *****
  • Posts: 13958
Re: Lidmotor's Penny circuit help needed.
« Reply #15 on: September 28, 2013, 04:25:27 AM »
It might be possible to use a "current probe" in your simulator, instead of looking at the voltage drop across a 1R, to see the current in the LEDs. Sorry I didn't think of this before.

Dark Alchemist

  • Jr. Member
  • **
  • Posts: 52
Re: Lidmotor's Penny circuit help needed.
« Reply #16 on: September 28, 2013, 05:50:56 AM »
Good for you. Now you are making higher voltages with your circuit, and you should be seeing more total light output.

I can't quite make out the scope trace in the video. It looks to me like your timebase is set to 1 millisecond per division, is that right?

I'd like to see the scope display the actual signal, three or four peaks across the screen, instead of the "comb" you are showing. Try changing the timebase to 0.1 ms or even 1 us per division. We should also look at the voltage signal directly at the base of the transistor on the other channel of the scope. The max rated Vebo for your BC337 is only 5 volts.

I hope you realize what I was saying about the LED current/voltage relationship. To see what the _actual_ current is that you are putting into your LED stack, you can put a 1R resistor in series with the LEDs/capacitor at the cathode end, and look at the voltage drop across this resistor with one of the scope channels. By Ohm's Law, the current in this resistor is I == V/R, and since R is 1, the current in amps will be equal to the voltage drop in volts that you see on this resistor.
Yep, I am familiar with the 1Ohm measuring technique but then I read it isn't that simple.

The resolution (gotta love how Youtube takes a perfectly clear 1280x1024 video and rapes it even for HD) of the time base is 1ms/div but at 1us a division I can't see what it is doing as it is too fast (I think this is a shortcoming of the scope in Multisim as it has no persistence at all and being digital what they are doing is clearing the screen of the scope each time the dot hits the end and retracing from the beginning which, at fast times, is really useless for real time work.

Scope is set X axis - 1ms/div and Y axis - 1v/div

Dark Alchemist

  • Jr. Member
  • **
  • Posts: 52
Re: Lidmotor's Penny circuit help needed.
« Reply #17 on: September 28, 2013, 06:05:54 AM »
It might be possible to use a "current probe" in your simulator, instead of looking at the voltage drop across a 1R, to see the current in the LEDs. Sorry I didn't think of this before.
I did use that in my previous JT that I posted about here (I think it was here) and people said not to trust those readings due to all sorts of issues.

TinselKoala

  • Hero Member
  • *****
  • Posts: 13958
Re: Lidmotor's Penny circuit help needed.
« Reply #18 on: September 28, 2013, 06:15:38 AM »
Well... I can't help you with your Multisim... I run a Linux system and I can't find any way to run it even under Wine. And it's not free either. But that doesn't sound like a normal, or useful, simscope behaviour to me. With a regular signal you should be able to display a regular, interpretable trace.

Let's see. JTs usually run at a fairly high frequency range, like 15-50 kHz or so. 20 kHz is a typical rate. So at 20 kHz,  in 1 ms, you would expect to see 20 peaks across one single division on the scope. Right? IOW, an uninterpretable comb like you showed in the video.

At 1 microsecond, there wouldn't be enough time for a complete period of a 20 kHz signal. Right? As you describe. The period is the inverse of the frequency, so the period is 1/20000 = 50 microseconds. Right?

So 1 ms is too slow, 1 us is too fast. Try 0.1 ms (100 us) per division. If the JT is running at 20 kHz you should see two peaks per division. Try 10 us per division, then you should see two peaks across 10 divisions... an ideal display of a 20 kHz signal.



TinselKoala

  • Hero Member
  • *****
  • Posts: 13958
Re: Lidmotor's Penny circuit help needed.
« Reply #19 on: September 28, 2013, 06:16:13 AM »
I did use that in my previous JT that I posted about here (I think it was here) and people said not to trust those readings due to all sorts of issues.

What were the issues? Who were the people?

ETA: Now that you have your voltage pumping up a little bit by using series LEDs, you can start adding parallel groups of LEDs for possibly more light output. Try three in series, parallel with three more in series, along with whatever parallel cap value you need.
Here's a video of a 1:1 JT like yours, running 24 LEDs in a 12-parallel, 2-series arrangement, running on a depleted AAA battery. This JT will light a single LED to good brightness... but it lights the full array much better.
http://www.youtube.com/watch?v=zM1qdATaiks

Dark Alchemist

  • Jr. Member
  • **
  • Posts: 52
Re: Lidmotor's Penny circuit help needed.
« Reply #20 on: September 28, 2013, 06:22:05 AM »
What were the issues? Who were the people?
A real Joule Thief operates off a single 1.5-volt battery and has one or more LEDs as the load.  Because of it's somewhat unusual operating characteristics and the fact that LEDs complicate the output power measurement because they are what is called non-linear electrical devices it's actually much more complicated and difficult than you think.

The real way to do it would be with a digital storage oscilloscope.  Preferably you would have one with a built-in math function.  You have to measure the instantaneous voltage and current output by the battery, as well as measuring the instantaneous voltage and current across the LED.

It just happens to be a circuit where multimeters and a basic scope will not cut it and you can't really make the power measurement unless you have the right equipment and you really know what you are doing.

MileHigh

TinselKoala

  • Hero Member
  • *****
  • Posts: 13958
Re: Lidmotor's Penny circuit help needed.
« Reply #21 on: September 28, 2013, 06:38:29 AM »

Well, MH is right, but that's not a fatal issue! Don't let him scare you away from doing the analysis properly. It's not complicated, just tedious. Your MultiSim will surely do the required stuff as far as simulating the DSO's storage and math capability. And I've demonstrated several times how to do it all using analog scopes, screenshots and spreadsheets.

The current probe, once the scope is set properly, will give you the instantaneous current waveform through the measurement point, and you already have the voltage measurement in the other channel. I would be very surprised if the software could not multiply those together point-by-point and then integrate the resulting instantaneous power curve over the time of the measurement. Or you can ask the scope to give you the "average" value of the current, if that's all you are looking for. The peak current can be read directly from the traces once you have the scope set to display the signal properly.

Dark Alchemist

  • Jr. Member
  • **
  • Posts: 52
Re: Lidmotor's Penny circuit help needed.
« Reply #22 on: September 28, 2013, 06:49:16 AM »
Well, I must be doing something wrong because as soon as I hook up a 1ohm resistor it will never start up.  I tried it in series from ground to the last LED's cathode and I tried it from ground to the anode of the first LED and no go with either.

Dark Alchemist

  • Jr. Member
  • **
  • Posts: 52
Re: Lidmotor's Penny circuit help needed.
« Reply #23 on: September 28, 2013, 07:59:30 AM »
Here is a screen capture of a moment in time with my settings which are: X - 1us/div, Y - 1V/div ch1, Y - 2mV/div ch2 R2 - .1ohm

I am seeing negative current it seems but if my math is correct I am seeing 2.3mv/.1oh, = +/- 23ma or is my calculation faulty?

TinselKoala

  • Hero Member
  • *****
  • Posts: 13958
Re: Lidmotor's Penny circuit help needed.
« Reply #24 on: September 28, 2013, 05:26:28 PM »
Here is a screen capture of a moment in time with my settings which are: X - 1us/div, Y - 1V/div ch1, Y - 2mV/div ch2 R2 - .1ohm

Oh, have mercy! It's really hard to make out the scope screen from here, at the size/resolution you are posting. But I can see a couple of things. The blue Current signal isn't like any JT signal I've seen, it's more of a symmetrical sinusoid instead of the regular more-or-less rectangular pulses or spikes I generally see. But at least we can tell the frequency. You have just over 30 peaks in 8 microseconds.. .call it 30.1 cycles per 0.000008 sec, or around 3.76 MHz, which seems really high to me. And I can't see any corresponding peaks in the red Voltage trace, so I'm wondering if you've actually captured the true JT oscillation here or are seeing some other noise source.
When using a very small CurrentViewingResistor (CVR or CSR) it's easy to pick up noise. Since your system won't oscillate with a 1R in there (also a bit strange) I would try eliminating the resistor and the voltage-current conversion completely by using a pure "current probe" there if your sim allows it.
Quote

I am seeing negative current it seems but if my math is correct I am seeing 2.3mv/.1oh, = +/- 23ma or is my calculation faulty?
Well, as I said before I'm not sure if you are really seeing the current signal yet. But what I see on the Blue trace is 1 full division plus about 1/3 or 2/5 more, so that would be just under 3 mV (1 1/2 divisions would be 3 mV). And the same on the negative side. Your math is correct but you are reading the trace incorrectly, at least from my blurry view here.
However, your scope has cursors that you can position. The numbers and arrow buttons in the T1 T2 box "time Channel A Channel B" at the top left of the parameters area appear to be the cursor positioning and values section. So you can read the exact voltage peaks by using the cursors and reading the values you get here.

Also, at the bottom right of the screen are the scope's Trigger controls. Please set this at "Auto" or "Normal" and a level of about + 0.25 volts, ascending slope,  and use the Red channel, the Channel A, as you are using. What happens to the trace when you do this?

The JT I showed in the video above has a 1:1 transformer (actually 13 : 13 turns) and each winding has an inductance of 388-390 microHenry, measured on my ProsKit RCL meter and confirmed by my Arduino-based L meter. I can't tell what the coupling coefficient is but it's probably in the range you are using, 0.5-0.8, since it's a good ferrite.   It certainly doesn't oscillate at 3.7 MHz, though!

I might be able to modify this unit to "simulate" your simulation results in hardware by changing the cap values and etc. Let me see what I can arrange, and I'll give you some news later this afternoon.

But meanwhile, see if you can try the trigger setting I recommend, along with using the scope cursors to get precise voltage values from the traces. Also, if the 0.1 R CVR is causing noise pickup, eliminate it and use a "current" probe in the same position, if your sim allows.

I would apply the changes in this order: First, the current probe. See if your signal still looks the same. Then apply the trigger settings I suggest. The signal might disappear! If it does then try changing the timebase to 10 or 100 microseconds per division and see if the signal comes back stably on screen. If the signal doesn't disappear when you set the trigger as I have suggested but remains looking like it does now... then I'll have to play around with hardware myself to see what is up.

Bottom line: since I'm not seeing the characteristic JT-like peaks on the Red trace, and your Blue trace is so symmetrical and so high frequency, I am not confident that we are seeing the actual JT performance on the scope yet.

I've attached an example scopeshot from one of LTseung's JTs. Top trace corresponds to your Ch A probe point and bottom trace to your Ch B. The values you get will be different of course - especially the spikes/pulses should be narrower -- but the basic shape of the signals should be similar, I think. The symmetrical sinusoid of your Blue trace is atypical and I'll have to reproduce it in order to understand it fully.

TinselKoala

  • Hero Member
  • *****
  • Posts: 13958
Re: Lidmotor's Penny circuit help needed.
« Reply #25 on: September 28, 2013, 05:44:42 PM »
Another thing: Looking at the specs for the BC337 and comparing to the 2n2222 I don't see much difference except in the voltage handling ability. What happens when you take the exact circuit you show above, with scope settings exactly as you have there, if you change the transistor to a 2n2222, or a 2n2369a, or MPSA18, my personal low-voltage favorite?

I'm asking because I don't have a BC337 on hand, and am not likely to be able to get one.

Dark Alchemist

  • Jr. Member
  • **
  • Posts: 52
Re: Lidmotor's Penny circuit help needed.
« Reply #26 on: September 28, 2013, 05:52:39 PM »
Another thing: Looking at the specs for the BC337 and comparing to the 2n2222 I don't see much difference except in the voltage handling ability. What happens when you take the exact circuit you show above, with scope settings exactly as you have there, if you change the transistor to a 2n2222, or a 2n2369a, or MPSA18, my personal low-voltage favorite?

I'm asking because I don't have a BC337 on hand, and am not likely to be able to get one.
I literally have a crap load of BC337, lol.  Let me run the sim with a 2n2222 using my changes as Lidmotor's original design would not oscillate.

As far as a current probe in the sim I do not have that but I do have a probe that just spits numbers at you but someone mentioned that wouldn't tell me anything.


Nope, the 2n2222, 2n2369a, or the MPSA18 will oscillate.

Dark Alchemist

  • Jr. Member
  • **
  • Posts: 52
Re: Lidmotor's Penny circuit help needed.
« Reply #27 on: September 28, 2013, 06:05:54 PM »
Using your suggestions.

TinselKoala

  • Hero Member
  • *****
  • Posts: 13958
Re: Lidmotor's Penny circuit help needed.
« Reply #28 on: September 28, 2013, 07:21:59 PM »
OK, so by setting the trigger as I suggested, the Blue trace vanished? You made no other changes, just setting the trigger?

Now try changing the timebase to 10 or 50 microseconds per division, making no other changes.

In your last post you said,
Quote
Nope, the 2n2222, 2n2369a, or the MPSA18 will oscillate.
So I'm confused again. The "nope" part doesn't seem to agree with the "will oscillate" part. Did you mean that "neither" will oscillate, or that they "will not oscillate" or what?
If they _do_ oscillate could you please post a scopeshot using one of them? If they don't oscillate, never mind, just please try the timebase change I suggested above.

ETA: Wait I see you have the "Blue" channel disconnected. Please go step-by-step. Replace your 0.1 ohm resistor, use the Blue channel to monitor the voltage just as you had in your previous shot. Change ONLY the trigger setting, to +0.25V on the Red channel, set to "auto" or "normal".  Then show the screenshot.
Then, change the timebase from 1 us/div to 10 or 50 us/div and show the screenshot. Please only change one thing at a time, because I confuse easily.

TinselKoala

  • Hero Member
  • *****
  • Posts: 13958
Re: Lidmotor's Penny circuit help needed.
« Reply #29 on: September 28, 2013, 07:38:09 PM »
I just checked the JT I showed in the video above. It oscillates at about 14.6 kHz, much much slower than the 3.76 MHz of the Blue trace you showed above. I'm not convinced that your Blue trace really is showing the main JT oscillation of your circuit, nor is the Red trace in your latest scopeshot.

Try this: using the exact setup you have just above, with only the Red Channel A Q1 collector signal shown. Set the trigger on Ch A, rising slope, +0.25 volts, "Auto" or "normal". Now change the timebase in steps. You are showing 1 us/div already, so now look at the signal at 10 us/div, 50 us/div, and 100 us/div, or whatever similar settings your scope allows.