Is joule thief circuit gets overunity?

[ltseung888]

Lawrence, I would recommend you try to find out why your scope consistently shows a 12-13mV offset when measuring the input current. I would like to see that scope shot duplicated and show the LED as well. It sure appears that the circuit is OFF.

Also, did you do any number crunching on my data from board #33? What were the results?

What is your comment on my scope results? How do you explain that the current never crosses the 0-ref line?

@poynt99

I just checked with Board 108.  I can turn the LED ON with 380mV.

May 1 is a big day in China.  Some organizations and individuals use that period to have long holidays.  They may go back to work on May 5.

I have not started on the number crunching yet.  Some people attributed my ATTEN results as due to "cheap China made" scopes have restrictions.  I am trying to get help from organizations with both high end scopes and expertise.  When the ATTEN scope shots are different from yours, I have to find out the reason first before making any conclusions.  I am by no means an oscilloscope expert. 

Please wait at least 1-2 weeks so that I can locate the experts with the right resources in Hong Kong.  Thanks for helping to solve the ATTEN Scope puzzles from half way around the World.

[poynt99]

Lawrence,

Borrow, buy, or rent an analog scope of any description and perform the same observation test to see if the current trace crosses the 0-reference line at any time while the input voltage is varied. Use different scope probes, preferably the ones that come with the analog scope.

I pulled out my 25 year old 20MHz two-channel Hitachi V-212, and performed the same sweep as before. The current trace did not cross the 0-ref line. The circuit oscillated this time down to an input voltage of about 0.38V.

Check out the video of this sweep:
http://www.youtube.com/watch?v=P2tETAwx3xc

[MileHigh]

Very nice clean analog waveforms Poynt.  Even though your new high-end DSO is amazing, if I had a bench I would still have an analog scope to compliment the DSO.  I know we have had this conversation before!  For example, at the bottom of the "fang" in the current waveform, you can see some tiny vestigial spikes.  So you crank up the intensity for a few seconds to make the vestigial spikes more visible.  Then you can hunt down the source of those spikes if need be.  You would have a hard time doing that with a DSO, you might note even realize that they were there.

I am not really following here but it would seem to me that the current only flows in one direction.  When you look at the schematic and analyze it, one would suppose that the logical conclusion is that the current should only flow in one direction.  So the bench and the circuit analysis are in accord.  I fear that Lawrence is chasing rainbows but the sky is clearing up now.

[TinselKoala]

See that little "B" next to the channel setting indication in the shot of the shorted probes? That means that "bandwidth limitation" is set for that channel, which is excessively noisy and shows a definite DC offset.

Lawrence, please DO NOT use the scope's "auto setup" or Default Setup function for these traces! Clearly your scope should be set to the EXACT same settings, including BW limit OFF, for the probe and DC offset tests as it is set when you are making your measurements on a board! Set your scope MANUALLY and use its functions to SAVE the setup, then use the SAME setup for your measurements and for your probe/offset testing. DO NOT USE BANDWIDTH LIMITING, unless you also are using it when making your live measurements !!

I think .99 has asked you to make some actual measurements of small DC voltages, KNOWN small voltages like those that come from a voltage standard source like a LM199 chip or this unit here:
http://www.voltagestandard.com/Home_Page_JO2U.html
A test like this will be better than simply shorting the probe, as the degree of offset clearly depends on the voltage applied to the probe (from your own data which show this effect.)

And YET AGAIN, I am asking you to tell us the cause for the LARGE voltage discrepancy from one probe-channel combination that you illustrated when we first started exploring the probe compensation issue. You have never told us how that discrepancy happened, how it was resolved, and you haven't shown traces that prove that this issue has been fixed.

Also... high-end DSOs and DPSOs automatically detect the probe attenuation setting, but the Atten DSO does not. There is no way on the screen to tell whether or not the probe is set to 10x attenuation or 1x attenuation or that the scope's setting matches the probe setting. Are we quite assured that the scope and the probe have the same attenuation setting for that May 1 set of scopeshots?

[TinselKoala]

Lawrence,

Borrow, buy, or rent an analog scope of any description and perform the same observation test to see if the current trace crosses the 0-reference line at any time while the input voltage is varied. Use different scope probes, preferably the ones that come with the analog scope.

I pulled out my 25 year old 20MHz two-channel Hitachi V-212, and performed the same sweep as before. The current trace did not cross the 0-ref line. The circuit oscillated this time down to an input voltage of about 0.38V.

Check out the video of this sweep:
http://www.youtube.com/watch?v=P2tETAwx3xc

This is in accord with my present results as well. I can only conclude that, for the one instance when I reported zero-crossing, I must have inadvertently flipped the channel coupling switch to "AC coupled" instead of DC coupled, after I switched the channel to Ground to check the reference zero level.