Testing the TK Tar Baby

[PhiChaser]

...by trying to INSIST that the function generator actually passes it's own current supply to a circuit is not only INCORRECT - it is UNBELIEVABLY incorrect.  It shows a want of precision that is entirely inappropriate to the study of science.  And it reflects - not on my competence - but on your own.  I am VERY well aware of what you can 'infer' by 'implying' that the function generator itself is passing its current to the circuit...

Um... Seriously?? I mean, REALLY ROSEMARY?!? A FG produces NO CURRENT??
That IS what you're saying right? A function generator doesn't put ANY current into a circuit AT ALL?? Can I quote you on that?? (Oh wait, I just did...)

I have an old FG that is coming in the mail (hopefully early next week) so I will definitely be testing that statement. My collection of analog uA meters (ranging from 0-15uA to 2mA) will show me the (un)truth of your statement in a moment... How to explain those moving needles? More magic??

You should be thanking TK (and the rest of the gang, you know who) for STILL trying to educate you on your own 'invention'... Wow...
I've only known about the RA circuit for less than a year and I'm already tired of her endless deflections and denials... Amazing patience you guys. Excellent work TK!! I can keep up with most of what you are doing and I'm no electronics expert.
Sorry to crap up your excellent research thread, I'm just irritated that RA continues to disparage your work when clearly (to me anyways) you have been trying to leave no stone unturned. Must be low on caffeine..

PC
A daily reader

[picowatt]

@All,

If the circuit draws under 200ma, and the batteries are 60A/Hr batteries, one would reasonably expect this circuit to run for in excess of 275 hours.

But, in reality, one would need to know the A/Hr rating of the batteries at this <200ma load.  Battery A/Hr ratings are usually specified using a specific load or discharge rate.  If a battery is rated at 60A/Hr with a 5 amp discharge rate, its A/Hr rating may far exceed 60A/Hr if the load current is very much less than the specified 5 amp load.  This is typically true for most battery chemistries.  Absent mfg data related to this discharge rate, a load test at this rate could be performed to arrive at the true A/Hr rating at this discharge rate.

Again, possibly the manufacturer has data available at this low 200ma discharge rate, but I would not be surprised if this circuit could run for well in excess of 300Hrs.  To expect as much as (or even more than) 550Hrs may not be unreasonable at this discharge rate with the 60A/Hr batteries utilized.

PW 

[picowatt]

TK,

If you have trouble maintaining the osc with the full battery string bypassed, consider coiling up some hookup wire and adding that in series with the battery.  Seems there was some noise a time ago regarding your Rload not having enough L anyway.  Coiling it up 3"-4" diameter will also help keep TB compact and portable.

Good luck,

PW

[TinselKoala]

@PW:
No, I have no trouble maintaining oscillations. I'm not _that_ old yet....   

But I have noticed a very interesting thing that I don't fully understand, and that's how the amplitude of the CVR oscillations changes radically as I move that channel's ground reference point up the negative battery wire towards the battery.

For this, I'm using freshly charged batteries, monitoring battery voltage at the battery itself, probe tip to most positive pole, probe ground to most negative pole, directly with no intervening wires and no decoupling caps.

I'm supplying a DC negative bias using the regulated PS and the pot/10R series resistor, no 50R, and adjusting so that the inline ammeter at the battery  negative  reads about 170-180 mA.

And I'm monitoring the CVR at the board, with the probe tip on the transistor side and the probe ground clipped directly to the other side of the CVR itself. (This is using the new 0.2 ohm 3 Watt  commercial current viewing resistor).

Done this way, the p-p amplitude of the CVR trace is about 4 V p-p.

Now, making no other changes, I move the CVR ground lead only, to the point where the negative lead from the battery clips to the board. This is less than 1 inch of buswire and a couple of solder joints and nothing else intervening. Now the CVR amplitude is about 5 v p-p and the inline ammeter is solidly at 180  mA.

Now, making no other changes, I move the CVR ground lead only, to the other end of the green negative lead from the battery, about 18 inches of wire, to where the inline ammeter is clipped in. Now the CVR p- amplitude is offscreen and requires an amplifier attenuation change (more volts/div) to display its new p-p amplitude of about 20 V. And the inline ammeter goes up too, to 250 or 300 mA.

Now, I simply disconnect the ground lead altogether for the CVR channel. The scope is still grounded by the other channel's lead at the battery negative pole. The magnitude of the oscillations goes up again to about 40 v p-p !! And the inline ammeter goes to 350 mA or so, the mosfets clearly get warm and the load clearly heats.

Before, I would have said that the apparent amplitude increase is due to the inductance, similarly to the Joule Thief measurement pitfall I've illustrated before. But since the inline meter and the load heating both indicate that the increased current is at least partially real.... I dunno.

I suppose the moral of this story is that the circuit is extremely sensitive to where and how it's measured, and even just moving ground leads from one "electrically equivalent" position to another can make radical changes in the measurements.

I'd like to know what happens in .99's sim under similar conditions. Of course since he's using current probes this effect might not be reproducible in the sim, unless he sets up CVRs and looks at voltage drop with a voltage probe referenced to the appropriate (or inappropriate) nodes in the circuit.

The presence or absence of the decoupling caps  that I've tried seems to make no basic difference in this. I noticed it in fact when I was trying a decoupling cap stack at the main battery, so I removed the caps and the effect remained.

[picowatt]

@PW:
No, I have no trouble maintaining oscillations. I'm not _that_ old yet....   

But I have noticed a very interesting thing that I don't fully understand, and that's how the amplitude of the CVR oscillations changes radically as I move that channel's ground reference point up the negative battery wire towards the battery.

For this, I'm using freshly charged batteries, monitoring battery voltage at the battery itself, probe tip to most positive pole, probe ground to most negative pole, directly with no intervening wires and no decoupling caps.

I'm supplying a DC negative bias using the regulated PS and the pot/10R series resistor, no 50R, and adjusting so that the inline ammeter at the battery  negative  reads about 170-180 mA.

And I'm monitoring the CVR at the board, with the probe tip on the transistor side and the probe ground clipped directly to the other side of the CVR itself. (This is using the new 0.2 ohm 3 Watt  commercial current viewing resistor).

Done this way, the p-p amplitude of the CVR trace is about 4 V p-p.

Now, making no other changes, I move the CVR ground lead only, to the point where the negative lead from the battery clips to the board. This is less than 1 inch of buswire and a couple of solder joints and nothing else intervening. Now the CVR amplitude is about 5 v p-p and the inline ammeter is solidly at 180  mA.

Now, making no other changes, I move the CVR ground lead only, to the other end of the green negative lead from the battery, about 18 inches of wire, to where the inline ammeter is clipped in. Now the CVR p- amplitude is offscreen and requires an amplifier attenuation change (more volts/div) to display its new p-p amplitude of about 20 V. And the inline ammeter goes up too, to 250 or 300 mA.

Now, I simply disconnect the ground lead altogether for the CVR channel. The scope is still grounded by the other channel's lead at the battery negative pole. The magnitude of the oscillations goes up again to about 40 v p-p !! And the inline ammeter goes to 350 mA or so, the mosfets clearly get warm and the load clearly heats.

Before, I would have said that the apparent amplitude increase is due to the inductance, similarly to the Joule Thief measurement pitfall I've illustrated before. But since the inline meter and the load heating both indicate that the increased current is at least partially real.... I dunno.

I suppose the moral of this story is that the circuit is extremely sensitive to where and how it's measured, and even just moving ground leads from one "electrically equivalent" position to another can make radical changes in the measurements.

I'd like to know what happens in .99's sim under similar conditions. Of course since he's using current probes this effect might not be reproducible in the sim, unless he sets up CVRs and looks at voltage drop with a voltage probe referenced to the appropriate (or inappropriate) nodes in the circuit.

The presence or absence of the decoupling caps  that I've tried seems to make no basic difference in this. I noticed it in fact when I was trying a decoupling cap stack at the main battery, so I removed the caps and the effect remained.

TK,  When you do what you do to increase Vpp of the osc, any noted changes in waveshape? (i.e., clipping, more distortion, etc)

As for maintaining oscillations, I am glad you can still dance.

PW