Testing the TK Tar Baby

[picowatt]

Stefan:

The second schematic you posted has problems.  The current flow from the Q2 array source is bypassing the shunt resistor so you are not measuring the current flow through the Q2 array.

The 10K potentiometer concept from TK is unfortunately fundamentally flawed.  The impedance going through the 10K potentiometer is too high and it will choke off the normal current flow.  There is supposed to be an impedance of 50 ohms at all times.

PW suggested a direct resistor connection between the 9-volt battery negative and the Q2 source node.  It would be lower than 50 ohms, perhaps 30 ohms.  This would allow approximately the same amount of current to flow through the circuit as compared to when you have a 50 ohm resistor connected to the function generator negative output.   However, even this has an issue.  This would change the "bounce up" voltage potential when current flows through the circuit as compared to the function generator/50-ohm resistance.

Like I said before, I personally don't see any point in going this route.  The function generator in negative oscillation mode injects power into the circuit, but it does not in any way inject power into the batteries.   The batteries will still discharge with the function generator in the circuit.

MileHigh

MH,

That "bounce up" as you call it is negative feedback.  And yes, changing the 50R with any other value will affect the amount of negative feedback (circuit gain) and the ability of inductances and strays to shift the phase of that neg. feedback (if phase shifted sufficiently, that negative feedback can become positive feedback, which will cause the circuit to oscillate).  There are other feedback mechanisms as well related to the CSR and a couple other sources that occur relative to AC.

I have always suggested keeping the source resistor close to the 50R similar to the NERD circuit.  A higher voltage battery could be used in place of the 9volt if 50R is used and a higher Ibias is desired (than a 9volt will allow).  But, again, we don't really know the NERD bias current.  I would temporarily go back to a bench supply and a 50R and vary the applied voltage (and therefore Ibias) until the waveforms have the same clipped appearance as in FIG 4 of the first paper, and then note the bias current/supply voltage.  From that a suitable battery voltage could be selected to provide a similar operating bias point.  With the bias current determined, the bias supply battery can be sized to provide the required run time.

PW

[picowatt]

@ALL,

Wow, everybody jumped at the same time regarding that ground point connection...

I am just too slow.

TK's on the ball!

PW

[TinselKoala]

TK,

That "floor" as you cal it is the gate to source turn on voltage.  Once you are up and into the turn on knee of Q2, the voltage at the source will vary only a small amount around that 4 volts unless you try to get the bias current well in the amps range.  If you look at the IRFPG50 data sheet at the turn on characteristics graph, the voltage observed at the source and the current measured will folllow that graph pretty much.  But in the 50ma to 300ma range, you may only see .5 to 1 volt of change at the source.

The series pot would be OK, but a wirewound ,though great for handling the current, will add more inductance.

In Stefan's schematic he has the bias source connected to real ground instead of the CSR, which will not allow the DC bias current to be accounted for and calculated into total power draw.  Also, even if Ibias is set to 25ma, the 9V will eventually drain even at that rate.  A good alkaline might get you a bit more than 500mAh of use, so around 20 hours at 25ma.  A 9volt lithium will provide around 1200mAh, so run time would be longer if used.  The additional load from the 10K pot, at just under 1ma would also add to the battery drain.

I do like the series pot idea you mentioned, as it would allow you to set current while observing the waveforms.  I would try to set the bias so that the oscillations resemble the clipped sines in the first paper FIG 4.

What's you "wild dog's" name?

PW

Maggie is her name. She's a feral dog that was rescued as an adolescent puppy from a pack that was roaming one of the local large greenbelt parks. I think she's part coyote and part chow. She's the most vocal dog I've ever met, not so much barking but talking. And she is totally not motivated by treats, since she knows perfectly well how to scrounge grubs and garbage and hunt lizards and mice for her food. After a full year of constant contact with me she finally is beginning to act like a dog rather than a wily coyote.

The Paper 1, Fig 4 waveforms are typical of what I see if I use higher drive voltages.

Stefan's schematic is wrong, probably my fault for having the color code on the battery wires mixed up. The 9volt system is entirely floating, too, the way I use it. If I also connect the _positive_ side of the battery/pot to the circuit negative rail common ground, it makes only a tiny difference in the gate signal and none in the oscillations on the common drains. If I connect the _negative_ side of the battery/pot to the circuit common ground, it kills everything.

Your explanation of the "floor" is also what I came up with, once I realized the relationship between the polarities and the actual potential difference between gate and source. I was confused by the "negative" going floor... when it's actually a positive going ceiling, so to speak !! It all makes good sense to me now.

Damn carbon pots in my lab always seem to develop holes in the resistive layer from current surges and cause all kinds of problems. Even the one I'm using now has a burned spot in it. A little inductance more or less from a wirewound pot probably won't turn out to be the reason my batteries discharge, though.

[picowatt]

Stefan:

The second schematic you posted has problems.  The current flow from the Q2 array source is bypassing the shunt resistor so you are not measuring the current flow through the Q2 array.

The 10K potentiometer concept from TK is unfortunately fundamentally flawed.  The impedance going through the 10K potentiometer is too high and it will choke off the normal current flow.  There is supposed to be an impedance of 50 ohms at all times.

PW suggested a direct resistor connection between the 9-volt battery negative and the Q2 source node.  It would be lower than 50 ohms, perhaps 30 ohms.  This would allow approximately the same amount of current to flow through the circuit as compared to when you have a 50 ohm resistor connected to the function generator negative output.   However, even this has an issue.  This would change the "bounce up" voltage potential when current flows through the circuit as compared to the function generator/50-ohm resistance.

Like I said before, I personally don't see any point in going this route.  The function generator in negative oscillation mode injects power into the circuit, but it does not in any way inject power into the batteries.   The batteries will still discharge with the function generator in the circuit.

MileHigh

MH,

I've always doubted the ability of the FG to charge the battery as well, unless some amount of charge pump action is going on.  The body diodes and inherent capacitances may to some degree allow this action.  A normal FG (not TK's, his is not "normal") can contribute around 2 watts max.  At a DC bias of 100ma, and Vbatt of 48 volts, that's 4.8 watts, so although charge pumping may contribute, if it were happening, I don't see it as enough to keep the battery charged.  Whether any charge pump like action is happening or not would have to be determined, but I lean towards agreeing with you.

Also, if the battery were being charged from the HF oscillation, I would think the AC at the battery would show evidence of clipping at closer to the battery voltage during the positive half of the cycle.  That is why I questioned the SLA battery's AC impedance.  I may hook up an FG to my vehice battery to see what the AC impedane of it is.  It appears that the AC is not being loaded much by the battery impedance.


PW

[picowatt]

TK,

Does Maggie have enough chow to have the purple tongue?  Great dogs.  I am trying to picture in the coyote.  Around here the coyotes are a skinny, somewhat scroungey looking critter.

After all that "independence", your lucky she's taking a shine to you...

PW

PS.. you'll hve to have Maggie narrate a video!!