Testing the TK Tar Baby

[TinselKoala]

Hi TK,
in your video
http://www.youtube.com/watch?v=fC7zJouJAoU

it is not possible that there are flowing 200 mA DC bias current,
as you have a 10 KOhm pot and you turn it up into the center and
then only the oscillations occure at around Minute 2:20 .

But then the pot has already around 2000 to 5000 Ohms in this position
so the maximal current then able to flow from the battery is only around 1.8
to maximum 4.5 mA  at 9 Volts...
Just Ohms law...

No, perhaps I wasn't being clear. The oscillations _begin_ at a relatively centered pot, but the magnitude increases as the pot is turned further, and the "robust" oscillations happen near one end of the pot's travel, at this point. Even with other sources, there is that "voltage floor" that indicates that, over about 4 volts negative, there is suddenly a low impedance path activated and the current goes up. You have to enter this voltage area for the oscillations to occur.

So I don´t see the 9 Volt battery go flat very fast at this current level....

The 9v battery has indeed gone flat fairly fast during my testing today, and it's gotten warm, too. I have been trying for a setting that gives me right at 100 mA from the 9v battery, and that gives me about 190-200 mA indicated on the inline ammeter at the main battery, which I know will produce measurable load heating in a reasonable time. And will deplete my batteries in a reasonable time... although.... they are STILL over 12 volts each, and I haven't recharged them since many working days.

So a 200 mA DC bias current seems to be not required....

It depends on what you mean by "required". To get the oscillations of great magnitude, which indicate that the Q2s are partially conducting... hence heating the load.... I need to be able to supply substantial current, as far as I can tell. Since the magnitude of the oscillations is set by the voltage level, and the current then flows according to the voltage, I'm not sure if one can get oscillations without the current, just with the applied voltage. .99 and PW seem to think that you can, but I haven't had a chance to try all of their ideas yet. Certainly whenever I've been able to get oscillations, the current flow goes with, in lockstep. There has been one exception and that was when I cap-coupled the FG to the very early 2n7000 circuit. It oscillated fine, switched fine, and no DC current was able to flow in that gate drive loop.
So I think the bias source should be able to source 200 mA if asked to, but it's not always necessary to ask it to.

Regards, Stefan.

[picowatt]

TK,

I was always curious about your cap coupled to the source early FG experiments.  What did you use for a cap?

I would think the same Ibias would flow through the cap and would be seen as the voltage across the cap discharging when the FG is negative.  You might not have noticed with faster/shorter switching periods, or with a very large cap, but that Q2 source DC current has to flow somewhere.  If you AC couple the 9V battery, as you did the FG, you could "flip" the polarity once (to simulate FG switching) and having now charged the cap, the circuit will oscillate until the cap discharges.

Regarding the other bias circuit, again, that was with Q1 removed, a 50R placed across the FG terminal connection points (Q2 source to CSR), and then applying a positive voltage to the gate of Q2 instead of having it at ground potential.  Without those mods, the use of a positive voltage from the main battery string can only be used to turn on Q1 (as you stated).

I would think that in your use of the 9V batery as in the first part of the video, the DC drain current, or observed DC current at the CSR, should be the same as what you would measure coming from the 9V.  Possibly the oscillations are affecting an ammeter, particular if not perfectly symmetrical or sinusoidal?

PW

 

[TinselKoala]

TK,

Regarding your 9V battery video:

First, when you show the 9V across the pot and adjusted for oscillation.  Do you still have your DC milliammeter in the loop, if so, at what current draw does the onset of oscillation happn?  That 10K pot would drop much more voltage for a given Ibias then your 555's 10R or the FG's 50R unless you were very close to the end of the pot's rotation.  Also, as your oscillation looks a lot cleaner than RA's, it does look like the circuit is being biased at a different operating point.

The problem with batteries is that they do run down. I realise now that that 9v battery is almost flat in that video, showing about 8.4 volts no-load. That's probably why I found that the 50R series resistor was too large and limited the Ibias to about 45-65 mA even when the pot was cranked to the stop. So I changed to 10R and this gives me just almost exactly 100 mA on the Hickock. Onset of oscillation happens at about 12 mA and increases in amplitude until the max pot stop is reached. Yes, my oscs seemed cleaner.... because of the flat battery and the lack of voltage. At more voltage the oscs become increasingly distorted from that nice sinusoid.

Second, as for hooking the first main battery across the pot, the best you would be able to do with that is turn on Q1.  Have you looked at .99's burst osc schematic?  I was referring to applying a positive voltage to the gates of Q2 and omitting Q1 altogether.  Look at .99's schematic and you will see what I am talking about.

Yes, you are right and I have seen the schematic and sim results, and I know exactly what you are talking about, now that I know what "the gate" means. I guess you know that I've shown the oscs happening while Q1 is physically removed from its socket. But remember... there is another operating mode of the NERD device, one which turns on Q1 almost correctly, while still oscillating the Q2s in the usual manner. This is the one you've been asking about from their anomalous scope traces: the positive gate pulse which SHOULD show a current in their "shunt" but in some cases does not.... while in my system with known good mosfets and heatsinks, it always does.

Third, regarding attaching the positive of the 9V to the main battery string negative terminal, I would think placing a 10R to 50R, possibly even 100R, between the 9V negative and the Q2 source terminal should start the osc.  Possibly use the MOSFET end of the CSR instead of "real" ground for the 9V plus terminal connection.

I'm not following you here. As soon as I touch the positive pole of the 9v battery to the negative pole of the main battery, the "second" type of oscillations begins, regardless of whatever else I have hooked up where, apparently. Are you thinking that making the connection you suggest will stop these 25 MHz oscillations and allow the other ones to be produced?

In the circuit I'm using, I'm putting the "FG -" point right where it is supposed to be by Ainslie in the schematic that she "has said she has said" is the correct one. That is, I am on the transistor side of the CVR. This is NOT how it was wired for the demo video, apparently, but after being asked repeatedly what is correct and having her not being responsive to the issue, fuzzytomcat and .99 and I seem to have agreed that the correct schematic is the one I'm using as posted below, with the "MOSFET end of the CSR instead of "real" ground for the 9V plus terminal connection" just as it is portrayed in the schematic... but I'd rather call it the "FG -" connection, even though in this case it is the +9v connection. I think. If there's another "official" schematic I can do that too.... and I'll be another increment pissed off.

If you think about it, the first and last part of the video are essentially the same, as in both instances the positive terminal of the 9V was tied to ground, but in the first part you used the MOSFET end of the CSR and not the battery ground.

Yes, I think that's right, and in the first case--- the case of the NERD schematic.... I think the current didn't go through the CSR but may have contributed to the power dissipation. Does it (the power from the bias supply) get to the load, or is it dissipated in the mosfets, or what? Where and how to account for it, if the hookup is made on the transistor side of the CVR?

Of most interest to me, is what amount of current are you having to pass thru the 9V to make the circuit oscillate?  If you have the pot set so that the wiper resistance is very high compared to the FG's 50R (or your 555's 10R), your Ibias will be much lower than was likely used in the RA tests with the FG or with your 555.

PW

The circuit begins perceptible oscillations at about 12-15 mA on the inline meter. As the pot approaches the end of travel, the oscillations increase in amplitude, and the Ibias goes up. I've replaced your 50R with a 10R and this gives me 100 mA, about, on the Hickock with the pot fully turned. But my battery is weak so this will all change.

I do have some 78 series regulators in the box over there.

[picowatt]

TK,

With as much as you've worked on this today, I guess you did indeed not get any "real work" done!

Your efforts are appreciated.

PW

[TinselKoala]

TK,

I was always curious about your cap coupled to the source early FG experiments.  What did you use for a cap?

It was a little bitty ceramic cap that worked the best. I don't remember exactly but maybe 0.1 mF or even 0.01 mF. I do remember that bigger caps screwed up the waveshape. And I remember also that I couldn't get that same cap to work with the bigger mosfets, neither the 830a or the PG50, and when I tried bigger caps they didn't work well either. I don't have a cap substitution box... wait a minute yes I DO, I just remembered... where is that thing...... argh it will take me a while to dig it out if I can even remember which box it's in.

I would think the same Ibias would flow through the cap and would be seen as the voltage across the cap discharging when the FG is negative.  You might not have noticed with faster/shorter switching periods, or with a very large cap, but that Q2 source DC current has to flow somewhere.  If you AC couple the 9V battery, as you did the FG, you could "flip" the polarity once (to simulate FG switching) and having now charged the cap, the circuit will oscillate until the cap discharges.

Yes, I've seen that happen too. Perpetual oscillations until the gate was grounded to get rid of the charge, which may have been resupplied by capacitive coupling elsewhere in the circuit to keep it going.

Regarding the other bias circuit, again, that was with Q1 removed, a 50R placed across the FG terminal connection points (Q2 source to CSR), and then applying a positive voltage to the gate of Q2 instead of having it at ground potential.  Without those mods, the use of a positive voltage from the main battery string can only be used to turn on Q1 (as you stated).

I would think that in your use of the 9V batery as in the first part of the video, the DC drain current, or observed DC current at the CSR, should be the same as what you would measure coming from the 9V.  Possibly the oscillations are affecting an ammeter, particular if not perfectly symmetrical or sinusoidal?

Yes, that's right as  long as the mosfets aren't very on. Once they start carrying current the DC drain current... what I'm looking at with the inline ammeter at the battery.... increases over the value indicated at the Ibias meter.  15 mA Ib on the Hickock gives me 10-20 mA on the inline DMM (resolution issues) and 100 mA Ib on the Hickok gives me 190 mA on the inline DMM.
I'm not trusting the absolute values given by the inline DMM but it is consistent and therefore useful nevertheless, even if its accuracy is affected by the RF.