• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Unusual amps..

Status
Not open for further replies.
Jump to Latest
#362
smoking-amp said:
Cathode resistor depends on what current you want the tube to run at. Its the auto bias. Plate resistor depends on B+ available and what load C its gotta drive (slew rate, bandwidth), and also determines linearity of the tube via loading effect.

"It would seem that we don't need the pot for balance."

All depends on how well you want it balanced. For Mu 20 tubes, 5% imbalance without it (unless you build it in with dissimilar divider resistors). How well matched the output tubes are also determines its usefulness. You can use the balance pot to bring mis-matched output tubes back into balance.
Click to expand...

Ok,

I'll keep the pot..🙂


So I guess the question is coupled or noncoupled cathodes on the PI?

Regards
M. Gregg
 
Last edited:
#363
According to MJ's book, V2 with its feedback over-rides any cathode balancing, so not much effect. Somewhere I read something to the effect that a common tail was OK for guitar amps to save the part, but separate tails were best for Hi-Fi. Had to do with positive feedback thru the tail. Crowhurst shows bypass caps on the cathode resistors for the floating paraphase, but not on the See-Saw. I don't see where they would be different, probably just a question of gain versus linearity objective, and output impedance.
 
Last edited:
#364
smoking-amp said:
According to MJ's book, V2 with its feedback over-rides any cathode balancing, so not much effect. Somewhere I read something to the effect that a common tail was OK for guitar amps to save the part, but separate tails were best for Hi-Fi. Had to do with positive feedback thru the tail.
Click to expand...

Sounds good to me 🙂

Well I guess its getting the values right and perhaps a triode/pentode or UL sw.

Does it show a bypass cap on both cathode resistors or just on V1...or is it a coupled cathode?

Regards
M. Gregg
 
Last edited:
#369
Hi Smoking Amp , maby it's wrong to take even an small amount of signal from V2 of the paraphase via a divider and place it in the grid of V2 , there is no reason to make this , you want to re-reverse that signal ?, when we know that V2 taking the signal from V1 via a divider and reverse it .
 
#372
So,

with the pot removed possible caps across the 1K cathode resistors..
As discussed by smoking amp..

This is all just a talking point at the moment..🙂




Regards
M. Gregg
 

Attachments

  • paraphase8.PNG
    paraphase8.PNG
    64 KB · Views: 255
#375
"The Gimp gave a hint in post #101 - The Valve Wizard"

Ah, that's where I saw the writeup about the single versus dual cathode resistors. And it gives the exact formula for the difference in value of the two feedback resistors if a variable pot is not going in:

Rf2 = Rf1 * [(1 + 2) / (Ao - 1)] (the subscripts for the nearest feedback resistors to V1 and V2)

What this is formula is using is the required V2 grid drive to get the observed difference voltages on the plates.

re: Dimitris AR:

What I'm trying to setup for the "Full" See-Saw is two consecutive stage 2's of the regular See-Saw (or floating para..), each one an inverter. The same formula given above gets used to derive the grid drives for each stage's grid. (well, because the two inverters are then connected in a loop, so the same plate V's get used each time but in reversed order.) The feedbacks then just insure the grids have the proper symmetrical differential voltages to develop the observed differential plate voltages.

Since two inverters in a loop nominally make an oscillator, we have to reduce the gain ever so slightly to avoid that. I would leave the second stage gain at -1 as usual and reduce the stage 1 gain to -.99. or: -(1-delta) Then the input signal has to provide the missing .01 piece to control things. The smaller the delta from -1, the smaller the input signal current needed to tilt the loop.

What is attractive here (the "Full" See-Saw case) is that the feedbacks insure that the output difference is directly related to the input difference (well single sided to 0 here) via resistor values. Whereas the usual LTP or differential stage develops its output difference thru the device gm's, leaving some residual odd harmonic distortion.

Which is not to say that the "Full" case would be entirely devoid of distortion. But variation of the gm's (or Mu's here if CCS loading is used) make the Ao vary slightly, causing the required Rf1, Rf2 values needed to vary. So end result is the delta term varies with signal slightly. This just varies the input impedance slightly. If delta is small enough, then a low input Z will overide the distortion factor.
 
Last edited:
#377
Yes, that will be a challenge. I think the (Full See-Saw) circuit behavior here is similar to the Hawksford Error Correction scheme used by the SS guys for output stage linearization. The obvious danger is that you have to live extremely near oscillation territory. But the SS guys get around the problem by enclosing global feedback around the whole thing (with plenty of loop gain to enforce control).

The Error Correction circuit behavior completely changes in that environment. As one tunes the gain pot up thru unity gain, instead of breaking into oscillation, the output error just goes thru zero and then grows again in an inverted form. They can look at the residual coming out of their THD meter on a scope and watch the error residual null and then re-expand inverted-wise.

So this strange inverter would be best enclosed in some, at least modest, amount of global feedback to prevent oscillation. Using this scheme on each stage in an amplifier would allow one to null out distortion throughout the amplifier. Then maybe someone really hits 1 ppm THD with a tube amp.
 
Last edited:
#378
So,

quote..Valve wizard Build out resistor could be a pot and adjust for "taste"..😕
LOL sounds like soup..(Dual pot ?) dual channel..

EL84's with 47uF /150 or dual 220. or mozener..

SS or tube rec...adjust for taste..LOL

Regards
M. Gregg
 
Last edited:
#379
smoking-amp said:
Yes, that will be a challenge. I think the circuit behavior here is similar to the Hawksford Error Correction scheme used by the SS guys for output stage linearization. The obvious danger is that you have to live extremely near oscillation territory. But the SS guys get around the problem by enclosing global feedback around the whole thing (with plenty of loop gain to enforce control).

The Error Correction circuit behavior completely changes in that environment. As one tunes the gain pot up thru unity gain, instead of breaking into oscillation, the output error just goes thru zero and then grows again in an inverted form. They can look at the residual coming out of their THD meter on a scope and watch the error residual null and then re-expand inverted-wise.

So this strange inverter would be best enclosed in some, at least modest, amount of global feedback to prevent oscillation. Using this scheme on each stage in an amplifier would allow one to null out distortion throughout the amplifier. Then maybe someone really hits 1 ppm THD with a tube amp.
Click to expand...

Look forward to your results..

Thank's for the help 🙂

Regards
M. Gregg
 
#380
Morgan Jones mentions the buildout R on page 403, 3rd ed. Valve Amplifiers. (and how to calculate it's value too.) He mentions that commercial amps routinely left out this part. What it does is equalize the output impedances of the splitter. Where it would matter most I think would be at HF and high power where one is running out of steam to drive the output tube's grid capacitance. Causing some slewing distortion. Whether that ever happens enough to hear anything, maybe not?
Doubtful one would want an adj. pot for it, just try a calc'd resistor some time to see if you can tell any difference.

Oh, and another thing, the buildout R doesn't fix the slewing problem on the high Z output, it breaks the low Z output to match it. Kind of like the old joke about asking for devine intervention for a broken arm by asking for both arms to be the same, then getting your other arm broken the same way. A better fix would be to put a source follower on the high Z output to actually fix the problem, if it's really an issue.

(the Full See-Saw has low Z outputs on -both- sides since local feedback is around each one)
 
Last edited:
Status
Not open for further replies.
Top Bottom