• 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.

Diff drive

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Looks like RC-30 p.696 with CCS loads on triodes instead of pentodes
and without the direct plate to plate feedback (which might not be
great with your triode drivers anyway). Have you calculated some
real-life resistor ratios yet?

Plate feedback is very interesting with pentodes. I just read O.H. Schade
"Beam Power Tubes" and got inspired to play with pentodes (well
actually transmitting tetrodes...) and thought of plate feedback
directly to the plate of a pentode or drain of a MOSFET driver...
What didn't they think of 70 years ago???

Don
"Re-analysis (assuming no grid currents in the driver)"

The feedback from outplate to driver stage doesn't do anything
to lower the driver impedance... should be OK with receiving pentodes.

Cheers,

Michael
 
"What didn't they think of 70 years ago???"

Not much apparently. But it seems that only a limited portion of the ideas ended up in commercial amplifiers.

Hawksford EC was invented by a tube guy (Hawksford references his patent) but it only showed up in commercial SS amps after Hawksford expanded on it in his subsequent journal articles.

A very hot topic lately in power supply design called ripple steering is explained in Electronic Designer's Handbook by Landee, Davis and Albrecht in 1957 and patented over and over again as it was re-discovered 4 times afterwards.

The Harmonic Equalizer from WE tube days seems like an idea that was never fully developed after NFDBK was discovered. I think there is still some interesting potential there using the driver stage as the balanced mixer. (or an external mixer a'la S. Bench)

If cathode feedbacks can get rid of the odd harmonic residual of a diffl. stage, and so it appears, this should be primo stuff for Hi-Fi P-P amp design. While current feedback in SS ampls has a somewhat dubious high distortion rep., with triodes we have firm control of the gain using a CCS. Different animal altogether.

Don

post edit:

"The feedback from outplate to driver stage doesn't do anything
to lower the driver impedance... should be OK with receiving pentodes."

I think the feedback has to be lowering the driver impedance effectively too, because it will desperately try to get the output tube to behave accurately. But one can just use a low rp triode for the driver anyway.

It may also be possible to mix the two schemes (plate feedback to driver pentode and the cathode feedback to the triode driver) by using a driver pentode with crossed plate feedbacks to the driver screens and direct plate feedbacks via resistors to the driver plates. Plus the cathode feedbacks. The driver tube now acting like a triode and a pentode at the same time.
 
I try to avoid too much feedback in my designs, and if I do use it, the loop is very short


I think the anode load CCS is the better of the two, as it forces the triode to be more linear and work with more gain

I must admit a lot of these replies are over my head a bit, so I'm not really able to partake in most of the discussion (though the replies are very much appreciated! :) ), my designs are pieced together with educated guesses and intuition into the way things might work
 
"I just realised "I try to avoid too much negative feedback" is such a clichéd phrase these days "

I think the CCS in the plate version (post 21) should be reasonably buildable with its simplicity. The amount of actual neg. feedback of course depends on the selected resistor values for the feedback.

I myself don't like using huge amounts of feedback either usually. But since the neg. feedback equation shows that error reduces with increasing neg. feedback (ie, higher loop gain), I like to cheat and try to find some sort of anti-error signal somewhere so that the loop gain can stay low. Thats the idea behind the Hawksford EC, just compute the remaining error and subtract it out. But its theoretically equivalent to high gain, it just doesn't require actual device (tube) gains.

The WE Harmonic Equalizer also computes some anti-error stuff using a different approach (balanced mixer and common mode signals). So it gets by without high gain also.

The RCA RC-30 50 Watt design may be using the additional plate feedbacks (beside the cathode feedbacks) to take advantage of the slightly different transfer functions of plate feedback versus cathode feedback to null/tweak some distortion out. But since the driver tube and controlled output tube are in anti phase regarding current flow, this generally cannot produce wide range cancellation of distortion. (leading to odd harmonic residuals just like the diffl. stage at large signal levels)

To get tracking cancellation, the two devices need to be tracking current wise. So some extra cross coupled feedbacks (beside the cathode fdbks) are more likely to succeed. This would entail using some high value feedback resistors cross coupled between the output plates and the driver grids. Its not really an attempt to increase the overall feedback, just to partition some of it differently. This, of course, is a completely un-explored area. Maybe.., maybe..... the Citation II got it right?

Don
 
A bit of help for those following trying to follow this thread.

The RCA RC-30 Handbook can be downloaded from Pete Millett's site.
http://www.pmillett.com/tubedata/RC30.pdf

The schematic from page 696 referenced above is this:
Apologies for the quality but best I could manage with the postscript snapshoot tool.


Cheers,
Ian
 

Attachments

  • rca50w.jpg
    rca50w.jpg
    65 KB · Views: 313
Ian, several decades ago I built that inverse feedback RCA 50W amp but despite it's more complicated design, I could never get the performance to warrant the cost of the extra components, so I re-designed it as a basic p-p amp, which actually sounded better. I think the problem was the 6CB6 pentode drivers could have performed better as triodes, but that implies a complete recalculation of the feedback components. Not worth it.
To be fair, a well designed conventional design can produce a similiar thd signature.
DePalma's Audio Amplifier design of the symmetrical phasesplitter diff driver is the bit I'm working on: and this works well but needs refinement. Although 6SN7 triodes are used; pentodes can be used and the screen grids fed from o/p stage anodes.

Loads of ideas.

richy
 
Hi Richy,

"DePalma's Audio Amplifier design of the symmetrical phasesplitter diff driver ...."

You mean that Circuit 1 Vacuum Tube amp, dated 1-16-97?
http://depalma.pair.com/Analog/analog.html

Looks to me like each top 6SN7 driver acts as a follower and the bottom one uses its Mu gain, with the splitter pickoffs providing coordinated grid signal amplitudes.

I'm not sure how you plan to put a pentode in there, but one idea would be to put P channel Mosfets for the top tubes (with adjusted - smaller - gate drives) and the pentodes for the bottom driver tubes. (forming complementary matched current drives on each side) Then crossed output plate feedbacks to the screen grids. This would end up as two hi-Z/hi-gain drives controlling the output tube grids, with the plate to driver screen feedback resistors setting the effective gain and effective drive Z.

That configuration requires crossed feedbacks though, which makes me uncomfortable about leaving class A operation. (the feedback signals having to go thru the xfmr then) That could be fixed though by using a CT'd CFB winding on the xfmr. for the feedback pickoffs instead of the plates. The extra output grid drive voltage required should be a breeze for this driver configuration.

Don
 
Hi Jaap,

Say, that Scott Reynold's headphone amp sure looks similar to Bigwill's first post. The feedback scheme seems more necessary for output pentodes than triodes though. As has been mentioned a few times, one can save a tube(s) easily by just routing the feedbacks to the driver cathodes instead of a full diffl. pair for each side.

Don
 
On second thought, my idea in post 33 above (for DePalma circuit mod) isn't working too well, since the screen feedbacks only control the bottom pentode tubes (and not the P Mosfets). The feedbacks would need to go all the way back to the splitter to work that way.

I guess the top driver tubes could be kept as N-followers, and the pentodes (bottom driver tubes) with screen feedbacks could be gain matched to them. Maybe thats what Richy had in mind?

Don
 
OK, back to my idea in post #33 again. The P-channel Mosfets are now set up so as to operate in "near CCS" mode. The drives (Ra, Rd pots) to the Mosfets are adjusted so that the lower driver tubes (V1, V2) continue to operate with constant current WITH their actual loads attached (which would presumeably be some output tube grids and their biasing resistors).

This is slightly different than a CCS load, since the P-chans are adjusted to actually provide the output current so that the drivers still operate at constant current despite the load. The drivers (V1, V2) are low impedance outputs though, so they maintain control of the output voltages.

This allows the driver tubes to operate in CCS constant Mu mode, even with load currents. So this is an improvement over ordinary CCS loading.

I have drawn triode drivers for V1, V2 in the diagram for simplicity, but these could just as well be pentodes with screen feedbacks from the output plates (crossed feedbacks).

The interesting unknown here is what happens if the P-chans are adjusted for a little more current drive than the output requires. Then the V1,V2 drivers would begin to operate with a little less current when pulling down the load.

Since tube distortion increases with more loading generally, this would now provide anti-loading (or neg. resistance loading) for the driver tubes. Maybe one could adjust out some residual distortion from the V1,V2 tubes that way (providing they don't oscillate).

Don
 

Attachments

  • newsplitter.gif
    newsplitter.gif
    10.1 KB · Views: 114
Don, interesting info. I've tried the existing DePalma circuit 1 and find the adjustments far to snappy and the big problem is swapping tubes (old habit of mine) all goes haywire/out of trim.
My first instinct is to improve the exisiting circuit and re-examine the levels and avoid the wretched balance adj potty (preset). I have my doubts if the 6SN7 is the optimum tube for the job, but I'm currently hacking at the math.
You mention the CGS an active load. This should give thd reduction advantage even as the impedance at diff cathode junctions is already low and forcing into constant mu mode should nicely linearise the lot. Sounds too easy.. My other intention is to provide some regeneration feedback from cathode tail to grids. All this takes time and some refreshment reading up.
As I see it, the 6SN7 spec falls sim to alot of other tubes but >I'm after an even lower mu. I'm working on the assumtion of the concertina is capable of 30+30V rms on each 1/2drive. I don't need alot of extra gain and may even go for ECC88 types (using parallel sections) as to keep B+ within limits. It should be easily possible to provide 70+70V Rms from this circuit using alot less than 400V B+.
I've got alot of ECL82's, the pentode strapped in triode mode is another alternative.
Someone mentioned class B ? we'll run into this another time.


richy
 
Jaap said:
perhaps it is possible to make a nice small amp with this idea if you use ecc40 concertina => ecl82 triode/mosfet => ecl82 pentode (?)

I think with small amps there's no advantage. The o/p stage Miller effects are too small to dump. Circuit 1 with the right tubes should be able to be optimised without tweaking.

Since I'm in the BIG game 150W++ with parallel o/p pairs; Circuit 1 made me bash my skull against the wall why it took me solong to get coined with that split winding UL transformer technique. There's alot going for it although the winding is more complicated. The reason why; snubbers are minimised. And thats a BIG plus as the heat goes upwards.
Incidentally as the subject is about diff drive, I also had another look at the E.F.Worthen (modified New Isodyne). On the bench I wasn't impressed and the article does mention tube selection of the fittest.

richy
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.