• 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.

CFB for tubes in parallel, PSE. Any surprises?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The most interesting, and annoying thing, about those twin anode valves was that the two anodes are not matched, and I dont know if they are just differently rated, despite being the same size (I cant speak for 2A3), and sadly, one plate always ended up red plating, while the other was fine.

Retrospectively, the single anode version (6P30B) is not as similar as one would think, and I planned to parallel these and compare.

But yes, perhaps twin anodes in a bottle, is already a PSE, so by the time I got to 3 valves, that's 6 anodes.
 
mondogenerator,

I am sorry for your bad experiences with parallel tubes.

I think there are others who have made parallel tubes work with results that are closer to the expected power, the expected distortion, and the expected reliability.

Poor performing amps, poor sounding amps, and also red plates have happened (ask me how I know).

I guess I will have to go back to designing and building a Parallel Single Ended amp, just to see if my memory of parallel amps is correct . . .
(I am getting old, memory escapes me, now what was that I said again?).
 
It's not a bad experience per se, just a reality that realising twice power with half impedance OPT and double the anodes, is, well....an over- simplification.

Now...this may be because of the twin anodes, which, like the triodes in a dual triode, are unlikely to be matched, other than by chance, or extreme accuracy of electrode geometry.

I guess the latter (geometry) is entirely dependent on the quality of assembly, and those Soviets were 90s production (so may be poor in comparison to other twin anode valves)

I can quite honestly say that I was able to halve OPT impedance, when I had 3, and not 2 valves in parallel, and achieve just 3x power, at half OPT Z.

4 valves in parallel was even better but I didnt pursue it for long, being as each biased at 15mA approached the DC rating of the OPT, and performance began to suffer. So I settled for 3 valves and almost 3W

Thankfully it was also at 70% of the THD of a single valve, less than 1%.

But paralleled valves concern me sometimes, as there is always the possibility (almost a guarantee) that one valve will redplate at some point.

But yeah, you're correct, I am just a hack....thanks for the kind reminder.

I am no expert, nor claim to be

Build others up, not tear them down.
 
Last edited:
I'm experimenting with SE amp circuits right now. Earlier, I had built a circuit that did ~10W with a single 6384 at low distortion. I wanted to scale this up to 20W.

The sensible thing to do would have been to make a parallel SE amp since scaling up presents an idle dissipation issue. 6384s can handle plenty of voltage.

I opted for the non-sensible solution of buying transmitting triodes to increase idle dissipation capacity. I did this because, well, the transmitting triodes look cooler than a 6384.
 
Spread spectrum, you're amps seems to be very low THD.

Far better than I have achieved - I almost exclusively use LNFB over one stage, and my only real foray into GNFB was from inverted secondary to the input stage cathode ina 3 stage amp.

It worked very well indeed, but proved difficult to maintain a nice level frequency response, with gain peaking above audibility, until I got it better controlled.

I would guess, again, that CFB is very dependant on either or both, an independent winding, and a good OPT free of ultrasonic resonances (or at least a single resonance that is easy to compensate; not the 3 or 4 that my OPT show, and cannot all be compensated for)

50AE is the person here to know, how to make his OPT perform well, and have the windings required to do this better than I.

Perhaps several CFB windings would be difficult to balance though, in terms of ratio, and parasitics to adjacent winding layers
 
Last edited:
Two Cfb windings for a PSE OPT can be perfect (i.e. identical windings) just using bifilar wire.

Regarding the 2A3 I have always found the 2A3 PSE more convenient than a 300B SE regarding PSU, driver and OPT. Distortion figures are basically the same, not much difference.
 
...almost exclusively use LNFB over one stage, and my only real foray into GNFB was from inverted secondary to the input stage cathode in a 3 stage amp.

I did experiments with nearly 100% NFB in the output stage and distortion still was higher than I wanted, so I decided to go for two stages and still keep the OT outside the loop and it ended up working pretty well. I thought putting an inexpensive SE transformer inside a global loop might be difficult.

Back to the original subject, as Kay Pirinha alluded to above, the additional hiccup here with a directly-heated tube is that if you use a shared filament supply independent self bias won't work. All filament supplies have to be independent as well. Since it is already going to be complex, I'd probably go all out and try to make fixed bias work with a bias servo but that's just how I prefer to make things.

Or I guess you could closely match tubes and check them every so often to see if they are drifting apart with age.
 
distortion increasing with tubes mismatch makes sense.

I will try going for pentode mode with UL + CFB, so input capacitance won't be much of an issue. Just looking for a 4P1L LTspice model to verify it. Some tubes aren't meant for CFB + UL.

Will probably go for a mixed (cathode + fixed bias) route, + voltage meter with switch to look across each cathode resistor's current.
Being a RF transmitting amplifier valve, there is an increased tendency for oscillation to be expected when putting those in parallel. BW limitation might be in order.
 
Personally, I would use LED self bias, which to my mind, is somewhere between self bias and fixed bias only because the Vgk voltage is pretty much fixed at the Vf of the diode string.

my mind might be totally wrong though :)

I'd then probably try a BJT cascade CCS anode load, one for each device.

Things become complex very quickly!

I may be completely off tangent, but that's just my 2 pence.
 
This will go tricky enough, due to the necessary individual current sensing resistors, won't it?
Best regards!

I have always wanted to see if I could use an optical isolator or something to make a good current sense on an electrode that isn't referenced to gnd.

The dual cathode windings that were discussed above could make this easier if the OP is making the transformer. But as you said, the directly-heated cathode complicates things.

I'm currently experimenting with directly-heated transmitting triodes as output tubes but I never considered combining with CFB, just because it seemed like it would really complicate things having the cathode of a directly-heated tube (running on DC) swinging around.
 
The dual cathode windings that were discussed above could make this easier if the OP is making the transformer. But as you said, the directly-heated cathode complicates things.
Yes, the CFB winding halves need to be isolated, not CT'ed. The urge of individual heater supply would remain anyway.
Directly heated valves really are a PITA to deal with, unless you restrict yourself on applying only a single one :rolleyes:.

Best regards!
 
1. Increase of chance of oscillation due to electrode intercapacitances increasing with the amount of paralleling?

2. Why the idea of separate CFB windings? No problem doing it trifiliar even, for 3PSE, but why do it? Separate heater supplies? Why?
 
1. Increase of chance of oscillation due to electrode intercapacitances increasing with the amount of paralleling?

2. Why the idea of separate CFB windings? No problem doing it trifiliar even, for 3PSE, but why do it? Separate heater supplies? Why?

Separate windings are mandatory if you use DHT or DHP. Not necessary for indirect type.

Oscillations? Not sure it will be a real problem.
 
Last edited:
As far as I can self explain, one should start with separate heater supplies if one wishes to have separate resistances in each cathode. Otherwise one heater supply low resistance will bind them all. Or you might get away with a CCS per each heater?

A similar idea comes in mind about the separate CFB winding perhaps?
 
As far as I can self explain, one should start with separate heater supplies if one wishes to have separate resistances in each cathode. Otherwise one heater supply low resistance will bind them all. Or you might get away with a CCS per each heater?

A similar idea comes in mind about the separate CFB winding perhaps?

Heaters with CCs? Why? You need a stable voltage if anything.

The CCs is a more refined circuit that also does voltage regulation like Rod Coleman's or TentLabs boards for directly heated types is different thing because the filament is the cathode as well.
 
As far as I can self explain, one should start with separate heater supplies if one wishes to have separate resistances in each cathode. Otherwise one heater supply low resistance will bind them all. Or you might get away with a CCS per each heater?
A similar idea comes in mind about the separate CFB winding perhaps?


One CFB winding for SE and two winding halves for PP is sufficient even with DHT's or DHP's. You need to connect an individual heater supply per tube to the CFB winding's hot end(s) via the bias RC combination (for cathode bias) or the current sensing resistor (for fixed bias).


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