I have a working SE output stage with Sovtek 6550 tubes in triode. OPT is NP Acoustics amorphous core 3.5K. I'm driving the tubes quite light - just 15.5W dissipation but I can increase that in future. The sound is quite nice - light, neutral, detailed. Maybe some kind of KT88s would give a fuller sound - what do you guys think?
The question now is what to drive the tubes with. I currently have 27 mesh tubes as input, but I could do with more gain than the mu=9 they provide. Given that the sound is quite light in tonality I could do with something fuller and richer. I can use type 56 tubes for instance, but I might sell this amp so I'm thinking that a 6SN7 would be more acceptable. Again - what do you guys think? Suggestions welcome.
The question now is what to drive the tubes with. I currently have 27 mesh tubes as input, but I could do with more gain than the mu=9 they provide. Given that the sound is quite light in tonality I could do with something fuller and richer. I can use type 56 tubes for instance, but I might sell this amp so I'm thinking that a 6SN7 would be more acceptable. Again - what do you guys think? Suggestions welcome.
Type 56 has mu of 13.8 according to Cunningham data sheet (https://frank.pocnet.net/sheets/029/5/56.pdf). That's not much different from 27.
A 6SN7 with its two triode sections in parallel should make a capable driver. Mu = 20, rp = 4k (approx.), gm = 5mA/V (approx.).
A 6SN7 with its two triode sections in parallel should make a capable driver. Mu = 20, rp = 4k (approx.), gm = 5mA/V (approx.).
Andy,
Are you sure, that it's the "best" operating point for trioded KT88?
If you choose another o.p., the required grid swing also changing.
Are you sure, that it's the "best" operating point for trioded KT88?
If you choose another o.p., the required grid swing also changing.
That's a good point. The "sound is quite light in tonality" could be due to the low current operating point chosen for the 6550-triode. You could try goosing the plate current a bit to see (hear) if the sound becomes more full that way.
Have you tried a 6L6 output tube? How does that sound? With those operating points, you could try a Russian 6P3C (not the "-E" version; the straight bottle one with the large black phenolic base).
Have you tried a 6L6 output tube? How does that sound? With those operating points, you could try a Russian 6P3C (not the "-E" version; the straight bottle one with the large black phenolic base).
The issue is the 3.5K primary. The 6L6 variants have a Ri of around 1.7K in triode so would prefer to see 5K. Of the 6L6 types the 5881WXT+ sounds the best of those I have and by some way, but that's into 5K.
I'm certainly going to put more current into the 6550 - I'll try different operating points.
Do you think it's necessary to parallel the 2 halves of a 6SN7?
I'm certainly going to put more current into the 6550 - I'll try different operating points.
Do you think it's necessary to parallel the 2 halves of a 6SN7?
Ah, right. I forgot about the rp in triode. Still, isn't 3.5k a bit more than 2X the 6L6 rp? That's often given as the optimal rp to load ratio for an SE OPT, at least for max power output.
Also, rp is not constant (even though it's called one of the "tube constants"). Running the triode with lower voltage and higher current (but within dissipation limits) lowers rp and increases gm, while mu remains constant (mu is the most constant of the 'constants').
With your B+ at only 346V and the output tube cathode biased, the plate-cathode voltage is going to be somewhere around 320V.
For 28W plate+screen dissipation, 28/320 = 87.5mA, which should be OK for a 6550, and even for a coin-base 6P3S-E. Can the power supply deliver that much plate current?
Re: 6SN7 --
It's not necessary to parallel the 2 halves of the 6SN7, but it does result in lower rp and higher gm, which are both useful attributes in a driver stage. I figure if you're already using a 56, you would have the holes made for one octal socket for each channel's driver tube.
Also, rp is not constant (even though it's called one of the "tube constants"). Running the triode with lower voltage and higher current (but within dissipation limits) lowers rp and increases gm, while mu remains constant (mu is the most constant of the 'constants').
With your B+ at only 346V and the output tube cathode biased, the plate-cathode voltage is going to be somewhere around 320V.
For 28W plate+screen dissipation, 28/320 = 87.5mA, which should be OK for a 6550, and even for a coin-base 6P3S-E. Can the power supply deliver that much plate current?
Re: 6SN7 --
It's not necessary to parallel the 2 halves of the 6SN7, but it does result in lower rp and higher gm, which are both useful attributes in a driver stage. I figure if you're already using a 56, you would have the holes made for one octal socket for each channel's driver tube.
Indeed - I can put more current through the 6550s. 60ma at least, maybe 65mA. It's probably a good idea to have 2 sockets for the input tubes. I can use 6J5s or 6P5s. Or swap for a 5 pin socket and use 76s of which I have plenty. I'll try those - 6.3V so I won't need an extra heater transformer. This looks to be coming together. As for Rp and the OPT primary, I always prefer 3x or 4x the Rp. Sounds cleaner and not so congested.
So for the output tubes there can be a choice of 6550 or KT88, or theoretically KT120 or KT150. Any recommendations for 6550 and KT88 types in triode? Fuller sound preferred.
So for the output tubes there can be a choice of 6550 or KT88, or theoretically KT120 or KT150. Any recommendations for 6550 and KT88 types in triode? Fuller sound preferred.
I tried out 76, 56 and 27 mesh as driver tubes and the 27 mesh were so utterly better sounding that they went straight back in. Since the 6J5 sounds no better than the 76 that rules that out as well. So we don't have quite enough gain overall, and that pushes the gain issue onto the preamp. A tiny bit more gain could be eeked out with a 2P29L but the hassle of filament regs and supplies seems hardly worth it. So it looks like it's going to be 27 mesh into 6550/KT88 after all unless a comparable driver tube emerges. I'm sceptical - the only tubes that sound as good as a 27 mesh to me are DHTs and then you don't have the gain either.
Back to the drawing board. After further listening the Sovtek 6550 isn't that good sounding in triode. Bass is one dimensional, treble lacks extension and delicacy and there's a kind of a midrange bulge which several users refer to. From what I read the KT88 is better but I don't have any to try.
So I'll try out PSE EL84, which at least will have enough gain for me to use my 27 mesh tubes as inputs. The EL84 in triode doesn't have any obvious vices I'm aware of. 6V6 is an alternative but the gain goes down.
So I'll try out PSE EL84, which at least will have enough gain for me to use my 27 mesh tubes as inputs. The EL84 in triode doesn't have any obvious vices I'm aware of. 6V6 is an alternative but the gain goes down.
Hi Andy,
Lots of people rave about the EL36 in triode mode, the so-called 'poor man's 2A3'. Have you ever tried that?
Lots of people rave about the EL36 in triode mode, the so-called 'poor man's 2A3'. Have you ever tried that?
I will second 6P41S. Nice tube. I have built three amps with them. I kept one and each of my children have one.
Tell me more about the 6P41S, what base is it exactly - noval, novar, magnoval?
What is the Rp in triode and the amplification factor?
What is the Rp in triode and the amplification factor?
Triode curves a couple of posts down here: https://www.diyaudio.com/community/threads/those-magnificent-television-tubes.211254/page-75
Thanks Fuling, I have the curves but I want to know what base is it exactly - noval, novar, magnoval?
What is the Rp in triode and the amplification factor?
What is the Rp in triode and the amplification factor?
The 6P41S uses a magnoval socket. Here are some measurements performed over the datasheet triode curves. Unfortunately the datasheet only shows curves for the low voltage region. For 130V/80mA, u=5.87 (simulated 5.63), rp = 500 ohms (10v/20mA, simulated = 578 ohms).
Last edited:
I am running mine at Vak=230V, Ia=60mA.
Interesting article comparing several soviet era tubes: Spectra
Interesting article comparing several soviet era tubes: Spectra
I have a prototype of PSE, 2x6P41S + 6N1P, 350Vak/56 mA per tube, measured and sounded pretty well.
After much thought I may go with PSE 4P1L. A pair would work at 250V a-k and 36mA each. Rp would be around 750mA and gain around 11. Just about perfect for 18W dissipation, and could be easily driven by a 27 mesh. I know PSE 4P1L very well since I built several in filament bias, but this time I'd just use normal cathode bias with a DC Link bypass. This would certainly sound better than PSE EL84 or 6V6. Also I have plenty.