Hi all,
I'm starting my first scratch-built tube amps, and after hearing one in the flesh, I've settled on a 5-20 as a base. Wow, what a lovely sound.
I'm a little stuck on optimizing the circuit for 6L6GC's (my preferred power tube), is there anything it would need to work best for these tubes?
I'm also looking at picking up some iron for them, namely a pair of 6.6k OPT's, probably Hammond 1650G's, and matching Hammond 374BX PT's.
Would these be suitable for the 5-20?
Any suggestions?
Thanks!
I'm starting my first scratch-built tube amps, and after hearing one in the flesh, I've settled on a 5-20 as a base. Wow, what a lovely sound.
I'm a little stuck on optimizing the circuit for 6L6GC's (my preferred power tube), is there anything it would need to work best for these tubes?
I'm also looking at picking up some iron for them, namely a pair of 6.6k OPT's, probably Hammond 1650G's, and matching Hammond 374BX PT's.
Would these be suitable for the 5-20?
Any suggestions?
Thanks!
It's a good tube choice. With a 6k6 load, 450V, and fixed bias, you can idle at 50mA and get decent tube life with low distortion. Expect 40W in ultralinear, a bit more in pentode.
The 5-20 can be improved a bit by using matched plate resistors on the phase splitter and a CCS in the cathode. But this is shaving things pretty fine- it's a good amp as-is.
The 5-20 can be improved a bit by using matched plate resistors on the phase splitter and a CCS in the cathode. But this is shaving things pretty fine- it's a good amp as-is.
Is there any truth in the idea that beam tetrodes, like 6L6, have a lower input impedance than pentodes (e.g. EL34) so can't be driven from high impedance valves like ECC83/12AX7? I have seen this stated in various places, but never managed to find any hard data or plausible explanation. Is it just a myth propagated by all you (present company excepted) US ECC83/pentode haters?
Lundahl Transformers - 30W PP amplifier designed by Claus Byrith
Hi, definitely IMO worth a read regarding the 520 (and 510), rgds, sreten.
Hi, definitely IMO worth a read regarding the 520 (and 510), rgds, sreten.
This is the first time I have ever heard this, so I am not sure who is propagating this idea. I've not found evidence to support it in any short search either or from direct experience.. Would not the KT-66 and 77 be the subject of a similar conjecture? (Kinkless Tetrode?) I suspect a myth.. I've seen plenty of 5-20 with KT-66, 5881, and 7581..
I like ECC83 and certain pentodes a lot... Mainly as error amplifiers in my voltage regulators.. 😛 Actually used properly the ECC83 is a very linear device, just needs a little help driving a load - not always an issue depending on circuit design..
The Mullard topology is excellent and will yield good results with O/P trafos of modest cost, like Hammond. However, given its high RP and low gm, the 12AX7/ECC83 is a POOR choice for long tailed pair (LTP) phase splitter duty. A 12AT7/ECC81 as the LTP will yield approx. = stage gain, while exhibiting low RP and high gm.
SY's advice about a LTP tail CCS is spot on. A 10M45S integrated circuit is quite adequate for the job, but a more sophisticated current sink is, of course, OK.
SY's advice about a LTP tail CCS is spot on. A 10M45S integrated circuit is quite adequate for the job, but a more sophisticated current sink is, of course, OK.
I think I may have seen the myth on Audio Asylum. Maybe I should take less notice of what I see there. I just wondered whether there was smoke without fire. Glad to be reassured.
My understanding is that ECC83 is a good choice for LTP phase splitter as it has high and stable mu. It is mu which primarily determines the ratio of differential-mode gain to common-mode gain, which is the main issue with an LTP phase splitter. In fact the mu is high enough that a tail CCS is not strictly necessary, whereas it is necessary with almost any lower gain valve. The snag with ECC83 is that it needs to be run rather too close to the grid current region so imbalance can result.
My understanding is that ECC83 is a good choice for LTP phase splitter as it has high and stable mu. It is mu which primarily determines the ratio of differential-mode gain to common-mode gain, which is the main issue with an LTP phase splitter. In fact the mu is high enough that a tail CCS is not strictly necessary, whereas it is necessary with almost any lower gain valve. The snag with ECC83 is that it needs to be run rather too close to the grid current region so imbalance can result.
This is a bit of a myth based most likely on the effect of interelectrode capacitance within the valve.
This capacitance is made worse by the effect of Cag times the mu (voltage gain) of the valve. This is known as the Miller effect. If say you have control grid to anode capacitance of 10pF and the voltage gain is 20 then the effective capacitance from grid to anode is 200pF. This capacitance is effectively 'seen' by the driving valve and becomes a loading impedance, but the effect is only really significant at radio frequencies.
The difference between a 6L6 and EL34 is much less then the different operating modes, ie, triode mode is much worse then pentode, with ultralinear somewhere in between, depending on tapping ratios.
Regards
Henry
This capacitance is made worse by the effect of Cag times the mu (voltage gain) of the valve. This is known as the Miller effect. If say you have control grid to anode capacitance of 10pF and the voltage gain is 20 then the effective capacitance from grid to anode is 200pF. This capacitance is effectively 'seen' by the driving valve and becomes a loading impedance, but the effect is only really significant at radio frequencies.
The difference between a 6L6 and EL34 is much less then the different operating modes, ie, triode mode is much worse then pentode, with ultralinear somewhere in between, depending on tapping ratios.
Regards
Henry
The 5-20 doesn't run the output tubes as triodes. The input capacitances are really, really small. 200pF is huge and VERY significant for audio purposes.
No I wasn't thinking of Miller effect. I assumed whoever started the myth was talking about grid current, but I can't think of any reasonable explanation why that would be worse in a beam tetrode than a pentode. Always nice to kill a myth!
Repeat, the 'X7 is not a good LTP choice. Differential or not, stage gain is a function of μ, RP, and net ZL. Throw "fixed" bias O/P tubes, where net ZL is likely to be under 100 KOhms, into the mix and both stage gain and linearity suffer. Remember, that the RP of the 'X7 triode is approx. 62.5 KOhms. Kevin made a good point about the 'X7 being a fine voltage amplifier, when properly loaded.
BTW, in a circuit with a GNFB loop, like Mullard topology, high gm reduces vulnerability to HF error correction signal induced slew limiting. At 1.2 mA./V., the 12AX7/ECC83 is anything but high gm. I am VERY fond of Mullard's topology, but not their small signal complement. Look at the excellent H/K Cit. 5, where Mullard topology is combined with high gm small signal types.
BTW, in a circuit with a GNFB loop, like Mullard topology, high gm reduces vulnerability to HF error correction signal induced slew limiting. At 1.2 mA./V., the 12AX7/ECC83 is anything but high gm. I am VERY fond of Mullard's topology, but not their small signal complement. Look at the excellent H/K Cit. 5, where Mullard topology is combined with high gm small signal types.
I completely agree with Eli on this one, and the rp in LTP connection is a good deal higher than would be the case for common cathode operation - it could be significantly >100K - further reducing the already limited capability to drive even the small amount of capacitance at the grids. The 12AT7 really would be a better choice, so would 6CG7/6FQ7 or ECC99..
Leak even recognized this and in some of their later amps used the 12AT7 in the 5-20 configuration instead of the 12AX7... The Mullard topology is superior, the choice of tubes in the original implementation not so much.. IMHO..
Leak even recognized this and in some of their later amps used the 12AT7 in the 5-20 configuration instead of the 12AX7... The Mullard topology is superior, the choice of tubes in the original implementation not so much.. IMHO..
I like Morgan Jones's comparison of the 5-20 and the Williamson- the Williamson trades stability for linearity, the Mullard trades linearity for stability. My last Mullard-ish build used a 12AX7 input tube and a 6SN7 phase splitter to try to get the best of both worlds.
Eli, the other thing to note is that with the rather low AC load you brought up (it doesn't have to be that low, depending on the output tubes and biasing scheme), the predominant nonlinearity will be second order, which is canceled in the output stage. So it's not as bad as one might think.
This the version of the 5-20 I'm building. I'm keeping the tube rectifier though and adding an EB91 full wave rectifier for negative bias for the EL34s.
Here's my chassis....
I'm adding to 50k pots (DC Bias) and a 3.5mm jack socket (for a meter) next to the EL34s. A 50k pot next to the ECC83 to balance the anode loads (AC Balance). An EM84 just for show and 700v polyprop caps on the HT line. Two sub min Toggle switches on the flap at the back will set the Mains Voltage. Gold plated Transformer end covers. The whole thing will be covered by a dummy chassis (or cover) to look like this....
No electrolytics or semiconductors. Russian Paper in oil caps. Transformers & choke are drop through types from VVT Transformers. A lot of time, effort and money is going into this project. That's a measure of my faith in the Mullard 5-20 amplifier.
Last edited:
As for the 6L6GC, I'll be giving them a go in my amps.
Another version worth comparing is the Radford MA. It uses an ECF82 instead of the ECC83. I don't believe CCS's are needed in these amps at all.
Another version worth comparing is the Radford MA. It uses an ECF82 instead of the ECC83. I don't believe CCS's are needed in these amps at all.
Last edited:
Is the splitter generated distortion cancelled? Even order distortion products that are internally generated by PP "finals" get cancelled. OTOH, what's present in the I/P signal is amplified.
Thankyou all for your insights, I'm taking notes!
Sounds like the 12AT7 is the way to go, I've got a few Blackplate RCA's around which would be perfect. Might be worth rolling similar spec tubes whilst I'm breadboarding it out.
Sounds like the 12AT7 is the way to go, I've got a few Blackplate RCA's around which would be perfect. Might be worth rolling similar spec tubes whilst I'm breadboarding it out.
Not if it is presented in common-mode, as second-order from an LTP will be. This is one of those rare cases when one stage can genuinely clean up after the one before.Eli Duttman said:
- Status
- Not open for further replies.