Thanks to the great feedback from 45, we found out that I made a mistake in modelling the LL1682 OT in my previous post. In a nutshell, I was getting half of the power, doh!
I should have started from scratch, looking at the push-pull curves and estimating at least the A1 power from a pair of 6C4C in push-pull. So following the B.J. Thomson method plotted the curves in push-pull:
This quick and dirty analysis doesn’t take into account the OT losses at all, just a theoretical 8K8 anode load from the 8Ω load reflected in the anode. The actual real power will be less of course. The anode power will be the triangle area formed below the Zaa/4 load line. So looking at the maximum power then:
In class A1, this output stage should deliver at least 8-8.5W (if we factor in the OT losses and that grid current should kick-in before grid voltage reaches zero) with a driving signal of about 65V peak. So clearly the inductance ratio of my LTSpice model was incorrect. Revising the model I found where the problem was as both secondaries are in parallel looking at the Lundahl specifications:
We can see above that the reflected secondary impedance in the anode is about 8K8.
The revised amplifier then is presented below:
The fixed-bias is provided by a MOSFET follower which also delivers the grid current required when approaching into A2. No further changes in the design compared to previous post. We can see now that the circuit can deliver about 8W in A1 and with a nice harmonic profile. The H2 component is provided by the 4P1L driver and the H3 and H5 is characteristic of the push-pull topology. It’s nice to see that the THD is about 0.2% @ 8W until A2 grid current kicks in and a sharp increase of odd harmonics arises given that the driver is not suitable for A2 operation (THD is about 1.1% for 9W). Yet, there is grid current growing before reaching to 0V as we would expect. The 4P1L driver imparts a clear harmonic fingerprint in this amp so I’d be keen to enjoy the sweetness of the 4P1L or even some other drivers like 6e5P, D3a that would work quite nicely in this place and provide additional gain in case you don’t have the preamp required to deliver 8V peak at the input to take this amp to maximum power.
Interesting experiment to continue refining…
Great work Ale!
Brilliant idea for the final stage bias.
PP triode and IT phase splitter is one of my favorite configurations. Very balanced harmonics distribution and consequently a sweet combination of SE and PP sound. I already tried and almost done building a PP parallel 4P1L (LL1663) driven by LL1660 Alt V. I love the sound of it. Driver tube is now 6H30p-eb, which I found to deliver best sound and lowest THD compare to 6E5, 6S45p,6S3P. I am not done measuring THD, but I’ll come back when have all results. But in the end IMO all that counts is how it sounds and less how it measures.
I want to thank you for all your inspirational work
Hi, thanks for your comments! Yes, I realised this topology could provide the right mix of SE and PP. I’m excited to read your comments about the 4P1L in PP. I’m keen to explore it at some point, as the 4P1L is hard to beat.
Unfortunately I don’t have 6H30p-eb and these are expensive. I’m sure they are great, but will have to go for the 6e5p/6e6p or other pentode triode-strapped.
Let me know when you get the measurements done, I’m very interested.
Thanks, Ale
You could use the 6J9P-E that is rather linear as well in triode connection. I have just bought a NOS pair made in 1978. The E version has got golden G1 and should be better when full swing capability is required because gold plating G1 will reduce (and shift towards zero or even positive bias) grid current. Moreover you have about 3x gain in comparison to the 6H30. More or less for 60 V peak-to-peak drive (up to +5V into positive grid) at max Pout you should get away with just 0.62-0.65V RMS input biasing the 6J9P-E at nearly full power (3W+0.7W) around 175V/20 mA (-2V bias) and using the LL1660/18mA (2.25: 2+2).
P.S.
with 2V you could use LED bias for the 6J9P-E. Looks like a Zen amp..!! 🙂
I have some 6j9p-e with golden grid as well. They test well and traced their curves recently. Good idea, will simulate the stage when I get time over the weekend
Thanks for the ideas and feedback!
Yes! Now I would try the PP drive to see how much you can get in positive grid. It should be possible to get 12-13W. You could try with the LL1660/10mA and 2x4P1L with 250V plate voltage. The anode current high enough for the drive 25, 30 or 35mA enough to get to +20-25 V in positive grid region. Distortion at 8-9W should also be lower, I guess.
I am curious to see how much I could get out of the 45’s using the LL1620 (11.5K nominal, 12.4K effective), biased at 320V/30mA (-70V with DC filament supply) and 170-180V peak-to-peak drive. I have never tried a load different from the standard 5K.