Head-Fi.org › Forums › Misc.-Category Forums › DIY (Do-It-Yourself) Discussions › The Heretic. A hybrid push-pull 4P1L DHT / SS / SMPS / thru-hole / SMT design.
New Posts  All Forums:Forum Nav:

The Heretic. A hybrid push-pull 4P1L DHT / SS / SMPS / thru-hole / SMT design.

post #1 of 6
Thread Starter 

A few months ago I bought some 4P1L tubes. They're Russian, directly-heated pentodes with a plate dissipation of ~7.5W. With octal bases, they're readily available at reasonable prices.

 

I always had a fancy for building a hybrid with the solid state at the front of the amp. The usual thing is to use a MOSFET as the power stage for a tube front end to avoid the need for an expensive, heavy transformer. I just like unusual juxtapositions of components. I had no idea what I was going to do with the tubes. Gradually, as I looked at the arrangement of the heater circuit, some ideas emerged.

 

I called it The Heretic because it runs against the grain of most SS/tube designs, most are a compromise designed to bring the benefits of tube amplification without incurring the cost implied by a transformer. I've also chosen to use a SMPS and SMT components in many instances. Both of these are frequently avoided by traditional tube enthusiasts. It's a push-pull design, all the excitement these days in low-power tube amps is for single-ended, so it cuts across the expected in more ways than one.

 

The 4P1Ls have centre-tapped filaments. I decided to tie these points to ground and make use of a low-power DC SMPS for the heater supply. I picked a 5V Meanwell supply for this. They only cost U$10~15. Each end of the filament is driven by a constant-current source built with an LM317. Since the supply required is only 2.1V there's plenty of overhead to keep the LM317s in regulation.

 

Power supplies are often a problem with tube amps, so nibbling away at the requirements like this can sometimes help provide a solution.

 

If the filaments are to be grounded then a bias supply will be required. Obviously, if there's going to be an opamp SS front end then +/- LT supplies will be required. The negative rail of this will conveniently provide bias for the tubes. If the grids are biassed to -8V then the tubes will run ~45mA. With 150V B+, this gives a plate dissipation of 6.75W, a little high, but we can make this variable.

 

Lastly, a gain stage and phase splitter is required. A non-inverting opamp drives one tube and the input of an inverting opamp, which drives the second tube. The non-inverting opamp provides a high input impedance, a shunt resistor of 1M defines the impedance at this point and a nod to keeping RF out of the amp is made by the inclusion of a small cap here. The non-inverting opamp is arranged with a high gain to limit its bandwidth and the global feedback loop from the transformer is closed round its inverting terminal.

 

Two back-to-back transformers provide isolated supplies of 150V and +/-18V regulated.

 

 

...here you can see the low-power distortion which is vanishingly small...  ...0.001%THD for 7.5mW in 8 ohms.

 

 

...and the distortion @ 3.125W into 8 ohms, which is not a huge power to extract from these tubes, but I think a well-chosen transformer will provide more than this. All-in-all it looks pretty good (0.2% THD), it's just a shame about the capacitor coupling.

 

 

Below, the FR.

 

 

Nicely accommodates the audio spectrum and gives no immediate concern about stability, gain is <1 @ 180 degrees

 

I don't know when I can get around to building this, I don't have the energy for doing much apart from working a keyboard and mouse ATM, although I don't seem to be short of ideas. The 4P1Ls are highly thought of in terms of their linearity, they're triode-strapped in this instance, directly-heated tubes and push a lot of the right buttons for many enthusiasts, and the amplifier will almost certainly drive headphones perfectly adequately in the bass with Hammond 125D universal transformers, about 23 ea. All-in-all with the intended power tx's, not too expensive. The simulation here uses a NE5532 model, but an LM4562 would probably be substituted.

 

w

post #2 of 6

This looks very fun! 

 

For a while (maybe still?) a few cray-cray DIYers were using relatively low power (5-20W) gain-clone-style op amps to drive SET DHTs well into grid current. good times, good times. 

 

For the push-pull-ness of the output stage, why not use a more symmetrical input stage? I'm sure its only an argument on paper, but it seems weird to not offer balanced inputs 

 

Could you maybe find some distortion cancellation by switching to a long-tail-pair output stage, in place of the grid biased one? 

post #3 of 6
Thread Starter 

Thanks nikongod, for the input. I'll try the LTP, although I think it gets very complicated keeping the heater supplies out of the picture, I was hoping to avoid having gyrators. Should be fairly simple in simulation, tho'.

 

w

post #4 of 6
Thread Starter 

I tried putting a depletion MOSFET CCS in the tail of the output LTP (as it became), but the distortion was slightly (minutely) worse than with the grid bias.

 

w

post #5 of 6
Quote:

Originally Posted by nikongod View Post

 

For the push-pull-ness of the output stage, why not use a more symmetrical input stage? I'm sure its only an argument on paper, but it seems weird to not offer balanced inputs 

 

 

Wakibaki's input stage has the strong merit of offering a high input impedance and an easy way to wrap feedback around the whole amp with a minimum number of parts. With balanced inputs, it might not be as easy to keep those benefits.

post #6 of 6
Quote:
Originally Posted by wakibaki View Post
 

Thanks nikongod, for the input. I'll try the LTP, although I think it gets very complicated keeping the heater supplies out of the picture, I was hoping to avoid having gyrators. Should be fairly simple in simulation, tho'.

 

w

 

Thanks! 

 

Quote:
Originally Posted by wakibaki View Post
 

I tried putting a depletion MOSFET CCS in the tail of the output LTP (as it became), but the distortion was slightly (minutely) worse than with the grid bias.

 

w

 

Bummer

 

Quote:
Originally Posted by 00940 View Post
 

 

Wakibaki's input stage has the strong merit of offering a high input impedance and an easy way to wrap feedback around the whole amp with a minimum number of parts. With balanced inputs, it might not be as easy to keep those benefits.

 

I did not consider the global feedback around the *whole* amp. Good call. 

New Posts  All Forums:Forum Nav:
  Return Home
Head-Fi.org › Forums › Misc.-Category Forums › DIY (Do-It-Yourself) Discussions › The Heretic. A hybrid push-pull 4P1L DHT / SS / SMPS / thru-hole / SMT design.