HT smoothing for valve PP amplifiers

[martin.m]

I intend to build a homebrew valve Push Pull amplifier very loosely based on the 1950s Mullard Five Ten circuit. I notice that in this and many other similar amps, the HT for the output stage is taken directly from the cathode of the valve rectifier. Presumably the effect of any 100Hz ripple is cancelled out by the action of the PP output transformer.

Is this done for cost cutting reasons and would there be any advantage in fitting a smoothing choke or resistor in the HT feed to the output stage?

[GrimJosef]

Some of the effects of the 100Hz ripple are cancelled by the PP transformer - quiescent hum, for example, as long as the two output valves are matched or balanced to draw the same quiescent current. But the HT variation at 100Hz modulates the output stage gain and one of the effects of that is to superimpose sidebands at +/-100Hz on the audio signal. Their amplitude can be significant (we went through the numbers in another thread quite recently). How offensive you find it is a matter of personal taste.

I imagine it was done for cost cutting reasons. There would be measurable performance advantage in reducing the 100Hz ripple but as well as paying for an additional smoothing filter (CR or CL) you'd also have to up the mains transformer HT winding voltage to compensate for the resistive losses in the filter.

In the particular case of the Mullard 5-10 there are two circuit options - Normal Loading where the EL84s' screen grids are fed from after the smoothing stage and Distributed Loading where they're fed from before it. In the first case the impact of the 100Hz ripple should be a good deal less than in the second.

Cheers,

GJ

[snowman_al]

You are correct about the 'cancellation' effect of the push pull circuit with respect to 100Hz noise.
Others will have different opinions I'm sure, but I don't think there is a lot to gain adding a choke before the output pair.
You might try the Leak method though. Replace the 50uF main cap and make a 32uF / 100 ohms / 32uF pi filter. Then take the HT+ feed from the 2nd cap. You get some added smoothing and an easy way to measure the current through the amplifier as well.

Alan

[kalee20]

Originally Posted by GrimJosef

In the particular case of the Mullard 5-10 there are two circuit options - Normal Loading where the EL84s' screen grids are fed from after the smoothing stage and Distributed Loading where they're fed from before it. In the first case the impact of the 100Hz ripple should be a good deal less than in the second.

In fact there are 3 options, Normal Loading, Low Loading, and Distributed Loading. The first two use the EL84's as pentodes, the screen-grids are fed from after the smoothing stage (as you say), the difference being the output transformer ratio and the bias to the EL84's. The last is an ultra-linear configuration with the screen-grids fed from taps on the output transformer primary and yes, in this case the screens necessarily have the HT reservoir ripple on them.

[G8HQP Dave]

As always in engineering, numbers are your friend.

I built an amp similar to the 5-20 design. The PSU has just a 250uF capacitor feeding the output stage. No problem with audible hum, and somewhere I did an estimate of the likely hum intermodulation. See the recent thread for another estimate, with someone doing some measurements too.

[bikerhifinut]

Originally Posted by snowman_al

You are correct about the 'cancellation' effect of the push pull circuit with respect to 100Hz noise.
Others will have different opinions I'm sure, but I don't think there is a lot to gain adding a choke before the output pair.
You might try the Leak method though. Replace the 50uF main cap and make a 32uF / 100 ohms / 32uF pi filter. Then take the HT+ feed from the 2nd cap. You get some added smoothing and an easy way to measure the current through the amplifier as well.

Alan

I'd concur, it's effective speaking as an owner of a stereo20.
I'd go so far as to say you can get worthwhile improvements by upping the caps to 50+50 or even 50+200 after the resistor if using a valve rect. 50 uF is a practical limit for the first cap off the cathode if using a GZ34 or EZ81. If you havent got a mains transformer then save a lot of money and go for a silicon bridge rectifier. For not a lot more money, which you'll have saved on the transformer and valve etc you can use a nice big reservoir cap and CRC or CLC filter and get noticeable improvements. Chokes aren't terribly expensive in the scheme of things and a 10H 100mA choke would typically have a DC resistance in the 100 to 150 ohms area.

If you havent got a circuit for the leak TL12+ or stereo20 amps PM me and I'll send you a scan, including the differences in component values between later upgrades, you'd be better off building the later circuit as it gives the EL84 and easier life for the same bang.

Good luck, and do check out other variations, the Mullard circuits good but it isnt the only way.
I'd also look into lowering overall gain, these old circuits invariably have sensitivities around 100mV which is way too high for modern sources.
There's an easy fix and theres a probably cheaper fix if prepared to do a bit of maths and research.

A.

[martin.m]

Thank you all for your comments and advice. I am building this amp just for fun using parts collected over the years and a few items that were kindly donated by another forum member. A big HT resevoir cap is out of the question as an octal valve rectifier will be used. I have a large smoothing choke so will include it in the supply to the output stage. Hopefully the finished project will appear on this forum in the next few months. Thanks for the tip about lowering the gain. A sensitivity of around 400mv will be enough.

Regards
Martin

[Diabolical Artificer]

In my experience of building valve amps the IP stages are where noise and hum can be a problem, any noise will be amplified and as GJ sais your OP can get modulated, rideing on the ripple so decoupling the HT supply is a good idea. Valve rectification limits your options, with SS rectification you can use bigger caps, better RC filters. A bigger cap means a lower cutoff point for a HPF, you are restricted what R value you can use by how much V drop it has, so less wiggle room.

Lastly valve rectification is inefficient, can't think of a good technical reason for using it, though that could be said for valves full stop : )

Have fun, look forward to reading your thread, Andy.

[kalee20]

I built a small valve P-P amplifier, 15W, 5Z4 rectifier, but I did use a choke input filter followed by 150μF. (I just wanted the better regulation, and the much easier life everything gets).

It's the peak current that needs to be kept down. If you have enough series resistance, you can increase C in a capacitor-input filter. Above a certain point, increasing C has a diminishing effect on peak current, though ripple voltage decreases without limit. So, if you have a current probe, you can experiment. You may be able to do better than you think!