6EM7 Headphone Amp

[Paul Barker]

https://ibb.co/4wr86Mp8

[Nick]

I would be even less happy having that connected to either side of my head.

[Cressy Snr]

OK,
I've applied DC heaters to the amp, using one of my (formerly Ed's) 12V SMPSs, which carry the heater current and enter the amp via a common mode choke and cap-filtered socket on the rear. The SMPS itself is powered from a captive cable that exits the amp via a cable gland. Silence reigns.

SMPS.jpeg (59.5 KiB) Viewed 10243 times

Amp.jpeg (70.05 KiB) Viewed 10243 times

I've kept the feedback as it was. There is only 0.9mV of DC offset per channel, measured with the headphones connected. That level is simply not worth worrying about.

The sound quality is absolutely bang on with the filtered DC heaters, especially in the treble region, where it is a lot smoother than the AC or the un-filtered SMPS supplied voltage.

I need to make up a decent case for the SMPS heater supply and acquire some Neutrik mains connectors and sockets for the supplies in and out of the SMPS and apart from the filtered socket I also need a Neutrik socket at the amp end.

This is a keeper, until I finally get round to doing a 45 HPA, which will be a 2 box job. It'll have to be, if I'm going to keep the noise down.

[Cressy Snr]

OK,
So the DC offset was bothering me.

So I've reconfigured the feedback by relieving the OPT secondary of any feedback duties and instead I'm running plate-to-cathode feedback from the 1626 anode to the cathode of the ECC82, with a 470nF series cap to keep DC off the ECC82.

Resonant freq of the series RC circuit is about 3.4Hz, well below bass frequencies, so phase problems ought to be minimised. But my circuit analysis skills are not good, so there's probably something lurking in parallel somewhere that'll feck everything up.

1626HPA_v2.3.png (39.62 KiB) Viewed 10182 times

[Nick]

Isn't it the 470nf and 1k that matters?

[Cressy Snr]

In that case we have a corner freq of nearly 400Hz. No feedback at bass frequencies then.

[Cressy Snr]

We'll have the cap out and I'll split the cathode resistor with a 100R under it and take the feedback to the mid point of the resulting potential divider. That should do it.

[Cressy Snr]

1626HPA_v2.4.png (39.42 KiB) Viewed 10100 times

[Cressy Snr]

The only problem with this arrangement is that hum levels have increased. Looking at the feedback path I’ve tested it and I appear to be injecting output stage ripple into the cathode of the input tube. The ripple is in phase with the grid so will be amplified. FFS you cant win! You sort out the heater hum with DC then inject ripple and feck up the noise floor with something else.

The best plan might be to just admit defeat, take out the feedback altogether and live with too much gain.

[Nick]

I still think the original feedback circuit is fine, just try a 100uf cap in the feedback link, if you prefer the sound without it, live with the DC if not why worry.

[Cressy Snr]

I think you’re right Nick.
After reading your post I’ve reinstated the original feedback scheme. I’ve left the 1k/100R potential divider in place. Without the cap I have 1mV of DC offset on both channels; not worth worrying over.

1626.png (39.68 KiB) Viewed 9993 times

[Cressy Snr]

New power supply:

1626PSU_v2.1.png (51.96 KiB) Viewed 9557 times

This and the schematic above is the version that will come to Owston.