Mystery amplifier

[Diabolical Artificer]

Cascodes are usually used where their better bandwidth

Apologies, a little behind & a tad off piste. Re bandwidth do you mean in terms of frequency or some other parameter? A regulated power supply operates in a very narrow frequency range - DC to a 100hz or so, or does it.

Have you tried powering the PSU up yet with or without the valves Will? do you know how to safely, need any guidance?

Andy.

[G6Tanuki]

A regulated power supply feeding anything other than a Class A audio (or HF/VHF linear) amplifier can rapidly get complicated.

When the regulated supply needs to cope with the varying current demands of any amplifier whose characteristics include the letter B then the regulator can rapidly become an ~interesting~ design exercise where you need to consider slew rates, undershoot and overshoot, loop gain, ringing and other things involving the sort of maths that cause some people to run away and hide.

It's all too easy to find that building a regulated supply for an audio power amp means you are essentially designing a second audio power ampl!

[Kokotoni Will]

Originally Posted by Diabolical Artificer

Have you tried powering the PSU up yet with or without the valves Will? do you know how to safely, need any guidance?

I need to get hold of a Bulgin mains connector first. I could attempt to drill new holes for a kettle plug, but would rather preserve it in its current state. I also need to replace R8 - it should be 48k, but measures 13M.

[Richardgr]

This could be a good time for the inaugural UK Vintage Radio Live Stream!

[PJL]

470 ohm cathode resistors is unexpectedly high, the datasheet says 330 ohms. Assuming each valve has its own cathode resistor, this would seem to make the anode current not much more than 25mA with -13V on the cathode. I would also remove the shielding cans from the 4 valves although the low quiescent current may help.

[Radio Wrangler]

There used to be some 'cans' for valves which amounted to springy, thin, blackened metal which wrapped tightly around their valve, with large area contact on the glass envelope. These were designed for improving the cooling of the glass rather than for electrical screening. Tgey confused some people, who got them mixed up with the ordinary traditional cans with their bayonet fitting onto a skirted socket at the bottom, and a conical spring onto the top of the valve.These were really electrical screening and made temperatures worse. So, back in the day, there was some wrong information circulating, and whoever made this amplifier might have picked up on it. If the screens on the output valves of the amplifier are the loose fitting type, they need to come off.

The thermal improver sort cropped up in some military stuff and some high performance commercial stuff. You unplug the valve with them still on and then have to unwrap them carefully.

David

[Kokotoni Will]

After re-soldering R8, it now consistently measures about 48k, which is odd, because it definitely wasn't before.

I've got hold of the correct Jones connectors for both PSU and amp, and have a 3-pin Bulgin plug coming from a forum member. As the 24-pin connector is missing its shell, I'll make one using Milliput.

The cans on the output valves are the loose fitting type. I'll take them off when it comes to testing the amp.

I'm looking into getting a variac - does anyone have a recommendation?

[Kokotoni Will]

After posting some photos on a Coventry Facebook group, I think I've managed to make contact with the son of the man who made the amp. His name was Arthur Buck. He was one of the very first computer engineers for ICT, and later ICL. He worked at Bletchley Park before that, which explains the military spec parts. The amusing but not strictly relevant attached photo is of him as a young boy, taking a rest during a motorcycle trip. I'll post when I have more info.

[Diabolical Artificer]

Well done finding out about the builder & that's a great picture; "ahhhh lovely fags" a la Father Ted.

Andy.

[Radio Wrangler]

Originally Posted by Diabolical Artificer

Apologies, a little behind & a tad off piste. Re bandwidth do you mean in terms of frequency or some other parameter? A regulated power supply operates in a very narrow frequency range - DC to a 100z or so, or does it.
Andy.

I've just realised that quoted quote must have been me.

In a cascode circuit, the capacitance from the output of the amplifying device back to its input is broken. There are now two capacitances in its place with a grounded node in between them. This seems similar to what a screen grid in a tetrode/pentode does, but it can be even more effective. The anode of the bottom valve sees a load looking like a cathode follower Zout. Very little anode voltage signal gets to drive Ca-g1 on this valve.

A single valve exhibits a single pole. A cascode pair replaces this with a pair of poles, which sounds very bad. BUT, the two poles of the cascode are both not scaled back in frequency by the Miller effect and the stage gain.

So the gain you get is still flat until well above the frequency where the simple stage has started rolling off, and well above the frequency where the simple stage has started showing serious phase shift. Win-Win! You don't start losing until you get far above where the simple stage rolled off, and then you get double the rate of roll-off (dB/octave), double the phase shift. You need to be sure that the loop gan is dead and buried before you get here.

Applies to valves, applies to semiconductors. Cascoding iis a powerful design tool.

David

[Diabolical Artificer]

Thanks David & others. Andy.

[Kokotoni Will]

There are 8 Visconol X "Metalpack" caps in the amp, 6x 0.1uf and 2x 0.05uf. They all look to be in good condition, and measure within 10% of their capacitance. Is it worth changing these out for modern equivalents?

[trobbins]

I'd recommend measuring leakage current, and ESR for each. I was hoping to retain 7x 0.1uF 500V Dubilier 470B from 1960 the other day that were in an instrument I'm restoring, but the variation in measured levels of leakage and ESR was quite wide, even though capacitance was nominal for all, so I've ended up replacing. It may seem long-winded to test them, but at least you have a good basis for either retaining, or replacing.

[Radio Wrangler]

10% is probably original tolerance.

David

[PJL]

I cant see the Metalpak capacitors in your pictures BUT they will almost certainly be leaky. Some are likely to be grid coupling to the output valves and it only takes a few uA to cause havoc and destroy the output transformers. If you plan on using the amplifier I would replace all Metalpak and electrolytic capacitors and any resistors that have drifted out of their tolerance range.

[Kokotoni Will]

There's two in the top corners (circled), plus six more hidden underneath the board all the resistors are mounted on. They are actually marked 20% tolerance.

I'm not sure if I'm reading the resistor colour bands correctly. I've added a photo of the left channel with their actual values - do they look correct? They're broadly the same across both L and R channels, so I guess they must be reasonably within tolerance.

[Richardgr]

I think it would be hard for anyone here to give you the assurance you are looking for.

Whether they are likely to have drifted and whether that could cause an issue depends on how they have been stressed and their role in the circuit.

I think you need to work with an abstract view of what you are dealing with, i.e. create a schematic, and work with that, maybe adding a layout view that links the components on the turret board to their identifiers on the schematic. The time invested will enable you to visualise the circuit and will make it much easier to get guidance.

If you download LTSpice, it is an extremely powerful tool that, once you have mastered it, enables you to emulate the operation of a circuit.
BUT, at its most basic, once you have got the hang of the drawing tool, it is an excellent aid for creating a schematic. Just add all the resistors, capacitors, valves and transformer windings. Then link them with connectors, and then the circuit is 'glued' together and can be moved around for readability.

Valve amplifiers are usually comprised of stages, linked with coupling capacitors, so focus on one stage at a time. Valve amps generally follow similar designs, so it will start to fall into place as you go along.

[Diabolical Artificer]

Leaving the caps in place without testing at full working voltage for leakage is like lighting a firework held in your hand, you may be ok, you may end up in A&E minus a few fingers.

Some of the resistors as you have seen have drifted in value,a well known issue with these resistors. It may not matter, depends where they are in the circuit, as Richard says, your flying blind without a schematic.

Hows the search going for a variac? Andy.

[Kokotoni Will]

The amp schematic is coming along - I've finished the left channel.

I think I will probably change the paper caps as I have no means of testing them properly without a 500v power supply. I should be able to reform/test the electrolytics with a 30v power supply - they're rated at 25v or less.

I've made a lamp limiter, and have made a video of powering up the PSU with it:

https://youtu.be/j7PPvMx1Xqo?si=JgW6GHJl6xI45e6I

The only valve fitted is CV378. The multimeter is measuring voltage across reservoir cap C2. Voltage rises to about 466-467 after about 5-10 minutes, then shortly after goes back down to 463-464. The unit makes a very quiet, low hum, quieter than I'd expect from something this size.

Does everything look ok so far?

[PJL]

Are C1 & C2 the large metal boxes mounted to the top? These will be paper in oil and are typically very reliable. You have marked them as 8uF 152V but I suspect the 152V is not the working voltage but a manufacturing batch code. You should look for other marks on them to confirm the working voltage is 500+V. before powering up again.