Valve tester HT PSU puzzler

[G8HQP Dave]

Originally Posted by TrevorG3VLF

Just a thought, are the power supplies going into oscillation.
I presume they are stabilised.

Good point. Stabilised power supplies are often essentially followers, and followers can become unstable with a capacitive load. What do people usually do: add a capacitor to the output of a stabilised supply "just to make sure"!

The usual solution is to add a 'stopper' (i.e. small build-out resistor) between the regulator and any capacitor following it.

[kalee20]

All true.

However, adding a capacitor across the output 'just to make sure' does at least stabilise the capacitance - if the thing is stable with this internal capacitance, then any external random capacitance should not destabilise it.

A bit of stopping on the follower's input is the better way to stabilise it.

However, the current thread seems to indicate more an oscillation in the Valve Under Test, rather than the supplies powering it (or possibly one provoking the other).

[PJL]

I think we need to understand a bit more about your test conditions. Looking at the regulator, it only has a very small reservoir capacitor so it would take no time at all for an 807 to empty that. And comparing with an AVO is also unwise as this is an AC tester and the source impedance is probably inductive.

How are you supplying the grid voltage? What voltage are you using? Can you try running it up with enough -ve volts to cut the 807 off.

Have you checked the PSU into a resistive load?

Are you confident the 807 is in working condition?

[ms660]

Out of interest would it be possible to post or link to a schematic of the valve tester?

Lawrence.

[G8HQP Dave]

Originally Posted by kalee20

However, adding a capacitor across the output 'just to make sure' does at least stabilise the capacitance - if the thing is stable with this internal capacitance, then any external random capacitance should not destabilise it.

True, but if someone is unlucky then some external inductance might just remove enough capacitance to tip it over the edge. The internal lags which make a follower potentially unstable are most effective at fairly high frequencies.

I suspect the situation here may be marginally stable PSUs connected to a marginally stable valve. The whole thing needs 'damping down'.

[Diabolical Artificer]

Thanks, some good points there. What do you mean by "build out resistor" David? Is this a resistor in series or parallel with the output and what value would you recommend?

PJL, the first schematic shown, IE anode supply has a bridge rectifier and 220u cap on the PCB before the schematic shown. I'm confidant that the 807 is functional. Yesterday I subbed the 1625 for an EL84; same problem. I put two meter's in series, one measuring anode DC I , the other set to AC I. Result if anything the AC current measured was a bit higher. I also scoped the PSU in question, it had about 20mA of ripple on the OP. The other simpler g2 PSU had no visible ripple, not to say there isn't any, but if it is, its not significant. My scope goes down to .1 v/div.

Re bias supply, see below and yes I can cut the valve off. At cut off all is well. As soon as I adjust bias more +, the problem occurs.

Sorry Lawrence havn't got a schematic, but its very simple. We have the two PSU's as mentioned, one LM317 based PSU with two TIP3055's to boost current for the heater supply, variable 0-30v capable of 3A @ 5v. The bias is another LM317 based PSU, variable 0-30v @ about 20mA. I'll try and draw a schematic later.

. I think from what you've all said, and what I've observed, that the anode PSU isn't up to snuff. I suspect the layout of the tester isn't helping things as it uses a patch panel. Unable to test with resistive load as I havn't got any R's that big of the right value. The biggest I have is an 8 ohm 200w dummy load, which is a bit low for this job. I've filmed all this so will post a link later so you can get a better idea.

Andy.

[ms660]

Out of interest, with your tester what would be the method of determining whether the valves emission is in the ball park?

Lawrence.

[PJL]

Building a dummy load for the HT PSU sounds like a good idea. Do you trust it enough to use an AC/DC radio? If not then lots of 2W resistors or anything else you can find.

[Diabolical Artificer]

Re used the old supply that was in it, IE the second schematic/circuit as used for the g2 supply. Tested the EL84 with no issues. So it looks like Art of Electonics supply is suspect. I'll test the big 1625 tomorrow with the tester as is and report back.

Lawrence, for a quick test I just look at the datasheet of the valve in question, find a bias line that cuts an anode current line and test at that. So for an EL84 for example, with the screen at 250v, bias at -4v and anode at 100v, the valve should pass about 70mA anode current. depending on how thorough and how many valves I have to test I also test at various a/g2 voltages and bias or alter the bias by 1v and do the sum for gm. For Htr/K shorts I use the DMM but at some point am going to build a proper test circuit. At the time I built I wanted something simple but something I can build other tests on top.

There's a simple DC load build on the EEV blog etc. I got some big heatsinks some time ago out of a welder with building a reasonably hefty PSU load in mind, after getting some advice here.

Andy.

[ms660]

Cheers for that.

Lawrence.

[PJL]

It would be interesting to find out if the PSU is still unstable when seeing a resistive load. A pentode is a constant current source but it's response to transients will be the more serious problem, particularly if your wiring is a bit all over the place. The Mullard valve tester uses a (valve) regulated HT supply but it does always include a small series resistor.

[Diabolical Artificer]

Ok, with 130k as the the load, the AOE circuit PSU was stable at 300v @ 2.6mA but had about 20 hz 200mA P-P ripple on the OP. It's hard to tell its exact f as it's low frequency so flashes on and off, but am sure that's about right.

I tested the 1625 with the two simple one mosfet PSU's and it was stable but had some low frequency low voltage ripple on the OP.

I'll add a few resistors to the suspect PSU, try taking the 2.2u cap off the output etc and report back.

Andy.

[G8HQP Dave]

What do you mean by "build out resistor" David? Is this a resistor in series or parallel with the output and what value would you recommend?

Resistor in series, possibly between the regulator and the capacitor. The value is typically a few hundred ohms, but it must be small enough not to disrupt circuit operation.

Under certain circumstances a tetrode screen grid can exhibit negative resistance (or even negative current). A true pentode is much less likely to do this. Can your g2 PSU sink current as well as source it, if necessary? A fat resistor to ground may be sufficient to allow for this, but it will get hot.

[G6Tanuki]

Originally Posted by G8HQP Dave

Under certain circumstances a tetrode screen grid can exhibit negative resistance (or even negative current). A true pentode is much less likely to do this. Can your g2 PSU sink current as well as source it, if necessary? A fat resistor to ground may be sufficient to allow for this, but it will get hot.

This is something which often gets overlooked - indeed I would argue that a "linear" amplifier using the Eimac 4x-series tetrodes where the screen-supply cannot both source and sink current doesn't deserve the title 'linear'.

See page 10 of this excellent treatise: http://www.cpii.com/docs/related/22/C&F2Web.pdf

[Diabolical Artificer]

I'm back to working on this and am still puzzled. I could save a lot of time and hassle and just put the other simple PSU back in, but i would like to try and get the Art of Electronics supply working.

We've established that it works ok with a resistive load so why is it unstable with a valve? I've tried a resistor in series, no improvement, so will try as Dave suggests one in parallel with the output which will mean some loss of OP but as it's only on for short periods this won't matter.

I presume it's the regulator that is unstable, but not sure how to test the circuit to establish this. I've read the datasheet for the LT3080 and one thing it suggests is to bypass the the set R with a 10-20pf cap. Looking at the load transient response graph, I see it mentions using a 10u ceramic cap, which I havn't got and sounds V expensive. the other thing it mentions is the use of low ESR output caps. It also says " a minimum OP cap of 2.2u of 0.5 ohm ESR is recommended. This being a HT PSU tantulum of other exotic caps are out, however I thought of trying a cap from OP to the 1k R. I suppose I could try a few bypass caps between pins on the regulator as per normal practice with 3 pin regulators.

I'll try a few things and report back.Thanks for the link Tanuki, looks interesting reading.

Andy.

[Diabolical Artificer]

I've just tried a few things without any improvement. First I put a 47p cap from the set pin to ground as well as a 1N1007 from OP to in - anode OP/K in. I then tried a big 10u/450v poly, I presume cap on the OP, again with no improvement and finally a 4k3 R in parallel with the 10u cap with the same result.

I'm getting a low frequency almost triangle wave on the OP at onset of the problem of about 200mv P-P, this rises in amplitude as I increase the voltage of g2.

I also scoped the gate of the mosfet, Q1 and am getting low F oscillation there. I have a ferrite bead on there as suggested.

So we seem to have a low frequency oscillation at first, then a higher frequency on top of this as g2 voltage increases and the condition gets worse. Can't really afford to spend that much more time on this circuit, but will try a few more ideas before I give up on it.

Andy.

[G8HQP Dave]

Solutions for LF oscillation may be quite different from solutions for HF oscillation. In both cases, however, you need to think of two possibilities: loop instability, and negative resistance (which may be caused by phase shifts in a servo loop). Can you measure currents and voltages to see where, if any, there is negative resistance? For example, add a low resistance to the output and look at the voltage on both sides; which side (PSU or g2) appears to be the source of the oscillation?

[Diabolical Artificer]

Re loop instability, I presume that is where positive FB is making the circuit oscillate? I'm just starting to try and learn about this area of electronics, it's still an area that I'm not too familiar with.

I'm also struggling with the concept of negative resistance, but suppose it's analogous to a current sink as opposed to a current source.

I'll do as you suggest Dave, and report back - thanks for the advice.

Andy.

[G8HQP Dave]

You will struggle to tame a regulated supply unless you know about loop stability.

Negative resistance is just like positive resistance, except that it is opposite: it generates power, instead of dissipating it; voltage and current are antiphase instead of in phase. In real life we can't have genuine negative resistance, but we can have negative slope resistance - sometimes over a limited frequency range. PSU current foldback is a form of negative resistance, as is screen grid behaviour under circumstances which maximise secondary electrons.

A current sink is something quite different, and will usually have positive resistance.

[kalee20]

Agree Dave - loop stability knowledge is pretty important.

However, as a brute-forse approach, try a big electrolytic across the output, possibly in series with 100Ω or so. The resistor puts 100Ω of damping across the output; the capacitor stops this being a steady DC drain.

A 'scope is almost essential - great that you have one!