problem with house earth/grounds - oscillation!

[Alex R]

I am in the UK.

Here is a strange thing I really don't understand. When certain pieces of old equipment (in this case a Beam-Echo 50s jukebox amp, and an early 60s Selmer Vanguard combo) are connected to the mains they oscillate at something like 150Hz. When I lift the earth (ground) the oscillation stops. On the juke amp the oscillation also stops when a strong signal is fed through, turn the sig gen input down towards mag cartridge levels and it starts up again.

My bench is in my house. The place was rewired to high standards less than a year ago, as we may let the place at some point and we wanted the latest certification. I use an isolation transformer, and I can connect or disconnect chassis earth on a piece of equipment on the bench. The oscillation happens on the bench with earth connected and on sockets throughout the house.

I first noticed it was something to do with earthing because if you pull the plug out of the mains the oscillation stops instantly, it doesn't wait till the equipment starts to cool and the caps discharge, it stops instantly.

I think it is oscillation because on the Selmer the tone controls affect its pitch. Pulling out the first stage valve stops the noise.

My scope is earthed and when I clip the earth lead to the chassis the oscillation stops or is much reduced.

Any ideas at all? It's a difficulty because I can't satisfactorily test either piece of equipment. Meanwhile other jobs go through fine.

[Jazz P Bass]

Schematics.



BEAM ECHO AMP0001.pdf

beam echo amp A3 FINAL.pdf

[pdf64]

Just a wild stab in the dark, but how about if the isolation transformer is out of the frame, ie the oscillating equipment connected directly to the mains?

[nickb]

What is connected to the input when this happens? Does it stop if you disconnect the input source.

[Alex R]

Thanks guys, a little more detail then and maybe you can see whatever it is I am missing.

Started with the Beam-Echo. Being a juke amp it needed every capacitor changed and some resistors that drifted too high. Put a deck on it and it amplifies ok to about 4 watts then distorts and is generally low on gain. Some stage clipping early I guess. I've a schem with signal and supply voltages but it's not quite the same and the amp came in with no bass and low output so there's a glitch somewhere to trace. The input signal voltage on the schem is 10 mV but as I dial down the sig gen below 100mV or so the oscillation cuts in. It doesn't oscillate with the (un-earthed) deck so I suspected a hum loop as my old sig gen is earthed. Tried a battery sig gen but it still oscillates. At about 150Hz, a loud buzz but pitch varies if you move wires about and if you ground the first stage grid (or most grids) with or without a cap it stops.

I should add the oscillation isn't a background noise, it drives the amp up to full output with massive clipping, immediately. Behaves in every way like oscillation except I don't usually see such low frequencies. It's because the low frequency is shared with the Selmer that makes me wonder if there is some connection between the two issues. Earthing also is implicated in both.

OK so the Selmer. Recapped and so on, fired it up and it has the same oscillation - pitch varying a bit when you move leads or tap the board, stops when you ground a grid or pull out the first stage ECC83. Lift the earth on the amp (replaced by a cap and resistor to drain static) and it stops oscillating and works ok. At this point I thought, lets try without the isolation transformer and plugged into a couple of other sockets in the house - same oscillation. Plugged in a guitar and the distorted guitar can be heard through the oscillation.

There's a ghost in ma house! Or a dimwit tech

[nickb]

If moving leads varies it, then that smacks of ultrasonic oscillation rather than LF. Have you put a scope on it to see what it looks like? If it is then add screening / small caps/ grid stoppers as required to fix it.

The only way I can think that the safety earth connection could affect it is that it might change how current circulates in the chassis. Try moving that connection to the chassis to different spot just to see if it affects things.

[R.G.]

Is it possible to lift the speaker output from chassis ground and connect it directly to the first filter cap negative lead? I like the idea that chassis currents might be doing you in, and one place chassis currents and oscillation intersect is the speaker return. I'm still groping for why *exactly* lifting safety ground would change this, but while I pound on it, you might get lucky with a likely suspect.

[Alex R]

Thanks for the input guys, will try this tomorrow. Or Monday maybe.

Ye I've scoped it and can see no trace of HF just the clipped 150-ish Hz. The tone controls on the Selmer change the pitch. They're simple caps to ground plus pots.

Maybe both amps have oscillation problems with different causes. But that low buzz is so unusual.

Can't help feeling there's an elephant in the room standing on my toe whilst I look earnestly the other way. You know the feeling.

[Steve A.]

I assume that both amps do not exhibit these symptoms when plugged in at a different location, with and without your isolating transformer.

And you have checked continuity between the grounds on your AC outlets and your cold water pipes? I would also compare voltage readings between the hot legs and the outlet ground vs. the cold water pipe ground.

In the UK you do have two hot legs, each measuring ~120VAC to earth ground, right?

Steve

[pdf64]

...In the UK you do have two hot legs, each measuring ~120VAC to earth ground, right?

No, it's 240V (well 230V nominal) live, 0V neutral and earth, neutral being bonded to earth at a local distribution point.

[R.G.]

Can't help feeling there's an elephant in the room standing on my toe whilst I look earnestly the other way. You know the feeling.

There probably is, but I don't see it yet either.

Here are some things to check that might lead to the elephant's hiding place. Is the 150Hz line synched, or is it just close to 3xline frequency? If it was 50Hz and line synched, it would be AC leakage or pickup. If it was 100Hz, it would be full wave rectified ripple. I suppose it could be 100hz ripple riding on 50Hz line pickup, and overdriving the amp, but that's just weird. It being related to whether the power line ground is attached or not is very strange, as well. So it may well be useful to know if the oscillation is really, no-question synched to the AC line, or just confusingly similar.

If it's not AC line synched, you could go back to the first steps on oscillators. An oscillator can only exist with gain, feedback, and the proper phase. If either one drops below the conditions needed for oscillation, the oscillation stops. The criterion for oscillation to exist is that there must be
(1) feedback - a way for the device to get a signal back to its input
(2) gain, and enough gain to make up for the losses from the output back to the input, so that the amplitude of the signal fed back, at the input point, is enough to drive the device to create itself. This is referred to as a loop gain of one or more, as the amplification has to be as big as or bigger than the losses in getting the signal back to the input.
(3) the right (if you're trying make an oscillator) or wrong (if you're trying to make an amplifier) signal phase as presented to the input. If the signal presented back to the input reinforces what's already there, that's positive feedback, and oscillation is easy to get. If the signal opposes what's already there, and oscillation isn't possible until something mucks around with the phase of the returned signal.

Amplifiers using negative feedback run into problems where the high or low frequency phase shifts add up to 180 degrees in addition to the 180 degrees of shift implied by "inverting". There's a single time-constant phase shift asymptotic to 90 degrees with every series capacitor on the front of an amplifier stage to block DC. Any two stages with capacitor coupling can't quite get to 180 degrees, but any three stages with overall feedback and capacitor isolators can easily oscillate at the frequency where the combined input capacitors get the overall phase shift to 180. Since almost all tube amps are cap coupled, any feedback around three or more stages will oscillate, and they will oscillate at the low end of thee frequency range.

That's the reason for using a resistor/capacitor isolator on the power supplies. If you hook more than two stages together on a power decoupling cap, there is enough feedback through the power supply to start up motorboating in many circumstances. Sometimes you get away with it, but it IS getting away with it, not good design.

Hmmmm. Talking through this makes me wonder if one or more of your bypassing/decoupling caps may be open circuited somehow by attaching the line ground. That would let motorboating happen. Likewise, does anything about line ground affect how the speaker is "grounded"/returned?

[Alex R]

Thank you all, lots to think about but as there's no obvious link between the two problems I do suspect it's not one elephant but two, one per amp. I have to get some other jobs through so it will be a week or so now before I start giving time to these two again. I will post the results, however embarrassing!

[Alex R]

Well I got out a battery sig gen with no earth and now I can get the signal as low as I like in the juke amp and there's no 'oscillation', if that's what it was.

Haven't looked again at the Selmer but I've a feeling it is just plain oscillating! Same room different elephant. More later on that one.

[Enzo]

You mentioned new caps, so have all the filter caps been replaced yet? Thinking about a lack of decoupling between stages.

[Alex R]

Thanks Enzo - yes I'd just finished a cap job on it. The board sits less than an inch above and parallel with the controls, and a lot of wiring passes to and fro, lead dress may be an issue though moving them around did nothing last time. I need to check I fitted all the caps right!

I was working on a Beard P100 stereo 4xKT88 power amp today which also oscillated - the reason was a mono conversion ran the wires that connected the speaker outputs together right past the unscreened input wiring. Whenever I plugged in my sig gen the oscillation stopped - the genny output impedance is 600 ohms, so I guess that anchors things down rather. Anyhow twisted speaker leads placed away from the inputs, and a screened input lead, stopped the oscillation.

Funny how things come together - oscillation seems to be my problem of the month.