Preamp input selection and ground

[ToneArm]

Hi folks,

My diy preamp does input selection with a 4 pole switch selecting signal and ground for each source. That approach was chosen to reduce the chance of ground loop hum and has done a good job at that for a long while.

However.. Now I have connected a CD player which is not earthed (class ii). When the cd player is selected theres no problem at all. But if phono stage is selected there's hum. Fairly loud hum, but i am using a low output MC into a 2:15k step-up transformer, quite a lot of gain there to sniff out stray fields. With no transformers and a MM cartridge it still hums, much quieter and perhaps may have gone unnoticed, until the day the wife is out and the volume goes up!

It is easily fixed by commoning all the grounds and just switching signal, but is that a retrograde step? I wonder if there are other methods to try. A better CD player might even be a preferable option..

Thanks
Glyn

[mhennessy]

Of course, you don't have an earth loop - it's crosstalk from the ground of the CD player which is probably (confirm with a 'scope) bouncing up and down at mains frequencies. Grounding it will fix it.

I've never bothered switching grounds - it's more trouble than it's worth, especially as it's hard to predict what sort of currents your switch might be expected to deal with.

This might sound glib, but earth loops are pretty much inevitable in any complicated system - they will happen! Far better to design your equipment to be immune from them, using differential input stages. But a workable alternative is to break the loops.

If you ground your inputs as they arrive at the pre-amp, you have the potential for currents to flow in the grounds if the output ground of the preceding equipment is earthed, and these are turned into voltages across any resistance encountered. Usually, this voltage ends up being in series with the wanted signal voltage.

For that reason, it's common to use an "earth breaker" circuit. It's very simple - you just need to tie your input earth to mains earth via a low value resistor. Anything between 10 and 100 ohms will do the job.

But to prevent this resistor from being overloaded, you also need to add a couple of diodes in parallel with it, pointing in opposite ways to each other. These limit the maximum voltage seen by the resistor to plus/minus 0.6V nominally. Unless something is very wrong, the differences in earth potential will be very much smaller than that. Usually when the voltages are large - as is the case for your class 2 CD player - the currents involved are very small. The resistor can be a standard 1/4W film type.

The diodes can be standard 1N400x types or whatever you have to hand. They won't ever pass large currents for longer than the 30mA it takes to trip an RCD (for which, these diodes will be fine, but there's no harm using larger ones if you're worried). They won't ever need to pass the 25A from a PAT tester (unless the PAT tester operative is ignorant) because the outside screen of the phono sockets are "functional earths" rather than safety earths. Provide a suitably labelled earth point for anyone who might wish to PAT test - which you probably already have if there's a phono pre-amp in there.

Also, it's worth bypassing the whole lot with a 100n or similar - just a ceramic will be fine. Again, it only sees the diode drops. This ensures a low impedance connection to the chassis at RF frequencies. You can sprinkle these around with impunity - some manufacturers use them between every pair of input/output sockets, and at every mounting pad if using metal hardware.

In most cases, allowing your signal earth to be loosely tied to earth via an earth break circuit will do the job. Providing the following power amp uses the same strategy, it will work - I've done it countless times. But if it is necessary for your signal earth to be tied firmly to mains earth, a differential input stage helps. The "cold" side is connected firmly to the ground of the source equipment, so the input stage is definitely responding to the difference between the signal and ground connections of the source. You still need the earth break at the input sockets, as it can be hard to predict where the I-to-V conversion will take place - better to kill the current before it can become a voltage!

Of course, there's a lot of work to do to avoid earth loops inside your equipment, and diff-amps can be very useful there. But there's lots of guidance in the books from Self and others.

A good reference is here: https://www.hypex.nl/media/fa/d8/a3/...20G%20word.pdf

There's a lot of other good stuff on the Hypex website: https://www.hypex.nl/support/downloads/

Cheers,

Mark

[ToneArm]

Hi Mark, thanks for the help and taking the time to point me in the right direction. I had done what I could to avoid ground issues and thougt maybe I'd mastered some dark art, as I didn't have any. Humble pie, I will do some more reading.

And experimentation - for now signal grounds are joined together at the input sockets and all seems to be ok. It doesn't have an earth breaker at the moment but i will add one for good measure.

Thanks
Glyn

[G6ONEDave]

This might be nothing to do with the problem but I trust that the turntable is not sitting on top of the CD player. My reasoning is that the tonearm cartridge could be getting the mains noise via induction from the CD power supply/mains transformer or anything else for that matter, if it's been recently relocated. Just a thought.

Dave

[Jez1234]

Yep that's a common issue and the MC step up transformer (SUT) can be very sensitive to this.

[ToneArm]

Hi Dave, no the CD player was a good distance away from the turntable. I dont usually use a CD player, it was actually my son's, abandoned here 12 years ago when he left home. It needed a new belt in the drawer mech but otherwise is like new so I thought I'd get some use out of it.

Cheers
Glyn

[knobtwiddler]

I think Mark is likely correct, although - as Self shows us (easy to prove correct with an FFT probe) - crosstalk is typically around -70dBV in switches with adjacent contacts, so unless the gain structure is oddly calibrated (i.e. you're getting a lot of gain after the switch, and things aren't as hot as they could be prior to it, for whatever reason), then it might make sense to draw out a block diagram of the setup and post it here.

One thing to be aware of: the negative terminal (or 'signal ground') for your cart / SUT should be as far away from any power ground as possible. High level line signals won't give problems if they meet at points where noisier ground currents are present, but negative phono output definitely will. Make sure that the -ve signal for the cart is insulated from the chassis and grounds itself through the RIAA preamp only, which gets its ground from its line out, i.e. as far away from the I/P as possible. Cart + SUT signal 0V needs to be isolated from overall 0V (aside from chassis ground used to shield them) until line-level output. Where you get your 'ground' in an RIAA preamp is important.

[mhennessy]

But do bear in mind his class 2 CD player could have significant leakage via filter components in the PSU, meaning the amplitude of the interfering signal could be massive. Depends on the topology of the CD player, of course. But we're likely talking 100-200V here. From a high source impedance, but it's there...

And I didn't explicitly state that it was crosstalk via the switch. It could easily be to other nodes. In this case, it sounds like the phono pre-amp, which of course has 30-40dB of gain for MM, more (via the transformer) for MC.

So for a source that might start out at around +50dBu, and a pre-amp that has a gain of 40dB, it doesn't take much imagination to see the problem...

Grounding the CD player, even weakly via an earth breaker to power ground, should solve it in this case. "Should". That's the joy of earthing - there's always a new scenario to discover

The "offness" of switches depends on surrounding impedances and frequency (as it's capacitive coupling). The figures Self quotes are at 10kHz, obviously the isolation rises at 6dB/oct below that. Taking 70dB as the starting point, at 50Hz, it's more like 110dB, roughly. That does suggest that some other coupling route might also be contributing...

Even at line level it definitely pays to keep signal grounds and power grounds separate - and if they have to join, do it at one point only, and consider how currents might flow and ensure they can't find their way onto the signal ground. It's easy if you consider it right at the start of the design process - not always easy to do after the fact. Mistakes here might not lead to obvious hum, but they will result in higher distortion levels than is necessary (though not necessarily audible, of course). All this gets worse when you include digital electronics! Another trap is single-rail designs, where you can easily end up with class B currents in the ground (from the output stages of op-amps).

All good fun!

[PJL]

Surely you need separate earth breakers on each input for it to eliminate earth loops? if it cannot be an earth loop would it be better to run a highish value resistor from each incoming screen to outgoing earth?

[mhennessy]

Yes, if you have multiple sources that are earthed, and are also willing to switch your signal earth to each input ground (as the OP has done) as well as the signal, then that's a good solution.

I haven't found it to be necessary though. Perhaps I've been lucky with the gear I've used domestically, which has been mostly class 2.

But just because a source is class 1, it doesn't automatically mean that the signal earth is connected to mains earth. If the sources use some form of ground breaker too, then no problem