Sequence in which grounds are connected

[Saurav]

Hi,

Questions about mystical grounding schemes. I'm sure this will make some people happy.

Let's assume I'm not going for a full star ground where everything connects to a single point. I'm trying to understand the correct sequence in which ground points should connect to each other, with the goal being to ensure that ground currents don't flow where they shouldn't flow. Let's assume a simple SE amp with a single tube driver stage.

* Starting at the power supply, the first filter cap becomes the main ground point. Safety earth connects here, chassis connects here, PS transformer secondary CT connects here.

* Any subsequent PS filter cap grounds connect back to this main cap. Once the PS filter stages split off for the two channels, they connect back separately. The same holds for the cap for the output tube B+, that attaches back to the cap just before it.

* What about the filter cap for the driver tube B+, should that attach to the output tube B+ cap ground? Seems like it should, since that's the way the currents are flowing.

* Do the output tube's cathode bias resistor and bypass cap attach to the output tube B+ ground? Or to the driver tube's B+ ground? I'd think the it should go to its own B+ cap, because that's the return path for the AC signal, right?

* The output tube's grid leak attaches to the cathode bias resistor/cap.

* Starting at the other end, the input RCA ground connects to the grid leak for the driver tube, which connects to the cathode bias resistor and bypass cap.

* Where does it go from here? Does the driver tube's ground attach to the driver B+ cap ground, or to the output tube's grid leak resistor? Is it more important for the driver tube to have a short loop with the output tube, or with its power supply cap?

Or, in general, is it more important to group all PS grounds together and all audio grounds together, or to group PS and audio grounds for each stage together?

Thanks,
Saurav

[fdegrove]

Hi,

Or, in general, is it more important to group all PS grounds together and all audio grounds together, or to group PS and audio grounds for each stage together?

On a general note I'd say group PSU grounds to one single star earthing point, ditto for the audio part.

Just as important is to find to "quiet" spot for both points and this is best determined by using a scope.

If pinpoint imaging is important you, grounding schemes are well worth investigating.

Busbar connection may give fine results too and your reasoning seems sound to me yet I always had the best results with the star scheme, YMMV.

Cheers,

[Saurav]

If pinpoint imaging is important you, grounding schemes are well worth investigating.

Well, it's not at the top of my priority list, but it's not something I'd mind having

I find my wiring becomes really messy whenever I try a star earth for every ground in the component, which is why I try to build a "ladder" grounding scheme, where small signal stages are further away (and therefore at a higher potential) from the lowest ground point, than large signal stages.

On a general note I'd say group PSU grounds to one single star earthing point, ditto for the audio part.

Which means that the driver tube grounds would be closer to the output tube grounds, but further away from the PS grounds for the driver. I think that's what I did when I built my amp.

Just as important is to find to "quiet" spot for both points and this is best determined by using a scope.

Could you please elaborate? What should I be looking for?

On a side note, my scope is uncalibrated, so while I used it to set the position of my hum nulling pots, I can't use it to get an idea of the actual values of the signals. My cheap RS DMM says 2mV AC at the speaker outputs. The hum I hear is 120Hz. How reliable would a cheap DMM be at measuring 120Hz near-sine-waves?

[EC8010]

If you have to connect the 0V line to chassis, do it as close as possible to the input valve. Mains earth and the mains transformer's E/S screen connect to the chassis as close as possible to the incoming mains, and is an entirely separate issue.

Otherwise, annotate your circuit diagram with the returning HT currents, and you will see that your 0V strategy should be "ground follows signal."

[Saurav]

If you have to connect the 0V line to chassis, do it as close as possible to the input valve.

I don't usually do that, and I've wondered why it should be near the input. Also, "as close as possible" - does that mean physically or electrically or both? In other words, if I pick a physical location that's near the input valve, do I then run the ground back from the PS filter cap to this point, or do I connect the input RCAs and grid leaks to this point? I usually have my 0V flow back to the first filter cap, which then connects to the chassis and the safety earth.

Mains earth and the mains transformer's E/S screen connect to the chassis as close as possible to the incoming mains, and is an entirely separate issue.

So are you saying that the signal wiring connects to the chassis near the front, then the chassis connects to safety earth near the back? In that case, the signal grounds shouldn't be directly connected to safety earth, or I'll have a ground loop through the chassis, right?

Otherwise, annotate your circuit diagram with the returning HT currents, and you will see that your 0V strategy should be "ground follows signal."

I think that's pretty much what I've been trying to do. I wasn't sure whether the driver tube's ground should connect to the output tube, or to the driver stage's PS cap. It seems like that tube is logically connected to both points, depending on whether you're looking at PS current flows or signal current flows. But it can't be a good idea to connect one ground to two other grounds, because that's a ground loop right there. If I keep audio and PS currents separate, that answers this question, and the driver tube connects to the output tube.

[EC8010]

Any amplifier amplifies the difference between its two input terminals. We tend not to think of the ground terminal as an input, but it is. In order to screen an amplifier, we surround it with a conductive casing/chassis and connect it to the amplifier's ground terminal. Inevitably, there is capacitance from the chassis to mains. Similarly, there is capacitance from ground signal wiring to mains. Both return currents to the point where they are connected together. The further we are from the bond, the larger the voltage developed across the (non-zero resistance). To minimise the amplified voltage drop caused by these currents (which will be amplified and cause hum), we must bond chassis and signal ground at the input valve, and must do so with the lowest resistance we possibly can.

So, to answer your question, yes, it means both physically and electrically close.

More significantly, we must only make one bond between signal ground and chassis (otherwise we make an earth loop). Following the earlier logic, this bond is made at the input of the most sensitive stage.

Yes, this strategy usually means that mains earth connects to the chassis near the back, and signal ground might connect to chassis (if at all) near the front.

I don't see that power supply and signal currents are separable. When you change the voltage between grid and cathode, you change anode/cathode current. How do you split audio from power?

[Saurav]

I don't see that power supply and signal currents are separable. When you change the voltage between grid and cathode, you change anode/cathode current. How do you split audio from power?

What I was trying to say was, I can see a loop from the driver tube through its B+ cap, and I can see another loop through the output tube. I guess the second loop is the one that's more important, so that's the one which should be minimized. So the driver tube's ground connects to the output tube's ground, which then connects to the PS cap grounds.

Thanks for the explanation about the chassis.

How does the same logic extend to separate audio and PS chassis connected through an umbilical (that's what my phono stage looks like)? Some PS caps are back in the PS chassis, some are in the audio chassis.

Following the same logic, the audio chassis should connect to signal ground close to the input stage. Then... I guess the umbilical shouldn't connect to the chassis. Run the signal/PS grounds through the umbilical back to the PS caps in the other chassis, and that chassis connects to safety earth. This can get a little complicated

[EC8010]

Originally posted by Saurav
What I was trying to say was, I can see a loop from the driver tube through its B+ cap, and I can see another loop through the output tube.

Yes, and that was why I wasn't happy about attempting to split grounds. If you do true star earthing or ground follows signal, this doesn't happen.

If you have two chassis powered from a common supply via an umbilical, then the chassis connection and ground must be kept separate until the bond. Bond the two chassis together as firmly as you can. The HT capacitors for the phono stage belong in the phone stage - as close as possible to their associated circuitry.

[Saurav]

Yes, and that was why I wasn't happy about attempting to split grounds. If you do true star earthing or ground follows signal, this doesn't happen.

Understood.

The HT capacitors for the phono stage belong in the phone stage - as close as possible to their associated circuitry.

I think I have the rectifiers and one capacitor per channel in the PS chassis, and the RC filter sections in the audio chassis.

[fdegrove]

Hi,

I think I have the rectifiers and one capacitor per channel in the PS chassis, and the RC filter sections in the audio chassis.

Which seems to imply that you carry quite a bit of AC current on the umbilical.
I'd advise against that even when current demands of a phonostage are relatively low they'd still interact with one another creating EMI and possibly crosstalk aswell.

Cheers,