Help with grounding in a power amplifier

[Elerion]

Hello.
I'm building my first stereo amplifier, and I would like to share with you the ground/earth scheme I designed, for anyone to point errors in it, and hopefully make a better approach. It is based on local star ground scheme.

The amplifier is splitted in three boards:
- Power supply
- L-Channel
- R-Channel

The original schematic shows four filtering capacitors, all of them shared between channels.


I decided to split each of the pream and output stage capacitors (C1 and C3, respectively), in two (one for each channel), and place them in their respective boards, near the valves they supply filtered voltage to. This way, I can use a local star ground scheme, very easily, with short wires. The idea is to wire all vale circuitry ground (cathode resistor and bypass cap) to the ground terminal of their filter capacitor.

An image is worth a thousand words:


Then, there's a main star ground point, where all local star grounds go, and which connects to the chassis.
As far as I know:
- It is better to have a single chassis connection, at a single point.
- It is better to not do it at the first filter capacitor, but better to do at the input.

But this is a stereo, so there're two amplifiers, powered by the same transformer/supply. If connecting both inputs to the chassis, could this make a ground loop? I don't think so, but I'm unsure.

So just to be sure, I first decided that a single connection to earthed chassis could be done at the common filtering capacitor ground which is between phase splitter and output stage.

I read lots of info about the matter, but being my first unguided project, first stereo, and chassis building from scratch (thus facing quite several challenges), would like to avoid basic (and very time consuming) mistakes.

Thank you so much!

(of course, there's a chassis connection at the IEC socket, to the third prong wall plug)

 

[audioguru]

You are making an amplifier with obsolete valves? With performance robbing output transformers that are not made anymore? Why?
I replaced my valves amplifier with a solid state amplifier about 53 years ago and I still use the solid state one today.
There are inexpensive excellent power amplifier ICs available today that almost everybody use.

 

[Elerion]

The truth? Ok, this one is for someone who likes valves.
I plan to build an stero power amplifier later, solid state, for myself.
By the way I just built a guitar valve amplifier, and I liked the experience, and I like to learn and experiment (I never saw a valve amplifier before in my life).

But forget about why valve. In solid state amplifiers there's the same problem with grounding.
These projects are for me to learn about power supply, grounding and high voltage handling, I never faced before (because of ten years of digital low-power projects).

 

[KeepItSimpleStupid]

Elerion:

The same general principles should follow of keeping the high current grounds together and the references together and tie it together at one point.

Where the trouble comes in, is when you have the "third prong".

I might be inclined to not use the chassis, except for Earth, but I also see advantages of tying everything together at the ground of the power cord. Usually you can;t tie EVERYTHING there, so you break it up. The grounds that are used as references, you can tie together and the grounds you use as returns you can tie together anywhere and tied to one location.

What also works is to use isolated RCA jacks, but tie the ground to chassis through small resistors like 2.2. Ohms.

Once you start introducing lots of stuff with a third prong, it can ensure trouble.

I did a solid state amp essentially that way, except it used a 2 prong plug. By Star ground, it really was a star. All the returns, literally went to one screw with the input through a 2.2 ohm resistor on the "ground" side.

I was very happy with the result.

The star ground is literally in the upper right hand corner of this picture (on the inside rear panel of the amp).

 

[Elerion]

I might be inclined to not use the chassis, except for Earth

Do you mean not earthing the internal electronics ground?
As far as I know, that is unsafe, because a fault on the secondary's side of the power transformer can suppose a hazard, or a fault between primary and secondary windings.

I understand about grouping the grounds, but I'm having trouble to fully understand the details.
I've seen many valve amplifiers that place the filter capacitor that filters a certain stage supply voltage, next to the valve, not in the power supply board.
But many other people insist in that they all must be placed in the PSU.

From aikenamps (https://www.aikenamps.com/index.php/grounding)

"...the individual local grounds should be connected to the cap ground lug for the cap that supplies their B+ (voltage supply) line, not back to the main star point. Only the cap grounds should fan back to the main star for a single-star system. This will keep the cap charging currents from modulating the stage supply relative to it's local ground."

"filter caps should be located at the stages where they are used. For example, the first preamp's filter cap should be physically located next to the tube's components. All other filter caps are located near the circuits they are decoupling. It is not a good idea to lump all the filter caps together in one spot (...) it is far better to locally decouple each stage with a cap at that location. "

As I understand, the ripple current should be better kept in the PSU, not in the local grounds of the amplifier devices. But in the other hand, all guitar amplifiers I've seen place the filter capacitors right next to the stage circuitry, to "keep the cap charging currents from modulating the stage supply relative to it's local ground."

I'm confused getting the basic cincepts clear. Thus, seeking for some help.

 

[KeepItSimpleStupid]

Do you mean not earthing the internal electronics ground?
As far as I know, that is unsafe, because a fault on the secondary's side of the power transformer can suppose a hazard, or a fault between primary and secondary windings.

Primary to case or an open primary are more common faults if the transformer faults at all.

What I'm saying is that chassis ground (protective ground) should not be a hard ground with respect to the signal. When this is plugged into a tuner with a ground, there's the possibility of a loop. The amp chassis is first and foremost protective ground.

If you look at this http://leachlegacy.ece.gatech.edu/lowtim/graphics/ckt.pdf solid state amp. The one mine is based on. C6, C7 and C8 are the most important. They alter the feedback loop and change the frequency response. The value of C6, C7 and C8 needs to be stable.

R51 is that (2.2 ohm resistor, I think) that doesn't make the input have a "hard ground" with respect to other components. Any DC loops can drop across that resistor. The AMP will reject DC, since it's frequency response starts at 0.5 Hz.

C1 is an RF filter. C25/R50 would make sense to return to a high current ground point, The "Power supply" capacitors, are not shown. In my amp, they are 9600 uf per 50 V rail and there are 4 rails. This http://www.hammondmfg.com/pdf/5c007.pdf is a worthwhile read. Tubes do better with an LC filter.

C13, C15, C21, C23 and the corresponding ones on the other rail, locally stabilize the power supply, R32 and R33 nd nearby capacitance "de-couple" the high current fluctuations from the pre-amp side of things.

This http://en.wikipedia.org/wiki/Decoupling_capacitor article and it's companion articles make sense to read.

EVERY component has "parasitic" values. Real life parameters that make the components non-ideal. A wire has "inductance". it's a real small number, but it has inductance.
Because of these numbers, it's a rare event that three 1K resistors in series cannot be replaced by a 3K resistor

So, in this particular amp, the negative speaker lead is the high current reference. You MIGHT like that reference to actually be at the speaker, but that isn't going to happen.

 

[Elerion]

KeepItSimpleStupid, thanks, very informative. Though, that amp is much more complex than any valve amplifier I've seen. The one I'm talking about is this "simple":
http://diyaudioprojects.com/Schematics/images/Dynaco-ST35-Tube-Amp-Schematic.png

Trying to understand and think by myself, I've drawn the current flow in the two grounding schemes I previously talked about. I really appreciate if you guys can point me if I say something which isn't correct.

The one with the capacitor located near the valves seems better to me, because the current loop is smaller and local, so should not affect any other part of the circuit.


The one with all capacitors in the same place makes current returning from the preamp flow into the ground portion where the output valves are attached (I'm in doubt about the main current flow at the interrogation mark, the one coming back from the preamp valve; is it going to the capacitor which supplies current, or to the transformer?)


Both of them using a local star ground scheme, connecting to the chassis at a single point, and optionally via a ground loop breaker circuit (as proposed by Rod Elliot, **broken link removed**)

Which one would you guys choose?
Without any more information, and maybe wrong, I'd choose the first one (though I know many people use the one below, and others even don't care much, and do whatever seems easier, without any caution).

 

[KeepItSimpleStupid]

The "Grounds" you have to care about is the com of the output transformer, the input ground, The 100 uF||95 ohms and the center tap of the B+ supply. the rest, you really don;t care about.

The input is AC coupled, but you still have a ground issue. I'd only use the chassis for protective ground. The chassis will act as a shield too. i.e. Don;t connect the input common to the chassis. If you do, only connect it via a low value resistor.

The other option is to connect the point in the first sentence to a single chassis point except the input. Connect that with a low value resistor;

 

[Elerion]

Thanks for your advices.
After lots of research, book reading, looking at many projects,... I'd preffer to ground the internal circuitry, if possible. Preferebly right at the input(s).

Though, instead of a simple resistor, I think it is better to use the loop breaker circuit (just a cheap 35A bridge rectifier and capacitor, adds safety in case of fault currents), or a variant of this, a 120 ohm resistor shunt by a 100nF x2 capacitor, recommended by Bruce Heran.

 

[Elerion]

I finally placed all the filter capacitors together in the power supply board.

I connected the preamp local ground, using a wire, right to the last filter capacitor's negative terminal, and the output stage local ground to the capacitor that powers the stage (two capacitor upstream towards the power transformer).

Then connected signal ground to chassis at the inputs. No other connection to chassis, except for the security earth ground, both close.

Dead quiet. No hum, no noise. Perfect.

 

[KeepItSimpleStupid]

Good. So, your happy. Thanks for the feedback.

 

[Elerion]

Well, just wanted to share, for anyone interested on this topic, to have another experience.
There're more sophiticated approaches, but mine worked great for me.
Anyone feel free to ask for details. I'll answer (if I can) with pleasure.