GND/Earth Wiring Management by Power Amplifiers - what is the royal Way

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
There are follow GND connections by discrete power amplifiers:

A: clean GND
1) enclosure/cabinet if it consists of metal
2) input resistor
3) RC highpass network of NFB loop
4) reference resistor of current source (LTP)
5) load resistor of predriver stage (second voltage amp)

B) dirty GND
1) Boucherot resp. Zobel network
2) load (loudspeaker)
3) neg. pole of power supply capacitor from pos. rail
4) pos. pole of power supply capacitor from neg. rail
5) center tap of transformer's secundary winding
6) main earth

If there are dual mono devices (i. e. separate and independend power supplies for both channels) the royal way for wiring is clearly to determine.

But what about normal power supply design's - i. e. one power supply and one cabinet for both channels?

In such cases I discover often very strange things like caps or resistors between the dirty and clean GND. Sometimes I observe serial resistors to some blocker capacitors.

The English brand "Moth" (Stan Curtis design), led the GND poles of the power supply capacitors about long separate cables to the neg. speaker terminals. Therefore the sound character was completly different to other models which uses exact the same power amp circuit topology.

Everyone of the individual commercial manufacturers (and diy'ers) here seems to prefer a different wiring.

Therefore the question for the royal way - thank you for your comments.
 
Last edited:
Everything you refer it is related with the effort of rejection of buzz, hum etc noise. If you could have a faraday case to include the amplifier modules, a seperate case to include the power x-former and a perfect power supply without ripple, then you would not need any of that.

Fotios
 
This URLs offer a great amount of imformations:
1) Circuit design (digital, analogue, switch-mode, communications), and choosing components
2) Cables and connectors
3) Filters and transient suppressors
4) Shielding
5) PCB layout (including transmission lines)
6) ESD, electromechanical devices, and power factor correction

Design Techniques for EMC – Part 1
Design Techniques for EMC
Design Techniques for EMC – Part 2
Design Techniques for EMC: cables and connectors
Design Techniques for EMC. Part 3
EMC Compliance Club – Design Techniques for EMC. Part 3 — Filtering and Suppressing Transients
Design Techniques for EMC – Part 4
Design Techniques for EMC - part 4
Design Techniques for EMC & Signal Integrity – Part 5
PCB DESIGN TECHNIQUES FOR LOWEST-COST EMC COMPLIANCE
Design Techniques for EMC – Part 6
Design for EMC Part 6: ESD, dips and dropouts, etc.

More URLs:
Banana Skins
EMC Testing Part 3 – Fast Transient Burst, Surge, Electrostatic Discharge
 
Good post :) Although I would label the GNDs differently though, as it's not necessarily dirty ground.

I'd call it "Signal Ground" and "Power Ground". After all, the amp output /speaker has to connect to a clean star point of this power ground, as does the Zobel. The capacitor leads themselves I would consider dirty. Due to the star topology however doesn't/shouldn't affect other returns, especially not the signal ground returns.
 
Excuse me, i have few question.
In the input stage, the clean GND (signal GND) is connected with RC filter input, input resistor, caps or resistors in the feedback line. But resistor of CCS and diode zener or resistor of cascode circuit must be connect in where to right way? Signal GND or Power GND?
 
They lead to their respective power supplies. If you use two supplies (one for the input/driver, one for the output) it goes to the Driver GND, if you use one supply they go to the Power GND.

No matter the GND group, each group has its own local star, all the stars come together in the main star between / after the buffer elcos. Though the Power GND leads that do not come directly from the board (speaker GND) go directly to the star at the buffer elcos.
 
It's your choice basically, if you want to star it with the signal ground, you want the cascode reference node current not to be dirty, it generally wouldn't be enough to worry about I think. For that reason, the actual Power GND may be better since the cascode's job is to relieve another transistor from voltage variations on its drain, and a minor absolute variation on its gate/base does not affect the current going through it. Personally I would pick the respective power GNDs for the cascodes, but try to route them to the signal GND and power GND meet-up on the board. If you use a GND separating resistor (10Ohm) as a meet-up, then I'd connect the cascodes to the Signal GND
 
With no lawyers involved in the design of my 1960's power amps, I don't main earth the power amp at all. Only the 1998 design CS800s has a 3 pin power cord. The hub of my system is the preamp or mixer, where the turntable headshell (green wire) is grounded. There is a wire going from that point to the 3rd pin of the wall plug, which is separate since neither the PAS2 or the RA88a have a 3 pin power cord. Installing that wire minimizes my hum, and provides a fig leaf of safety since the RCA cables do the grounding of the power amp. With 1960's RCA cables there is probably enough copper in the wrap conductor to take a significant fault current, but I can't say anything for cost reduced 2000 cables.
Interesting, these power amps do not have the dual transistor input structure of every modern amp design (or op amps) and don't seem to need it IMHO.
The other details of internal star design are interesting. thanks for the review.
 
Last edited:
Okay guys!!!
So, we have impedance resistor, the RC filter network, the feedback line connected to Signal Ground.
Another as resistor of CCS, voltage reference cascode, caps in pos & neg pole, zobel network... to connect with Power GND.

I have a design power amp, i'm worry about the ground of input stage components, to get lowest noise.
Another one think about what?
Thanks!!!
 
Last edited:
Good post :) Although I would label the GNDs differently though, as it's not necessarily dirty ground.

I'd call it "Signal Ground" and "Power Ground". After all, the amp output /speaker has to connect to a clean star point of this power ground, as does the Zobel. The capacitor leads themselves I would consider dirty. Due to the star topology however doesn't/shouldn't affect other returns, especially not the signal ground returns.
there are a new paper (at least for me) from Bruno Putzeys in this matter:
http://www.hypex.nl/docs/papers/The G Word.pdf
Linear Audio | your tech audio resource
check out also this:
http://www.diyaudio.com/forums/anal...eamp-well-balanced-volume-control-really.html
 
Okay guys!!!
So, we have impedance resistor, the RC filter network, the feedback line connected to Signal Ground.
Another as resistor of CCS, voltage reference cascode, caps in pos & neg pole, zobel network... to connect with Power GND.

I have a design power amp, i'm worry about the ground of input stage components, to get lowest noise.
Another one think about what?
Thanks!!!

My solution for the resistor of CCS, resistor of cascode voltage reference etc, is not to connect to the ground, but the opposite power rail, if possible. The dissipation of the resistors are bigger, but it's not a big problem, if you take care about it.

Sajti
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.