Lets not bring device-to-device ground loops into the picture, for now just assuming the amp byself with the following components on seperate boards, each with it's own ground.
Power Supply
Low current input stage
High Current output Stage
RCA input
Speaker output
So we have 5 grounds here, staring all 5 of these components to a common point results in 120 hz hum.
What arrangement of star grounding(s) would result in the least amount of hum? I have been playing around with different combinations for about a hour and a half now and I can't eliminate the hum.
Power Supply
Low current input stage
High Current output Stage
RCA input
Speaker output
So we have 5 grounds here, staring all 5 of these components to a common point results in 120 hz hum.
What arrangement of star grounding(s) would result in the least amount of hum? I have been playing around with different combinations for about a hour and a half now and I can't eliminate the hum.
bonsai, i have two questions for you.
is there any benefit to bypass the 15R at the amps with 100nF as some other designs do?
why have you not routed the bridge rect 0V's directly to junction 1, instead it goes via the main caps wiring?
is there any benefit to bypass the 15R at the amps with 100nF as some other designs do?
why have you not routed the bridge rect 0V's directly to junction 1, instead it goes via the main caps wiring?
1. No benefit as far as I can tell. The cap would short the resistor out at HF. I would need to test this to judge the effectiveness
2. Because you must eliminate the capacitor charging currents from the 'T' or the star if that's what you are using. Charging currents are many amps, and just a few mV of resistance and you have a common impedance coupling problem.
I wanted to say I tried this layout and got some improvement. However a huge improvement was moving the amp pcb away from the power supply filtering caps and transformer... that results in almost zero noise. If I put my ear directly on top of the speaker, I could barely hear anything at all.
Anyway, it looks like some of my issues are spacing. My amp is tight so I'm going to look into redoing my PS layout to move the filtering caps away form amp PCB and maybe get some insulated wires for the inputs and outputs.
Anyway, it looks like some of my issues are spacing. My amp is tight so I'm going to look into redoing my PS layout to move the filtering caps away form amp PCB and maybe get some insulated wires for the inputs and outputs.
Also, another question, what is the advantage of not CT the transformer but instead using two bridge rectifiers? You would still get the same +/- rail values right?
In case anyone cares, I completely redid the wiring in the amp, used really thick ground wires (14 awg), twisted and isolated the inputs and outputs and the ground loop is 100% gone. Zero noise.
Also, I had worse noise with the isolation resistor, so I removed it.
Also, I had worse noise with the isolation resistor, so I removed it.
There are two ms in reasons
1. Prevent DC balance issues in the transformer - this is when one of the rails draws more current or has a higher voltage than the other and can be so issue on split rail supplies that only use one rectifier bridge. It can cause the core to buzz. See Nelson Pass who has commented on this. This is how I build my big amps now
2. The caps charging currents on an amplifier run into 10's of amps often. If you rectify each separately, then join the - of one and the + of the other to form a dual supply, you keep the charging currents isolated from the signal part. However, for this to work properly, you MUST T OFF the 0V junction of the two main caps.
Glad to see you've solved your problem.
On the isolation resistors, often they are no needed in that case, indeed the correct thing to do is to link them out.
I just wanted to mention that the updated slides which are available on your site are imho maybe the most valuable document available for DIYers . @Bonsai: thank you so much for this!
Nevertheless one question:
Do you already have the information reagrding the directly connected shield approach available?