How to wire up an Amplifier

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
www.hifisonix.com
Joined 2003
Paid Member
I've started this thread so the various approaches to wiring up an amplifier can be discussed and debated.

There are two primary concerns when wiring up an audio amplifier:-

1. minimizing loop areas to prevent magnetic coupling - of which the classic ground loop is the best example, but cross-channel ground loops are a related phenomena as well.
2. common impedance coupling - an example of this would be failing to use a 'T' or a 'Star' grounding scheme to connect the rectifier(s), reservoir capacitors and the amp module grounds together.

Clearly, some have their preferences which will most often be driven around what they consider to be the worst of the various evils.

To get the discussion going, here are some presentations - others can be added to the list so that there is a more complete overview on the subject.

We should set ourselves the goal here on DIY audio of building amplifiers with hum and related noise levels of at least -100 dB ref 2.83 volts pk to pk out, but a better target is better than -110 dB.

http://hifisonix.com/wordpress/wp-content/uploads/2019/02/Ground-Loops.pdf

http://hifisonix.com/wordpress/wp-c...B-Design-Guidlines-for-Minimizing-Hum-1-1.pdf

Here are DIY audio member ilimzn's Excellent Posts #50 and #51 on Ground Loops

http://hifisonix.com/wordpress/wp-content/uploads/2019/04/ilimzns-Excellent-Posts-on-Ground-Loops.pdf

(I've hosted this on my website for the time being since I don't seem to be able to load the doc up into this first post. Maybe one of the mods could help out here with some directions?)
 
Last edited:
  • Like
Reactions: 1 user
www.hifisonix.com
Joined 2003
Paid Member
The first topic to discuss is what do you do with the speaker return (ie the speaker -ve wire).

Do you take it back to the amplifier module and let it share the same 0V line back to the PSU as the amplifier module 0V wire (easy to minimize loop areas with this approach)

or

is it better to run it directly back to the power supply 0V. You avoid voltage drops across the common 0V wire back to the PSU with this method so the 0V to the module essentially only carries the module bulk decoupling capacitor smoothing currents
 
is it better to run it directly back to the power supply 0V. You avoid voltage drops across the common 0V wire back to the PSU with this method so the 0V to the module essentially only carries the module bulk decoupling capacitor smoothing currents

If you do that you move the speaker return reference away from the signal ground reference and the PSU return currents through the umbilical ground wire cause a noise voltage between the references.
 
How about both? Run it close coupled back to the amplifier module, but don't connect it to PSU 0V there, instead run it close coupled with the PSU supply leads back to the PSU 0V. Just a thought.

I would say he addressed this by writing

(easy to minimize loop areas with this approach)

In other words it could be done, but it's harder to do and he wants an easier solution.

BUT, while you reduce the loop area the PSU return currents still create a noise voltage across the impedance of the PSU return conductor. One step forward, two steps back.
 
www.hifisonix.com
Joined 2003
Paid Member
But, if you decouple suitably on the amp module PCB and the loops are smaller, the only currents in the 0 V wire back to the amp module are LF.

How much signal coupling do you expect between the speaker return and the amplifier module?

The question is, do you want wideband currents being radiated around inside you amplifier or do you think the common impedance between the amp board and the PSU is a more manageable problem?
 
Well let's say the onboard decoupling is 10mF 10mohm ESR capacitors, and the umbilical wires are 10mohm (much larger than would actually be expected). HF current is equally split between the onboard capacitors and the umbilical, minus a small amount from the umbilical inductance, which will be very low if it's routed correctly. Even a local 1uF film cap only kicks in at 15KHz across the 10mohm ESR.

Furthermore, the rectifier pulses aren't perfectly symmetrical, there is a residual ground pulse especially on asymmetrical music or frequencies below 120Hz.
 
www.hifisonix.com
Joined 2003
Paid Member
In fact if HF currents are confined to the amp board, that just proves that the umbilical resistance is enough to cause a noise voltage between board ground and the main reservoir tee.

They are confined to the amp module board because the loop is smaller through the local on board decoupling.

I would argue that as long as any noise voltage across the umbilical ground is below the noise floor, you’re ok.

I control the umbilical 0V connection by keeping it short and very thick (easy to do) and then the radiated wideband field from the speaker by ensuring the signal and return are twisted and the return goes back to the amplifier board.
 
www.hifisonix.com
Joined 2003
Paid Member
That's why the decoupling on the PCB has to be wideband - you should do this with either method (this applies to small signal stuff as well). And I don't think disregarding loop area above some frequency is good advice - if you have AC current flowing through a loop, its always best to minimize the loop area.

If you have HF currents flowing in the 0V umbilical rather than the local loop, the on-board bulk decoupling is inadequate.

For an extreme example, look at a class D amp. Would you want HF switching hash and HF signal related garbage on the 0V umbilical?
 
Last edited:
Does my example of a 1uF film cap not qualify as wideband? What do you use then? 100uF motor run capacitor? The 1uF film cap only hits 10mohm at 15KHz. That means 15 harmonics of 1KHz are entirely unhindered by it.

EDIT: My mistake, 1uF has a capacitive reactance of 10mohm at 15.6MHz, not 15KHz. So that's 156000 harmonics of 1KHz.
 
Last edited:
I hope you aren't thinking about the diagram I posted in Cordell's thread, because I'm not. I was just answering your question. That diagram was something totally different from what I'm discussing here. I am assuming you mean a typical amp, where you just move the speaker return off the PCB. I didn't expect you to carry context over into a new thread without explicitly saying so, that would just confuse everyone.
 
www.hifisonix.com
Joined 2003
Paid Member
Here is a look at Poldaaudio's set-up and how I'd approach it.

I have an amp with a similar set-up to that shown in the diagram. But, I located the input connectors next to each other and bonded the connector grounds together and then ran the screened cable directly to each amplifier module input - so the shortest route from connectors to modules. If I ran the screened cables together to the fist module and then continued the second screened cable around the front of the amp all the way around to the second module, it was much quieter (talking -110 dBV vs -90 dBV). In the second approach, the loop area between the two amp modules in minimized, so the cross channel loop current is minimized. I use 15 Ohm HBR's but also got good results with a 3.3 Ohm.

If you bond the input connector grounds together, and locate them next to each other, you trap cross channel ground loops inside the amplifier. In the attached diagram, having them separated and not bonded means you have a large loop area for noise pick-up.
 

Attachments

  • IMG_0988.jpg
    IMG_0988.jpg
    437.2 KB · Views: 637
Last edited:
Member
Joined 2017
Paid Member
Bonsai (and Poldaaudio) - What are your thoughts on Douglas Self's recommendation is chapter 25 of his "Audio Power Amplifier Design" book? He suggests tying the signal input ground to the audio-chassis ground at the input connector. Self says this causes ground loop currents to flow to the chassis ground and then mains ground and not through the loop in Poldaaudio's diagram presented above (Assuming I'm reading it correctly).
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.