In my case, where I have 4 channels and the preceding signal stage in the amp ground sensing (differential signalling) made for an easy decision. A HBR is about half of the ground sense already, so it makes total sense to add the other half and enjoy the benefits having the ground levels between the boards explicitly separated but as local references. Factors like crosstalk are improved too.
Of course for those who have the amp only in the one chassis or choose to avoid opamps, this might not be so easy.
Hi Fred,
The technique of twisting the + and - rail wires together is not to cancel currents, but rather to sum the two class AB half-wave-rectified currents back into a sinewave (when the signal is a sinewave) so that its net field is more linear.
Cheers,
Bob
I don't understand this statement here. What does harmonic distortion have to do with phase or phase shift?
Harmonic distortion arises from wave shape distortion of any kind.
I'm lost, do you have a schematic?
That reads incorrect as formulated. Phase shift is not directly related to distortion because of the very interesting mathematical properties of sine waves. The integral (inductor current) and derivative (capacitor current) of a sine wave, are also sine waves.
Poldaaudio,
Rather than the ‘crying out aloud’ bit, how about some scope shots or a diagram showing why your approach is better? Measurements?
I would thing that in a multi-channel system, terminating the speaker return back at each amp module would be even more important.
Try thinking about the star ground on the amp module as the reference. You then have two loops - both terminating at the star ground on the module star ground. As long as you provide sufficient wideband bulk decoupling on the module PCB and the supply wires from the PSU are twisted (minimize loop area and EM radiation) you are good to go.
Rather than the ‘crying out aloud’ bit, how about some scope shots or a diagram showing why your approach is better? Measurements?
I would thing that in a multi-channel system, terminating the speaker return back at each amp module would be even more important.
Try thinking about the star ground on the amp module as the reference. You then have two loops - both terminating at the star ground on the module star ground. As long as you provide sufficient wideband bulk decoupling on the module PCB and the supply wires from the PSU are twisted (minimize loop area and EM radiation) you are good to go.
Comments relate to postBob Cordell's Power amplifier book. You are correct about distortion of sine waves. What I meant to say is that without phase shift there is only a small amplitude error.
In my opinion, the 19.99V is incorrect as the 10mV is outside the feedback loop.
Last edited:
Ok - I accept that you are left unsatisfied after reading the presentation. But unless you are much more specific, I cannot address your concerns.
What is it that you do not like about the presentation, or that you feel is lacking?
Here it is, page 58:
http://hifisonix.com/wordpress/wp-content/uploads/2019/02/Ground-Loops.pdf
What I don't like is the reliance on ground breaking resistors which only add 15db or so of rejection. But that seems to be a necessary evil.
I made my own diagram. I wouldn't be surprised if this has been proposed on DIYaudio before, but it's easy to pass over this EMC stuff without seeing the significance. The 10nF capacitors entirely break the internal ground loop between channels and between PSU returns and input. This layout is no longer susceptible to magnetic interference when the RCA shields are shorted, as there is no need for any loop area before the 10R resistors. The resistors are no longer as critical as they were, but they still help with common mode hum from the source.
I'm not sure 10nF is enough, I think 330nF would be a fairly safe value.
http://hifisonix.com/wordpress/wp-content/uploads/2019/02/Ground-Loops.pdf
What I don't like is the reliance on ground breaking resistors which only add 15db or so of rejection. But that seems to be a necessary evil.
I made my own diagram. I wouldn't be surprised if this has been proposed on DIYaudio before, but it's easy to pass over this EMC stuff without seeing the significance. The 10nF capacitors entirely break the internal ground loop between channels and between PSU returns and input. This layout is no longer susceptible to magnetic interference when the RCA shields are shorted, as there is no need for any loop area before the 10R resistors. The resistors are no longer as critical as they were, but they still help with common mode hum from the source.
I'm not sure 10nF is enough, I think 330nF would be a fairly safe value.
Attachments
Last edited:
The diagram is correct. The feedback reference is on the other side of the ground resistance, as it is blocked by the 10R resistor. If the feedback reference were at speaker return before the ground resistance, you would be correct. However this would likely result in a number of other layout problems.
The voltage across the speaker would be 19V. The voltage to ground at the amp output would be 20V.
How does the 10mV across the ground resistance change any of the feedback voltages? That 10mV is no different than if you put a resistor in series with the speaker and measured the voltage drop across it.
You didn't answer my question of how the 10mV affects the feedback loop. The 10ohm resistor is only shunting the ground resistance. They share the same nodes. One pin to speaker return, the other pin to ground. The resistors are in parallel. There is no voltage between the two ground symbols, they are the same node.
On the other hand if you disconnect the input return from ground, you will get 20V across the speaker because the signal reference is floating at the speaker return voltage.
On the other hand if you disconnect the input return from ground, you will get 20V across the speaker because the signal reference is floating at the speaker return voltage.