Hi,
That's why you must do everything to ensure that the signal ground holds a steady reference and why you should try to reduce any modulation of the ground voltage by carefully routing speaker returns, electromagnetic radiation loops, charging spikes from the rectifier/smoothing cap combination, glitches from the decoupling caps etc. to a common point with least effect on the ground reference.
If you go for the separate terminal lugs solution it allows experimentation, you can follow your own scheme or look at MikeB's or heaven forbid, even my recommendation with it's separate Safety, Power and Signal grounds.
I suppose the definition of a good grounding scheme is one that WORKS for the widest range of amplifiers. Go experiment and then tell us what DOES NOT WORK, as well as what works adequately and what works very well.
absolutely yes.
That's why you must do everything to ensure that the signal ground holds a steady reference and why you should try to reduce any modulation of the ground voltage by carefully routing speaker returns, electromagnetic radiation loops, charging spikes from the rectifier/smoothing cap combination, glitches from the decoupling caps etc. to a common point with least effect on the ground reference.
If you go for the separate terminal lugs solution it allows experimentation, you can follow your own scheme or look at MikeB's or heaven forbid, even my recommendation with it's separate Safety, Power and Signal grounds.
I suppose the definition of a good grounding scheme is one that WORKS for the widest range of amplifiers. Go experiment and then tell us what DOES NOT WORK, as well as what works adequately and what works very well.
Cortez said:
Often the solution is a bridged configuration, classically with two of same chipamp or alternatively anything that provides a gnd potential (reference) and can handle with low impedance, dump the current back into the +- rails... but then you have to deal with the noise on those rails instead but it's often regarded as the lesser of two evils since effecting the signal during amplification (with chipamp reasonable PSRR) instead of before it (in the case noise which we ultimately wanted to filter out anyway before it gets to the amp chip).
Hi Cortez,
I just lost a long reply re your last post. I am giving up and going home. Talk to you soon.
But first a quick reply:-
The modulation that appears as hum and noise on the ground is brought about by varying currents and their voltages flowing between the various parts of the circuits.
I just lost a long reply re your last post. I am giving up and going home. Talk to you soon.
But first a quick reply:-
definately not.
The modulation that appears as hum and noise on the ground is brought about by varying currents and their voltages flowing between the various parts of the circuits.
Hi Cortez,
The modulation of the grounds is due to impressed currents and finite resistances. The modulation implies varying current (and voltage) flows. This seems a better way of putting it.
If you let the PSU charging system pass it's high current peaks back and forth between transformer, rectifier and smoothing caps using it's own short, compact (for low loop area) and thick wires (plates or copper strip if you prefer) then those currents and their voltages are confined to their own loops.
Now tie the PSU common sides together, current flows across the common link when some asymmetrical current is drawn from the PSU, otherwise no current passes between the PSU commons.
Connect the PSU common link to a star ground serving only the power returns, this is the Power Ground (PG). The charging pulses and voltages never enter the PG and PG becomes the reference voltage for the PSU commons that otherwise would float. Only when your amplifier draws unbalanced current from the PSU do you have a current and voltage through the PSU common link. But the PG reference is what matters, this separation of duty has very effectively removed the charging pulses from the PG. Now use the PG as a star ground for ALL the power returns plus two spare legs for later. Bringing all the dirty returns back to PG again constrains all the potentially big currents and big voltages to within the PG star.
Use one spare leg of the PG star to tie in the Safety Ground if you want, using that list of disconnecting network components.
Use the second spare leg to connect to the Signal Ground star (SG). By bringing all the clean returns to the SG you are excluding all the big contaminating currents from the SG in a similar manner that separating PG from PSU common had. SG is a clean ground for those parts of the circuit that NEED a clean reference. The connection between SG and PG carries only VERY small currents and thus suffer very small voltage differences from the PG voltage. These signal currents are all signal related and should not (indeed must not) contain any glitches resulting from sudden turn on or turn off of devices.
But, as usual, a short thick wire connection is better than the other options. Laying out the amplifier components that allows short routes between PCBs and grounds and from ground to ground and to PSU common, all helps reduce these errant voltages. I don't think, but I may be wrong, that any layout or connection scheme can eliminate the voltage drops, but it can reduce the effects relative to the voltage signals appearing at each ground to below audibility.
This is where MikeBs scheme seems to fall down. As soon as you direct connect multiple mains powered products together each with a common ground and each connected to the mains earthing system, you are guaranteed to introduce ground loops and the loop areas will induce hum way above audibility. No matter how much you reduce impedances the voltage induced in the loops just pulls more current to compensate.
All these returns can be bolted together at their separated locations using terminal lugs, we are now back to the experimentation I suggested earlier.
The modulation of the grounds is due to impressed currents and finite resistances. The modulation implies varying current (and voltage) flows. This seems a better way of putting it.
If you let the PSU charging system pass it's high current peaks back and forth between transformer, rectifier and smoothing caps using it's own short, compact (for low loop area) and thick wires (plates or copper strip if you prefer) then those currents and their voltages are confined to their own loops.
Now tie the PSU common sides together, current flows across the common link when some asymmetrical current is drawn from the PSU, otherwise no current passes between the PSU commons.
Connect the PSU common link to a star ground serving only the power returns, this is the Power Ground (PG). The charging pulses and voltages never enter the PG and PG becomes the reference voltage for the PSU commons that otherwise would float. Only when your amplifier draws unbalanced current from the PSU do you have a current and voltage through the PSU common link. But the PG reference is what matters, this separation of duty has very effectively removed the charging pulses from the PG. Now use the PG as a star ground for ALL the power returns plus two spare legs for later. Bringing all the dirty returns back to PG again constrains all the potentially big currents and big voltages to within the PG star.
Use one spare leg of the PG star to tie in the Safety Ground if you want, using that list of disconnecting network components.
Use the second spare leg to connect to the Signal Ground star (SG). By bringing all the clean returns to the SG you are excluding all the big contaminating currents from the SG in a similar manner that separating PG from PSU common had. SG is a clean ground for those parts of the circuit that NEED a clean reference. The connection between SG and PG carries only VERY small currents and thus suffer very small voltage differences from the PG voltage. These signal currents are all signal related and should not (indeed must not) contain any glitches resulting from sudden turn on or turn off of devices.
But, as usual, a short thick wire connection is better than the other options. Laying out the amplifier components that allows short routes between PCBs and grounds and from ground to ground and to PSU common, all helps reduce these errant voltages. I don't think, but I may be wrong, that any layout or connection scheme can eliminate the voltage drops, but it can reduce the effects relative to the voltage signals appearing at each ground to below audibility.
This is where MikeBs scheme seems to fall down. As soon as you direct connect multiple mains powered products together each with a common ground and each connected to the mains earthing system, you are guaranteed to introduce ground loops and the loop areas will induce hum way above audibility. No matter how much you reduce impedances the voltage induced in the loops just pulls more current to compensate.
All these returns can be bolted together at their separated locations using terminal lugs, we are now back to the experimentation I suggested earlier.
Hi Andrew !
Thanks for your detailed post, specially typing it already 2nd time, but
I am a very visual type, couldnt you maybe draw a picture to emphasize
your main points ? You know a picture worth of hundred words.
Explaining would be a lot simpler in this way, I guess.
For example:
- eventually whats your suggestion, where to put the PS star ground ?
- why is a ground plane not a good solution ?
- in case of ground plane how would you place the Power GND and the Signal GND ?
> Connect the PSU common link to a star ground serving only the
> power returns, this is the Power Ground (PG). The charging
> pulses and voltages never enter the PG and PG becomes the
> reference voltage for the PSU commons that otherwise would float.
Why wouldnt the charging pulses get into the ground ?
If we put our main GND between to smoothing caps, when they are charging,
the real ground will be at the trafo through the rectifiers, but there is
a very high current namely the charging currents, so the voltage at our
main gnd at the caps will depend on this current and on the resistance
between the trafo and the caps.
(There will be a shared current path: charging current, speaker current, other small currents)
> Only when your amplifier draws unbalanced current from the PSU
> do you have a current and voltage through the PSU common link.
But an amp tipically draws unbalanced power, doesnt it ?
Thanks for your detailed post, specially typing it already 2nd time, but
I am a very visual type, couldnt you maybe draw a picture to emphasize
your main points ? You know a picture worth of hundred words.
Explaining would be a lot simpler in this way, I guess.
For example:
- eventually whats your suggestion, where to put the PS star ground ?
- why is a ground plane not a good solution ?
- in case of ground plane how would you place the Power GND and the Signal GND ?
> Connect the PSU common link to a star ground serving only the
> power returns, this is the Power Ground (PG). The charging
> pulses and voltages never enter the PG and PG becomes the
> reference voltage for the PSU commons that otherwise would float.
Why wouldnt the charging pulses get into the ground ?
If we put our main GND between to smoothing caps, when they are charging,
the real ground will be at the trafo through the rectifiers, but there is
a very high current namely the charging currents, so the voltage at our
main gnd at the caps will depend on this current and on the resistance
between the trafo and the caps.
(There will be a shared current path: charging current, speaker current, other small currents)
> Only when your amplifier draws unbalanced current from the PSU
> do you have a current and voltage through the PSU common link.
But an amp tipically draws unbalanced power, doesnt it ?
Hi,
it takes me too long to type, it would be even longer to draw as well. Sorry if you can't follow my description, but look back at my only diagram posting (2b for zip, 2a for RAR) the diagram and my description follow exactly.
Try to follow the charging current and for simplicity assume initially there is no load connected. You will find it follows around a figure of eight pair of loops. There are a pair of loops for both the plus and minus sides. To charge up the smoothing caps and recharge if some current is drawn there is NO NEED for the common link. There is NO current in the common link when no load is connected. The two PSU supplies are isolated from each other and isolated from any other part of the circuit. They are floating.
Now add the common link, the two supplies are floating with respect to the other grounds and other circuits but are tied to each other. As the caps charge there is still no current in the common link.
Now connect the common link to Power Ground, there is still no current in the common link nor in the connection to PG. No current means no voltage drop. This implies that PG and the whole of the common link are exactly at the same voltage. The charging currents can be any size, any shape of waveform and still the voltage drops in the common link and the connection to PG are ZERO.
Do you follow this part?
Do you agree with this part?
it takes me too long to type, it would be even longer to draw as well. Sorry if you can't follow my description, but look back at my only diagram posting (2b for zip, 2a for RAR) the diagram and my description follow exactly.
Try to follow the charging current and for simplicity assume initially there is no load connected. You will find it follows around a figure of eight pair of loops. There are a pair of loops for both the plus and minus sides. To charge up the smoothing caps and recharge if some current is drawn there is NO NEED for the common link. There is NO current in the common link when no load is connected. The two PSU supplies are isolated from each other and isolated from any other part of the circuit. They are floating.
Now add the common link, the two supplies are floating with respect to the other grounds and other circuits but are tied to each other. As the caps charge there is still no current in the common link.
Now connect the common link to Power Ground, there is still no current in the common link nor in the connection to PG. No current means no voltage drop. This implies that PG and the whole of the common link are exactly at the same voltage. The charging currents can be any size, any shape of waveform and still the voltage drops in the common link and the connection to PG are ZERO.
Do you follow this part?
Do you agree with this part?
> it would be even longer to draw as well
Painting is so much faster, and you just have to write 1-2 sentences as comment.
Sorry, but English is not my first language, and to understand a lot of rows
is sometimes confusing, as it is often on my own language too on forums.
I guess the essence should always be a very simple and brief something.
Lot of explanations are often confusing instead of getting closer to the point.
> Do you follow this part?
> Do you agree with this part?
Well, its hard to imagine for me precisely the effect.
But check this out:
http://www.falstad.com/circuit/
File / Import and then Copy-Paste this:
$ 1 5.0E-6 8.531194996067258 55 5.0 48
v 224 160 224 16 0 1 40.0 5.0 0.0
d 224 16 288 96 0
d 224 160 288 96 0
d 160 96 224 16 0
d 160 96 224 160 0
w 160 96 160 192 0
w 288 96 336 96 0
w 160 192 336 192 0
c 336 96 336 192 0 0.0010 3.5346078318991223
w 336 96 416 96 0
w 336 192 416 192 0
r 416 96 416 192 0 430.0
x 451 152 457 152 0 16 load
v 224 304 224 416 0 1 40.0 5.0 0.0
d 224 304 304 352 0
d 224 416 304 352 0
d 160 352 224 304 0
d 160 352 224 416 0
c 336 432 336 352 0 0.0010 -3.5346078318991347
w 304 352 336 352 0
w 160 432 160 352 0
w 160 432 336 432 0
w 336 432 416 432 0
w 336 352 416 352 0
r 416 352 416 432 0 430.0
w 336 192 336 352 0
o 25 64 0 3 5.0 9.765625E-5 0
I hope it will work... Here, you can see the voltage as colourization.
And the common link is floating, but no current flows.
Thats exactly what you are talking about ?
But look the sine wave down on the scope, its very strange, isnt it ?
Or the simulation isnt correct somewhere ?
Painting is so much faster, and you just have to write 1-2 sentences as comment.
Sorry, but English is not my first language, and to understand a lot of rows
is sometimes confusing, as it is often on my own language too on forums.
I guess the essence should always be a very simple and brief something.
Lot of explanations are often confusing instead of getting closer to the point.
> Do you follow this part?
> Do you agree with this part?
Well, its hard to imagine for me precisely the effect.
But check this out:
http://www.falstad.com/circuit/
File / Import and then Copy-Paste this:
$ 1 5.0E-6 8.531194996067258 55 5.0 48
v 224 160 224 16 0 1 40.0 5.0 0.0
d 224 16 288 96 0
d 224 160 288 96 0
d 160 96 224 16 0
d 160 96 224 160 0
w 160 96 160 192 0
w 288 96 336 96 0
w 160 192 336 192 0
c 336 96 336 192 0 0.0010 3.5346078318991223
w 336 96 416 96 0
w 336 192 416 192 0
r 416 96 416 192 0 430.0
x 451 152 457 152 0 16 load
v 224 304 224 416 0 1 40.0 5.0 0.0
d 224 304 304 352 0
d 224 416 304 352 0
d 160 352 224 304 0
d 160 352 224 416 0
c 336 432 336 352 0 0.0010 -3.5346078318991347
w 304 352 336 352 0
w 160 432 160 352 0
w 160 432 336 432 0
w 336 432 416 432 0
w 336 352 416 352 0
r 416 352 416 432 0 430.0
w 336 192 336 352 0
o 25 64 0 3 5.0 9.765625E-5 0
I hope it will work... Here, you can see the voltage as colourization.
And the common link is floating, but no current flows.
Thats exactly what you are talking about ?
But look the sine wave down on the scope, its very strange, isnt it ?
Or the simulation isnt correct somewhere ?
Hi, sorry to disappear in the middle of a discussion but real life decided to pay me a visit and somethings need to take a back seat.
To answer Cortez's question, the hum pickup will show up in battery powered devices, my magnetic field sniffer is battery powered. It's the same mechinism as picking up a radio station signal that is referenced to the earth. The body is just a channel for the mains radiation that is all around us. Touching the heatsink allows the couple signal to find a path through the amp circuitry. Some part of the amp layout allows it to become part of the signal path. I haven't thought about this too deeply since it's a side thread; I'm just commenting based on observation.
On the comment about the speaker storing voltage and feeding it back later, actually the motor (magnet and voicecoil) in the speaker generates a back EMF (electro-motive force) (as soon as it starts to move) that has the speaker acting as a motor and a generator simutaneously; this is the source of the reverse currents I alluded to. This effect is real.
I guess we have to agree to disagree on this. Your comments make me believe we are saying the same thing with only slight differences as to execution.
I don't see any reason to separate the grounds as long as the ground reference to the power transformer windings is solid.
It's should attempt to be the perfect virtual ground and as I described ground in an earlier post. The return currents should be like throwing a bucket of water into the ocean, they should not affect the oceans level. By connecting all of the returns to this point there should be no interaction. By isolating the grounds to separate islands, as I interpert your suggestion, there is the possibility the high current or RF signals could modulate the ground across the intervening wire.
This is what I have been attempting to do. My background given me access to an amazing variety of equipment to experiment with.
Anyway, it's good to be back.
Mike.
To answer Cortez's question, the hum pickup will show up in battery powered devices, my magnetic field sniffer is battery powered. It's the same mechinism as picking up a radio station signal that is referenced to the earth. The body is just a channel for the mains radiation that is all around us. Touching the heatsink allows the couple signal to find a path through the amp circuitry. Some part of the amp layout allows it to become part of the signal path. I haven't thought about this too deeply since it's a side thread; I'm just commenting based on observation.
On the comment about the speaker storing voltage and feeding it back later, actually the motor (magnet and voicecoil) in the speaker generates a back EMF (electro-motive force) (as soon as it starts to move) that has the speaker acting as a motor and a generator simutaneously; this is the source of the reverse currents I alluded to. This effect is real.
AndrewT said:
I guess we have to agree to disagree on this. Your comments make me believe we are saying the same thing with only slight differences as to execution.
I don't see any reason to separate the grounds as long as the ground reference to the power transformer windings is solid.
It's should attempt to be the perfect virtual ground and as I described ground in an earlier post. The return currents should be like throwing a bucket of water into the ocean, they should not affect the oceans level. By connecting all of the returns to this point there should be no interaction. By isolating the grounds to separate islands, as I interpert your suggestion, there is the possibility the high current or RF signals could modulate the ground across the intervening wire.
AndrewT said:
This is what I have been attempting to do. My background given me access to an amazing variety of equipment to experiment with.
Anyway, it's good to be back.
Mike.
Welcome back Mike !
> the hum pickup will show up in battery powered devices
I just asked, because I had a chipamp, wich stoped humming
after I pulled out the mains and had just the caps in the PSU.
Thats why I thought only the trafo might produce it.
> back EMF (electro-motive force)
Since now, I thought by some reason, that this affects only the
amps output and the rails, but the ground can polluted also I guess.
AndrewT, are you still with us ? How did you like "my" simulator ?
I like to see these things in animation, it helps me a lot to better understand.
Its also a magnetic field simulator, it shows for example how a twisted pair
cancel eachothers fields, etc...
> the hum pickup will show up in battery powered devices
I just asked, because I had a chipamp, wich stoped humming
after I pulled out the mains and had just the caps in the PSU.
Thats why I thought only the trafo might produce it.
> back EMF (electro-motive force)
Since now, I thought by some reason, that this affects only the
amps output and the rails, but the ground can polluted also I guess.
AndrewT, are you still with us ? How did you like "my" simulator ?
I like to see these things in animation, it helps me a lot to better understand.
Its also a magnetic field simulator, it shows for example how a twisted pair
cancel eachothers fields, etc...
Cortez said:
In your example the transformer is completing the loop back to the mains. I thought you were asking if it was necessary to have that connection.
The back EMF is generated by the speaker, therefore the rails try to absorb/sink the energy, which causes charging currents in the caps with the return loop being back to the speaker ground (not the transformer ground). I became aware of this working with motion control electronics and trying to accurately control high-speed motor motions.
Have you tried grounding the heatsink yet?
Regards, Mike
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.