Turns out it did disappear for a while, along with at least one other. It's back. Very worthwhile: http://www.diyaudio.com/forums/diya...udio-component-grounding-interconnection.html
Hi DF96 I just want to clarify something... are you saying that the problem is with dual bridges (one for each rail) X 3 or with separate rectifiers and filter caps (with only one transformer) for each amp period?
That is regardless of the rectifier configuration, that if you have one PS per channel it should have one transformer per channel as well?
Just that I have purchased the caps for re-doing the PS on my hum ridden amp and was planning on doing a similar supply to the one I posted near the beginning of the thread, ie separate recitifier and caps for each channel, but the 0V being to the transformer secondary's mid point.
The PS layout I posted in my first post is for my chipamp (not the amp I have hum problems with) and is absolutely dead quiet apart from a tiny amount of hiss you can only hear with your ear right against the tweeter. I thought maybe Elbert's problem was the dual rectifiers per channel and not having a common (through the transformer windings) return, but maybe I just got lucky with my implementation!
Tony.
That is regardless of the rectifier configuration, that if you have one PS per channel it should have one transformer per channel as well?
Just that I have purchased the caps for re-doing the PS on my hum ridden amp and was planning on doing a similar supply to the one I posted near the beginning of the thread, ie separate recitifier and caps for each channel, but the 0V being to the transformer secondary's mid point.
The PS layout I posted in my first post is for my chipamp (not the amp I have hum problems with) and is absolutely dead quiet apart from a tiny amount of hiss you can only hear with your ear right against the tweeter. I thought maybe Elbert's problem was the dual rectifiers per channel and not having a common (through the transformer windings) return, but maybe I just got lucky with my implementation!
Tony.
You have almost convinced me that the main problem is grounding.
This has been made much more difficult to find/cure because you have a multi-channel amplifier.
That is why I keep building monoblocks and why I recommend that everyone else learns how to build amplifiers via the monoblock route before they start extending to multi-channel.
Read Davenport, the whole of it. Then re-read the part you think where his advice applies to your situation. That should solve it.
One thing I am finding is that a two channel amp must have ONE common Audio Ground
or
complete separation using 4secondaries to create two quite separate Audio Grounds that float from the chassis. Then make it safe with two Disconnecting Networks. I can't make any of the in between these two states work properly.
This has been made much more difficult to find/cure because you have a multi-channel amplifier.
That is why I keep building monoblocks and why I recommend that everyone else learns how to build amplifiers via the monoblock route before they start extending to multi-channel.
Read Davenport, the whole of it. Then re-read the part you think where his advice applies to your situation. That should solve it.
One thing I am finding is that a two channel amp must have ONE common Audio Ground
or
complete separation using 4secondaries to create two quite separate Audio Grounds that float from the chassis. Then make it safe with two Disconnecting Networks. I can't make any of the in between these two states work properly.
Yes, this looks like grounding, just from the fact that one channel alone works..
The good thing is that this means the other channels are supposed to work as well!
Just need to figure out how..
After sleeping on this a bit, the next step in the faultfinding process will be to systematically disconnect/ connect/ short etc.. yesterday I only shorted various inputs, which would still leave any unfortunate ground paths in place.
What's this Davenport?
Just out of curiosity, with two mono-blocks as you describe, what happens when two of those are connected to a pre-amp that has one common ground reference for both channels?
Wouldn't that cause a connection between the two floating grounds of the mono's?
The good thing is that this means the other channels are supposed to work as well!
Just need to figure out how..
After sleeping on this a bit, the next step in the faultfinding process will be to systematically disconnect/ connect/ short etc.. yesterday I only shorted various inputs, which would still leave any unfortunate ground paths in place.
What's this Davenport?
Just out of curiosity, with two mono-blocks as you describe, what happens when two of those are connected to a pre-amp that has one common ground reference for both channels?
Wouldn't that cause a connection between the two floating grounds of the mono's?
Hi,
a pair of monoblocks do not share a common Audio ground.
Connect a pair of monoblocks to a two channel source with a common audio ground and you have no new ground loops. No added hum, no added buzz. Each amplifier performs as it did alone.
If you are building dedicated active amplifiers that can only drive one model of speaker then it makes sense to build a monoblock and mount the speaker driver almost directly to the amplifier terminals.
I cannot understand the logic of building a 5.1 amplifier inside one chassis, when the speakers, by definition, must be spread about in at least 5 different locations.
a pair of monoblocks do not share a common Audio ground.
Connect a pair of monoblocks to a two channel source with a common audio ground and you have no new ground loops. No added hum, no added buzz. Each amplifier performs as it did alone.
If you are building dedicated active amplifiers that can only drive one model of speaker then it makes sense to build a monoblock and mount the speaker driver almost directly to the amplifier terminals.
I cannot understand the logic of building a 5.1 amplifier inside one chassis, when the speakers, by definition, must be spread about in at least 5 different locations.
If you want to keep control of ground currents then the rule is:
one secondary feeds one rectifier bridge which feeds one reservoir cap (if secondary is CT then the resultant supply can be CT too, with a pair of caps).
You can have as many of these as you need. Keep the ground returns of the charging pulses separate from each other, and well away from any star point. The onward-going ground (to the amp circuit) is then clean; it carries DC and perhaps a little ripple current but no worse than the +- connections.
If you want multiple supplies from one secondary then you can add as many extra RC or LC smoothers as you like to the one reservoir cap, but you need to be careful about grounding - only have one ground connection to the reservoir cap. If you have two or more connections then they will never be at the same potential so if they ever then get connected together elsewhere in the circuit (e.g. via signal grounds) you will have some circulating charging pulses. I suspect this is what is happening.
Don't just go round disconnecting grounds. If your circuit was correct then disconnecting any ground would stop it from working, as there would be no loops. Instead, draw an accurate diagram of exactly what you have done. Then look for ground loops, and ask yourself where the charging pulse loops are.
I'm not sure what else to say. I have said the same thing in several different ways now. Maybe someone else could explain it better than me?
one secondary feeds one rectifier bridge which feeds one reservoir cap (if secondary is CT then the resultant supply can be CT too, with a pair of caps).
You can have as many of these as you need. Keep the ground returns of the charging pulses separate from each other, and well away from any star point. The onward-going ground (to the amp circuit) is then clean; it carries DC and perhaps a little ripple current but no worse than the +- connections.
If you want multiple supplies from one secondary then you can add as many extra RC or LC smoothers as you like to the one reservoir cap, but you need to be careful about grounding - only have one ground connection to the reservoir cap. If you have two or more connections then they will never be at the same potential so if they ever then get connected together elsewhere in the circuit (e.g. via signal grounds) you will have some circulating charging pulses. I suspect this is what is happening.
Don't just go round disconnecting grounds. If your circuit was correct then disconnecting any ground would stop it from working, as there would be no loops. Instead, draw an accurate diagram of exactly what you have done. Then look for ground loops, and ask yourself where the charging pulse loops are.
I'm not sure what else to say. I have said the same thing in several different ways now. Maybe someone else could explain it better than me?
Now I disconnected the signal input cables from two of the amplifier cards whulst leaving them connected to power. I then short circuited the input pin in these cards with the input ground pin. (the input ground pin is connected to the amplifier-card 0V, in turn connected to the PSU 0V)
Virtually no hum or buzz in the speakers. ( a very faint hum audible in the midwoofer when putting my ear right next to it)
So, clearly a ground problem. Now I need to sort out how to elliminate any ground loops etc, whilst making sure all amplifier cards and X-over card signal ground is maintained..
The first approach that springs to mind, is to connect the shielding of the x-over to amplifier card signal cables in one end only. This should prevent loops forming between the amplifier cards through the x-over card.
I must then provide a different signal ground reference on the x-over card. This could be a separate ground wire from the 0V point in the PSU.
As far as I can see, this should follow the guidelines explained in the grounding article and the explanations provided here in this thread..
Lets see what happens..
Virtually no hum or buzz in the speakers. ( a very faint hum audible in the midwoofer when putting my ear right next to it)
So, clearly a ground problem. Now I need to sort out how to elliminate any ground loops etc, whilst making sure all amplifier cards and X-over card signal ground is maintained..
The first approach that springs to mind, is to connect the shielding of the x-over to amplifier card signal cables in one end only. This should prevent loops forming between the amplifier cards through the x-over card.
I must then provide a different signal ground reference on the x-over card. This could be a separate ground wire from the 0V point in the PSU.
As far as I can see, this should follow the guidelines explained in the grounding article and the explanations provided here in this thread..
Lets see what happens..
just attach a DVM to the speaker terminals.
If you have 3 DVMs then you can monitor the noise on all three channels at once as you try the different grounding tests.
It would help me a lot if I knew you were talking about 0.7mVac of hum and buzz or 2mVav or 5mVac, and a change from 2.0mVac to 1.8mVac tells me a lot.
Disconnect your speakers. There is absolutely no advantage in blowing them up.
If you have 3 DVMs then you can monitor the noise on all three channels at once as you try the different grounding tests.
It would help me a lot if I knew you were talking about 0.7mVac of hum and buzz or 2mVav or 5mVac, and a change from 2.0mVac to 1.8mVac tells me a lot.
Disconnect your speakers. There is absolutely no advantage in blowing them up.
First of all what is a DVM??
Anyway, I got the old scope hooked up and did some measurements.
It's not the best scope or digital camera, but hopefully, the images can give a clue about what's going on..
I measured on the output of the first amp.
First measurement is with the first card fully connected, the other two connected to PSU, but signal screen disconnected.
The scope is set to 5mV V/div and 2mS time/div.
The first measurement looks OK, about 4mV peaks, this is the hookup that gave virtually no audible noise in the speaker..
For the second measurement, I connected the signal cable screen on the second card.
Peaks allmost doubled and changed to positive sign only..
Third measurement; all cards now have their signal cable screen connected, peaks increase even further.
what could be going on here??
Anyway, I got the old scope hooked up and did some measurements.
It's not the best scope or digital camera, but hopefully, the images can give a clue about what's going on..
I measured on the output of the first amp.
First measurement is with the first card fully connected, the other two connected to PSU, but signal screen disconnected.
The scope is set to 5mV V/div and 2mS time/div.
The first measurement looks OK, about 4mV peaks, this is the hookup that gave virtually no audible noise in the speaker..
For the second measurement, I connected the signal cable screen on the second card.
Peaks allmost doubled and changed to positive sign only..
Third measurement; all cards now have their signal cable screen connected, peaks increase even further.
what could be going on here??
Attachments
For consistency, I performed the same set of measurements on the second amplifier card as well..
Same tendency, the noise adds on for each addirtional card that gets connected to the signal cable screen.
But the interesting thing here is that the peaks change sign compared to the first card...
Same tendency, the noise adds on for each addirtional card that gets connected to the signal cable screen.
But the interesting thing here is that the peaks change sign compared to the first card...
Attachments
Update;
This is starting to get interresting...
I have now disconnected and removed th X-over card, so for the moment we can leave that and its power supply out of the equation.
With the sope hooked up to the output of amplifier 1, I short circuit the signal wire and signal screen at the end of the cable previously attached to the card.
A perfectly flat line on the scope..
I then connect the screen from the cable leading to the second amp.. I get noise, even more so when i connect the screen from the third amp..
So far, the results are comparable to what I tried earlier on.
This just demonstrates that the signal grounds/screens are not at all happy to be connected together, be it directly or on the x-over board..
Then i did soemthing interresting.. I repeated the experiment, but this time i put a 39 ohm resistor between the screen from amp 2 and the shorted screen/signal to amp 1.
This time, there was no noise, the scope was as flat as when the second amp screen ewas not connected!
Could it be this simple? just insert some resistors to curb ground currents??
Of course, I didn't just come up with this all by my self, Had I been that clever I wouldn't be here asking stupid questions in the first place.
No, I picked up on this from the section on "ground loop supression" on page 5 in Dave Davenports grounding article.
Not wishing to get my hopes up too high yet, but this looks very interresting..
PS
Just tried with a 1 ohm resistor in stead of 39 ohms, and even with that, the ripple on the scope was barely visible.
This is starting to get interresting...
I have now disconnected and removed th X-over card, so for the moment we can leave that and its power supply out of the equation.
With the sope hooked up to the output of amplifier 1, I short circuit the signal wire and signal screen at the end of the cable previously attached to the card.
A perfectly flat line on the scope..
I then connect the screen from the cable leading to the second amp.. I get noise, even more so when i connect the screen from the third amp..
So far, the results are comparable to what I tried earlier on.
This just demonstrates that the signal grounds/screens are not at all happy to be connected together, be it directly or on the x-over board..
Then i did soemthing interresting.. I repeated the experiment, but this time i put a 39 ohm resistor between the screen from amp 2 and the shorted screen/signal to amp 1.
This time, there was no noise, the scope was as flat as when the second amp screen ewas not connected!
Could it be this simple? just insert some resistors to curb ground currents??
Of course, I didn't just come up with this all by my self, Had I been that clever I wouldn't be here asking stupid questions in the first place.
No, I picked up on this from the section on "ground loop supression" on page 5 in Dave Davenports grounding article.
Not wishing to get my hopes up too high yet, but this looks very interresting..
PS
Just tried with a 1 ohm resistor in stead of 39 ohms, and even with that, the ripple on the scope was barely visible.
Last edited:
Hi Elbert I think this comment from Antonio looks spot on based on your scope measurements and your insertion of resistors. The only thing I'm a bit uncertain about is why the return path is favouring the signal shield rather than your heavy duty PS wiring.
I thought you had tied together the three grounds of the three PS's, but I just had another thought... Are the connections from the diode bridges to the transformer secondaries all the same exact length and making equally good contact?
I've taken part of your original schematic and put coloured dots on it. assuming all diodes conduct equally (which won't necessarily be the case) then if there is no potential difference between any of the points with a same coloured dot then there shouldn't be any current flow between those spots... Small differences in wire length or joint resistance between any of these points could cause uneven current flow.
I've just done this to try and get a better image in my head of what might be going on... I'm uncertain at this point exactly how your power and signal grounds are connected with respect to the larger black dots in the image... as to me it would seem that there would have to be a lower resistance path (via the cable screen) to one of the black points other than the one it is supposed to return to, in order for the current to take that path.. If all of the black points are connected together at a common star point and each signal ground goes to the common star point I'm not sure how this would be happening.... inserting the 1ohm resistors makes that path no longer more desirable...
Tony.
Attachments
Well, first of all, Inserting 1 ohm resistors in series with the screen on the signal cables going from X-over to amp-cards didn-t work. it actually made matters lightly worse.. so much for that..
Now back to You Tony,
The connection points at the coloured dots are done pretty much according to the schematic. The rectifiers are physically next to each other in two groups of three, so I've simply soldered the leads from the rectifiers together, about as short as it gets really.
It is of course unlikely that the diodes conduct exactly similarly, but the rectifiers are rated for 25 A (or thereabouts if I remember correctly), so there is little chance of one overloading due to imbalance.
The black dots are simply where the 0V wires to the capacitors are soldered to the rectifiers. (relatively short lengths of 6 square mm wires, i.e. thick ones)
I then interconnected the 0V points at the capacitors, something that was realised with 2 cm pieces of 6 square mm cable. All I had to do was to make two such bridges on the cards the capacitors are soldered to.
I hope this clarifies things?
Such a nuicance that my scanner isn't working with the new PC, a real pain not being able to scetch and scan to explain things clearly...
Anyway,
Been trying to understand this issue in small steps..
Individually, the amplifier cards work well. Short the input on one card, and there is no noise on its output.
Non- shorted , there is a clean sawtooth on the output.
Connecting the input cable screens toogether, there is the same clean sawtooth on the output of all 3 cards.
Short all 3 inputs and screens together, and there is only a very faint ripple on the outputs.
If there was some serious isue with the power-supply, I would expect to see noise appearing at this stage as ground currents could now start to flow between the amplifier cards through the signal grounds and signal inputs.
But when the screens are connected to the x-over card, where they become interconnected through the ground tracks of the card, and the signal wires are connected to the X-over card, things get messy. Perhaps this is because the signal screns do not connect at a single point but on points across the card?
Tomorrow I'll see if I can devise a set-up that allows me to experiment and measure a bit to shed some more light on this..
Now back to You Tony,
The connection points at the coloured dots are done pretty much according to the schematic. The rectifiers are physically next to each other in two groups of three, so I've simply soldered the leads from the rectifiers together, about as short as it gets really.
It is of course unlikely that the diodes conduct exactly similarly, but the rectifiers are rated for 25 A (or thereabouts if I remember correctly), so there is little chance of one overloading due to imbalance.
The black dots are simply where the 0V wires to the capacitors are soldered to the rectifiers. (relatively short lengths of 6 square mm wires, i.e. thick ones)
I then interconnected the 0V points at the capacitors, something that was realised with 2 cm pieces of 6 square mm cable. All I had to do was to make two such bridges on the cards the capacitors are soldered to.
I hope this clarifies things?
Such a nuicance that my scanner isn't working with the new PC, a real pain not being able to scetch and scan to explain things clearly...
Anyway,
Been trying to understand this issue in small steps..
Individually, the amplifier cards work well. Short the input on one card, and there is no noise on its output.
Non- shorted , there is a clean sawtooth on the output.
Connecting the input cable screens toogether, there is the same clean sawtooth on the output of all 3 cards.
Short all 3 inputs and screens together, and there is only a very faint ripple on the outputs.
If there was some serious isue with the power-supply, I would expect to see noise appearing at this stage as ground currents could now start to flow between the amplifier cards through the signal grounds and signal inputs.
But when the screens are connected to the x-over card, where they become interconnected through the ground tracks of the card, and the signal wires are connected to the X-over card, things get messy. Perhaps this is because the signal screns do not connect at a single point but on points across the card?
Tomorrow I'll see if I can devise a set-up that allows me to experiment and measure a bit to shed some more light on this..
Hi Elbert,
I would not bridge between the capacitors at all.
I would get an insulated bolt and run equal length heavy gauge wires (no need to go to extremes anything heavier than 8 gauge gets very difficult to work with, the gauge of the wires in your original pic looks fine) from the centre point on each cap bank to that bolt. Make these three the first three lugs on the bolt. I would then run three similar wires from your amp cards zero volts points to the bolt. Next (depending on how your speaker returns are done) run the three speaker returns wires to the bolt (if they go to the amp pcb you can ignore this). This should hopefully take care of any asymetry in the PS. The reason I suggested an insulated bolt is because generally your safety earth (chassis) conenction is close to the power in, and also you may need an earth loop breaker depending on what other equipment you are connecting to.
Now the next question is the connection from your x-over card to the amp card. My feeling is that you should connect the shield to the pcb at the xover end, but not connect the shield to the pcb on the amp end, don't leave it floating though, run an additional wire (and heat shrink at the amp) to the bolt from each amp end of the signal cable.
Hope that makes sense, I think that should minimise the possibility for unwanted current flow paths. I'm sure Andrew or DFX96 will jump in if I'm way off the mark!! You might want to wait till they comment before you try and modification like this.
Tony.
Yes that does and I have a suggestion
I would not bridge between the capacitors at all.
I would get an insulated bolt and run equal length heavy gauge wires (no need to go to extremes anything heavier than 8 gauge gets very difficult to work with, the gauge of the wires in your original pic looks fine) from the centre point on each cap bank to that bolt. Make these three the first three lugs on the bolt. I would then run three similar wires from your amp cards zero volts points to the bolt. Next (depending on how your speaker returns are done) run the three speaker returns wires to the bolt (if they go to the amp pcb you can ignore this). This should hopefully take care of any asymetry in the PS. The reason I suggested an insulated bolt is because generally your safety earth (chassis) conenction is close to the power in, and also you may need an earth loop breaker depending on what other equipment you are connecting to.
Now the next question is the connection from your x-over card to the amp card. My feeling is that you should connect the shield to the pcb at the xover end, but not connect the shield to the pcb on the amp end, don't leave it floating though, run an additional wire (and heat shrink at the amp) to the bolt from each amp end of the signal cable.
Hope that makes sense, I think that should minimise the possibility for unwanted current flow paths. I'm sure Andrew or DFX96 will jump in if I'm way off the mark!! You might want to wait till they comment before you try and modification like this.
Tony.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.