That is the INPUT to the PSU. It goes on the input side of the smoothing capacitors.
The Chassis to Main Audio Ground (MAG) can be in many locations.
What is much more important is the ROUTE that Fault Current needs to follow must not rupture any part of that route and thus prevent the mains fuse blowing if, or when, there is some catastrophic failure.
That route passes only interference current in normal operation and so should only create a small emi field. Controlling the location and length of that route should make the interfence field small enough to have no measureable consequence to the performance of the amplifier.
A short route to chassis that is well away from open loops in the input circuit would seem to be desirable.
Last edited:
Okay so I've been busy for the last hour or so with this..
I have put the now very tatty paint diagram together for 6 channels, I believe it should be exactly as it should now be?
From this I have applied it to the cad drawing (crudely), does everything look suitable?
Steel shielding okay?
IEC and centre tap grounds in suitable places?
Have I chosen the best method in minimising the untwisted parts to and from the caps?
Is it okay to "layer" the twisted triples into the amp of have them run along side each other for neatness?
It is okay to have the output triples to the amps running near or along side some of the input triple for neatness?
Preamp PCB with 50mm clearance above the caps suitable from rfi point of view?
General placement of everything seem reasonable?
Many thanks again.
Edit: I've forgotten then 15r and ground lift but I'd try and put it near the centre tap terminal.
I have put the now very tatty paint diagram together for 6 channels, I believe it should be exactly as it should now be?
From this I have applied it to the cad drawing (crudely), does everything look suitable?
Steel shielding okay?
IEC and centre tap grounds in suitable places?
Have I chosen the best method in minimising the untwisted parts to and from the caps?
Is it okay to "layer" the twisted triples into the amp of have them run along side each other for neatness?
It is okay to have the output triples to the amps running near or along side some of the input triple for neatness?
Preamp PCB with 50mm clearance above the caps suitable from rfi point of view?
General placement of everything seem reasonable?
Many thanks again.
Edit: I've forgotten then 15r and ground lift but I'd try and put it near the centre tap terminal.
Attachments
Last edited:
Looks good.
Now test it on a mono build on the bench.
Then test it in the chassis and only after passing those first two stages of testing, add on the extra channels.
I am still having trouble after adding a mono bench test into the chassis.
The noise goes up from 0.0mVac on the DMM to 0.2mVac to 0.3mVac.
I can see the problem on the scope. I can't isolate what is causing the problem.
I have changed the PSU.
I have changed the location of the PSU.
I have changed the transformer.
I have changed the location of the transformer.
I have tried the second amplifier channel.
The only "trick" that significantly reduces the scope pic and the DMM measured noise is to move the transformer+PSU out of the chassis and lay it on the floor 4' away and connect with a long triplet cable. That gets the noise down to 0.1mVac. That is still >6dB more than I expect from a good amplifier. 0.3mVac is >15dB more noise than a good amplifier. By comparison 0.049mVac is equivalent to Ein~11.5nV/rootHertz. 0.3mVac is equivalent to Ein~70nV/rootHertz.
I think I know where the problem is, but it will take extensive reworking of the PCB to prove whether I have guessed right.
I have been at this for over a month.
Now test it on a mono build on the bench.
Then test it in the chassis and only after passing those first two stages of testing, add on the extra channels.
I am still having trouble after adding a mono bench test into the chassis.
The noise goes up from 0.0mVac on the DMM to 0.2mVac to 0.3mVac.
I can see the problem on the scope. I can't isolate what is causing the problem.
I have changed the PSU.
I have changed the location of the PSU.
I have changed the transformer.
I have changed the location of the transformer.
I have tried the second amplifier channel.
The only "trick" that significantly reduces the scope pic and the DMM measured noise is to move the transformer+PSU out of the chassis and lay it on the floor 4' away and connect with a long triplet cable. That gets the noise down to 0.1mVac. That is still >6dB more than I expect from a good amplifier. 0.3mVac is >15dB more noise than a good amplifier. By comparison 0.049mVac is equivalent to Ein~11.5nV/rootHertz. 0.3mVac is equivalent to Ein~70nV/rootHertz.
I think I know where the problem is, but it will take extensive reworking of the PCB to prove whether I have guessed right.
I have been at this for over a month.
Last edited:
Really? I expected there to be at least something I have proposed that wasn't ideal/best practice?
"Is it okay to "layer" the twisted triples into the amp of have them run along side each other for neatness?
It is okay to have the output triples to the amps running near or along side some of the input triple for neatness?
Preamp PCB with 50mm clearance above the caps suitable from rfi point of view?"
Please could you advise on those points?
"Is it okay to "layer" the twisted triples into the amp of have them run along side each other for neatness?
It is okay to have the output triples to the amps running near or along side some of the input triple for neatness?
Preamp PCB with 50mm clearance above the caps suitable from rfi point of view?"
Please could you advise on those points?
Last edited:
Damn, I'm dreading problems such as yours which is why I'm trying to nail down as much as possible. I'm finally confident enough to start the build. Will post pictures when completed without wiring and amp boards (likely to be a few weeks).
I still haven't chosen what amplifier design to use as the site is swamped with so many.
For sense of scale this thing is HUGE. The transformer is 23kg rated at 5kVA supposedly continuous but I have doubts.
450mm W 522mm L 250mm H
I'm going for around 250W 8ohm and plan to almost double into 4 ohm.
Not quite enough heatsinking for all channels driven 100% but good for at least 70% before thermal cuts will take over.
I'm trying not to think about how ridiculous the weight will be.
I still haven't chosen what amplifier design to use as the site is swamped with so many.
For sense of scale this thing is HUGE. The transformer is 23kg rated at 5kVA supposedly continuous but I have doubts.
450mm W 522mm L 250mm H
I'm going for around 250W 8ohm and plan to almost double into 4 ohm.
Not quite enough heatsinking for all channels driven 100% but good for at least 70% before thermal cuts will take over.
I'm trying not to think about how ridiculous the weight will be.
Last edited:
I might try and get hold of some braided metal sleeving to thread the triplets into which I could terminate at the centre tap stud. Could well be more trouble than it's worth though. Besides using 1 single centre tapped toroid I'm trying to create a no compromise design.
Last edited:
What tech issues make (one shared capacitor bank) less desirable then (six individual capacitor banks)?
The need for a MONSTER high current rectifier bridge is the only negative I find, against several positives. My Krell KSA250 has one 4.2kVA transformer and uses one negative and one positive rectifier bridge without reliability issues.
matt09: "The transformer is 23kg rated at 5kVA supposedly continuous but I have doubts. 450mm W 522mm L 250mm H"
----------
Building a single shared-bank of capacitors, with a single metal ground plate connected directly to every power supply capacitor, with one-positive rectifier bridge and one-negative rectifier bridge attached to this ground plate, would allow any-or-all amplifiers to "time-share" transient power from the total-store.
If AMP_X is unused, the other AMPs can benefit from reduced charge demands on the total-store shared-bank.
If AMP_X drives a subwoofer with high current demands, and AMP_Y drives a low-demand side-ambience speaker, AMP_X can pull more charge from the total-store shared-bank.
-------------
What problems have been found with taking Bonsai's stereo wiring schematic and just attaching more channels to one common power supply, on one common chassis, with one common AC-plug?
The need for a MONSTER high current rectifier bridge is the only negative I find, against several positives. My Krell KSA250 has one 4.2kVA transformer and uses one negative and one positive rectifier bridge without reliability issues.
matt09: "The transformer is 23kg rated at 5kVA supposedly continuous but I have doubts. 450mm W 522mm L 250mm H"
----------
Building a single shared-bank of capacitors, with a single metal ground plate connected directly to every power supply capacitor, with one-positive rectifier bridge and one-negative rectifier bridge attached to this ground plate, would allow any-or-all amplifiers to "time-share" transient power from the total-store.
If AMP_X is unused, the other AMPs can benefit from reduced charge demands on the total-store shared-bank.
If AMP_X drives a subwoofer with high current demands, and AMP_Y drives a low-demand side-ambience speaker, AMP_X can pull more charge from the total-store shared-bank.
-------------
What problems have been found with taking Bonsai's stereo wiring schematic and just attaching more channels to one common power supply, on one common chassis, with one common AC-plug?
My first Power Amp build ~1978 used the manufacturer's wiring layout. They recommended a pair of rectifier+smoothing cap PSUs from the one set of centre tapped secondaries.
IT DOES NOT WORK ! But it took me about 15years to find out why.
The alternatives are
two sets of centre tapped secondaries to two PSUs,
or one set of centre tapped plus one PSU,
or one set of centre tapped secondaries feeding multiple rectifiers plus smoothing caps, with appropriate loop current attenuation.
The latter two options require the centre tap to be the SOLE PSU Zero Volts. Both versions can work.
The first option gives isolation between channels and isolation between the Zero Volts. This works.
IT DOES NOT WORK ! But it took me about 15years to find out why.
The alternatives are
two sets of centre tapped secondaries to two PSUs,
or one set of centre tapped plus one PSU,
or one set of centre tapped secondaries feeding multiple rectifiers plus smoothing caps, with appropriate loop current attenuation.
The latter two options require the centre tap to be the SOLE PSU Zero Volts. Both versions can work.
The first option gives isolation between channels and isolation between the Zero Volts. This works.
Last edited:
I got a significant improvement last night.
Changing to a 3rd transformer but without the DC blocking capacitors, I am consistently measuring noise @ 0.1mVac changing to 0.2mVac just occasionally. The encapsulated transformer is bigger and there was no room for the DC blocker on top of the transformer.
The scope confirms the slightly lower level of pulsing of the output noise signal.
This does seem to confirm that there is radiated emi that is getting into the amplifier.
Matt: 5kVA trafo...?! Wow, this will be a real monster then... 
If I might ask: what kind of environment it will be used in?
With this size it's almost capable to fill a huge movie room with SPL...
Regarding the size of this project I would really and strongly recommend to 1st
build just your modules and make your tests on the bench as Andrew told you.
It will be a horror to optimize this and rearrange your case when it's not perfect for the 1st try.
On the bench you can use even thin, flexible, long wires just to test the effect
of interaction between modules and distances: PS, amps, wire triplets, etc...
If there is a hum it's also good to localize it (moving the trafo far away few
meters for example) just to be sure wether it's caused by some grounding
issues or by the radiated field of this huge monster trafo.
Andrew: ooh, so you are dealing with a similar multiple channel layout just right now?
BTW: with my last dual mono layout amp I had the similar finding: there are 2 trafos
being the same in theory but one of them was humming and that channel produced a small
but noticeable hum as well could be heard close to the speakers. The mechanical humming
could be solved with a DC blocker circuit but the radiated field is still there causing some noise.
If I might ask: what kind of environment it will be used in?
With this size it's almost capable to fill a huge movie room with SPL...
Regarding the size of this project I would really and strongly recommend to 1st
build just your modules and make your tests on the bench as Andrew told you.
It will be a horror to optimize this and rearrange your case when it's not perfect for the 1st try.
On the bench you can use even thin, flexible, long wires just to test the effect
of interaction between modules and distances: PS, amps, wire triplets, etc...
If there is a hum it's also good to localize it (moving the trafo far away few
meters for example) just to be sure wether it's caused by some grounding
issues or by the radiated field of this huge monster trafo.
Andrew: ooh, so you are dealing with a similar multiple channel layout just right now?
BTW: with my last dual mono layout amp I had the similar finding: there are 2 trafos
being the same in theory but one of them was humming and that channel produced a small
but noticeable hum as well could be heard close to the speakers. The mechanical humming
could be solved with a DC blocker circuit but the radiated field is still there causing some noise.
Not quite. While I was investigating the turn off oscillation I noticed that the output noise changed now and again.
I didn't know why. Kept looking and discovered that when I set trigger to mains and set to 2mV/div I could lock onto and see the blipping superimposed on the noise.
I added the second channel to the PCB and it showed the same anomally.
At the moment I am testing with just one channel, which has been in and out of the chassis at least 15 times (I am not exaggerating) and trying to get to why there is this 2ms blip that repeats with the 100Hz mains recharge pulses into the smoothing capacitors.
The amps have been taken off their heatsinks at least a dozen times to try small modifications not related to this noise.
I am going to fit the second channel back in and see if the improvement I measured last night is the same for both channels.
It is not a hum, but it repeats at the same frequency as a hum.
Last edited:
One of the primary reasons I decided on separate rectifiers and capacitor banks was the ability to turn off each channel through using SCR's for rectification. I plan on having a 5 minute power down timer after which time it will take the gates of the SCR's for that channel low, cutting the power to the amp board and output transistors. I'm lazy and don't like to have to turn things on and off; the waste power on an amp like this through standing current would be enormous across 6 channels.
Each channel will have 22mF + 4.7mF per side anyway since the enclosure allows room for it. That's over 340mF in total and certainly more than enough for transient demand without needing to cap share. I based the uF/Watt figure on amps from Classe.
Matt: 5kVA trafo...?! Wow, this will be a real monster then...
If I might ask: what kind of environment it will be used in?
With this size it's almost capable to fill a huge movie room with SPL...
Regarding the size of this project I would really and strongly recommend to 1st
build just your modules and make your tests on the bench as Andrew told you.
I want this to be a reference amp I will keep for the rest of my life where the only upgrades will be re-capping / maintenance or swapping out the amp board for new designs. The power and size isn't so much to fill a huge room as have the capability to drive any set of speakers competently in the future. At the moment I'm only relatively young but will be using this amp on my 703's and MA GR floor standers.
Yes I will definitely be bench testing first.
Thanks
Last edited:
Can you share the scope shot maybe? And do you have any clue about the root cause now?
It's wiring, grounding, EM radiation?
Building a single shared-bank of capacitors might become more favorable after you study and develop your Protection Circuits, and decide how you will modularize 6-amplifiers. Do all amps have the same power(watts), OR a mix of high wattage(subwoofer) and low wattage amps.
DC-SERVO: If you decide to include an opamp based DC-Servo for each channel you will need +/- 15V on each channel, and this simplifies putting analog protection circuits on each channel PCB, PLUS one simple common Protection PCB to SUM faults and manage common soft-start and quick-stop for the main Transformer+caps.
Most soft-start plus protection logic boards have the functionality to execute a timed power-down. You can add a push button which turns the amp off after a defined time like 60minutes. There are also input signal detection circuits like Eliot sound "Signal Detecting Auto Power-On Unit Project 38" which can be added to the protection logic board, but a simple time delay is more dependable.
============
"Push a few buttons" power management idea.
(1) Install one mechanical switch on each amplifier which connects that amp to the main shared (+/-) DC power supply and the (+/-) 15V protection power supply. The user can switch in just two amplifiers for stereo, or all six channels for a HT session, or any other combination. Each amplifier is also always connected to a Common Sum "Protect and Control" logic board.
(2) One mechanical MAIN power switch turns on the low voltage "Protection and Control" logic power supply, which then relay connects the main transformer to the main AC, and starts the Soft-Start process. Easy to disconnect the main transformer here with a tme delay.
(3) The "Protection and Control" logic includes speaker current monitoring to protect one failed amp from destroying the shared powr supply, or burning down the house.
------
Your decision on including DC servo for each channel drops the first Domino in your FlowCharts.
DC-SERVO: If you decide to include an opamp based DC-Servo for each channel you will need +/- 15V on each channel, and this simplifies putting analog protection circuits on each channel PCB, PLUS one simple common Protection PCB to SUM faults and manage common soft-start and quick-stop for the main Transformer+caps.
Most soft-start plus protection logic boards have the functionality to execute a timed power-down. You can add a push button which turns the amp off after a defined time like 60minutes. There are also input signal detection circuits like Eliot sound "Signal Detecting Auto Power-On Unit Project 38" which can be added to the protection logic board, but a simple time delay is more dependable.
============
"Push a few buttons" power management idea.
(1) Install one mechanical switch on each amplifier which connects that amp to the main shared (+/-) DC power supply and the (+/-) 15V protection power supply. The user can switch in just two amplifiers for stereo, or all six channels for a HT session, or any other combination. Each amplifier is also always connected to a Common Sum "Protect and Control" logic board.
(2) One mechanical MAIN power switch turns on the low voltage "Protection and Control" logic power supply, which then relay connects the main transformer to the main AC, and starts the Soft-Start process. Easy to disconnect the main transformer here with a tme delay.
(3) The "Protection and Control" logic includes speaker current monitoring to protect one failed amp from destroying the shared powr supply, or burning down the house.
------
Your decision on including DC servo for each channel drops the first Domino in your FlowCharts.
I think I will use 6 seperate rectifiers from the one "control toroid" with 0-12 x2 to supply 6 sets of isolated DC for each amp board and all it's functions. This should avoid ground loops via the control circuitry? This includes a thyristor controlled sets of rails, preamp/ switchable bridge circuit, timed power down and SSR output protection with soft start. A thermal event will power down a master relay and cut the AC in to the main power toroid.
This is still easier than trying to place a common plate across 12 capacitors that cannot all be lined up in "twos". All amps will be identical.
Protection will be on the output side while the timeout power save will have to operate at the power input side to save on transistor static power loss so these will be seperate functions.
Thanks
This is still easier than trying to place a common plate across 12 capacitors that cannot all be lined up in "twos". All amps will be identical.
Protection will be on the output side while the timeout power save will have to operate at the power input side to save on transistor static power loss so these will be seperate functions.
Thanks
Last edited:
All,
I thought this was an unusual toroid with a centre tap but in actual fact it does in fact have separate secondary windings, i.e 4 cables rather than three. This would allow me use separate 0v's derived at the T of the capacitors rather than connecting it to the 0V of the transformer. It does however mean I now need 12 thyristor bridges and more space..
Are there any REAL world gains in changing the design to the above?
Also which one of the T's would I connect the PE to if I used this design as otherwise they would all join together through the PE point.
Thanks
I thought this was an unusual toroid with a centre tap but in actual fact it does in fact have separate secondary windings, i.e 4 cables rather than three. This would allow me use separate 0v's derived at the T of the capacitors rather than connecting it to the 0V of the transformer. It does however mean I now need 12 thyristor bridges and more space..
Are there any REAL world gains in changing the design to the above?
Also which one of the T's would I connect the PE to if I used this design as otherwise they would all join together through the PE point.
Thanks
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.