Hi
I'm building a power amp using the carcass/PS of one that died. I'm using Symasym5-3 kits that want +/- 20-45 DC power (schematic attached).
The PS of the carcass has DC 75 x 2 and I'm planning to use one for each channel.
However, the 75 DCV supplies are not referenced to ground - there is one cap between +/- from the rectifiers (and not two that are grounded at the connection between the two caps).
The DC voltage is within the acceptable range for the amp modules (ie +/- 37V if the voltage is notionally halved) but the +/- would effectively be referenced to ground through the amp circuit.
Is someone able to tell me whether this would be a problem? Or do I need to get new PS caps so I can incorporate a ground reference between two for each supply?
Thanks, Jon
I'm building a power amp using the carcass/PS of one that died. I'm using Symasym5-3 kits that want +/- 20-45 DC power (schematic attached).
The PS of the carcass has DC 75 x 2 and I'm planning to use one for each channel.
However, the 75 DCV supplies are not referenced to ground - there is one cap between +/- from the rectifiers (and not two that are grounded at the connection between the two caps).
The DC voltage is within the acceptable range for the amp modules (ie +/- 37V if the voltage is notionally halved) but the +/- would effectively be referenced to ground through the amp circuit.
Is someone able to tell me whether this would be a problem? Or do I need to get new PS caps so I can incorporate a ground reference between two for each supply?
Thanks, Jon
Attachments
Short answer: don´t suffer unnecessarily, get an ampboard that is happy with +/-75V , period.
Mind you, if you had a plain +75V power supply and nothing else, it might be worth building or maybe getting a capacitor output single supply 100W@4 ohm board, empty or stuffed, but as-is you are (mis)using a huge boat anchor, capable of hundreds of Watts, for a meager part of its capacity ... still having all of its disadvantages.
Even worse if you want to convert a split supply amp to be used there.
Mind you, it is technically possible, but very impractical.
Compare it to fitting a V8 engine to a grass mowing machine .... you can, but ......
Mind you, if you had a plain +75V power supply and nothing else, it might be worth building or maybe getting a capacitor output single supply 100W@4 ohm board, empty or stuffed, but as-is you are (mis)using a huge boat anchor, capable of hundreds of Watts, for a meager part of its capacity ... still having all of its disadvantages.
Even worse if you want to convert a split supply amp to be used there.
Mind you, it is technically possible, but very impractical.
Compare it to fitting a V8 engine to a grass mowing machine .... you can, but ......
There are few ways of handling this issue. The caps are not the biggest problem. You need a DC reference that is half the 75V, which you make ground, but you do not connect the speaker return to that ground. The speaker return goes to the junction of a couple large caps, which span the 75V in series. Similar is a bridged output. The virtual ground should not be used for internal bias currents. If you add an active voltage divider, you can connect the speaker normally, but the amplifier may overpower the divider causing a bias upset.
1. Replace R2 (22K) with two 47K to each filtered rail (C10 and C12 ).
2. Change R7 (22k) to 47K and move the grounded end to the filtered +rail at C10. Note that both sides of the 75VDC are floating.
Do not ground -75V ! The amp should follow the 2x47K divider floating the +75 to +37.5 and the negative to -37.5, which becomes the DC on each supply capacitor.
The two channels can share the two supply caps, but there are benefits to having them separate.
A +/-75V=150V total is a bad design decision and the reason you have to rebuild this amp, unless you are using MOSFETs and lots of them, ie a 700Watt+ amplifier. It's only reasonable for BJT outputs if you make a class H amp, which requires another +/-50V supply.
1. Replace R2 (22K) with two 47K to each filtered rail (C10 and C12 ).
2. Change R7 (22k) to 47K and move the grounded end to the filtered +rail at C10. Note that both sides of the 75VDC are floating.
Do not ground -75V ! The amp should follow the 2x47K divider floating the +75 to +37.5 and the negative to -37.5, which becomes the DC on each supply capacitor.
The two channels can share the two supply caps, but there are benefits to having them separate.
A +/-75V=150V total is a bad design decision and the reason you have to rebuild this amp, unless you are using MOSFETs and lots of them, ie a 700Watt+ amplifier. It's only reasonable for BJT outputs if you make a class H amp, which requires another +/-50V supply.
Thanks for the reply.
I take your point that the power supplies are probably surplus to requirement. I thought this might be an advantage in supplying peak demands.
The power supply is not +/- 75V. There are two completely separate transformers/rectifiers/supply caps that each put out 75V. Because they have separate transformers, I could configure them to +/- 75V but that's too much for my amp boards. What I wanted to do was use each supply for each channel.
In a Phase Linear amp, the +/- leads of the rectifier are connected to the +/- leads of two caps and the other -/+ leads of the two caps are grounded. The +/- leads of the two caps supply the + and - Bus voltages of about +/- 75V which appear to be referenced to ground by the grounding of the other poles of the caps. I have checked the schematic and now realise that the secondary of the power transformer has a centre tap that is grounded.
I thought that I might be able to use a similar arrangement to the Phase Linear so that the 75V becomes +/- 37V relative to ground. However, my transformers don't have a centre tap, so I'm wondering whether it is necessary to ground a centre tap in addition to a pole of each of the caps to get that result?
Otherwise, I would like to know is whether I can simply connect the +/- terminals of the amp board (ie +/- 45V max) to the 75V supply. The voltage difference is fine but the 75V hasn't been ground referenced to be +/- 37V. The 75V can be referenced however I like (if I ground the negative it will be 0/+75V, if I ground the positive, it will be -75/0V). Can I connect a floating power supply to an amp board that expects +/- voltages?
Cheers
I take your point that the power supplies are probably surplus to requirement. I thought this might be an advantage in supplying peak demands.
The power supply is not +/- 75V. There are two completely separate transformers/rectifiers/supply caps that each put out 75V. Because they have separate transformers, I could configure them to +/- 75V but that's too much for my amp boards. What I wanted to do was use each supply for each channel.
In a Phase Linear amp, the +/- leads of the rectifier are connected to the +/- leads of two caps and the other -/+ leads of the two caps are grounded. The +/- leads of the two caps supply the + and - Bus voltages of about +/- 75V which appear to be referenced to ground by the grounding of the other poles of the caps. I have checked the schematic and now realise that the secondary of the power transformer has a centre tap that is grounded.
I thought that I might be able to use a similar arrangement to the Phase Linear so that the 75V becomes +/- 37V relative to ground. However, my transformers don't have a centre tap, so I'm wondering whether it is necessary to ground a centre tap in addition to a pole of each of the caps to get that result?
Otherwise, I would like to know is whether I can simply connect the +/- terminals of the amp board (ie +/- 45V max) to the 75V supply. The voltage difference is fine but the 75V hasn't been ground referenced to be +/- 37V. The 75V can be referenced however I like (if I ground the negative it will be 0/+75V, if I ground the positive, it will be -75/0V). Can I connect a floating power supply to an amp board that expects +/- voltages?
Cheers
Hi Steve
Thanks for the input. Our posts crossed.
Would I be right in thinking that the 47K resistors replacing R2 connect between the Base of Q1 and filtered rails C10 and C12, respectively?
I see now that I cannot ignore the ground referencing of the rails.
Do you have any thoughts on the two caps for the supply filter grounded at one pole? They could possibly have resistors in parallel with them to help reference the supply rails to ground with a voltage divider.
It seems like there could be instability with possible solutions. The input into the circuit is class A. The draw on the +/- rails would be mostly symmetrical but there could be exceptions that might then produce feedback.
An option might be to use the carcass for another purpose that uses +/- 75V. Would I be better off thinking along those lines?
Cheers.
Thanks for the input. Our posts crossed.
Would I be right in thinking that the 47K resistors replacing R2 connect between the Base of Q1 and filtered rails C10 and C12, respectively?
I see now that I cannot ignore the ground referencing of the rails.
Do you have any thoughts on the two caps for the supply filter grounded at one pole? They could possibly have resistors in parallel with them to help reference the supply rails to ground with a voltage divider.
It seems like there could be instability with possible solutions. The input into the circuit is class A. The draw on the +/- rails would be mostly symmetrical but there could be exceptions that might then produce feedback.
An option might be to use the carcass for another purpose that uses +/- 75V. Would I be better off thinking along those lines?
Cheers.
You could have said this in the first post 😉
That makes it more reasonable.
1) if they are completely separated and can be grounded at will, yes, now you can get rid of the heavy, bulky but "still there" -75V supply, halving weight, bulk, etc.
Or simply having it connected to mains "doing nothing" but still powered.
2) you will have no advantage by using each +75V (after rewiring) supply for each channel, just one is already overpowered for both.
As estimated above, those can easily feed a 700W amplifier while I guess your amps will provide some 100W@4 ohm, tops, and maybe some 60W @ 8 ohm.
In any of these cases a single transformer + filtering will be just idling 🙂
3) the above suggestion of using 2 series capacitors between +75V and Ground is fine ... your amp will now be "output capacitor coupled" as much as if it were a classic single supply capacitor out type (just follow the speaker current path and you´ll see it always goes through a cap, one way or the other).
4) I suggest you build EXCELLENT and WELL PROVEN Apex Audio AX6
Hundreds successfully made, lots of data, PBs available, you can ask anything you want in its very active thread ... what´s not to like?
HAPPY with +75V 😉
https://www.diyaudio.com/community/threads/retro-amp-50w-single-supply.236256/
You can do what QSC did with some of their early power supplies - make a pseudo-ground for AC signals. This does turn the amplifier into a cap-coupled design, but not a problem if you just make the caps big enough. 22,000 uF is about right - since the effective value is half that. Too small and the “ground” bounces around all over the place when playing heavy bass.
Even though your amplifier topology does not require separate power supplies for each channel, I would if I were doing this. That will prevent any ground bounce from one channel from affecting the other.
Even though your amplifier topology does not require separate power supplies for each channel, I would if I were doing this. That will prevent any ground bounce from one channel from affecting the other.
Attachments
Yes, you could do the QSC circuit, but there is no ground connection between the QSC power supply and ground. The normal "output" is ground and the whole power supply is driven, the center of which is the speaker output. This simulation shows two possibilities using op-amps:
Attachments
Thanks for the comments.
wg_ski - that was the circuit I had in mind. I should have drawn a picture. I hadn't realised it had been used before, so that gives me some comfort.
JMF - I take your point that the return current from the speaker will need to pass through the supply caps. Presumably it wouldn't if the transformer had a grounded centre tap. I'll look into the AX6. Output coupling caps are something I remember from vintage units that had a 'warm' sound. I had thought that the sound went with the use of output coupling caps and I hadn't seen the circuits attracting interest in modern builds. Perhaps that's not the case if people are going back to them.
wg_ski - that was the circuit I had in mind. I should have drawn a picture. I hadn't realised it had been used before, so that gives me some comfort.
JMF - I take your point that the return current from the speaker will need to pass through the supply caps. Presumably it wouldn't if the transformer had a grounded centre tap. I'll look into the AX6. Output coupling caps are something I remember from vintage units that had a 'warm' sound. I had thought that the sound went with the use of output coupling caps and I hadn't seen the circuits attracting interest in modern builds. Perhaps that's not the case if people are going back to them.
Hi Steve
Our posts crossed again. Thanks for the analysis. However, the op-amp simulation is a little above my pay grade. Are you saying the QSC power supply circuit has significant limitations that point away from using it with the type of amp circuits I have in mind?
Cheers
Our posts crossed again. Thanks for the analysis. However, the op-amp simulation is a little above my pay grade. Are you saying the QSC power supply circuit has significant limitations that point away from using it with the type of amp circuits I have in mind?
Cheers
The QSC circuit requires a separate PS for each channel, but that is not a problem for you, right? Another advantage of the QSC circuit was that the drivers need not span the supply rails, so an op-amp IPS was a great fit. But I don't know of any QSC style kits.
The QSC amplifier circuit requires separate power supplies for each channel. The older models also used the power supply in post #7. Newer ones used a conventional center tapped transformer - that was a requirement for the class H models.
The amp circuit and the supply with the pseudo ground don’t necessarily have to be used together, they just happened to be in the older QSC amps. That same power supply will work with most amplifiers, but it does effectively convert it to a cap-coupled design. The reason QSC used it was the effective cap coupling. A shorted output transistor would not send DC to the speaker.
A regular cap coupled design can be simpler, and is easier to understand. The circuit in post 6 is a lot easier to build and tolerant of mistakes. I’d increase the output cap to 10,000 uF, and use the same in the power supply, but thats me and my bass talking.
The amp circuit and the supply with the pseudo ground don’t necessarily have to be used together, they just happened to be in the older QSC amps. That same power supply will work with most amplifiers, but it does effectively convert it to a cap-coupled design. The reason QSC used it was the effective cap coupling. A shorted output transistor would not send DC to the speaker.
A regular cap coupled design can be simpler, and is easier to understand. The circuit in post 6 is a lot easier to build and tolerant of mistakes. I’d increase the output cap to 10,000 uF, and use the same in the power supply, but thats me and my bass talking.
Sorry but this is not what QSC does.
They ground the speaker out and supply caps centerpoint is actually "hot" speaker out.
Of course you can not ground that point under penalty of shorting speaker out.
Sorry, it does not work as shown, see above.
In any case, nothing wrong with using "old style" capacitor coupled amps, in fact there are a few modern versions (Yamaha comes to mind, I was surprised by such a circuit short ago), main reason for them not being popular anymore (in my view) is that large high voltage caps fpr main filter are relatively scarce and expensive,, that´s a deal killer.
I should check Mouser, I bet 1 100V cap for +V plus a 50V one for output are more expensive than two 50V ones for split supply, in a same performance amplifier.
There´s also the perceived advantage of split supply amps reaching DC or very low frequencies if you wish, not a real deal maker/breaker in the Real World , until speakers also reach DC, or at least 5-10Hz flat, but hey .... 😉
Either works equally “well”. Same strengths and weaknesses apply. Without the center tap connection you are at the mercy of voltage division between the two caps. For DC, the bleeder resistors enforce balance. For AC, the caps’ impedance needs to be “low” or that center connection moves around. It was good enough for 1980’s-vintage QSC’s. But they were not considered Uber-hi-fi, just fairly reliable.
I dont have anything against cap coupled amps per se. My biggest beef is the turn on/off thumps which can be substantial. Build one at 600 watts and it can launch a 4“ woofer cone into the middle of next week, not to mention what it could do to a compression driver. If you go and add a muting relay, you’re sort of defeating the purpose of going cap coupled in the first place (DC fault immunity withOUT a relay). I build them, but keep them on the smaller side. Biggest I have is 150 wpc, but it wont get used with wimpy speakers because it does thump. The others are all 40w and under. A split supply amp built with a pseudo-ground would tend to produce less of that nonsense. If the caps get out of balance enough as they age, you will start picking up turn on/off transients.
I dont have anything against cap coupled amps per se. My biggest beef is the turn on/off thumps which can be substantial. Build one at 600 watts and it can launch a 4“ woofer cone into the middle of next week, not to mention what it could do to a compression driver. If you go and add a muting relay, you’re sort of defeating the purpose of going cap coupled in the first place (DC fault immunity withOUT a relay). I build them, but keep them on the smaller side. Biggest I have is 150 wpc, but it wont get used with wimpy speakers because it does thump. The others are all 40w and under. A split supply amp built with a pseudo-ground would tend to produce less of that nonsense. If the caps get out of balance enough as they age, you will start picking up turn on/off transients.
Attachments
In the 1 or 2 output transistor class (60-100 watts/channel) turn on thump is NOT A PROBLEM in my ST120 or my AX6 upgrade to that chassis. Plus I abandoned 4" bass drivers about 1971. Both my early speaker purchases, when I was making $3 an hour, had 10" woofers: for real bass. Plus my AX6 has no "warm" sound. It does have a regulated power supply, ~75 v regulated to 69 v by 5 parallel TIP147 darlington transistors. AX6 sounds exactly the same at 1/4 to 60 watts as my CS800s which is split supply direct coupled to speaker with a 30 part DC protection circuit.
Advantage of the speaker cap to newbies, is that if a solder joint pops loose, DC burnout or rips of the bass driver suspension are prevented by a $3 capacitor - one part. The extra $.50 of a 80v cap versus a 63 v cap is trivial when it takes $10 to get a box or parts shipped in. Pennies matter when a company is trying to sell 10000 products, not to the home hobbiest. Biggest upgrade to my ST120 I did was putting in a MUCH BIGGER heatsink, to avoid O.T. meltdowns that were so common on that product.
Splitting the single supply with a 2.2 k or 3.3 k resistor divider is pretty wiimpy. Speaker ground wanders all around. As that happens speaker current is limited which may have its own sound. When Frank Beard whaps that bass drum you want a big current surge to the speaker. AX6 senses the center voltage and pushes it towards the middle with feedback to the input transistor , at a cost of a couple of resistors.
Page 21 of thread quoted above shows my build, which was on a point to point board because the standard AX6 board won't fit in a ST120 chassis. Too tall. Plus all sources of PCB's I'm aware of are in a country I don't buy things from. Unfortunately all ON plastic transistor production has been moved there. ST transistors are 100% from there, even the TO3 packages. However 12 transistors (two channels) are a lot cheaper than a whole amp from that country.
Phase Linear of the 1970's was an amp with a bad reputation. Great power, good sound, lots of failures on stage. The output transistors of those days had really bad gain, 2 to 5, is why the +-75 volts was necessary. SOA testing hadn't been invented yet (by Motorola) to predict transistor meltdowns. Later class AB amps with +-80 power supply like my PV-1.3k used 10 output transistors per side, a 10"x2" heat sink, a high flow fan, and much tougher transistors with higher gain (MJ15024/25) than were available in 1970's.
Last edited:
Quad use an active ground in the 306, 606 etc
http://quadrevisionspot.blogspot.com/2009/12/quad-306-monoblock-virtual-earth.html
Stuart
http://quadrevisionspot.blogspot.com/2009/12/quad-306-monoblock-virtual-earth.html
Stuart
True.
There is also a similar 70´s Revox 40W amp which drove me crazy back in the day because it had split supply but no transformer center tap.
Both killer amps but quite away from mainstream.
Really the easy/practical/tested solution is a classic amp specifically built for single supply and with a plain classic output capacitor.
No surprises.
There is also a similar 70´s Revox 40W amp which drove me crazy back in the day because it had split supply but no transformer center tap.
Both killer amps but quite away from mainstream.
Really the easy/practical/tested solution is a classic amp specifically built for single supply and with a plain classic output capacitor.
No surprises.
I used the split supply using transistors to get the ground for a circuit I designed.
I didnt want to use two wall warts so used one 24 volt DC supply and split it into two 12v supplies.
Worked well, but I dont know how well it would work for a power amp that takes a lot more current.
I would LTSPICE it to be sure.
The resistor divider isn’t what really splits it - it’s the job of the caps which handle the amps of ripple current. In that old QSC I had, the “ground”, or rather speaker out, did bounce all around with the woosy 3300 uF caps. Put a zero on the end and presto - no more bounce. The resistor divider just enforces voltage balance for DC - same as when you stack caps for voltage when making 600 volt supplies.
I wouldn’t do this with anything putting out a crazy amount of power. But it is a workable solution for an amp which wants split supplies and draws a symmetrical quiescent current.