ChrioN
Well-known member
Is it possible to get a +12v 0 -12v power supply with a "single tapped" transformer? Using 7812 regulators or whatever..
Making a +/- power supply out of a 0-24vac transformer?
- Thread starter ChrioN
- Start date
Help Support GroupDIY Audio Forum:
ChrioN
Well-known member
50mA maximum. probably around 20mA
Svart
Well-known member
http://www.groupdiy.com/index.php?topic=32448.0
> http://www.groupdiy.com/index.php?topic=32448.0
Well, that Alesis plan starts with 9 VAC or 14VAC, and ChrioN apparently only has one 24 VAC winding.
The Alesis goes on to double-double the 9VAC to give 48VDC as two 24VDC points around the same center.
Done-over with 24VAC, with the double-doubler, he'd get about 128V total DC, which is is excess of requirement. Or with the simple doubler, about 64VDC as two 32VDC rails, almost three times more than the 12V+12V=24VDC he needs.
It seems silly to waste 2/3rds of your voltage. It triples the heat in the box. (Yes, switchers can transform voltage with low heat; but if ChrioN has only one 24VAC winding he may not have a full assortment of switchers handy.)
Waste is inevitable. The question becomes: how much do you need? The answer will let you figure if the waste is acceptable or obscene.
The obvious thing to do with 24VAC is rectify it onto a cap for about 35VDC. This is only 150% of required voltage. Loss in power supply is less than full-load heat in working-parts.
However, with a center-tap, 24VAC CT gives +/-17VDC, which is just-maybe enough to keep LM317 etc in regulation. And without a CT, there is no hard "center", which is usually expected when someone asks for " +12v 0 -12v power". There may be a way to have two 3-pin regulators in a tug-of-war and derive a "center". You can certainly do it with an additional amplifier, with additional complexity and sonic corruption. I do not like "active grounds", certainly not in a build which is apparently cost-constrained. (If ChrioN were building with gold-plated amplifiers and vellum capacitors, he would not strain at the cost of one of the many fine power supplies available.)
The most-simple "solid" +/-12V supply, with the "24VAC (no CT)" constraint, is to flood a maximum current through a couple 12V Zeners. The no-load Zener current must be larger than the actual load. This is a poor plan for "probably", it would be best (and cool) to know the actual load. But ChrioN's "50mA-20mA" would work with a bit over 55mA in the Zener string. Each Zener throws over 0.5W but less than 1W of heat. 1W Zeners are readily available and cheap. The resistor must flow a bit over 55mA with the voltage difference between 35V raw and 24V final. 11V/55mA= 0.2K= 200 ohms. Heat is 55mA*11V or 0.6 Watts. Double the steady dissipation for safety, round up, buy a 2W resistor.
If he pulls much less than 50mA, the voltage will not rise much over 12V. If he pulls 60mA, it will sag to 11V, but still run. If he applies a dead-short, only the $0.50 resistor smokes. (Indeed a $1 10W resistor would make it utterly short-proof.)
The total dissipation (including load) is 2 Watts. Considering the load is "probably" 24V*20mA= a half-watt, we could do a whole lot better. But 2W is not a huge strain on the planet. And unless he runs it 24/7 for years, the total energy to transport a nice 9VAC part over the seas and up the road to ChrioN's house is probably greater than the "waste" of using the "wrong" part.
HeathKit used similar plans a LOT. It works, and it has a very low solder-joint count.
Well, that Alesis plan starts with 9 VAC or 14VAC, and ChrioN apparently only has one 24 VAC winding.
The Alesis goes on to double-double the 9VAC to give 48VDC as two 24VDC points around the same center.
Done-over with 24VAC, with the double-doubler, he'd get about 128V total DC, which is is excess of requirement. Or with the simple doubler, about 64VDC as two 32VDC rails, almost three times more than the 12V+12V=24VDC he needs.
It seems silly to waste 2/3rds of your voltage. It triples the heat in the box. (Yes, switchers can transform voltage with low heat; but if ChrioN has only one 24VAC winding he may not have a full assortment of switchers handy.)
Waste is inevitable. The question becomes: how much do you need? The answer will let you figure if the waste is acceptable or obscene.
The obvious thing to do with 24VAC is rectify it onto a cap for about 35VDC. This is only 150% of required voltage. Loss in power supply is less than full-load heat in working-parts.
However, with a center-tap, 24VAC CT gives +/-17VDC, which is just-maybe enough to keep LM317 etc in regulation. And without a CT, there is no hard "center", which is usually expected when someone asks for " +12v 0 -12v power". There may be a way to have two 3-pin regulators in a tug-of-war and derive a "center". You can certainly do it with an additional amplifier, with additional complexity and sonic corruption. I do not like "active grounds", certainly not in a build which is apparently cost-constrained. (If ChrioN were building with gold-plated amplifiers and vellum capacitors, he would not strain at the cost of one of the many fine power supplies available.)
The most-simple "solid" +/-12V supply, with the "24VAC (no CT)" constraint, is to flood a maximum current through a couple 12V Zeners. The no-load Zener current must be larger than the actual load. This is a poor plan for "probably", it would be best (and cool) to know the actual load. But ChrioN's "50mA-20mA" would work with a bit over 55mA in the Zener string. Each Zener throws over 0.5W but less than 1W of heat. 1W Zeners are readily available and cheap. The resistor must flow a bit over 55mA with the voltage difference between 35V raw and 24V final. 11V/55mA= 0.2K= 200 ohms. Heat is 55mA*11V or 0.6 Watts. Double the steady dissipation for safety, round up, buy a 2W resistor.
If he pulls much less than 50mA, the voltage will not rise much over 12V. If he pulls 60mA, it will sag to 11V, but still run. If he applies a dead-short, only the $0.50 resistor smokes. (Indeed a $1 10W resistor would make it utterly short-proof.)
The total dissipation (including load) is 2 Watts. Considering the load is "probably" 24V*20mA= a half-watt, we could do a whole lot better. But 2W is not a huge strain on the planet. And unless he runs it 24/7 for years, the total energy to transport a nice 9VAC part over the seas and up the road to ChrioN's house is probably greater than the "waste" of using the "wrong" part.
HeathKit used similar plans a LOT. It works, and it has a very low solder-joint count.
![Soldering Iron Kit, 120W LED Digital Advanced Solder Iron Soldering Gun kit, 110V Welding Tools, Smart Temperature Control [356℉-932℉], Extra 5pcs Tips, Auto Sleep, Temp Calibration, Orange](https://m.media-amazon.com/images/I/51sFKu9SdeL._SL500_.jpg)
ChrioN
Well-known member
Thank you so much for your time!
Svart
Well-known member
well I didn't mean for him to straight copy it, I meant it simply as a point of research..
ChrioN
Well-known member
A new pretty important question popped up; Is there enough filtration in this circuit for a headphone amplifier? Don't want excessive noise...
SSLtech
Well-known member
Headphone amplifiers... PRR is the absolute LORD of all questions related to that subject...
But he may be better served with some information: what impedance, and how loud does it have to go? -Also, is it always going to be the SAME pair of headphones? -or will different engineers be plugging in their favourite sets on different days?
The reason which I ask is that all the aformentioned conditions are significantly related to current consumption in my mind, and thus pertinent to the supply load...
..But I'd trust a reply from PRR better than from me. -If you can supply the answers to those questions, I think he'd give you a trustworthy reply.
Keith
But he may be better served with some information: what impedance, and how loud does it have to go? -Also, is it always going to be the SAME pair of headphones? -or will different engineers be plugging in their favourite sets on different days?
The reason which I ask is that all the aformentioned conditions are significantly related to current consumption in my mind, and thus pertinent to the supply load...
..But I'd trust a reply from PRR better than from me. -If you can supply the answers to those questions, I think he'd give you a trustworthy reply.
Keith
ChrioN
Well-known member
The circuit has a gain of 8, so it can go reasonably loud. 32-300 ohms cans will be used.
rlaury
Well-known member
Hey Group
Here's a circuit i've been using for several years in racking modules ect.
It uses a single secondary for Bipolar and phantom output.
You can leave off the phantom if it's not needed.
Power output is determined by the regulator and the heatsinks
and the transformer rating. Adjust R1 & R2 for about 4 volt headroom
under full load and the lowest primary voltage you might ever expect under
normal operation. Also use good wire-wound resistors for R1, R2, R3.
I'll leave this up for a few days.
RonL
http://www.nashaudio.com/public/PowerSupply.pdf
EDIT:
PM me for schematic.
Here's a circuit i've been using for several years in racking modules ect.
It uses a single secondary for Bipolar and phantom output.
You can leave off the phantom if it's not needed.
Power output is determined by the regulator and the heatsinks
and the transformer rating. Adjust R1 & R2 for about 4 volt headroom
under full load and the lowest primary voltage you might ever expect under
normal operation. Also use good wire-wound resistors for R1, R2, R3.
I'll leave this up for a few days.
RonL
http://www.nashaudio.com/public/PowerSupply.pdf
EDIT:
PM me for schematic.
> If you can supply the answers to those questions
Thanks, but ChrioN can only post one maybe two facts at a time, and I won't play 20 Questions.
On the facts so far:
> 32-300 ohms cans
24V supply on two 32-ohm loads can be 0.75 Amps peak, and full-power sine could be nearly a quarter-Amp average. That's ASS-uming a simple AB output stage... if class A, the sky is the limit (actually your heatsink budget is the limit).
Yes, such levels in headphones would probably blow your brains out through your nose. Whether this is desirable or not is a fact I do not have.
If the fact is that users might come anywhere close to brain-blowing levels in lo-Z phones, then you will want 1,000u caps across the Zeners. The proposed plan will deliver "50mA", but NO more (maybe 55mA), not even for an instant. Which is a safety feature for builders who may not know their true demand; it is more likely to punk-out than to smoke stuff. And while the caps won't support a full power sine test, on speech/music signal they will probably support all you could expect from +/-12V supply.
> has a gain of 8, so it can go reasonably loud.
Does not follow. Gain is not loudness until you know source. If the source is a 10mV low-line-level, it won't be real loud. If the source is a +18dBm bus, gain less than unity will be "loud" in 32 ohms.
My own HP amp has maximum gain of 50, smoothly controllable to less than unity, so that I can tap anywhere in an audio system and bring it to desired level. This includes bringing "silence" (room rumble and mike hiss) up to medium-loud so I can judge noise quality. Not normal listening.
FWIW, both my HP amps start from a single 24VAC winding. But I did not then arbitrarily throw a "+/-12V" spec on it.
The older one rectifies 24VAC to 36VDC, drops 3V in an R-C filter, and dumps that on BIG resistors to the collectors of a couple TIP120 transistors and the + pin of a TL072 op-amp. This is ample filtering for that topology. The box hums faintly because the PT is 2 inches from the input jacks.
This one runs capacitor-output. The collectors sit 12V above common and that must be blocked. A 470u 16V cap is not expensive. Some may say that cap-coupling is bad... well, there is NO way to mix DC and AC for a wall-powered amplifier without either a cap or a choke in the path somewhere. In "capacitorless" designs it typically lurks in the power supply, and its flaws may be reduced by NFB. Although in many real cases, capacitor currents dumping into signal traces do more harm than a simple cap-coupled output. And when "capacitorless" outputs blow short, they usually take a speaker with them... cap-coupled usually doesn't.
The other one starts with a saggy 24VAC wall-wart, doubles that to +/-31VDC. This raw voltage is applied to the several power transistor collectors. It also runs through 3906-like transistors with 2*9V Zeners to give +/-17.4V (actually +17.2V/-17.6V; a little unbalance should never hurt) to run opamps, then diode-droppers to give +/-16.8V on the second stage (a work-around for the phase catastrophe of overdriven Bi-FET opamps).
I don't feel like regulating Main Power when it is not needed. If you hold the base of a PN3055 or TIP120 firmly, inside overall NFB, fairly raw power does not ripple the output; indeed this connection has roughly the same ripple rejection as a regulator. The heatsink I wudda needed on a regulator can be added to the power transistor, saving pennies, parts, and work. Then the little stuff which could not stand 62V supply can be regulated with a 13-cent plastic transistor.
The second amp has considerably more output voltage than something run on "+/-12V", yet will not cause nose-bleed on some 300 ohm phones. It is not uncommon for serious users to want +/-18V even +/-24V to completely eliminate transient clips on some 300r cans. So I'm thinking, if you have 24VAC=35VDC, you may want to use it all. There are plenty of 36V chips. However lacking a center-tap or some other elegant way to get a mid-point, you will need a mid-reference and a capacitor-coupled output. Non-elegant ways include an "passive rail splitter" which leads to DC offset and low bass separation, or "active ground channel" which demands an amp twice as big as either channel and adds unnecessary cross-channel intermodulation.
Thanks, but ChrioN can only post one maybe two facts at a time, and I won't play 20 Questions.
On the facts so far:
> 32-300 ohms cans
24V supply on two 32-ohm loads can be 0.75 Amps peak, and full-power sine could be nearly a quarter-Amp average. That's ASS-uming a simple AB output stage... if class A, the sky is the limit (actually your heatsink budget is the limit).
Yes, such levels in headphones would probably blow your brains out through your nose. Whether this is desirable or not is a fact I do not have.
If the fact is that users might come anywhere close to brain-blowing levels in lo-Z phones, then you will want 1,000u caps across the Zeners. The proposed plan will deliver "50mA", but NO more (maybe 55mA), not even for an instant. Which is a safety feature for builders who may not know their true demand; it is more likely to punk-out than to smoke stuff. And while the caps won't support a full power sine test, on speech/music signal they will probably support all you could expect from +/-12V supply.
> has a gain of 8, so it can go reasonably loud.
Does not follow. Gain is not loudness until you know source. If the source is a 10mV low-line-level, it won't be real loud. If the source is a +18dBm bus, gain less than unity will be "loud" in 32 ohms.
My own HP amp has maximum gain of 50, smoothly controllable to less than unity, so that I can tap anywhere in an audio system and bring it to desired level. This includes bringing "silence" (room rumble and mike hiss) up to medium-loud so I can judge noise quality. Not normal listening.
FWIW, both my HP amps start from a single 24VAC winding. But I did not then arbitrarily throw a "+/-12V" spec on it.
The older one rectifies 24VAC to 36VDC, drops 3V in an R-C filter, and dumps that on BIG resistors to the collectors of a couple TIP120 transistors and the + pin of a TL072 op-amp. This is ample filtering for that topology. The box hums faintly because the PT is 2 inches from the input jacks.
This one runs capacitor-output. The collectors sit 12V above common and that must be blocked. A 470u 16V cap is not expensive. Some may say that cap-coupling is bad... well, there is NO way to mix DC and AC for a wall-powered amplifier without either a cap or a choke in the path somewhere. In "capacitorless" designs it typically lurks in the power supply, and its flaws may be reduced by NFB. Although in many real cases, capacitor currents dumping into signal traces do more harm than a simple cap-coupled output. And when "capacitorless" outputs blow short, they usually take a speaker with them... cap-coupled usually doesn't.
The other one starts with a saggy 24VAC wall-wart, doubles that to +/-31VDC. This raw voltage is applied to the several power transistor collectors. It also runs through 3906-like transistors with 2*9V Zeners to give +/-17.4V (actually +17.2V/-17.6V; a little unbalance should never hurt) to run opamps, then diode-droppers to give +/-16.8V on the second stage (a work-around for the phase catastrophe of overdriven Bi-FET opamps).
I don't feel like regulating Main Power when it is not needed. If you hold the base of a PN3055 or TIP120 firmly, inside overall NFB, fairly raw power does not ripple the output; indeed this connection has roughly the same ripple rejection as a regulator. The heatsink I wudda needed on a regulator can be added to the power transistor, saving pennies, parts, and work. Then the little stuff which could not stand 62V supply can be regulated with a 13-cent plastic transistor.
The second amp has considerably more output voltage than something run on "+/-12V", yet will not cause nose-bleed on some 300 ohm phones. It is not uncommon for serious users to want +/-18V even +/-24V to completely eliminate transient clips on some 300r cans. So I'm thinking, if you have 24VAC=35VDC, you may want to use it all. There are plenty of 36V chips. However lacking a center-tap or some other elegant way to get a mid-point, you will need a mid-reference and a capacitor-coupled output. Non-elegant ways include an "passive rail splitter" which leads to DC offset and low bass separation, or "active ground channel" which demands an amp twice as big as either channel and adds unnecessary cross-channel intermodulation.
ChrioN
Well-known member
PRR said:
Offcourse, but, as you said, thats very loud into 32z. And pretty loud into 300z too. Its all relative though.
PRR said:
How come the need to capacitor couple the output when running on a virtual ground? Thanks!
edit: oh, you mean to get protection from dc-offset?
ChrioN
Well-known member
What happens when doing this instead?:
Svart
Well-known member
http://www.linear.com/designtools/software/ltspice.jsp
LTspice.
This way you can see for yourself what would happen. You can't beat a free simulation tool when you simply want to play around to see what would happen.
;D
LTspice.
This way you can see for yourself what would happen. You can't beat a free simulation tool when you simply want to play around to see what would happen.
;D
CJ
Well-known member
single tapped, or center tapped?
ChrioN
Well-known member
Single tapped transformer, +0- dc output.
I actually simmed that circuit, with a very simple program, it looked good but don't know if it'll work in practice...
I actually simmed that circuit, with a very simple program, it looked good but don't know if it'll work in practice...
CJ
Well-known member
sometimes it is best to spend a few bucks more for a xfmr than chase a mis app with parts, suffer the possible problems down the line.
and it might actually save time.
heck, you can get on digi key and charge up a pwr supply, one of those self imploding condors, just rip out the regulator board, save the 2n2055 neve transistor from mexico, and then add a plus and minus reg.
or wait for the "power one" to break, then fix it to a pure analog, bullet proof design.
you want your power supply as simple as possible, as it is responsible for 90 percent of all electronic problems. look at HP, their power transformers are so over built that you can short out all the tube sockets and it won't even blink.
barley even gets warm with a healthy chassis.
so design your supply like a locomotive, and you will have years and years of trouble free service.
zeners are ok, but they can be noisy, and they are running like a car with no oil, it is only a matter of time before the avalanche boundrys start to deterioate, the voltages drift apart, the...
oh, never mind, half the fun of electronics is building weird circuits, you learn the most thaqt way.
and it might actually save time.
heck, you can get on digi key and charge up a pwr supply, one of those self imploding condors, just rip out the regulator board, save the 2n2055 neve transistor from mexico, and then add a plus and minus reg.
or wait for the "power one" to break, then fix it to a pure analog, bullet proof design.
you want your power supply as simple as possible, as it is responsible for 90 percent of all electronic problems. look at HP, their power transformers are so over built that you can short out all the tube sockets and it won't even blink.
barley even gets warm with a healthy chassis.
so design your supply like a locomotive, and you will have years and years of trouble free service.
zeners are ok, but they can be noisy, and they are running like a car with no oil, it is only a matter of time before the avalanche boundrys start to deterioate, the voltages drift apart, the...
oh, never mind, half the fun of electronics is building weird circuits, you learn the most thaqt way.
ChrioN
Well-known member
Thanks CJ. The headphone amp I want to power only has a 2 conductor power inlet, thats why my needs are specific :-\
