# tailor output voltage of power supply

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.

#### Evaas

Hi All

I'm building a cross of a Pass Aleph 4 and 5. Anyhow I want a rail voltage of 30 to 35 V DC. For my filter I was just planning on using tons of caps. However, I don't have a transformer yet, but I found a good deal on 33v transformers. If I just use the caps, it would be around 46V. My question is can I use a choke to limit the voltage? Whats the formula to calculate the voltage drop. Also, how does this change when its used in a pi filter (i.e. how is an inductor fed by the transformer different from an inductor fed by a cap which is in turn fed by the transformer) Finally, what is the power loss for such an inductor. I want an efficient psu, and don't want to waste any more power then I have to.

Thanks for any help you can give, as I've searched the site and found nothing to help so far. I also want to brush up on my formulas, but I'm afraid that they'll be sold out before I figure it out on my own.

Evan

#### GRollins

Evan,
All things being equal (they never are), a choke input will give you something on the order of .9x your secondary voltage. Not quite what you had in mind. Granted, you can scale the inductor, but...
Keep in mind is that high current inductors are hard to come by.
Another point is that the current draw of the circuit will pull down the rail a bit. How much? Depends on the current capability of the transformer vs. the current draw.
Nelson has stated that the Aleph circuit is pretty scalable. You might consider running your circuit at the higher voltage, but pulling back the idle current to keep device dissipation under control.
Or you could go ahead and shoot for the 100W target, since the Aleph 2 uses a 45V rail.

Grey

#### Evaas

Grey, when you say .9x the secondary voltage, I assume you mean the RMS voltage. That would be 33*.9 = 29.7 V. A little low for me, but it might be ok. For my design, I want to be able to idle as much as 1A bias through my four output transistors. I will add the ability to change the bias so that it will produce less heat in the summer, and because I'm not exactly sure of my surplus heatsink's thermal resistance (done some approximations based on area and other ones that are similar). I think my target will be more like .7A but I want to leave a little margin.
So... at .7A and 35V, each transistor is dissipating 24.5W. I'd go up to 35W each, but thats pushing it and I'm not sure how hot it will get. My problem is that I'd prefer to reduce the voltage (and output power) than to reduce the bias current, which will have a bigger impact on the sound quality. Therefore I don't want to go above 35v. Well ok, so if I use 46V rails and .7A bias, the power per device would be 32W, which is still acceptable. However if it runs too hot then I'll have to lower the bias. If I use .5A (the lowest that any Aleph uses) then I'll dissipate 23W, which I'm confident will not be too hot. However, the only designs that bias at .5A also use six devices, which serves to reduce the distortion that is addes with lower bias. If I add more devices then I increase my power again... So basically I'd be ok using 46V rails if I could be sure of my heat sink. Oh well maybe its time to actually test the darned thing...

oh yeah.. I'm also leaning more towards Aleph 5 specs (which I think is around 33v) as NP claimed that had the punchiest bass, and I do enjoy that sort of thing

Thanks again,

Evan

#### Evaas

oops!

Looks like I did more spewing of the mouth than thinking..

if the choke gives .9 times the secondary voltage, does refer to the rectified voltage? if there's still some sinusoidal component (which there is, but I don't know how much), then the fully filtered component would be higher. Or does that .9 refer to the voltage after filtering?

#### rljones

Evaas,

33VAC will rectify to about 46VDC (_roughly_ 1.4 x VAC; depending upon which type of diodes you use, some lose more volts/diode than others) with a capacitor to ground immediately after the rectifier diodes as in NP's schematics.

If you run the rectified DC directly into the inductor (before using a capacitor to ground), you'll lose more volts, than if you have a capacitor before the inductor. In other words you'll have a lower voltage output with VDC -> series inductor -> caps, than if you use VDC -> caps -> series inductor -> caps.

Another way to drop the voltage is to use a regulated power supply ala NP's design in his balanced preamp schematic. This method is cleaner and will give the lowest ripple. Another method is to use a shunt with a zener, if it can handle the power. Example, if you drop the 6 volts with a 4 A load, the zener will need a 4*6 or 24 watt rating, which is pretty big.

Another method is to use a resistor-capacitor filter as the tube guys use. The resistor is in series with the voltage and the capacitor goes to ground. This will help reduce ripple. The resistor, like the zener, will need to dissipate power. For a 3 ohm resistor, with a 4 A load, it will drop 12 volts (3 * 4); the heat (power) dissipation will be 12 * 4, or 48 watts. In this application, I would use a resistor rated for 100 watts and put it on a heat sink.

Keep in mind all of these methods waste energy. It is best to use the minimum voltage for your desired power swing and then adjust the current, rather than the other way around. This keeps your heat generation to a minimum, and lowers cost: fewer components and smaller heat sinks.

Also, if you run the current too low, you won't be able to drive a 4 ohm load very well (or even a 6 ohm load). Your drive capability is best with lower voltage and greater current when running a class A power amp. (This isn't quite true: the best would be high voltage and high current, but this is practical for most of us.)

Robert

#### grataku

Evaas,
what drops the DC voltage down is R (resistance). Usually valves amps use chokes with lots of turns of thin wire. It is not uncommon to see 12 henry inductors used in Pi filters with 150 ohm resistance and 200mA current rating. That works well to drop and filter the DC voltage. In your case you can probably find a 50 mH inductor with hi current capacity but low internal resistance. That may be effective as filter but not as voltage reducer.
Just get a transformer with the right secondary voltage 2x ~22-24V and be done with it. Then if you want to be cool you can use a high current inductor in a pi filter to get rid of most of the ripple noise. Would have to be 4 caps and two inductors, right? One filter for the + and one for the - rail.

#### Evaas

Ok, so I'm getting the feeling that this is a bad idea. I definitely want to stay away from the restistor and diode ideas, and even the thin wire inductor, as those are very inefficient. However, I thought that inductors that are wound with low resistance wire will not dissipate much heat, but this does not seem to be what you guys are saying. Let me walk through an example:

I have a transformer with 8.5VAC secondaries driving a bridge rectifier(assume 2v loss through the rectifier), so the rectifier has 10v peaks. (8.5*1.41-2=12-2=10v). If I then use a large capacitor bank, my dc rail will be close to 10VDC. However, if I use a choke between the rectifier and the capacitors, then it will serve to lower the peak voltage out of the rectifier, and raise the bottom voltage. Whereas the peak was 10v and low was 0v without the choke, the peak my be, say 8v, and the low will be 2v with the choke. Is this basically correct? Furthermore, the area under the curve of the circuit with the choke will be similar to the area under the curve for the circuit without the choke. That is to say that the average voltage in both scenarious will be similar, but it is distributed differently. Based on this, the power dissipated by the choke should be small because in essence, its chopping off the peak voltage and distributing it later on (90deg later) rather than wasting it as heat. While the average voltages may be the same, what matters to me is the peak voltage, since that is what my caps will hold and the peak voltage of the inductor circuit will be lower.

Since nobody else seems to think that this is a good solution, can you tell me what I'm missing?

as for getting the right transformer, I'd love to Unfortunately in two months, and with trips to all the local surplus stores and all over the web, I still haven't found a good one less than \$70, and I need two because I'm building monoblocks.

Grataku, yes I'm thinking 4 caps (actually 4 banks of caps) and two inductors

Robert, when you refer to difficulty of driving 4Ohm loads with lower current, you're talking about current limited by the power supply and not about bias current, right? I don't see how bias would affec tthe ability to drive harder loads, but I know it would be a problem if the psu ran out of steam

Evan

#### rljones

Evan,

I've simulated and built Alephs with low and high bias currents. By low, I'm referring to 0.5 to 1 A per channel (over however many devices you wish) versus 1 or more amperes per channel. The bias current is fairly constant with a constant load like 8 ohms (the amp is biased for class A). The 1A or less setups seem to be able to only drive 8 ohm loads, while the 1.5 to 2.5 A setups can better drive lower impedance loads.

Keeping everything the same, if you drop your load to 4 ohms, the current delivery will need to double if you're to maintain the same power output. If your bias current through the output devices cannot double, but remains the same, your power will be halved compared to the 8 ohm load. All of this current is delivered by your power supply; if it cannot deliver the current, then the voltage will sag, the transformers will overheat (if they're too small for the load) and distortion will increase.

Not to prolong this discussion, but the power supply is probably the most important aspect of any amplifier. If it is not properly laid out or of inadequate design, you won't be able to drive real world speaker loads. Basically, you need to know what power output you want, what loads you wish to drive (2, 4, 6, 8 or 16 ohms for example), and then calculate the current versus voltage to get there (volts x amps= VA). If you want to run at +/- 35 V with 1A per channel, and use one transformer for both channels, you'll be drawing 70V x 1A x 2 = 150 VA. A 150 VA transformer used here would run hot; I would double this and use a 300 VA transformer (you can't overrate too much except for size and cost considerations).

Personally, I feel that if you spend so much time and effort building an amp, you don't want to skimp on the transformers. In any event, you need to know what current and voltage you want to run at so that you can choose the properly rated transformer.

Robert

#### Evaas

Robert, I need to apologize because I looked at my first post and I wasn't entirely clear. When I gave the bias current values, I was talking about current per transistor. So since I'm using four output mosfets, that would be a total of 4A bias per channel; and .7A bias *4 = 2.8A per channel. Also, I'm building a separate power supply for each channel. So lets take your calculation and assume 4A bias per channel, and remove the x2 because I'm only driving one channel with each transformer.
so lessee.. 70V x 4A = 280VA double that and get 560. I'm looking at a 750 VA transformer, so I think that should be able to source enough current. I agree with you completely and I want to use a good transformer.

Evan

#### rljones

Evan,

If you're going to bias at 0.7 A/device with 4 devices, that would be 2.8A x 70, or about 200 VA. I happen to have an Avel catalog (860-355-4711) right here. They make a 500 VA (25+25 @ 10A), model Y236801. I think you could get by with a 330 VA (25+25 @ 6.6A), model Y236750. The 25+25 should rectify up to about +/- 35 VDC.

But if you might bias at 1A/device, then I would go with the first transformer. I would guess about a \$10 to \$15 price difference per transformer between the two models.

Robert

#### jbateman

Evaas said:
Ok, so I'm getting the feeling that this is a bad idea. I definitely want to stay away from the restistor and diode ideas, and even the thin wire inductor, as those are very inefficient. However, I thought that inductors that are wound with low resistance wire will not dissipate much heat, but this does not seem to be what you guys are saying. Let me walk through an example:

I have a transformer with 8.5VAC secondaries driving a bridge rectifier(assume 2v loss through the rectifier), so the rectifier has 10v peaks. (8.5*1.41-2=12-2=10v). If I then use a large capacitor bank, my dc rail will be close to 10VDC. However, if I use a choke between the rectifier and the capacitors, then it will serve to lower the peak voltage out of the rectifier, and raise the bottom voltage. Whereas the peak was 10v and low was 0v without the choke, the peak my be, say 8v, and the low will be 2v with the choke. Is this basically correct? Furthermore, the area under the curve of the circuit with the choke will be similar to the area under the curve for the circuit without the choke. That is to say that the average voltage in both scenarious will be similar, but it is distributed differently. Based on this, the power dissipated by the choke should be small because in essence, its chopping off the peak voltage and distributing it later on (90deg later) rather than wasting it as heat. While the average voltages may be the same, what matters to me is the peak voltage, since that is what my caps will hold and the peak voltage of the inductor circuit will be lower.

Since nobody else seems to think that this is a good solution, can you tell me what I'm missing?

as for getting the right transformer, I'd love to Unfortunately in two months, and with trips to all the local surplus stores and all over the web, I still haven't found a good one less than \$70, and I need two because I'm building monoblocks.

Grataku, yes I'm thinking 4 caps (actually 4 banks of caps) and two inductors

Robert, when you refer to difficulty of driving 4Ohm loads with lower current, you're talking about current limited by the power supply and not about bias current, right? I don't see how bias would affec tthe ability to drive harder loads, but I know it would be a problem if the psu ran out of steam

Evan

Go look at standard toroids at the Plitron site.
http://www.plitron.com/

300Va are selling for around \$45, 500Va for \$60.

#### grataku

jbateman,
I don't mean to start a flaming war but can you avoid quoting someone's message for no particular reasons other than wasting server space? All the relavant info in your post was contained in one line and there are other ways to "beef up" one's postings.

grataku

#### Evaas

accepting defeat

Well I guess I can't use my surplus transformer, so I'll just bite the bullet and buy one. Are there any real differences between the major brands? Avel, Plitron, and Toroid of Maryland?

Avel - I like the Avel Y236801, and I'd go for even more power if I could. However I couldn't find prices on the web. Do they sell direct, and if not, where can I get it?

Plitron - hmm.. it looks like the 750VA transformers are only 6 bucks more than the 500VA, and they have the same voltages for each. \$66.85 for 750 VA, 25V secondaries

Maryland - the only ones they have with similar ratings are > \$100

Surplus - \$40, 750 VA but the voltage is too high - 33v <sniff>...

I've also decided that I'm going to have to simulate the whole circuit, because there are still a few things that I don't understand. Can anyone recommend a good free simulator? should I stick with spice, or is there something thats graphically oriented? that would be nice..

I did a little checking before sending the message and I found this: http://www.duncanamps.com/psud2/index.html
Its a power supply simulator and it looks pretty good. Has anyone used this? I'll let you know what happens after I try it out.

Evan

#### rljones

Evan,

I gave you Avel's phone number in my previous post; it is: 860-355-4711. They do sell direct and typically ship the same day you order by whatever method you wish to pay for. My guess is that it'll run you around \$75 apiece for a 500VA transformer, but give them a call. I've found them very helpful.

I now only buy from Avel. Plitron seems to have nice stuff but EVERYTIME I've tried to place an order they have no stock and lead times are 6 to 8 weeks. Forget it.

Maryland seems a bit overpriced. Like I said, I buy from Avel. You'll save a couple of bucks on shipping if you buy both transformers at the same time.

Good luck,

Robert

[]

#### Petter

Duncan can tell you all you need to know

http://www.duncanamps.com/psud2/index.html

A side effect of using inductor loaded supplies is to increase the charge period and thus make the transformer and look bigger than it is (more iron if you wish).

Watch out for resonances, though.

In tubed circuits, some people use a smallish cap at the front of the inductor to adjust voltage upwards. Others use pi filters which are not really choke loaded at all. The more current you have passing, the smaller the choke you need (Energy=.5L*I^2)

http://www.transcendentsound.com uses something in the range of 50mH for an OTL tube amp. I plan to use 150mH custom gapped toroids.

Petter

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.