Power Supply Design - Off My Rocker?

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

This is my first actual design, so I'm looking for input.

It's a power supply, but I am curious if it'll work. What I tried to do was convert the 120V 60Hz AC to DC, then convert it back to a 120VAC at a faster frequency.

The frequency is defined by the 555 timer running off a different power supply. The values I included set the 555 timer to oscillate at about 200Hz.

Am I off my rocker here, going after something that really won't work? My reasoning for doing this is so I won't need so many caps for the 15 amps my amplifier is requiring, ideally so the faster cycles will keep the caps charged.

For those who understand electronics better, I have a few questions:

1) Will this work?
2) If so, is it worth it?
3) Will this be too hard on the transformer/transistors/caps?
4) Will this provide better isolation from AC fluctuations than a typical unregulated power supply?

I found transistors rated at 150V @ 8 amps, which should be more than an enough.

All comments/suggestions welcome, including those related to how to draw schematics better! ;-)

Reece
 

Attachments

  • power supply.jpg
    power supply.jpg
    60.8 KB · Views: 1,340
ah, you may want to check those specs again... 150V at 8A is 1200W, which may be above the SOA of your transistors... what transistor are you talking about?

Generally, If you are willing to use big fat transistors in your power supply anyway, and want to be sure it will work, just use big pass transistors on a regulator. Hardly even any figuring out to do.
 
Well I'm not quite sure the H bridge you have driving the transformer will work. You'll have the pair of transistors on the positive rail that will need some form of level shifting to turn them on.

I think you'd be better off with just more caps if you're really worried. I read your schematic as 4x10,000uf on each rail. That's really not bad for an amp that draws 15 amps. You could even take those 2x10,000uf from the line voltage side, and put one extra on each rail. Save yourself the power transistors and heatsinks, and buy yourself another pair of caps, for 6x10,000 on each rail. That's really a fairly realistic value, and a good foundation for an amplifier.

I see you also have a housekeeping transformer that provides + and - 12V. Since your 555 circuit only needs the +12, I am assuming you're going to use that +12 and -12 V supply for your audio circuits. I highly advise against anything that could possibly connect line power to your low voltage ciruits, no matter how many improbable failures it would take.

If you still want to do this, check out this H bridge controller from IRF:
http://ec.irf.com/v6/en/US/adirect/ir?cmd=catProductDetailFrame&productID=IR3220S:
They even have modules that include IGBT's.
 
hummhoom said:
I think you'd be better off with just more caps if you're really worried. I read your schematic as 4x10,000uf on each rail. That's really not bad for an amp that draws 15 amps. You could even take those 2x10,000uf from the line voltage side, and put one extra on each rail. Save yourself the power transistors and heatsinks, and buy yourself another pair of caps, for 6x10,000 on each rail. That's really a fairly realistic value, and a good foundation for an amplifier.


Yeah, I think you're right. It was a nice idea, but throwing the caps together would probably produce the same results with a lot less work.

Thanks for your help!

Reece
 
No problem.

There's really very little to be gained in quality in the power supply, except make it big enough.

Power switch, fuse, ransformer, rectifier bridge, and filter capacitors, and that's it. And make sure the power to any low level signal stages are regulated.
 
Check the SOA chart again. The 15030 is rated at 150V and 8 A continuous, but not both at once. At 150v it is rated at less than an amp. The 8A rating is good only up to maybe 15v. It is only rated at 50 watts.

Are you running 240VAC into the thing? It looks like it with 120VAC on either side of the bridge with the cap common grounded. If you are running plain old 120VAC to the bridge recto, remember one side of that is already at ground, so that ground in the center of the filters will not cooperate.
 

PRR

Member
Joined 2003
Paid Member
> convert the 120V 60Hz AC to DC, then convert it back to a 120VAC at a faster frequency. ...I am curious if it'll work.

No. Aside from practical details like SOA, your "200Hz" is still heavily modulated with 120Hz ripple. You have more than doubled costs, for no advantage, and probable extra annoyance from having two sets of ripple spectrums. Plus you have a 160V square-wave, which is probably a very bad idea around audio.
 
PRR said:
> convert the 120V 60Hz AC to DC, then convert it back to a 120VAC at a faster frequency. ...I am curious if it'll work.

No. Aside from practical details like SOA, your "200Hz" is still heavily modulated with 120Hz ripple. You have more than doubled costs, for no advantage, and probable extra annoyance from having two sets of ripple spectrums. Plus you have a 160V square-wave, which is probably a very bad idea around audio.


Well, I don't fully agree with that...

The safe operating area of the transistors was 2400 watts (16A @ 150V). I was looking at doing 3A @ 84V, so that wasn't really an issue (assuming heat was taken care of). With the capacitors on the AC line, I was getting about a 2.5 volt ripple at 84VDC, so that wasn't really bad. The signal coming out of the transformer would have been squarish, but that's all the better as rectified full wave that would have produced a near perfect DC on it's own, even without the 4 10,000uf caps I had smoothing each rail out!

The transistors cost $1.68 each, extra bridge rectifier a couple bucks, 555 timer maybe a buck or two, and everything else was negligable. I was already going to use the extra transformer, so that wasn't an issue. It would have cost maybe $10 more, which unfortunately is a lot less than the cost of 10 good quality 10,000uf caps.

But in the end, I probably don't want the extra trouble of it all. Fun thought though.
 
Hi !

Hmm, i am not sure what exactly you try to do, but it looks like you
want to feed a squarewave into a transformator ?
If yes, DONT ! A transformator only works fine with sinus-signals,
a squarewave fed into a transformer results in VERY high voltagepeaks
at secondarywindings. Depending on the quality of the squarewave
these peaks will be several kilovolts. Also feeding the primary winding
with a squarewave results in very high voltagepeaks, also several
hundred volts. This will definetely blow your transistors at once, you
are dealing with severe inductive loads.
Also switching with transistors is not too easy, as transistors open
faster than they close, you always get a small short everytime you
switch, immediately blowing these transistors...
There are special transistors designed for switching, the mje15030
is designed to operate as audiodriver.

If you are generating a sinewave, the losses will be very high, this
means you are dealing with several 100 watts, needing the most
powerful bjts and heatsinks. So, using some more caps will be
smaller and less expensive.

If you are looking for a "ripplefree" psu, consider a switchmodesupply,
but this is also not too easy...

Mike
 
Enzo said:

Are you running 240VAC into the thing? It looks like it with 120VAC on either side of the bridge with the cap common grounded. If you are running plain old 120VAC to the bridge recto, remember one side of that is already at ground, so that ground in the center of the filters will not cooperate.

Furthermore your earth leakage switch will be triggered and the whole circuit will be disconnected from mains.

Steven
 
rjon17469 said:

Well, I don't fully agree with that...

The safe operating area of the transistors was 2400 watts (16A @ 150V).


you need to learn how to read the datasheets before you design anything.

at 100microseconds, the SAFE OPERATING AREA is right at 2 amps. 3A 84V is pushing it. Don't be surprised to see things blow up if you specify them that close to their SOA ratings.

In any case, if you try to get more than the specified 50W out of one of these, it will very likely go BANG. If you want to get that kind of power, you need to use LOTS of TO-220 devices... or use TO-247 or TO-3.
 

Attachments

  • 15030soa.jpg
    15030soa.jpg
    29.3 KB · Views: 358
I was already planning on doing a case fan, as having amps that pull 15 amps generates some heat by itself.

When I originally posted this, I was looking more for opinions on the general design rather than individual component selection. Granted, the transistors would generate a lot of heat, but with a decent heatsink and a fan, I don't think it'd be too much of a problem.

But I think I'm going the route of a typical unregulated power supply.

Reece
 
Clean power helps amplifiers sound better. Your idea is basically a good one but you have significantly oversimplified the design task.

Driving transformers with square waves can be made to work fine. Its done all the time, but if you make the effort why stop at 200 Hz? Typically these things are done at 20KHz to as high as a Megahertz. A significant design and debug task, expect a small pile of burned out stuff before you get it right. Now you need to find a way to keep all those fast edges out of your audio circuits.

Pure sine wave power can also be done but expect to waste AT LEAST 33% of your amplifier rated power as additional waste heat in the sine wave power section. Most of this power is wasted whether or not music is playing!

For less total effort and cost and better results, regulate the DC power fed to your speaker driver output stage. By doing this you waste a little heat but can significantly reduce the total storage capacitor size. You can lower the effective source impedance of your output stage into your speakers, you can lower the noise floor and present your ouput stage with a low source impedance. These are all very good things to do.

Look up a capacitance multiplier, this again converts some power to heat but lowers the effective source impedance a lot and reduces storage capacitor size by a fair amount.

Why do these things work? Here is an imaginary case: Lets say you have decided that your speakers will need 10 Amps and you want a ripple of 0.5V at your power supply. This means (Amps X seconds = Volts X Farads) or 160,000 uF. (60Hz power refreshes the caps every 8.3 mS). With a regulator you can allow 5 V of ripple AHEAD of the the power stage regulator, you can use 16,000 uF and loose 5V in the regulator stage, at 10 Amps this is 50 watts but only while the peak load lasts.

Obviously many trade-offs exsist but regulated power for output stages is often done in expensive power amps because among other advantages listed above it also allows you to lower global feedback since power supply ripple does not need to be cancelled out in the final stage.
 
oh boy, high current goodness.

External transistors (pass transistors) can be used to increase output current from a voltage regulator as high as required.

Caps after the regulator would help on transients, yes.

I just noticed that SOA chart's time ratings... the line to look at falls somewhere between DC (0Hz) and 5mS (200Hz) (1/0.005) unless *I* am off my rocker. That makes it 0.4A 100V and 0.6A 80V, with the thing working at 25*C, nevermind the heatsinking... i.e., 50 watts and (much) less.
 
The regulators don't need a lot of gain so one or two transistors will do the job. The transistors will need to be rated to handle peak currents and max voltage so a TO-3 or a couple of TO-220 case devices seem like a decent bet.

There are few transients on the output side of a rectifier the large capacitors mostly take care of them.

And yes there would still be a need for capacitors on the regulator output but their job is different. They don't act as current reservoirs the regulator supplies the big currents between the 60 Hz recharge the smaller caps are needed to keep the regulator honest at say 20KHz.

Also a good capacitor is needed on the regulator reference voltage.

A simple example of a capacitor multiplier, bad because it needs 15 watt resitors, a darlington would help a lot.
 

Attachments

  • capmulti.jpg
    capmulti.jpg
    51.3 KB · Views: 381
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.