[wap32]

I'm building a simple amp with an opamp gain stage followed by a szekeres ve mosfet buffer and I was hoping to power it using some old transformers I have lying arround.

They're iron core, single primary 230VAC, single secondary 24VAC, 50VA.
Using both in a center-tapped configuration, after rectification, results in about +/-35.7VDC.

My first try was to use 79/7812 regulators, even tho the max input voltage is 35V.
Well, they regulate, but I only get to listen to about 30 seconds of music before one of them goes into thermal shutdown.
Even with heatsinks they get crazy hot.

I searched arround a little, but didn't find anything else besides adjustable regulators (LM317/337).
Output voltage needs to be +/-12 and from what I could measure, the amp draws about 0.3A, but I'd prefer something that can supply 1A or so just to be safe.

Do you guys have any advice about other fixed regulators I could use?

 

[mugdecoffee]

You're not going to be able to drop 23V over a regulator without some crazy heatsinking. You'd be much better off getting a different transformer and using the 317/337 regulators. If you'd rather keep the transformer you can make a virtual ground that will split the rectified voltage from one of the trasformers so you aren't dropping nearly as much voltage. Tangentsoft.net has a page on virtual grounds as well as a handy power supply checker that tells you if your PSU is reasonable or not.

 

[Juaquin]

Unless you are really set on using those old transformers, you'd have much less of a headache and a better quality PSU by using a transformer with a lower output voltage and then using a 317/337 or similar.

 

[wap32]

Thanks for the input guys.

I guess I'll be buying a new transformer soon and stick with the 78xx/79xx regulators.

 

[tomb]

Quote:

Originally Posted by wap32 /img/forum/go_quote.gif Thanks for the input guys. I guess I'll be buying a new transformer soon and stick with the 78xx/79xx regulators.

Or, you can simply regulate down to 24V and split the power with a TLE afterward - something most amps do around here, anyway - M3, PPA, PIMETA, etc.

 

[Rescue Toaster]

Just for discussion's sake... if you had a 30v circuit w/ virtual ground, and a power supply in the 35-45v range, are there any things to really watch out for in the regulator?

A 317 or 1085 can handle 30v in-out but I always figured there would be startup/shutdown issues depending on how fast/slow the bias current comes up.

Should clarify, I'm talking about a single regulated 30v supply. (virtual ground and such handled later, assume power trans is single-secondary and thus not grounded)

 

[tomb]

Quote:

Originally Posted by Rescue Toaster /img/forum/go_quote.gif Just for discussion's sake... if you had a 30v circuit w/ virtual ground, and a power supply in the 35-45v range, are there any things to really watch out for in the regulator? A 317 or 1085 can handle 30v in-out but I always figured there would be startup/shutdown issues depending on how fast/slow the bias current comes up. Should clarify, I'm talking about a single regulated 30v supply. (virtual ground and such handled later, assume power trans is single-secondary and thus not grounded)

Most of the linear-regulated power supplies around here use the regulator to float a voltage difference between the unregulated voltage and the regulated voltage. That way, there is no 30V limitation with the LM317. The only thing that people have had trouble with is the current limitation. The LM317 is good for 1.5A, I believe, but some high-capacitance amps can exceed that on startup.

 

[AudioCats]

or you can do a two step regulation, first drop the voltage down to 24V with 78/7924, then use the 78/7912 to drop it to 12V. It will take two more regulators but it gives (arguably) the side benefit of a cleaner DC.

 

[wap32]

Quote:

Originally Posted by tomb /img/forum/go_quote.gif Or, you can simply regulate down to 24V and split the power with a TLE afterward - something most amps do around here, anyway - M3, PPA, PIMETA, etc.

I'd prefer to keep the design as simple as possible, I think a virtual ground would add unneeded complexity and limitations.

Quote:

Originally Posted by AudioCats /img/forum/go_quote.gif or you can do a two step regulation, first drop the voltage down to 24V with 78/7924, then use the 78/7912 to drop it to 12V. It will take two more regulators but it gives (arguably) the side benefit of a cleaner DC.

That would be actually be a simpler (albeit uncommon?) solution and much cheaper than a new transformer.

 

[00940]

Quote:

Originally Posted by wap32 /img/forum/go_quote.gif I'd prefer to keep the design as simple as possible, I think a virtual ground would add unneeded complexity and limitations.

The szekeres VE lends itself to be used with a virtual ground. The complexity of a dual power supply is unnecessary.

A virtual ground needs nothing but a pair of resistors in this case...

 

[wap32]

If I was powering the mosfets alone, I guess a resistive divider could suffice, since not much current needs to flow through it, but I'll also be feeding the opamp.
In that case, would it not be a lot less reliable?

In my view, using a dual power supply is not complex at all.
I already have the transformers, it is just a matter of adding a pair (or as AudioCats suggested, two) of regulators.

However, I am by no means an expert, I'm still in the early years of my Master's, studying the basics.

 

[dsavitsk]

The VE has a lovely midrange and the ability to cancel out PS ripple, but it has frequency response issues -- if you search at headwize there is a post from amb explaining this. It can sound quite nice, but it is really more of a tone control than a ruler flat amp.

You would likely do better either using 4 fets for true balanced output -- though you'll need to either transformer couple it, or use balanced phones -- or, there is a design from PRR for a mosfet within the feedback loop of an opamp that should do what you are looking for w/o coupling caps. HeadWize - Project: Designing an Opamp Headphone Amplifier (A HeadWize Design Series Paper)

There was a thread started about this the other day I think.

 

[AudioCats]

you can also add a pair of 12v/500mA light bulbs in series to the existing 78/7912's input, and use them to burn voltage.

They also serve as current indicators.

 

[wap32]

Quote:

Originally Posted by dsavitsk /img/forum/go_quote.gif You would likely do better either using 4 fets for true balanced output -- though you'll need to either transformer couple it, or use balanced phones -- or, there is a design from PRR for a mosfet within the feedback loop of an opamp that should do what you are looking for w/o coupling caps.

I was afraid this would happen.

There's just so many circuits and amp designs...it's a little overwhelming.
And when I decide, "well, I'm going to try and build this", someone points out "actually, design xyz is a much better choice" and trashes my plans all over the place.

Please note I'm not in any way ungrateful for your post, quite the opposite.
The design you presented actually appears to be what I am looking for, and I will certainly take a more thorough look at it.

It's just that after having all the parts ready and a working circuit, it's a little disappointing knowing that you might have made the wrong choice and need to refresh your ideas and plan all over again.

Quote:

Originally Posted by AudioCats /img/forum/go_quote.gif you can also add a pair of 12v/500mA light bulbs in series to the existing 78/7912's input, and use them to burn voltage. They also serve as current indicators.

That would actually be a pretty creative solution, but wouldn't the bulbs dissipate a quite a bit of heat as well?

 

[grenert]

Quote:

Originally Posted by wap32 /img/forum/go_quote.gif That would actually be a pretty creative solution, but wouldn't the bulbs dissipate a quite a bit of heat as well?

Sure, but light bulbs can handle the heat

You could also drop some volts with resistors. They'll have to be pretty big ones (high wattage) if your circuit really draws 300 mA. Put a capacitor afterward, and you'll really get rid of ripple, too.