[peranders]

Today I have made the layout ready of a super regulator based power supply, 230/115 VAC, +- anything out. Basic voltage is 15 volts but up to 24 volts seems to be possible. Output current at 15 volts is 200-300 mA but this is only a guess.

The size of the board is 137 x 102 mm, 5.4 x 4"

I have snubberized it. I'll guess this the first super regulator in the world which is snubberized. Over at some other forum there has been a hot debate wether adding a small resistor in series with decoupling caps is good and/or important. It's extremely rare in linear applications and I wonder if any high-end equipment has it. I'm pretty certain that I can't test it but at least the implention is good. I have this feature as a source of evaluation for the golden ears.

Even in the old days there has been snubbers, both the magazine Elektor had it in at least one design and also Mr. Ottala in 70's in his low TIM amp.

If you want to read more about snubbers, you can start here:
http://sjostromaudio.com/smf/index.php/topic,11.0.html

The super regulator pcb look like this in my Mac:

I'll guess the pcb will see the day of light in August or September. Right now it's summer and vacation.

If a few people want to be "beta builders" with free*) pcb's drop me a note. Remember though that those people need to be very skilled in order to get the parts in the right spot and be able to do some troubleshooting.

*) under certain conditions.

 

[Nixie]

Additionally, there should be snubbers on the unregulated side. Diode switching causes RF, which is too high frequency to be heard when induced into nearby circuits, but it is modulated by the power frequency, and that modulation makes sensitive circuits behave like AM radios.

Most people put capacitors across the diodes. This is not the correct approach; an RC (or RC paralleled with C) snubber should be instead placed across the transformer secondary, as shown here:
http://www.hagtech.com/pdf/snubber.pdf
Scope needed to make sure you are using correct values.

 

[JWFokker]

What, may I ask, is snubberization?

 

[Nixie]

Damping of oscillations. I don't think it's an actual word. He derived it from the word snubber.

For example, a diode's parasitic capacitance, combined with the transformer parasitic capacitance and inductance as well as the filter, forms an LC oscillator that is excited each time the diode switches. This is bad if you have sensitive audio circuits nearby, and NFB connections are especially prone to RF interference such as this can cause.

You shouldn't be using too much NFB in the first place. Positive feedback, within a frequency dependent overall error correction strategy described by Hawksford (see paper C35 on his page) is the best approach to distortion nulling (as opposed to the asymptotic and bandwidth limited reduction that NFB does). OK, I guess that was offtopic enough...

 

[peranders]

Nixie is right in theory, also Mr. Hagtech but in real life it needs to be verified. Simulations aren't everything.

I did thought about using real snubbers across the diodes or just caps. I chose to only have caps, mostly on space reasons. As you may have noticed I have a resistor in series with each transformer winding. Remember that snubbers in power supplies are pretty uncommon even in high-end gear.

I'm also aware of that high-end doens't imply good engineering. It's only someone is willing to pay much for. Sometimes high-end really is good, sometimes it's only expensive.

 

[peranders]

Quote:

Originally Posted by JWFokker What, may I ask, is snubberization?

I invented the word from snubber which carlosfm over at www.diyaudio.com used but in his application it's wasn't a snubber really, still it need to be called something.

What should it be called? Any suggestions?

If you want to read more about snubbers, you can start here:
http://sjostromaudio.com/smf/index.php/topic,11.0.html

 

[bhjazz]

Snubber is good. The same term is used at the Bottlehead camp, and is a simple upgrade to many of their products. Doc B explains a snubber (at least for their uses, anyway) as this:
"...We prefer the term reverse recovery spike filter, because it really tells you what the circuit does.

All SS rectifiers exhibit a reverse recovery spike when the current thru them reverses and they switch off. This is reflected back thru the power trans and up the power line to other equipment. The filter specified does a great job of eliminating the Mhz range spike peculiar to the UF4007 fast recovery soft start diode. It makes it cleaner than a HEXFRED.

Sonically it is reputed by various folks to improve image width, smoothness, noise floor, etc. And it takes about $2 worth of parts and about 10 minutes to install. Hard to come up with a reason not to try it!"

 

[Nixie]

Linear application note 70 for example shows how to build an EMI probe. I strongly suggest you build it to check the performance of any snubber you build, to make sure it's doing what you expect. It might be, as you say, $2 in parts, but the specific part values that would work well differ widely depending on the rectifier, filter, and transformer used.