Could any electronics guys look over my schematic?

[Amphiprion]

I am designing a dual regulated power supply +/- 11V to +/- 16 V for my +/-12 and +/-15V opamp stuff. It will also do a lot more current than necessary as I will use it for other stuff should I ever need to (hence the big filter caps).

Anyway, here is the schematic:

http://www.hayenga.com/mark/electronics/PS.mdi

I would like recommendations as to what cap types to use (electrolytic or tantalum where applicable) and what my bypass caps should be (right now 1uF polypropylene + 10nF ceramic C0G). Maybe I could just use a 100nF X7R ceramic instead? Or 1 or 10uF ceramic X7R with 10nF C0G? I dunno. I'm just shooting for <1mV ripple at full output after the regulators, and making things uber-overdone on a DIY scale isn't really that much more costly. But I would like to avoid making it look silly at the same time.

Thanks for any help!

[mazurek]

I would suggest exploring a circuit simulator, I don't remember why, but I use the free version of SiMetrix. There's others available too. You can model the filter sections as a transfer function, and explore the effects of L,C,and R, in the filter, and maybe include an idealized transformer in there too. I think the Texas Instruments op amp guide book has a brief section on types of capacitors and the frequency range over which they are capacitive. I believe ceramic and tantalum are used for high frequency, and tantalum is popular in power supplies because they have higher equivalent series resistance, which can damp oscillations. I didn't see polypropelene in their guidebook, but many on diyaudio swear by it. As long as you are going brute force, you can just bypass a lot as you show. Layout is a non-trivial problem which is equally important, remember to keep the feedback paths short.

[JonMarsh]

I'll have a look, probably tomorrow on an airplane if I can download tonight- analog supplies are my bread and butter- I'm really just a wires and sparks guy, not a speaker builder.

[Amphiprion]

I'm starting to be a wires and sparks guy. It's like Lego's, but with Digi-key as my lego box. Cheaper than high-end speakers too.

The poly's are in there just to ask others if they are worth it, normally at work I would just stick a .1uf X7R 0805 ceramic on it and be done with it. But I actually did have problems with bypassing on the last board I made, I had a programmable delay line that I was using to make 10-500ns pulses when combined with some logic gates and a square wave generator. An isolated DC/DC converter on board really messed things up (a cheap TI part, DCP020515 iirc, little 2W 14pin DIP footprint thingy). Anyway, jitter in my pulse shot up to > 50ns. I had to put 220uF caps on everything in addition to 10uF ceramic 0805's in order to get jitter <5ns. Never thought bypassing was that big a deal until then. Of course, nothing that noisy going on this board, so maybe not so much of an issue. I'm still pretty green at this stuff.

Oh, diodes are 1N4002 unless you suggest something else. At 750ma out I drop not quite 2.5V across the regs, I need to keep that number above 2 for the LM317/337 IIRC.

[cjd]

Looks good at first glance, though I've only done fixed output so not seen the pot where you have one. You have capacitance at the device consuming the power? I'd be surprised if you need the 100uF after the regs if so... I've only built supplies with the LM338.

C

[Amphiprion]

I have a bunch of LM338K steel TO-3's for sale now that you mention it. 10 or 12 pieces. Any interest?

The pot is a 1k unit, with the additional resistor it keeps the min/max voltage of the output in the 11-16V range, assuming my calcs are right.

I've been reading some at work today because my computer is a POS and my schematic and layout programs won't export/import properly thus greatly infuriating me. IT is coming by tomorrow to fix it.

Anyway, I read quite a lot about caps. Apparently, Kemet (in a reference paper) solidly recommended against X7R and lower grade ceramics in audio ckts. I knew they were piezoelectric, but Kemet says the effect is the worst in the 30Hz to 30kHz range and it specifically said not for use in audio ckts, save for very low voltage/signal applications. I can't seem to find NP0 or C0G in anything >10nF either. I started looking at tantalums and Kemet makes one (for $16 a pop) that's 100uF that stays under 20 milliohms up to 1MHz. I thought that was kind of impressive.

How much capacitance I need after the rectifier, I don't really know, have to go back and read the app note on the 317/337.

[cjd]

You don't need to read up on the 317/337 to know what kind of capacitance after them, you need to look at the device you're powering!

I use various electrolytics (I really like the Panasonic FC caps for audio stuff, great mix of price and performance) with small value caps being small polys.

I'd go with fixed resistors if you can, pots tend to wander over time (or so I have heard... that seems odd to me, but hey). Any reason you want adjustable voltage?

338's I use the TO220 package. A lot easier for me to sink. Most I'm running off a pair (one pos, one neg) is a pair of LM3886's.

C

[Amphiprion]

The pot is a Bourns 3296W cermet trimpot (25 turn). The adjustability is mostly for those opamps (like some of the new Analog Devices parts) that won't take more than +/-12V. I could just do it with fixed resistors and a jumper, but the obsessive-compulsive in me likes the idea of being able to trim it to 15.000 volts

Another question: when you use a horizontal mount heatsink for a TO-220 part, how do you bend the leads? A 'hard' 90 degree angle, or just stick it through the pads and push it over? Such practical things they never teach you in college...

[cjd]

I always, ALWAYS grab the leg in a needle-nose pliers up against the package, and bend *after* that (it makes sure NO stress from the bend is transferred to the case). So I get nice clean bends every time. Though I sometimes use a pair of tweezers for stuff with smaller leads (resistors, etc.)

C

[mmoeller]

I have a bunch of LM338K steel TO-3's for sale now that you mention it. 10 or 12 pieces. Any interest?

The pot is a 1k unit, with the additional resistor it keeps the min/max voltage of the output in the 11-16V range, assuming my calcs are right.

I've been reading some at work today because my computer is a POS and my schematic and layout programs won't export/import properly thus greatly infuriating me. IT is coming by tomorrow to fix it.

Anyway, I read quite a lot about caps. Apparently, Kemet (in a reference paper) solidly recommended against X7R and lower grade ceramics in audio ckts. I knew they were piezoelectric, but Kemet says the effect is the worst in the 30Hz to 30kHz range and it specifically said not for use in audio ckts, save for very low voltage/signal applications. I can't seem to find NP0 or C0G in anything >10nF either. I started looking at tantalums and Kemet makes one (for $16 a pop) that's 100uF that stays under 20 milliohms up to 1MHz. I thought that was kind of impressive.

How much capacitance I need after the rectifier, I don't really know, have to go back and read the app note on the 317/337.

The caps your talking about, "after the rectifier", if I'm thinking correctly, are not nessassarily what Kemet was talking about. If they are associated with setting the time constant for the power supply, supply regulation sorta thing, they are not really in the audio path. Although still very important, but for a different effect.

If your talking about input filter caps and noise reduction devices within the amplifier stages the use of X7R caps might have a more profound effect.

I may be way off base though. I've worked with small signal stuff for a long time, but the higher voltage amplifcation stuff is fairly new to me. The princeples are still very similar however.

edit removed.

[mmoeller]

Did a little poking around and found some articles that might interest you.

"The sound of capacitors", Steve Bench

http://members.aol.com/sbench102/caps.html

Looks at hysterisis in differnet material caps, as well as biased and non-biased electrolytics.

http://waltjung.org/

Check out classis papers on Walt Jung's webpage. Picking capacitors 1 and 2. Good reading albeit alot of op amp stuff.

Richard N. Marsh does alot of work with power regulation, hard to find any papers other than the one with Walt Jung, mentioned above.

[Amphiprion]

cjd,

Is there a reason you prefer the FC over the FM? I see the FC is more expensive but the FM is lower ESR/longer life. I'm wondering if the FC is better in a way that I am missing. Thanks,

Mark