A plethora of power supply advice

[R.G.]

We seem to be in the middle of a rash of power supply questions. Here is some distilled advice, all in one place.

1. Yes, you need power supply decoupling and bypassing caps.

Just do it. Sometimes you can get by without it, but there are so many funny situations that can come up, you'll be a lot happier with the results if you do. In fact, think of power bypassing and decoupling as something you always do, and have to have a very good argument NOT to do. National Semiconductor used to state in some of their application notes that most application schematics do not show power decoupling caps, but that does not mean you don't need them. They recommended a 0.1uF or 0.01uF ceramic to ground from **each** power pin of **each** opamp or other IC to the ground pin of that IC as close as you can get them to the IC body, plus a 10uF or larger electrolytic cap no more than a couple of trace-inches from the power pins of each IC.

This provides local buckets of electricity for the IC to draw from while waiting for the remote power source to send reinforcements. ICs are better about ignoring the power supply  (that is, have higher power supply rejection ratios) than they used to be, but caps are very cheap insurance against having to post "Do I have to have decoupling caps?" or "Why is my pedal [insert anti-social behavior]-ing?"

2. Use enough voltage rating on your caps. For 9V pedals, you need 25V. It is certain that some boy-genius will eventually stick an 18V supply on the pedal. 16V caps may or may not live through this.

3. Go check your power adapters for 9VAC outputs, and figure out some way NOT to connect these to your pedals, or worse yet, a pedal daisy chain. This will kill most pedals, dead, faster than you can unplug.

4. A three terminal regulator is often the simplest, fastest way to get clean power at a fixed DC level.  These things are REMARKABLE in terms of power supply goodness returned for the effort and money invested. It is quite difficult to do better than one of these wonders in other ways.

5. If you use a three terminal regulator, put in the protection diode from output back to the input. Not doing this makes the regulator vulnerable to dying if you short the input voltage, like with an unused plug on the daisy chain. A diode is remarkably cost effective insurance.

6. Things with voltage across them and current through them heat up. If you use things with lots of voltage and/or current, touch them with your index finger tip. If they burn the maker's logo into your fingertip, they're too hot. The 78L09 is a remarkably good solution to powering a single pedal, as it can supply up to 100ma of regulated 9V without a heat sink if it's not supplied from more than about 20Vdc input. But test it with the fingertip test. If you're running a small power amp chip or regulating the power to one of these, you probably need to heat sink the power amp chip and the regulator, if used.

7. Reversing the power supply on most modern chips can kill them very quickly. Many modern chips have bonding wires which can burn open at 50-100ma. That's not much, and a 9V battery can do it. Use polarity protection.

8. Go read "Power Supplies Basics" at GEOFEX if you even think you want to make a transformer-rectifier-filter-regulator power supply for your stuff.

9. Don't mess with AC power line wiring unless you already know how to do it safely. Reading about it on line is not enough. Get a real, live person to instruct you and help you learn to do it safely. You can die, burn down your house, or kill someone else who runs into the latent issue you wired in when you're not around. This is much, much less fun than it sounds like here.

10. Running a pedal at a power supply bigger than it was intended for is not a neato mod for the uninformed. Blind Urge to Mod Syndrome needs to be leavened with some caution.

[armdnrdy]

All good sound advise!

Maybe Aron can pin this thread to the top of this forum as he did with the debugging thread.

[Kesh]

11. The pinout on LM317 regulators and similar is not what you think it is. Consult the datasheet.

[R.G.]

11. The pinout on LM317 regulators and similar is not what you think it is. Consult the datasheet.

Good point. I'd say it another way though: the pinout on all ICs actually mutates and changes as you work on installing the chip unless you have a printed hardcopy of the manufacturer's datasheet lying on the bench next to the board you're working on. The pins are forced to be in agreement with the datasheet by the nearness of the printed datasheet, and will change around on you otherwise, even if you have memorized the pinout.

... at least that's how it works for me.   

[electrosonic]

For the bypass caps - specifically ceramic caps or are plastic caps OK?

Andrew.

[Kesh]

the pinout on all ICs actually mutates and changes as you work on installing the chip unless you have a printed hardcopy of the manufacturer's datasheet lying on the bench next to the board you're working on. The pins are forced to be in agreement with the datasheet by the nearness of the printed datasheet, and will change around on you otherwise, even if you have memorized the pinout.

Yes

The 317 is particularly tricksy because EVERY schematic has its centre pin as adjust. Even the ones on the datasheet! Yet the chip itself stubbornly refuses to cooperate.

[armdnrdy]

There is a big difference between a schematic and a pinout....

Always check the data sheet and then double check that with another manufacturers data sheet and then hope it's right! 

[PRR]

> pinout on LM317 ...is not what you think it is

Also 79xx pinout is very different from 78xx pinout. Important to know if you swing bipolar.

[roseblood11]

12. ...and 78Lxx pinout is different from the bigger TO-220 versions.

[Lizard King]

Where can one buy LDO 9v regulators?

[Levispeights]

Where can one buy LDO 9v regulators?

I was just looking this up yesterday... Here you go.

http://www.mouser.com/Semiconductors/Power-Management-ICs/Low-Dropout-Regulators-LDO/_/N-5cgac?Keyword=regulator&FS=True

[Fender3D]

12. ...and 78Lxx pinout is different from the bigger TO-220 versions.

BTW
do not forget xxLxx usually dissipates 0.7W in a TO-92 package, then if you wanna drain 100mA, you'd better have no more than 7V input voltage respect output regulated voltage...

[R.G.]

For the bypass caps - specifically ceramic caps or are plastic caps OK?

Whatever other faults they have, ceramic caps are better bypass caps than other types. A plastic film cap is OK if that's all you have; it's certainly better than nothing. But ceramics are still cap-like up in the RF regions where you want them to be bypasses. Plastic caps are more of a maybe.

Use ceramic if you possibly can.

[Lizard King]

Where can one buy LDO 9v regulators?

I was just looking this up yesterday... Here you go.

http://www.mouser.com/Semiconductors/Power-Management-ICs/Low-Dropout-Regulators-LDO/_/N-5cgac?Keyword=regulator&FS=True

Thanks.  I never order from Mouser because I can't work their search engine.  I put in LDO regulator  and there were no 9v dvices available.

[scratch]

using the finger to check how hot a chip is getting, and getting the logo burned in, done that ...

now when I suspect a chip might be doing that, I wet the finger first, and have been saved by the 'sizzle' ... ;-)

[electrosonic]

How about a ceramic cap paralleling the first on board filter cap to shunt (switching) power supply noise to ground?

Andrew.

[R.G.]

I never order from Mouser because I can't work their search engine.  I put in LDO regulator  and there were no 9v dvices available.

All Products » Semiconductors » Integrated Circuits - ICs » Power Management ICs » Low Dropout Regulators - LDO

8 devices in through-hole packages, of which the most promising+cheap seems to be the 512-KA78R09CTU

It's worth learning how to work their search engine, although it does have some quirks.

How about a ceramic cap paralleling the first on board filter cap to shunt (switching) power supply noise to ground?

That's OK. Better to put in a series R or L, then a ceramic cap to ground. Or one on each side of the R/L.

Even better to use a power supply which, although perhaps switching in nature, does not produce enough switching power supply noise to hear, having been designed to be as quiet as possible.

[therecordingart]

Thank you for the info, R.G.

In regards to 3 and 7. I've been considering putting a diode bridge in future builds. Overkill?

[R.G.]

In regards to 3 and 7. I've been considering putting a diode bridge in future builds. Overkill?

Good idea - just not practical, for reasons to do with how pedals are grounded.

I messed with this a ways back. Aron may remember us pinging back and forth about making a power supply entry circuit that was polarity and voltage insensitive, a kind of uber-jack that took in whatever you plugged in and put out 9Vdc, the right polarity.

The fly in the ointment is that pedals are still using the old single-ended power scheme where signal ground is tied to one or the other polarity of the DC power supply, instead of being somehow floated free of them. If you feed a full wave bridge with AC or DC, you get out the right polarity, OK, but both the positive and negative sides are offset from the incoming power by one diode drop. That's arguable for the non-ground side, but on the ground side, the incoming DC power is now one diode higher (for negative-ground pedals) than the power supply negative side.

This is fine as long as you only have one pedal, and don't care where the negative side of the power supply actually sits with reference to signal ground. The problem starts when you plug in another pedal that doesn't do this, and it ties the signal ground to the power supply negative with a wire. Now your two pedals are trying to force the power supply negative side to be at two different voltages, separated by a diode. This is not so good.

I messed with MOSFET bridges and active rectification, trying to get the ground differential down to insignificant, but that gets complicated, and is likely beyond the average DIYer.

The real answer is one I've tinkered with a bit, falling loosely under the header of the "wrapper", the whole mess of semi-standard stuff like jacks, power inlets, housing, etc.; more on this at geofex. The wrapper could take whatever is put in and make whatever DC is needed out of it. There are power supply controllers that do this today, where a voltage between perhaps 3V and 30V would be converted to 9Vdc - or 12, or whatever. The problem is that this needs to be a switching supply, and now the complexity of making it well behaved and quiet is large. Again it's probably beyond the average DIYer, as it's more complicated than the pedal in most cases. Not that it's difficult if you always use exactly the same circuit, but things are still more costly and complex.

I put it down to "just because you can do something doesn't mean you should" and left it.

[davent]

using the finger to check how hot a chip is getting, and getting the logo burned in, done that ...

now when I suspect a chip might be doing that, I wet the finger first, and have been saved by the 'sizzle' ... ;-)

Laser thermometer works great. At around 75°c reflex demands you pull away but mind over matter won't result in a burn.

dave