MC7915a reg. Unstable w/out load ?? (+schemo)

[Freq Band]

For me, with no load at all, negative side's volts will climb....16v to 17v to 18v, etc....
Positive is ok...(14.9v, stable)

Here is a standard PSU regulator setup.
Except I have added capacitance- post regulator. (220uf NP)
I also have the standard extra protection diodes across regulator's in/out (not shown).

I have built and used several of these, and always added a resistor+led at output, for front panel "on" indicator. No problems so far.
This is the first time I tested it w/out the led+resistor (6k)....
and neg side will climb rather quickly.
Is this normal when adding large-ish capacitance after the regulator?

=FB=

 

[bcarso]

It's not very prominent, but most datasheets do list a minimum amount of loading for the proper operation of these regulators.

(it's not the capacitor)

 

[clintrubber]

I do seem to recall some nastyness with the cap-value for neg-regulators though. Try for instance 1uF i.s.o. 220 uF, how does that work ?

IIRIC there's some info on this 'sensitivity' in the datasheet, but I'm not sure.
I do know that not all brands of 79XX are equal; some worked fine while other brands went ma-ad.

Never had such problems with 78XX.

 

[Freq Band]

I've built about 20 PSU's using this regulator topology ,with and without the cap(s) in question. I've got it memorized, if you will. After about the 10th build , I just assume they will work now. :shock: :green:
(because they do)

Yesterday, I built 2 of these circuits (p2p) at the same time.
Same results.
So I checked again the wiring and diodes...all OK.

It is possible the other psu's I built (in-service) can be behaving this same way....with the diode's resistor keeping them in check. I just fear the day the diode dies.....but maybe the circuit being powered has enough resistance in play to prevent the neg. reg. from "running away".

Maybe I'll unattatch the 220uf caps, and see if the behaviour disappears.

=FB=

 

[JohnRoberts]

While I got some disagreement the last time I posted this I have routinely put 1000uF on the output of 78/79 series 3 terminal regulators to improve their output impedance in the audio band.

Running with no load is kind of like that tree falling in forest, if nothing is hooked up, it doesn't much matter what it does. :wink:

JR

 

[Guest]

I use 2,200 mkF there (I bought a box on ePay couple of years ago) rated for input voltage because if the IC fails I don't want a firework with soaked in chemicals paper spread.

 

[enthalpystudios]

[quote author="JohnRoberts"]Running with no load is kind of like that tree falling in forest, if nothing is hooked up, it doesn't much matter what it does. :wink:
JR[/quote]


!! that's really funny... during a long drive the other day I was reading a great recording book from back in the day, tons of cool info on api 2488's and what not, anyway, i was thinking about electronics as i'm apt to do, and thought that... more about inputs and outputs and what not, but how without a load there is nothing... i dunno, silly things that i hope help me understand electronics better, but its really funny you say that not a day later!

 

[Samuel Groner]

www.groupdiy.com/index.php?topic=21613

Samuel

 

[clintrubber]

[quote author="Samuel Groner"]www.groupdiy.com/index.php?topic=21613

Samuel[/quote]
Right, that's it, load (and it pops up from time to time, a typical FAQ-candidate). My cap-remark was neg-reg related but another thing.

 

[audiox]

There is a minimum load that regulator requires to work properly. For standard type (TO-220) it is about 5 mA and for low current type (TO-92) about 1 mA. Just put a resistor to the output that draws say 10 mA during your testing.

There can be problems with 79/79XX and large output caps. Some types believe that large cap is a short circuit and activate the over-current protection. There are differencies between internal circuit of different manufacturers.

For audio use I strongly recommend to use LM317/LM337. They can handle large capacitive loads better. And with 10uF filtering cap in the reference pin they provide much much lower output noice than 78/79.

Why are you using non-polar cap in psu? It is just waste of money. It is also good to know that polarised electrolytic in polarised circuit is more reliable than non-polarised. Save your non-polarised caps to the audio path.

 

[JohnRoberts]

I have actually encountered some weaseling from 3 term regulator manufacturers about performance not specifically defined on a data sheet. One way to brag about high quality and low reject rate is to just not reject some less than desirable parts if there is not a demonstrable spec violation.

Long story short I had a specific issue with unusually high 1/F noise, that because of a circuit with less than wonderful PSRR was corrupting the audio path. This noise was problematic in parts from only one manufacturer, all other (identical PN) chips from other manufacturers were clean. When I brought this IMO problem up with the manufacturer, who BTW advertised their remarkable high quality and low reject rate, they blew me off saying they don't spec output noise on those parts. I wanted them to be aware in case the 1/F noise was symptomatic of process impurities, but in hindsight it may have also been the result of a die shrink (to reduce cost).

While I ended up removing them as an approved vendor on those parts, it is interesting to reflect upon what is meant by six-sigma quality (low PPM reject) levels. All it means is meeting defined standards that could be wide enough to drive a truck through (or in my noise case, no standard).

Another observation worth noting, even identical part numbers from different vendors may be different parts in some characteristic so caveat emptor. The obvious example is what do you actually get when you buy a 2N3055 these days?

Note: I also had another vendor trying to get approved as a second source on a widely used bifet quad opamp who declined to guarantee their noise spec... See ya, I don't believe in luck when dealing in high volume manufacturing.

JR

 

[mmatrix]

If the 7915 neg. regulators you are using are ST Micro, I have found them to be very unstable and will no longer use ST Micro's....I have had problems with the neg rail in about 4 different bipolar supplies and as soon as I replaced the ST Micro's with Motorola's, JRC's, Nat'l Semi's, etc., the negative side came right up and was rock solid....that's my 2 cents.

good luck.

 

[SSLtech]

I've seen manufacturer sheet information recommending that there should NOT be excessive output capacitance on 78/79 series regulators. Local to the unit, I never go over 1µF. All other big rail smoothing gets a resistor between it and the regulator.

Also, 79xx regulators tend to over-volt when insufficiently loaded... see the several 'SSl 9k PSU' threads on the subject.

I have a noisy 78 or 79 regulator in a preamp that I built, so I'm presently rebuilding the PSU with 317/337 regulation, AND some pretty serious L/C output filtering...

Keith

 

[bcarso]

[quote author="SSLtech"]
I have a noisy 78 or 79 regulator in a preamp that I built, so I'm presently rebuilding the PSU with 317/337 regulation, AND some pretty serious L/C output filtering...

Keith[/quote]

Be careful about clamping possible ringing in the L...

If you find the refs on excess output C I'd like to see it to see what their reasoning is. I would think that, if the transient response isn't screwed up, as long as there is a reverse protect diode for cases when the unreg goes down faster than the output, all that would happen would be momentary current limiting on power-up.

The low-dropout regs are quite sensitive to output C IIRC, but I haven't seen problems with 78XX ones at least. The way things often go though, I'll probably have my nose rubbed in it shortly now---I'm looking at a schematic as we speak with 220uF's on the outputs of 7815/7915s :roll:

 

[SonsOfThunder]

>>>I don't believe in luck when dealing in high volume manufacturing.

John, I feel your pain on that one, dude! Its one thing to get a proto working, its something else to get ten pcs working. And its altogether a differnt matter to make a hundred or more pcs work. Now usually by the time you get to 500 pcs, you can rest a little bit...as long as the parts don't get changed behind your back!

My experience is to avoid ALL ST Micro and Fairchild parts. Both have been nothing but trouble in my experience. Stick with what mmatrix says... though I don't think we have ever used JRC regulators. Their opamps are reliable.

HTH!

 

[AudioJohn]

On my PSU modules I have a 4K7 resistor load present for each regulator - as I have observed the output voltage floating up when unloaded. This value seems to be OK for any value of 78 or 79xx regulator.

I guess that it is fairly rare for there not to be any load on a PSU - obviously even a LED would normally be enough!

 

[JohnRoberts]

[quote author="mmatrix"]If the 7915 neg. regulators you are using are ST Micro, I have found them to be very unstable and will no longer use ST Micro's....I have had problems with the neg rail in about 4 different bipolar supplies and as soon as I replaced the ST Micro's with Motorola's, JRC's, Nat'l Semi's, etc., the negative side came right up and was rock solid....that's my 2 cents.

good luck.[/quote]

FWIW one of the good vendors you mention happens to be the one I disqualified (just for those two part numbers). But I killed them because of high 1/F noise, otherwise the parts were solid. Since they never cared enough to resolve whether it was a batch issue or intrinsic to the design, I'm not going on record as saying they still exhibit that noise today.

I suspect there are perhaps subtle design variants trying to share the same part number, to get the sales volume of industry standards while running across a different fab or using slightly different process technology so caveat emptor.

The old industry standard 78xx and 79xx parts I messed with were based on 741 grade opamp technology so neither very quiet or low impedance at 20kHz. My personal experiments determined that a 1,000uF electrolytic cap across the output was a nice complement for the part's output impedance that was rising 6dB/octave from well down in the audio band. Note: A newer technology regulator, with wider bandwidth process would probably exhibit a higher pole frequency for any break in output impedance so wouldn't need such a large cap for the same low and flat output Z result. My experience 20 years ago, may need to be revisited in case the parts have changed or if using different part. YMMV

Of course if using local RC power supply decoupling this is academic. While that approach is more focussed on low impedance at only HF rather than flat wideband low impedance PS.

That said I find it difficult to fathom a voltage regulator that isn't engineered to tolerate capacitance (or whatever) on it's output terminal and most are unhappy if there's not enough. Hell, we designed audio power amps to drive capacitors JIC.

Whatever, I wouldn't be surprised if there weren't counterfeit 3 terminal regs in the supply channel. There must be some decent volume in those PN.

JR

 

[bcarso]

JRC I find to be an odd duck. Certain parameters they "kaizen" like mad and things are much tighter than you'd expect from the spec. But I also have seen excess noise on 431's, and their opamps have been iffy at times. I have heard horror stories from others.

This one vendor (I'm not sure what GL stands for---maybe Goldstar??) sent a tube of LM358s that weren't unity-gain stable. Now I thought that took the cake. This tech kept telling me his circuit itself, a very-low-frequency twin-T notch filter, was o.k. and I flat-out disbelieved that it could be a part problem, but he was right.

 

[SonsOfThunder]

>>>But I also have seen excess noise on 431's

We had a couple of products where I decided to eliminate this part because it was a dang thermometer. THD wouldn't play well til the unit warmed up several tens of degrees above ambient. Not certain they were made by JRC, but very possible.

But I don't know of any problems with opamps other than the occasional wacko piece. We have used the NJM2068 in the 100s of k over the last 4 yrs.

HTH!

 

[gar381]

I've had problems with ST Micro 7915/18s as well.
Just about any other brand is much more stable.

GARY