Simple voltage regulator get crazy... and so am I

[Malefoda]

Hi folks,
Maybe I deserve to be treated as an idiot but I can't figure it out.

Take a TX, some schottkys, a 2200/35V Rubycon ZL to smooth that.
Then I have 17V.
Well that is too much for the numerous LM317@5V after (heat!).

So i fitted, just after the cap, a µA7810 from TI, followed by a 2200/16 ZL.
With output leg in air it's 10V, when soldered with the cap... madness, my DMM can't auto-scale, it happened once that I saw 10V for few seconds but also that the TX started to humm...

I just don't understand, did I over rated the later cap? I think it is better with, else there will be nothing between my 7810 and many 7805 at 10/20 cm away... help!!!! When I see a 4700µF after a 7805 in my CDP I don't understand why that stupid thing goes wrong...

Matthieu wich is used to have troubles only with simple things...

[janneman]

Originally posted by Malefoda
Hi folks,
Maybe I deserve to be treated as an idiot but I can't figure it out.

Take a TX, some schottkys, a 2200/35V Rubycon ZL to smooth that.
Then I have 17V.
Well that is too much for the numerous LM317@5V after (heat!).

So i fitted, just after the cap, a µA7810 from TI, followed by a 2200/16 ZL.
With output leg in air it's 10V, when soldered with the cap... madness, my DMM can't auto-scale, it happened once that I saw 10V for few seconds but also that the TX started to humm...

I just don't understand, did I over rated the later cap? I think it is better with, else there will be nothing between my 7810 and many 7805 at 10/20 cm away... help!!!! When I see a 4700µF after a 7805 in my CDP I don't understand why that stupid thing goes wrong...

Matthieu wich is used to have troubles only with simple things...

Two possibilities I can think of: the 7810 is oscillating, or it is connected the wrong way around. Did you connect the ground pin of the 7810 to the samr ground as the caps? How long is the wire between caps and 7810?
But if you could post a circuit diagram it may jump out at you/us.
I know, the circuit should be easy to visualize, but then if it was so easy it would work, right?

Edit: any chance that the last cap is the wrong way around? That would explain the tx hum and the 10V when unloaded. It would also explain the multimeter going crazy: the cyclic current limiting of the 7810 causes the Vout to fly all over the place.

jd

[Khron]

The 7810 oscillating is still a possibility. The ZL is a low-ESR series, and linear regs don't really get along with low-ESR caps on the outputs.

You could try and see what happens with a smaller cap after the 7810, or at least with a non-low-ESR cap.

[Malefoda]

Thanks gentlemen,

the cap is on a test board with the 7810, maybe 3mm after And yes I've checked many times and can say it's the right wiring.
I will try the same but a LM317 and/or a 220/50 ZL. In datasheets they say a 0.1 or 1µF is recommended in most cases before the filter cap if it's far away (what does "far-away" stands for in mm???) and after (I have loads of Wima's MKS if needed...). But what in the case of "bewteen" pre-reg 7810 and regs LM317?

Matthieu

[Khron]

Once again, even if only as a test, try a non-ZL cap, if you have any, at hand At least to see if that's the problem or not.

[Malefoda]

Unfortunately my testboard will not survive that I guess...(bended cap legs to make links...) anyway it's faulty.
Until this answer I read on Internet many opinions... like input//filter cap (same in my case) smaller or as big as the output (2200//2200µF), lack of tiny caps (1/0.1) around or such...
In fact I'm also concerned by the best solution, AFAIK LM317/Low ESR in audio (CDP) do better than 7810/Non-Low ESR? If the 7810 is a part of the problem I may remove it
I'll tell you what I've found tomorrow, no light at the work table right now.
Matthieu

[Matt.B.H.]

I have had a few 7810s oscillate like mad before. Is it an old one? I find the older 7810 and 7809 to be particularly troublesome. I guess they just fool a standard 5V one by floating the ground pin with a few PN junctions, a sure way to make them unstable. The newer ones seem fine however (better, more accurate trimming of an actual 10V reg).

The 78/79 series of regs are very variable from manufacturer/manufacturer and batch to batch. So try another if you have one and or lose/lower the 2200uF cap. Or if you want to keep it, first try the circuit without it, then add some R in series with it untill it is OK.

Cheers Matt.

[kevinkr]

Probably a stupid question but what are you using for a heat sink on the regulator? Also most LM317 are not happy if the load current is significantly less than 10mA which you can usually cater to by setting the reference current to 10mA. (121 ohm resistor)

In terms of mm if the supply cap is within 50mm of the regulator there generally should not be a problem.

The 317 generally works fine with 30V supplies and can be operated at much higher voltages provided that the differential across it does not exceed about 35V. (I've used them in 300V supplies - similar to the Maida without problems, I believe something SY does to this day.)

[ikoflexer]

Humming transformer? Sounds like a short to me. Try a light bulb in series with the transformer primary. If it lights up, it's a positive.

[zebra100]

I would second the comment on heat. Is the regulator a TO-220 package? If it is, these are suitable for current up to 1A, which should be fine for your application. You could push it even higher with suitable heat sinking.

If it is in a TO-92 package, it will be unsuitable for your application, as the heat generated will cause it to become unstable after a few seconds of operation.