[Megaptera]

How common is it for old, unregulated wallwarts to be above their rated voltage? I'm building a LM317 voltage regulating circuit, and three of the wallwarts I have sitting around all measured high -- a 15V measured at ~18V, a 4.5V at 6.05V, and a 12V at 15.5V. The Elpac WM080 measures a steady 24.3V out of the same power strip, so I'm pretty sure it's not my meter. Why would the others be high?

Also, could this be affecting the LM317's stability? With the 12V (really 15V) wallwart, it's putting out ~13.8V, but slowly rises by a mV or so every couple seconds. The 4.5V gives me a little over 5V, and it fluctuates up and down a mV every couple seconds. The bypass and filter caps are in place, with diodes, but I did settle for a 220uF electrolytic instead of the tant, and neither of them are as physically close to the LM317 as I'd have liked (I had to relocate it after, err, dremeling it off by mistake). Does this sound like a whole different second problem, or just a byproduct of whatever's raising the wallwarts' voltages?

 

[Garbz]

That's the whole point about an unregulated wallwart transformer. It's unregulated. 15V is RMS i believe so it's not uncommin for it to go up to 18. I believe as u approach it's current limit it will drop down to 15v or thereabouts.

That's also exaclty what the lm317 is designed for, to regulate this fluxuation in voltage. Since any change in ur mains voltage (brown out, or surge etc) will be reflected on the secondaires of the transformer.

I do believe but i'm not sure about it. The lm317 could safely regulate 15v with a 12v (unregulated) transormer providing there's a lot of capacitance across the input. The 15v transformer would begin to heat up the lm317 a bit. But that's why they should be on heatsinks in the firstplace, they burn the voltage drop in heat.

 

[bpribadi]

The LM317 needs about 5 volts higher than the output. Means, if you need 15 volts regulated, you need to give LM317 input at least 20 volts.

Use LM2941, it only needs 0.6 volts higher, so you can get 15 volts regulated from your 15 volts wallwart.

 

[Megaptera]

Quote:

Originally Posted by bpribadi Use LM2941, it only needs 0.6 volts higher

I'm really not too picky about the output voltage. I'll be running an AD8610 Mint with it, so anywhere from ~10-18 and I'll be happy.


Quote:

Originally Posted by Garbz 15V is RMS i believe so it's not uncommin for it to go up to 18.

Interesting, I knew an unregulated supply would fluctuate a bit, but I was assuming it would be on the order of 10s of mV, not something like +/- 20%. Though I guess I'm not surprised.

So would it make more sense for me to test the supply with a load more demanding than a voltmeter?

 

[Garbz]

ideally you would want to test you supply with the load you intend to power.

But this still only gives u a very close estimate. Any fluxuations in mains will still be present without regualtion even with a demanding load.

That said most opamps can easily live without regulation. I run an opa2604 based amp from just an unregulated wallwart.

 

[Megaptera]

I'll see if I can get my old Cmoy working again and try that -- should be a close enough approximation of the Mint, loadwise. I know it doesn't need it, so if it's jumping around a few mV post-regulator I don't really care, so long as it doesn't crap out on me. I'm mostly putting it in there because I can

Thanks for the help

 

[tangent]

Quote:

How common is it for old, unregulated wallwarts to be above their rated voltage?

Every single unregulated supply ever made will have a higher unloaded voltage than their rated voltage. It happens because transformers aren't 100% efficient, so when you draw more and more current, the secondary voltage drops. The rated voltage for an unregulated supply is where the voltage will be when you draw the full rated current. At lesser currents, the supply voltage rises. Only an ideal unregulated supply would behave otherwise.

Quote:

could this be affecting the LM317's stability?

No. That's what the LM317 is designed for.

Quote:

but slowly rises by a mV or so every couple seconds

That sounds like a temperature coefficient issue rather than instability.

Just as a wild guess, let's say your OUT-to-ADJ resistor isn't high enough wattage, so it's really heating up; 40C rise, say. Let's also say it's a carbon type, with, say, 350ppm/deg.C tempco. That's a 1.68 ohm change. Let's say the value is 120 ohms at room temperature, and the adjust resistor is 2.2K, giving a nominal ~24V.

With the tempco taken into account, the output voltage could rise by about 500mV, or drop by about 320mV. And it will do so slowly, as the resistor heats up. Sound familiar?

Solution: use higher wattage resistors where heat is a problem, and make them lower tempco types.

Quote:

The LM317 needs about 5 volts higher than the output

At full rated output current, dropout for the LM317 is 2.5V at worst. At lower currents, it's well under 2V.

Quote:

15V is RMS i believe so it's not uncommin for it to go up to 18.

RMS vs. peak voltage differences are only significant in the rectification stage. Once it's filtered, it's very close to pure DC, so RMS concerns (an AC issue) mostly go away.

 

[Megaptera]

Quote:

That sounds like a temperature coefficient issue rather than instability.

It could be, but neither of the resistors are getting the slightest bit warm to the touch, and using my finger as a heatsink doesn't have any effect. Same goes for the LM317. The fixed resistor is 1/4W, the trimpot is 3/4W.

I think at this point it might make sense to redo the circuit. The voltage fluctuations isn't a huge problem (but I was even getting pretty big swings with a 9V battery just now), but between that, the slightly botched layout, and the "ooh, let's see what I can do today" approach I took when I put it together, I just don't know that it'll be the most sensible idea to use this one.

 

[DIAR]

I have this exact problem with LM317 and I don't know what to do about it. Has anyone solved this mystery yet?

 

[Megaptera]

I traced mine down to a backwards capacitor. Well, the second build had a similar problem due to that -- the first build I never actually fixed, but I suspect it was a cap problem.

 

[DIAR]

Thanks. I'm designing a PCB module for LM317 which will have space for the resistors, safety diode and caps. I really hope it works. I plan to use it to replace LM7805 in my CDP. It feed DAC board so the voltage quality is very important.

 

[Megaptera]

Have you considered other regulators? The LM317 is versatile, but you may find your needs met by a simpler, non-adjustable reg -- your DAC probably wants somewhere between 3.3-5V, right? I don't know of any off the top of my head, but I'll bet there's a pair of steady voltage +/- regs that would reduce the complexity of the board and eliminate the need for a rail splitter.

 

[DIAR]

Here's the module I've been working on. In addition to these small caps on the PCB, there's 2000 uF before and 1000 uF after the regulator (already in the original CDP configuration).

I really hope it works this time...

EDIT: The PCB is designed totally wrong !!! Don't use it!!!

 

[Megaptera]

I've never translated schematics into a PCB, so I'm not sure if what I've noticed is really of any consequence, but I hope I can offer some constructive comments:

Is the space with the two circles for a cap? Why is it going from the adjustment path to the output path? If it's for the large cap bypassing the adjustment path to ground (10-100uF OS-CON) it should be placed before the 240Ohm resistor and connect to the ground section.

Speaking of resistors, what kind will you be using? You'll want at least 1/4W here, I believe, which may be a bit large if you're not doing surface mount.

Also, will it matter that the 750Ohm resistor is placed after the 240Ohm resistor and A-O diode? And if I'm remembering the LM317 datasheet correctly (my coffee hasn't quite kicked in yet), the I-O diode placed after the A-O diode should not matter.

I assume the blue shaded section is a heatsink, right? The LM317 will be dropping a good amount of voltage here.

It looks really nice, overall -- very compact. Please take my comments as genuine questions and helpful notes

 

[00940]

For the dac section, I wouldn't use the LM317. Too noisy. What DAC chip is that ? If consumption is low (<100mA), consider a pcb for the REG101-5 from TI. It's a soic chip. With a small cap on the CNR pin, the noise is only of 7µV/VRMS, to compare to the 3mV/VRMS of the LM317. You probably won't get such low noise but you can at least get better than with a LM317. REG102 and 103 can give you more output but noise is slightly higher.

Furthermore, no need for 1000uf after the regulator. A 100 or 220uf oscon near the DAC is all that's needed.