Hello,
I'm trying to make a small power supply using a common schematic found on the web which has been around for many years. For comparison, it is the smaller diagram in the blue box at the link below.

Due to availability issues, I had to use 2 transformers in series to get the voltage I wanted. I had to substitute 68 uF, 200 V caps for the 4700 uF caps, and use LM317 and 337 voltage regulators with their associated circuitry for adjusting the voltage (highlighted in purple.)

I'm trying to get +15 and -15 V, and only need <100 mA. The circuit is supposed to be good for up to 1.5 A. The expected voltages are shown. The actual measured, open circuit, voltages are in red, in parentheses.

I can't find anything wrong, or any bad parts, but when I put the voltage regulators in the circuit, the maximum voltages I get are +8.7 V and -7.0 V, when the regulators should be good to +/- 37 V.

I tried different, also new, regulators, and changed R1 from 240 ohms to 120 ohms (what it should be depends on where you look.) I still get the same output.

All parts are brand new except the 68 uF caps which I scrounged from a fluorescent light circuit. I tried 4700 uF, 35 V caps temporarily (the voltage they see is about 38 V) but that didn't make any difference either.

Does anyone see anything I'm missing?
Thanks in advance.

P.S. I want to power a INA121 instrumentation amp, and small FET circuit.

 

The circuit is all good.

It is possible that the one u constructed has errors.

You have to build step by step. Eliminating false connections.

By the say, did you check the data on the LM, the R1 & R2 plays an important role in setting the voltage

 

The filter caps are way too small. Odds are you have pulsating DC on the output of the regulators.

Measure the voltages across the diode bridge to see if you have the expected voltages (a wiring error check). Referenced to ground you should have 25VAC on each side of the diode bridge. Measured across the diode bridge you should have 50VAC.

 

The datasheets for the LM say R1 should be 240 ohms and that is what I used. I have also seen the same posted circuits showing R1 being 120 ohms, so I tried that too. It should have given me a larger voltage, but I got exactly the same voltage.

I know the filter caps are too small, but that's all I have at the moment that can take the voltage. I tried 4700 uF, 35 V caps, even though they are underrated for the voltage, but there was no change.

My transformers are connected correctly and their output, as well as the bridge output is correct, as shown by the red numbers in parentheses (open circuit values.)

I have been over and over the circuit, checked the caps and diodes and they all check out.

There is one possibility. On most electrolytic caps I have (axial leads), the negative side is marked by arrows. The particular 10 uF caps I used have radial leads. One side has arrows and a shorter lead so I assumed that was the negative side just like the axial lead caps. Could they be marked differently?

 

Connecting Caps in reverse will heat them up and cause the voltage to sag.
Check them. If you have reversed them, do not use them. They will be faulty

 

A way to be sure is to remove all the caps besides the smoothing ones.
If Voltage rises then u know. But in order to use the supply, you will need caps.

 

Does the voltage move when you adjust it? I'm wondering if you've mixed up the pins on the regulator, since the pinout is different than the schematic layout. How could I possibly think that? Been there, done it. Simple errors are easy to make and hard to find.

 

Wayneh, I'm aware of the pinout difference and was careful about that. I can adjust the pos and neg voltages. When R2 is 0 ohms, I get 1.25 V out. This is a correct voltage. Polarities are correct. When R2 is 5K ohms, I get +8.7 V and -7.0 V. I should get +/-15 V when R2 is about 2600 ohms. I should be getting about +/-27 V when R2 is 5K ohms. I can't seem to get above +8.7 V and -7.0 V.

I tried bypassing the regulators and adjustment pots, so it was a circuit of only caps and diodes, and I get about 74 VDC out. No caps overheated.

 

I do not think this is related to your problem, but note that the 1N4003 diode and 10μF capacitor associated with the LM337 in your diagram appear to have been interchanged.

This rather smaller PU shows the configuration. http://www.elecpod.com/powersupply/2010/06062564.html

 

Note that if the 10μF capacitor is in its correct position, but leaky (faulty, too low working volts, or reversed) this could explain your problem. Leakage in this capacitor, in parallel with the 5kΩ pot, could limit the maximum voltage.

 

Rough calculation; 1.25v/240 Ohms = ~5.2mA, across 5k would mean ~26v @ the ADJ pin, +1.25v ~= 27.25v out. That's not counting for Iadj, which would add a bit.

I suspect that your "5k" pots are actually ~1k Ohms, which would explain the difference.
If that is the case, replacing the 240 Ohm R1 resistors with 91 Ohm resistors should get you up to around 15v.

 

As far as the cap and diode for the 337 goes, I just copied their positions for the 317. For some reason the datasheet for the 337 does not show a circuit for adjusting the voltage, even though it is adjustable.

I just did an experiment where I removed R1 and R2 and the connection to the adjustment pins on both of the LM regulators. Everything else is in place. My output voltage jumped from 15.7 V total to 36.7 V on the neg side. This is good. This is the regulator max. I get no pos voltage any more - I think I burned out the 317 with a short. However, this tells me that, at least without any adjustment circuit, the 337 is working. So, this narrows down the location of the problem. I'm changing the 317 with a spare now.

I'll try varying R1 and R2 to find a combo that works (though this doesn't really solve the problem yet for why the original R1 and R2 didn't work.)

 

Sgt.
My 5K pots do measure 5K. They also adjust fully from zero to 5K. The 240 ohm resistors also measure correctly.
But something in the adjustment part of the circuit is bad. Maybe it is a leaky 10 uF cap. The caps I bought are new, but cheap.
I'll replace them next after fixing the burned out 317.

 

Your transformers will output 25.2vac @ their rated current, but *1.41 - 0.7v for DC, or ~34.8v for each unregulated rail, or about 69.6v from the negative unregulated rail to the positive unregulated rail (on the filter caps). This is at the rated transformer load; it may be significantly higher when there is no load on the transformers.

The LM317 and LM337 are rated for a maximum differential from the IN to OUT of 40V; at high voltages the output current is reduced in an attempt to keep them from getting fried due to excessive power dissipation as heat.

The tabs are electrically connected to the OUT terminal, therefore the heat sinks must be electrically isolated from Ground and other non-OUT voltage circuits.

You will need heat sinks on the regulators, as even light loads will result in significant power dissipation in the regulators. The lower voltage the output (due to the voltage differential across the regulators themselves), the worse the power dissipation in the regulators will be. If you have a ripple-free 40v in and 15v out with 100mA current, your load will be dissipating 15v*0.1A = 1.5 Watts, and your regulator will be dissipating (40v-15v)*0.1A = 3.5 Watts of power.

Your filter capacitors are very low in value, as has already been mentioned. With a 100mA load, 68uF caps will have about 14.7v of ripple on them. You should have at least 2,000uF to get down to 1v of ripple at 100mA current. Excessive ripple will heat your capacitors rapidly, and may cause them to self-destruct.

 

As far as the cap and diode for the 337 goes, I just copied their positions for the 317. For some reason the datasheet for the 337 does not show a circuit for adjusting the voltage, even though it is adjustable.

The circuits should be symmetric: cap across pot for one > cap across pot for the other.

 

Sgt.
Everything you mentioned is close to what I'm actually seeing except that the 317 tab is connected to the Out terminal, but it is the In terminal for the 337. Both heat sinks are isolated (except for the recent temporary short which has been fixed.)

My no load DC voltage from each rail to gnd is 38V and within the LMs capability.

My estimate of <100 mA was high, it'll probably be about 50 mA.

I just tested the unit with no adjustment resistors, and all 10 uF caps have been removed. The voltage between pos and neg rails starts out about 91 V (wow) when first plugged in and drops to about 73V within 20-30 seconds, and stays there. This is suspicious. For such a well used circuit to have such a potentially damaging output seems screwy, but maybe it is because there is no adjustment connected at the moment.

I'll be buying new, higher voltage filter caps tomorrow. For now, I'm going to try some new 10 uF caps from the batch I just bought. Then I'll look for the R2/R1 ratio that works.

Adjuster.
I'll be making the cap & diode arrangement symmetrical now that I'm replacing the caps.

 

Why don't you try it with just ONE transformer instead of two? You'll need larger filter caps, and you might have to bump your output voltage down a bit (say, +/-14v or +/-13v) but you will have very little stress on your regulators.

 

Well, it looks like you guys have helped solve another problem. Apparently I must have had one (and maybe 2) leaky caps because after swapping out the existing 10uF caps with new ones I got the unit to work exactly like it should. Both pos and neg.

Tomorrow I get bigger filter caps, but it's working as is for now.
Also, no more 91 V when I plug it in. It jumps right to where it should be and stays right there.

Thanks for the quick help guys.