Hey guys-

I have built this power supply circuit:

I have eliminated the LED leg and the voltmeter leg. The transformer is 120VAC to 20VAC. After the rectifier, I have ~20VDC. If I leave out the large filter cap, I still have 20 volts at the output- until I ground P1. (If I leave the filter cap in, I get no voltage.) The pot (B5k) measures out fine and does not appear to have a problem, but once I connect it, everything dumps to ground, even when turned to maximum resistance. I replaced the 5k with a variety of heavier pots, which progressively raised the output voltage. With a 1M pot, I got ~5VDC, not nearly enough. Can anyone offer suggestions as to what is happening here, and how I can fix it? Thanks.

 

Sounds like you have the connections wrong on the regulator, post a picture of your wiring.

 

The minimum load current required to guarantee regulation is 10mA.

In the 40 years I've been using LM317, I've only encountered one LM317 that wouldn't regulate at 10mA.

 

If you are getting 20 VDC across the filter capacitor with no load and a transformer with a 20 VAC secondary, something is definitely quite wrong. I would expect something around 30 V on the filter cap under these circumstances since an unloaded transformer typically will produce a voltage a bit above nominal and an unloaded or very lightly loaded filter cap will charge to nearly peak voltage, which is about 1.4 times the RMS voltage.

 

Don't leave out the large filter cap.
It's required to smooth the rectified voltage to DC (with some ripple voltage depending upon the load).

With 20Vac into the rectifiers, the output should be over 26Vdc.

 

until I ground P1

Have you wired P1 correctly? Not all pots have the wiper pin between the other pins.

 

The transformer is 120VAC to 20VAC. After the rectifier, I have ~20VDC

Not possible. You should have 1.4*20 VDC across C1 and C2.

Check the voltage drops across the individual diodes in BR1.

You may need the -HV version of the regulator. Check the max input output differential with the datasheet. http://www.ti.com/lit/ds/symlink/lm317.pdf OK.input/output differential voltage is 40, so you should be OK.

C2 and C3 should be close to the body.

There is an extra protection diode that should be used.


You should have about 1.2V across the 220 ohm resistor no matter what the output is.

Shorting the ADJ terminal to ground will make the regulator output 1.2V

A an unregulated high voltage output can point to the lack of a minimum load of >=10 mA.

Watch the pin out. Watch how the LM317 is mounted. The output is the metal tab. You need a TO-220 transistor mounting kit to attach it to aheat sink. The tab stays electrically insulated from the heat sink and thermally connected.

Later, suggest an appropriately sized ZNR across C1 at C1.

Big things to check is pin out and check for shorts to the case because of not using a mounting kit.

The rest are basic improvements for a bench supply.

==

Seems like we all are seeing the rectified 20 VAC not supposed to be 20 VDC. That's a common mistake.

Check the grounds.

Measure the 220 ohm resistor. make sure it's 220 ohms.

Look at Fig 9 of the datasheet.

The TO-220 mounting kit contains a mica or silpad washer, a bellville washer (conical shaped), a screw and nut and a step collar. The Silpad doesn;t require heat sink grease.. The conical shaped washer applies a constant force. The step collar insulates the metal screw from the device.

Maybe you need to take some pics. Lots of construction problems are possible.

 

After thinking about this some more, I'm suspicious that the problem is in the transformer or the bridge rectifier. The symptoms seem to be consistent with the input to the regulator being a high-impedance source, so that the a little loading causes the voltage to drop dramatically. The fact that the voltage drops when the big filter capacitor is connected supports this idea. Electrolytic capacitors have a certain amount of "leakage current" - a small DC current will flow through (which would not be the case with an ideal capacitor). A big cap, particularly one that hasn't had voltage applied to it for quite some time, will often be quite leaky when voltage is first applied, then over a few hours the leakage will drop and stabilize to a low level, assuming the cap is OK.

For the following, you'll need a resistor of something in the range of about 5k to 10k. It should be able to handle a quarter of a watt. I'm also assuming that you can disconnect some things without too much difficulty.

Start by disconnecting the transformer and checking the voltage on the secondary. Check it first with nothing connected, then with the resistor connected across the secondary. There should be no more than a tiny voltage difference. Reconnect the transformer.

Disconnect everything on the output side of the bridge rectifier. Measure the DC voltage at its output first without and then with the resistor across the rectifier output. You might get a bit bigger difference here, but it shouldn't be more than a volt or so (the "forward voltage" of a biggish diode can be very low if there is only enough current flowing to provide that for the meter input, but with a few milliamps it usually is in the 0.6 V range). Most meters will respond to something close to the average voltage. You are getting pulsing DC and the expected average without the filter capacitor would be a bit less than what you measured as AC directly from the transformer. If the small capacitor is in the circuit there will be a fairly big difference between the voltage without and with the resistor.

Reconnect the big filter cap but not the regulator. The voltage across that capacitor should be about 1.4 times than the AC voltage you measured from the transformer. A bit less is to be expected because of the voltage drop across the diodes in the rectifier. There won't be much difference between measuring with the resistor across the cap or without.

If everything is OK up to this point, then a close visual inspection of the rest of the circuit is necessary. The LM317 pinout is quite different from most 3-terminal regulators.

 

I appreciate everyone's responses. I'm not at home right now, but I will go through the circuit with your suggestions and report back.

 

I appreciate everyone's responses. I'm not at home right now, but I will go through the circuit with your suggestions and report back.

It would be helpful if you posted voltages at the three terminals of the regulator.

 

UPDATE:

Testing the transformer as ebp suggested, with a simulated load of 10k ohms, dropped the output voltage from 20V to nearly nothing, .068V. I guess I need to dig up another transformer. I just wasn’t aware that a transformer could behave this way.

 

That is certainly a very strange thing for a transformer to do. I've never seen such a thing.

I suspect that the secondary winding may be open circuit at some point but that there is sufficient capacitance between the primary and secondary to produce some coupling ("high isolation" transformers use special winding design and/or a shield between the primary and secondary to prevent capacitive coupling, which is quite undesirable especially for high-frequency noise). But it might be something like a bad solder joint between the windings and a connector pin or lead wire.

 

In addition, the 0.1Mfd capacitors right across the regulator need to be there to prevent possible oscillation. Transformers can have some strange problems, including an open condition of an internal over-temperature device. I have come across a few of those.