At least the output is lower than the input now!
Now measure the voltage across each of R4 and R5. That is, connect the meter, on the component side of the board, to the wires of the resistors.

 

At least the output is lower than the input now!
Now measure the voltage across each of R4 and R5. That is, connect the meter, on the component side of the board, to the wires of the resistors.

The |absolute value| across the resistors is:
R5: 2.54 VDC
R4: 7.95 VDC

 

OK, I think the problem remaining is that the output current is below the minimum specified. To test this connect a resistor between 100Ω and 500Ω between the output and 0V and see if the output voltage is then correct (about 3.5V).

 

OK, I think the problem remaining is that the output current is below the minimum specified. To test this connect a resistor between 100Ω and 500Ω between the output and 0V and see if the output voltage is then correct (about 3.5V).
View attachment 127705

OK, I went back and looked at the data sheet and found the graph you posted. It would seem that we need at least a 5+ mA load to trigger regulation.

I hooked up a 330 ohm resistor across the output (~16mA) and here are the results:

OUT: -5.3 VDC
ADJ: -4.1 VDC
IN: -6.0 VDC
Hmmm? Somewhat of an improvement (I think) but......
What caused the drop in the input voltage?
The output should be ~ -3.5 VDC, why isn't it?
ADJ should be around 1.9 volts, why isn't it?

confused & frustrated in western PA

 

OK, the difference between output and adjust is now 1.2V which is what it should be but the output voltage is wrong.
Now, with it switched off, measure the resistance between the adjust pin and the right hand end of C2 and between the adjust pin and the output pin. Try each of those with the meter leads both ways round and report the higher reading for each position.

 

OK, the difference between output and adjust is now 1.2V which is what it should be but the output voltage is wrong.
Now, with it switched off, measure the resistance between the adjust pin and the right hand end of C2 and between the adjust pin and the output pin. Try each of those with the meter leads both ways round and report the higher reading for each position.

Thanks for your help Al. Will do tomorrow, the Grandkids just got here.

 

Thanks for your help Al. Will do tomorrow, the Grandkids just got here.

It's my sleep time anyway.
Zzzzzzzzz....

 

After some simulations, I think I may have located the difficulty.
If you are taking the output for the negative supply from the transformer at the input to the bridge rectifier, then you need a load on the positive output of the bridge equal to the negative current drawn.
Otherwise there's no path for the negative current and you will get flakey results.

Below is the LTspice simulation of the circuit.
The Vin voltage magnitude starts to drop when the bridge output current through R5 drops to ≈10mA and the regulator output, V(out), drops out of regulation when the current drops to ≈5ma.

 

After some simulations, I think I may have located the difficulty.
If you are taking the output for the negative supply from the transformer at the input to the bridge rectifier, then you need a load on the positive output of the bridge equal to the negative current drawn.
Otherwise there's no path for the negative current and you will get flakey results.

Below is the LTspice simulation of the circuit.
The Vin voltage magnitude starts to drop when the bridge output current through R5 drops to ≈10mA and the regulator output, V(out), drops out of regulation when the current drops to ≈5ma.

View attachment 127716

Does the voltage across C1 rise above the peak AC voltage with larger R5 values?
And that still doesn't account for an output voltage higher than that set by the resistors on the LM337L adjust pin.

 

OK, the difference between output and adjust is now 1.2V which is what it should be but the output voltage is wrong.
Now, with it switched off, measure the resistance between the adjust pin and the right hand end of C2 and between the adjust pin and the output pin. Try each of those with the meter leads both ways round and report the higher reading for each position.

ADJ > GND: 225 ohms
ADJ > OUT: 167 ohms
Meter reads the same when leads are reversed

 

Does the voltage across C1 rise above the peak AC voltage with larger R5 values?
And that still doesn't account for an output voltage higher than that set by the resistors on the LM337L adjust pin.

Then there's something flakey about your circuit or the LM337.
Post the complete circuit you are testing with part values.

 

Those are correct if you still have the 330Ω across the output so I am out of ideas as to why the output voltage is too high.

 

It's a standard half-wave rectifier and VR circuit straight out of the book. The PCB traces line up with the schematic. I still have the 330 ohm resistor tied from the output to ground.

 

So is the board fully populated now, because initially you stated you were starting to populate the board?

 

So is the board fully populated now, because initially you stated you were starting to populate the board?

In particular, are C1 and R1 fitted?

 

In particular, are C1 and R1 fitted?

No, it is not fully populated. My plan was to start with the 33 volt bus, the 24 volt bus and the -3 volt bus and check them as I went along. The 33 volt and 24 volt systems are fine, but the negative 3 volt bus is driving me nuts. C1 is connected to the top left terminal board. R1 is the large resistor behind it. I am using Diptrace for the PCB design and it is virtually foolproof unless you use the wrong component footprint so I am confident that the traces are correct. You can see my 330 ohm dummy load off to the right.

 

In particular, are C1 and R1 fitted?

Would a snapshot of the waveform at the OUT pin be helpful?

 

Then I repeat my request.
Show the circuit diagram with the parts crossed off that are not on the board (leaving the circuit you are testing).
Without knowing exactly what the circuit is, we are spinning our wheels.

 

Then I repeat my request.
Show the circuit diagram with the parts crossed off that are not on the board (leaving the circuit you are testing).
Without knowing exactly what the circuit is, we are spinning our wheels.

Everything in the enclosed areas is connected either on or off the board. The only component not shown on the schematic is the 330 ohm load resistor for the -3 volt bus.

 

So have you tried adding a load to U5's output to see if that makes a difference?

Are the two transformer output windings in parallel?