Strange issue with a bipolar power supply

[CurtZHP]

I've got this bipolar power supply that uses adjustable voltage regulators.  As built, it's supposed to put out +/- 18VDC as well as 48VDC phantom power.

I've built this same supply before and used it successfully.  This time around, something's not right.  Or maybe I'm just not testing it properly.

The phantom section is just fine.  It puts out about 52VDC.  The bipolar rails are screwed up.  Each regulator (LM317 and LM337) sees about +/- 38VDC on their input pins.  On the output pins, I get about 35VDC on the LM317, and I get about -10VDC on the LM337!

Thinking something was up with those chips, I replaced both of them with brand new ones.  Still the same problem.  I double-checked the adjustment resistors to make sure I had them in the right places.  I double-checked the orientation of the protection diodes, D103 and D104. 

The LED's and their respective resistors are not installed yet.  Made sure none of my caps were in backwards, even though I didn't get any fireworks.  I figure BR1 is fine since each regulator is getting the identical +/- input voltage.

I removed C116 and C117.  Here's where it gets interesting.  Once I did that, LM337 started giving me -22VDC output.  That's more like it.  But LM317 still shows over 30VDC on the output.  That's a head-scratcher.

Mind you, this supply is not currently connected to anything.  Being unloaded, I would expect to see voltages a little lower or higher than specified, but not this far off!

Now I'm starting to worry about the last one I built!  Unfortunately, I no longer have it to compare to.  (That gadget's now in someone else's studio...)

 

[Audio1Man]

C117 is BACKWARDS. + should be to common and - to ref terminal.
Duke

 

[abbey road d enfer]

The phantom section is just fine.  It puts out about 52VDC.  The bipolar rails are screwed up.  Each regulator (LM317 and LM337) sees about +/- 38VDC on their input pins.  On the output pins, I get about 35VDC on the LM317, and I get about -10VDC on the LM337!

Th eresistor values are very wrong! The LM317/337 require 5mA to regulate correctly. If you had connected the LED's, maybe you wouldn't hace noticed this issue.
The datasheet recommends 240 ohm resistor between Out and Adj, which draws 5ma so the circuit is somewhat loaded.

 

[CurtZHP]

C117 is BACKWARDS. + should be to common and - to ref terminal.
Duke

No wonder it worked better when I took that out!

:-[

 

[CurtZHP]

Th eresistor values are very wrong! The LM317/337 require 5mA to regulate correctly. If you had connected the LED's, maybe you wouldn't hace noticed this issue.
The datasheet recommends 240 ohm resistor between Out and Adj, which draws 5ma so the circuit is somewhat loaded.

240 seems like an oddball value.  Would 220 be close enough for rock-and-roll?

 

[abbey road d enfer]

240 seems like an oddball value.  Would 220 be close enough for rock-and-roll?

Actually I've often used somewhat higher values, like 470r. As long as there is a significant load, it works well enough.

 

[CurtZHP]

I also noticed that, for the LM337, the recommended R1 value is 120Ohms.

But, like you suggested, as long as I'm in the neighborhood....

(and 3.2K is definitely NOT in the neighborhood!)

 

[CurtZHP]

Went digging through the parts bins and managed to find some resistors whose values were at least in the same ZIP code as what the spec sheet called for.  Swapped them out and retested.  +/- 19.2VDC.  That's unloaded, mind you; so I think I'm good.

Thanks again, guys!  I shudder to think what I would do without you.

 

[Whoops]

Reference your schematic to the JLM PSU schematic it uses the same regulators and it's really well done and prooved