I have this circuit on a breadboard and it works exactly as I expect. If you put in -5V you get out +5V, if you put in 0V, you get out -5V, and if you put in +5V you get out 0V. enter image description here

The problem I'm running into is that I initially built this and several other similar circuits with the first bunch of diodes I had to hand without particularly looking at them. It turns out I was using some germanium diodes from an old radio project. They are 1N276 and something in a glass case with a green stripe but no part number. I only had a few so I couldn't continue with this project until I got more diodes. I tried 1N4001 and 1N4003 because I had them on hand, but didn't expect them to work (they didn't). I then bought some 1N4148's which I did expect to work since they are "small signal switching diodes". Unfortunately, the resulting outputs just oscillate wildly.

I'm kind of baffled because the diodes I would expect to work perfectly for this application don't, and the diodes I would never have specified for this circuit work perfectly. Any ideas as to why this is the case and a proper replacement part that will work as expected?

Thanks much!


2 Answers 2


The circuit can't work exactly as shown, there needs to be a load from the output to -5V that is much higher than 2K, say 100K, There should also be bypass capacitors across the two supplies, near the chip. Something like 100nF each, or you could get oscillation.

It should work with either of those diode types you tried, or with Ge or Schottky diodes, the only difference will be slightly different output voltages. With the two you tried, the output voltages should be a bit (~0.4V) lower than than with the Ge diodes at 0V and +5 (perhaps -0.8 and +4.1V with a light load to -5V, rather than -0.4 and +4.5V). To get similar behavior to the Ge diodes, you could use a Schottky diode such as the BAT54C.

  • \$\begingroup\$ I know it doesn't look like it should handle the -5V correctly, but that's a peculiarity of the LM319. Apparently the path to Vee is handled internally somehow. It definitely works as I described above. I've never found another part that acts that way. \$\endgroup\$ Commented Aug 5, 2015 at 20:44
  • 1
    \$\begingroup\$ I know what the comparators do- that allows split supplies with logic level output. It's not a matter of the comparators- the voltage on the other side of the diodes is indeterminate without a load resistor to -5V. \$\endgroup\$ Commented Aug 5, 2015 at 21:58
  • \$\begingroup\$ Wow. I have no idea why that worked. I put in the 1N4148's and added the 100K resistor but it still didn't work. However, I noticed that the oscillation was slower so I bumped the resistor up to 470K and it works perfectly. I honestly don't get why I'm providing a path to -5V here... \$\endgroup\$ Commented Aug 7, 2015 at 8:24

Germanium diodes have rather low forward voltage. Normal silicon diodes can get up to 0.8V, while germanium is a fraction of that.

The 1N400X is specified to have 1V drop (at 1A though), which probably plays with your +/- 1V references in some unwanted way.

I think the modern equivalent are Schottky diodes, which easily go down to 0.3 Vf.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.