I am looking to make something that takes a pulse of floating voltage (say from a feedback coil of on a transformer), and maintain a differential output voltage depending on the sign of the amplitude of the pulse. Then when another pulse comes along with a different sign, the output would change signs as well. I naturally looked into flip flops/latches, but after some experimentation and closer reading, I realized they do not really provide a differential output, and they do not seem to work with floating inputs. They seem to work instead with pulses of non-floating voltages and send a high to one of two nand/nor gates.

In summery: My problem is I am given an isolated voltage in the form of a pulse. I want something that will "remember" the sign of that pulse in the form of a voltage difference between two output leads.

The reason I emphasize the "differential" part is that the two output leads of a flip flop/latch will not form a complete circuit when connected through a load or something that senses voltage differences. It is just an open switch and a closed switch.

I may just be doing this wrong, or misunderstanding, but based on what I have tried, this seems to be nontrivial.

  • \$\begingroup\$ It's not clear what you are asking about, the flipflop of the isolation. It sounds like a set/reset flipflop would do for the first, and a opto-isolator for the second. What "differential" has to do with any of this makes no sense. \$\endgroup\$ Commented Jul 1, 2012 at 22:29
  • \$\begingroup\$ My problem is I am given an isolated voltage in the form of a pulse. I want something that will "remember" the sign of that pulse in the form of a voltage difference between two output leads. I will add that in the bottom of the post as a summery. Thanks for telling me I was not being clear. \$\endgroup\$ Commented Jul 1, 2012 at 22:36

2 Answers 2


If I understand your question right, a simple centre tapped rectifier like this should do:


L1 is the primary winding, L2 and L3 are the secondary windings. OUT1 and OUT2 are to your FFs (or whatever you are capturing with) Ignore the 1Meg R3, it's just there to keep SPICE happy.
You can use the correct windings ratio to adjust the levels as desired (may be useful if high voltages are involved), and add e.g. a couple of zeners to protect your inputs. There are also many other ways to do this, depending on exactly what you are trying to do.

Simulation of above circuit (light blue is input pulse waveform, blue is negative pulse output, green is positive pulse output) EDIT - to make it clearer I simulated with a pulse file rather than a square wave as shown in the schematic:


  • \$\begingroup\$ Thanks for the answer, but the problem I see with this is that it would not have the 'memory' functionality I was looking for. I wanted something that could remember the sign of the pulse even after the pulse died down. Then when another pulse (most likely in the opposite direction) came along, the outputs would switch signs. I apologize for not being more clear. \$\endgroup\$ Commented Jul 2, 2012 at 0:46
  • \$\begingroup\$ I just realized FF=flip flop. So you are suggesting I use a center tapped transformer to add a reference frame (ground) to the outputs for the flip flop. This, presumably, would eliminate the floating voltage problem. Do I understand correctly? If so, I will have to try that. \$\endgroup\$ Commented Jul 2, 2012 at 0:50
  • \$\begingroup\$ Yes, FF = flip-flop. The outputs send pulses to the flip-flops, one captures negative pulses, the other positive. \$\endgroup\$
    – Oli Glaser
    Commented Jul 2, 2012 at 0:54
  • \$\begingroup\$ @Feynman - I changed the simulation picture to show pulses rather than the previous square wave. So out1 goes to your positive capture FF, out2 to your negative capture FF. Since the input is isolated, it can float at whatever voltage it wants (within transformer isolation specs). Hopefully all is clear now. \$\endgroup\$
    – Oli Glaser
    Commented Jul 2, 2012 at 0:58

As close as I can get to understanding what you're asking for, you might want something like this:

enter image description here

XFMR1 provides isolation from your inputs. If your input is actually single-ended, "IN-" will be the reference (ground) of the source circuit. If you are using single-ended instead of split supplies, you could tie the reference of the transformer secondary to a mid-supply voltage instead of to ground.

OA1 and OA2 are being used as comparators. R1, R2, and R3 should be chosen to give thresholds corresponding to the minimum positive and negative pulses you want to detect.

LTCH1 is an SR latch that provides the latching behavior you requested. Depending on how OA1 and OA2 are powered, you might need voltage dividers or zener circuits between the comparator outputs and LTCH1 to give the correct logic levels. Or you might use comparator ICs with open-collector outputs, in which case you'd add pull-up resistors to LTCH1's supply to get the correct voltages.

OA3 is a differential op-amp, for example maybe THS4521. Feedback will need to be provided to give the output amplitude you want. Note that this op-amp will be attempting to force its common-mode output voltage (following an additional input that's not shown in my diagram). If you want a circuit that applies a differential voltage but allows its load to determine the common mode voltage, this will not work.

Final part selection of the comparators, latch, and differential op-amp will depend on the speed and repetition rate of the input signal, available power supplies, etc.

  • 1
    \$\begingroup\$ Thinking about this more, you might be able to reduce the comparator/latch part of the circuit to a single comparator with hysteresis...Unfortunately I don't have time to work out the details and update my answer. \$\endgroup\$
    – The Photon
    Commented Jul 2, 2012 at 3:38

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.