IC switch

Hello all,
I want to design an isolation circuit with low current leakage (< 500 pA). For my application I do need to have/inject positive and negative pulse injected to the DUT and after injection the DUT isolates from input signal circuit. I selected two ICs and connected in series
MAX321: (https://www.digikey.com/product-detail/en/maxim-integrated/MAX321CPA-/MAX321CPA--ND/1937453)
and MAX320: (https://www.digikey.com/product-detail/en/maxim-integrated/MAX320CPA-/MAX320CPA--ND/1513794)
As shown in the attached schematic, to have positive (negative) pulse I connected Pin#5 (pin#1) of MAX321 (Max320) to positive (negative) DC signal and using digital input and selecting the defined DC signal and toggling in pin# 6 or 2 (depends on which positive or negative signal selected), we have positive or negative pulse at the output (DUT). Here are my questions:
1- My circuit does not work and I have just a linear decay at the output (DUT= 1K-ohm resistor). Note that the logic input is defined by function generator (pulse amplitude will determines the 0 or 1 (pin#3 in MAX 320)) and the required pulse width at the output should be equal to pulse width by function generator.
2- Does this circuit work for my application?

Thank you!
University of Pittsburgh

Hi Mohammad,

At first look, it seemed like the circuit should work. However, I see a potential critical flaw. Specifically, there is no way to guarantee that the switches connected to “+Vwrite” and “-Vwrite” will not both be closed for some period of time when you switch “Logic Input S”. This means there is a potential for a short circuit between the two, which would likely damage one or both switches. If you were to replace the MAX321 with a MAX322 and re-wire the circuit appropriately, you can eliminate that risk. The MAX322 has one normally open (NO) switch and one normally closed (NC) switch, and they are guaranteed to break before make (open the closed switch before closing the open switch) if both are controlled by a single signal.

A few other things to check and to keep in mind:

  1. Obviously, double check all of your connections.
  2. Your signal passes through 2 switches before reaching the DUT, so all capacitances and resistances are increased, and inductances can start to become and issue. These can affect the signal at DUT.
  3. I assume you are using a breadboard, based on your use of through-hole parts, which means there is a high likelihood of significant stray reactances in your circuit. This means that the circuit will begin to behave in a non-ideal manner at a much lower frequency than if placed on a well laid-out circuit board, due to long wire leads and such.
  4. Regarding frequency, under ideal conditions, your circuit might function up to around 300KHz, but that would be extremely optimistic. I would suggest running it at a much lower frequency first (like below 1KHz) and if it behaves properly, then start to increase frequency.
  5. Make sure you apply power before any signals, as described on page 5 of the datasheet. Failure to do so will likely damage the parts.
  6. Make sure your logic inputs are within spec. Logic Hi should be no less than 3.5V and logic Low should be no more than 2.5V
  7. Be sure the output impedance of “+Vwrite” and “-Vwrite” sources are much less than 1K Ohm. If not, you may need to buffer them with op-amps.
  8. Put bypass caps of 0.01uF to 0.1uF between both V+ and V- to ground. (See Figs.5-8 as examples)
  9. Take a look at all of your signals, including V+ and V-, on an o-scope to look for clues.

Hello David,

Happy New Year! Thank you very much for your feedback and in detail diagnostic. I will go through your fruitful suggestions. I have a few questions regarding your some comments:

  • I have not applied/used -Vwrite and just applied +Vwrite to see if the circuit works for +Vwrite. So maybe there is no short circuit as I have used just Vwrite in the circuit. For this circuit, I got the result shown in the attached Fig. 1. In
    addition, I skipped MAX321 and just used MAX320 with only +Vwrite signal (shown in the Attached Fig. 2) and got the similar result. I could not figure out the linear decay at the output? (the blue curve is the output (DUT= 1K-Ohm) and yellow curve is the logic
    input P (went to pin#3 of MAX320).
  • In your comment 4 you talked about the frequency, I did not get the point. My +Vwrite signal is a DC signal (+1V) and I applied one single square pulse (4V, 10 us) as logic input to get a single pulse shape at the output. I supposing that the
    frequency hat you are talking about should by for multiple pulses (?) as in that case I can define frequency (testing multiple pulse is the next step of my experiment).
  • For the comment 7, I am using sourcemeter to apply +Vwrite I have not checked its output impedance Thanks for this important note. I will check it and change the output impedance to less than 1 K Ohm.
  • As you correctly mentioned, it is better to use one IC instead of two to get rid of issues that you correctly mentioned. Do you have any suggestions for the ICs that work for this application? I tried to consider two factors while choosing the
    ICs: 1- Low leakage current (<500 pA) and 2- having high on/off speed. As you mentioned, I will use PCB board instead of breadboard. The question is can I mount through-hole IC on the PCB by soldering?
  • For the oscilloscope, I selected 1M-Ohm impedance for the oscilloscope to see the output signal coming from DUT. Do you think this decay can be due to the RC effect as I selected high Z for the scope?

David, thank you so much again for providing these important and helpful points and comments!

Looking forward to hearing from you! Thanks!


Hi Mohammad,

Regarding frequency, yes I was referring to your “Logic Input P” signal, which you stated was to come from a function generator. For a single pulse, this is not relevant. I would, however, use a much longer pulse initially, to see if that affects your results. Based on your o-scope image, I would expect the output voltage to still drop to zero after 10us, but it would be worth checking.

For your simple circuit using only the MAX320, if everything is connected correctly and the MAX320 has not been damaged, it should work. The output you are showing is more representative of a capacitively coupled signal rather than a resistively coupled signal, represented by the circuit below, except that I would expect it to have more of an exponential decay look to it.


If this capacitor were placed on the output of COM2, it would give a similar result to what you are seeing. In a capacitively coupled circuit, the voltage across R would initially pull up to +Vwrite and then decay to zero as the capacitor discharged. If your MAX320 were damaged, the switch might behave in such a way by only capacitively coupling to the output. I would suggest trying Switch 1 or another MAX320 if you have one available to see if you get the same result.

As I mentioned before, if you have ever applied a voltage to any of the pins before V+ and V- were powered, that could damage the part.

Regarding the +Vwrite analog input voltage, when you say “source meter”, I assume you mean a DC power supply with a display to set the output voltage. If so, its source impedance would almost certainly be just fine. However, be very sure that V+, V-, and all of your other inputs and outputs are referenced to the same ground potential. You should have a ground point somewhere on your board (preferably very close to the MAX320 to minimize ground loop size) and the ground lead of each signal; V+, V-, +Vwrite, Logic Input Pulse, and DUT should be connected to that same ground point to ensure that all of your voltages are referenced to the same ground potential.

Regarding possibly using alternative parts, other than replacing the MAX321 with the MAX322, I think the MAX32x series should work. There may be a single IC that can do the whole job, but I have not looked into that yet.

You can certainly solder these through-hole parts into a circuit board as long as you have a board with the proper holes drilled. One can typically maintain signal integrity to higher frequencies using surface mount parts on a well laid out board, but one can make a very functional board with through hole parts, too.

For the oscilloscope, using 1M-Ohm test leads should work just fine. Just be sure to connect the ground lead to that same ground point as I mentioned for all of the voltage sources.


I just noticed something from the screen shot of your o-scope; it shows a 50 Ohm termination rather than 1M Ohm. You might want to check that.

That 50 ohm setting seems to be for channel 1.

Would be nice to have a measured waveform shown from pin 5 (+Vwrite in that reduced circuit setup having only one MAX320). Is it identical to pin 6 waveform (DUT)?
It may be a good thing to measure with DMM that the load resistor is actually 1kOhm (instead of something low ohmic or an inductor). What is the waveform when there is no load resistor at all?