I am switching an op amp input between two sources, one is ground, the other is a piezo material. I have an issue with switching transients which cause the op amp to saturate since I have a pretty high gain (x50). To keep things short, Here is the switch. I have a piezo element connected to s2, s1 is grounded. op amp's +input connected to D. Is there a way to prevent the transient from reaching the amp's input?


Here is what it looks like. Full Circuit

The hydrophone receives a sinusoid burst while the switch is in a logic 0 state, then once the burst is sent, it is switched to logic 1 and connected to the op amp. The idea is to read any acoustic reflections seen by the hydrophone. This has all been successful, but the transient is the one piece I would like to remove from the output signal. Unfortunately this is as far is my knowledge has gotten me so I have no clue where to go next.

Circuit is actually backwards my apologies. I just put it together in ppt since I don't have my files available with me. The hydrophone side should be S2 and op amp should be D.


I have some of the signals and here is what it looks like:

this is the switching noise Noise from the switch, with and without the hydrophone

Reflected signals

Here is what the reflected signal looks like. This is what I'm interested in, but as you can see it is quite low in amplitude, and thats why I need to amplify it.

@sperophany: As for the cable, I'm using Alpha wire 3302 twisted pair with a 50ohm bnc connector. It's about 6' long, 38pF/ft capacitance. more detailed specs can be found here http://media.digikey.com/pdf/Data%20Sheets/Alpha%20Wire%20PDFs/3302.pdf

Another thought I have as I was looking at the signals was to filter the switch's transient instead. It is pretty high frequency and well out of my system's range (<500kHz) so I was wondering if it would be possible to completely remove it with a lp filter. Any thoughts on this? Any drawbacks or new problems?

Overall, I just want to send a pulse and then listen for echos with the same transducer. I'm open to any and all suggestions - even if I have to scrap my current design and start from scratch.

Thanks again for your time, guys.

  • 1
    \$\begingroup\$ How is your circuit constructed? (A photo would help) --- construction details like the inductance between the switch and op-amp could be contributing to the transient as much as the switch itself. \$\endgroup\$ – The Photon Jan 9 '15 at 22:31
  • 3
    \$\begingroup\$ It is almost always a bad idea to insert a switch at a high-impedance node of a circuit, such as the + input of an opamp, for a number of reasons. Show us your circuit and explain why you need this switch. We can almost certainly suggest a better approach that achieves the same ends. \$\endgroup\$ – Dave Tweed Jan 9 '15 at 23:07
  • \$\begingroup\$ What would happen if you left D,S2 connected and use the switch just to ground D? \$\endgroup\$ – copper.hat Jan 10 '15 at 4:50
  • \$\begingroup\$ Sorry to resurrect an old thread, guys. I had to abandoned this project shortly after posting this, then suddenly the images show up on a random google search. @copper.hat I think I will try your suggestion and see what happens \$\endgroup\$ – user64054 Nov 4 '16 at 17:52

I don't see how a serious transient that would affect a relatively slow op-amp is being caused directly by that switch, nor do I see any problem the way you're using it in general.

Do you have a resistor directly across the piezo (not from the non inverting input to ground)? If not, then leakage may be causing the voltage to rise across the piezo while it is disconnected, and when the switch closes you get a transient as the piezo capacitance discharges.

| improve this answer | |
  • \$\begingroup\$ (+1) my answer, R to ground, whatever you can afford. (I've used analog switches on opamp inputs. a simple +/- lockin for instance) \$\endgroup\$ – George Herold Jan 10 '15 at 0:54
  • \$\begingroup\$ Interesting point about the resistor, I'll have to try it across the piezo. I have tried it to ground for all the terminals and it reduced the decay time, but the transient still persisted. Do you have any suggestions on how to select the size? Or does any arbitrarily large resistor do the trick? \$\endgroup\$ – user64054 Jan 12 '15 at 14:17
  • \$\begingroup\$ As George said, whatever you can afford (as low as you can go without unduly reducing the signal). \$\endgroup\$ – Spehro Pefhany Jan 12 '15 at 14:24
  • \$\begingroup\$ Given your edit, I would suspect the transient has little to do with the switch and everything to do with what is happening in the hydrophone after you whack it. What happens if you don't whack it? Still a transient? How long is the cable to the hydrophone? There will be some charge injection, so if the cable is very long you might get a disturbance. If that is the case you could try a switch with less charge injection. \$\endgroup\$ – Spehro Pefhany Jan 12 '15 at 14:53
  • 1
    \$\begingroup\$ @EmmanuelColeman The actual frequencies and damping information is useful. It's also useful to know what the cable looks like. \$\endgroup\$ – Spehro Pefhany Jan 12 '15 at 17:34


simulate this circuit – Schematic created using CircuitLab

Nice pictures!
I'm adding the schematic as an answer.
If the ~0.5V is a real signal, you might want two diodes in series (4 total) You have to choose the resistor (to save the diodes) depending on voltage level and time...
How much power do the diodes absorb? (from the stimulus pulse)

I'm guessing it's charge injection from the switch, maybe?
What's the hydrophone impedance? or if you hang 10k ohm across it what happens?

On thing your schematic doesn't show is ground? That may be your problem. (?)

| improve this answer | |

Since your switch is Form C (break-before-make), there's a short interval in there, during switching, where everything is floating and charge can accumulate where you don't want it to.

I'd change the switch/switching action to Form D (make-before-break) thus assuring the piezo will be grounded before the ground is lifted from the opamp's + input to see whether that fixes it.

| improve this answer | |
  • \$\begingroup\$ I doubt that this is the problem -- after all, we're only talking about approximately 30 ns of "open circuit" during switching. \$\endgroup\$ – Dave Tweed Jan 10 '15 at 1:18
  • 1
    \$\begingroup\$ Not knowing anything about the circuitry surrounding the switch, and since you're free to doubt, I doubt whether your doubt isn't premature. \$\endgroup\$ – EM Fields Jan 11 '15 at 3:28
  • 1
    \$\begingroup\$ @EMFields The ringing is of 40ns period, so it indeed may be a factor. \$\endgroup\$ – Spehro Pefhany Jan 12 '15 at 17:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy