0
\$\begingroup\$

I'm building a circuit that uses a 32:1 mux (ADG732) to feed the selected channel into an amplifying stage so I only need one amplifying circuit. The input signals are from pressure sensors which have a capacitive nature and produce ~10mV of amplitude. The problem is that when I try to feed the channels through the multiplexer the sensors connected to the off channels are still affecting the output of the multiplexer; essentially bleeding across the channels in low-frequency operation. I think it might have something to do with the high impedance of the sensor. Is there any way to deal with it or is there a better mux for this purpose? I've tried to separate the input and multiplexer from the rest of the circuits (amplifiers, filter) but there is still crosstalk.

Edit 1: I'm not switching channels at the moment so I doubt the settling time would be an issue. Also, the original plan is to do the amplifying after mux so I don't need to install 32 amps. The schematic is quite messy but it is basically sensor inputs through header are fed straight into mux then to the amp/filter stages.

This is the circuit with mux Circuit with mux

Amplifying/filtering circuit only Circuit for amp/filter stages

Edit 2: All sensors are on separated sheets of PVDF. The electrodes are attached to copper tape contacts on both sides of the PVDF with a little bit of solder. I do not have sufficient knowledge with ground design so I just tie all the things I want to ground to the GND pin on my microcontroller. Let me also note that when the system is tested with a function generator (20mV sine wave of 5Hz) everything works just fine. The multiplexer has no problem switching and the amplifying and filtering stages also perform as expected.

enter image description here

For example, say channel 1 and 2 are PVDF inputs and channel 3 is the 20mV sine wave at 5Hz. If ch1 is selected and no pressure is applied to ch1 sensor, the output of the multiplexer would just be flat as expected. However, if pressure is applied to ch2 sensor while ch1 is on and ch2 is off, we can observe variation in output caused by ch2 as if ch2 is also on. The same effect applies vise versa. The exception being channel 3 (sine wave) which does not create interference or get interfered by the first to channels.

\$\endgroup\$
  • \$\begingroup\$ How quickly are you changing the multiplexer from one channel to the next? \$\endgroup\$ – Andy aka Jul 5 '18 at 17:52
  • \$\begingroup\$ Need a schematic for this question. You should pre amplify the 10mV signal before sending it through the mux. \$\endgroup\$ – laptop2d Jul 5 '18 at 18:16
  • \$\begingroup\$ Can you post a link to pressure sensors you are using? Also, can you quantify the effect of "bleeding" you see? \$\endgroup\$ – Ale..chenski Jul 5 '18 at 20:49
  • \$\begingroup\$ Where is the reference ground to your sensors? \$\endgroup\$ – Ale..chenski Jul 5 '18 at 20:54
  • \$\begingroup\$ It's some generic PVDF film with electrode attached for testing so there isn't a spec sheet for it. The whole circuit is powered by Teensy 3.6 which offers a 5V output. One of the electrodes of the PVDF film is connected to the header that leads into the mux while the other electrode is connected to the ground of Teensy. The effect of the bleed is quite significant since almost all voltage from any off channel appears at the output with little or no attenuation. \$\endgroup\$ – Johnny Jul 5 '18 at 21:02
1
\$\begingroup\$

Between each multiplexer input line (whether on or off) and the common drain (output) there is a capacitance of around 150 pF. This is round-about stated on page 4 of the data sheet - it talks about On Switch Drain, Source Capacitance of 350 pF and it will be similar when the switch is off because internally it's a JFET or MOSFET. It might be as low as 100 pF when off but, for a ball-park figure I'd go for 150 pF.

So you have 31 unused channels all injecting a bit of charge through these 150 pF capacitors and, on the common drain pin you have a virtual earth amplifier responding accordingly. I understand it's this way because you are trying to amplify charge and convert it to voltage.

Do you start to see the problem you have? All 31 together can be regarded as a single input with a capacitance directly to the D pin of nearly 5 nF and this will potentially be a significant noise source and cross-interference source.

Also, on your schematic I don't see where your 0 volts comes from.

\$\endgroup\$
  • \$\begingroup\$ I can see that Cs and Cd would be problematic due to charge injection, but I am not switching channels at all and my input signals are at best 50Hz. So I guess even if the 31 off pins are combined, due to the low-frequency signals the impedance of the capacitor should be massive. (I'm referencing the figure on the left at 5:41 by Texas Instrument) \$\endgroup\$ – Johnny Jul 6 '18 at 18:37
  • \$\begingroup\$ I’m not talking about switching channels. Your feedback capacitor is 3.3 nF and that is the same order as the combined capacitance of the multiplexers. \$\endgroup\$ – Andy aka Jul 6 '18 at 23:54
  • \$\begingroup\$ Where does the 3.3nF come from? \$\endgroup\$ – Johnny Jul 9 '18 at 14:26
  • \$\begingroup\$ C1 is 3n3 I presume - it has 332 in the part number and this usually means 3300 pF = 3.3 nF. \$\endgroup\$ – Andy aka Jul 9 '18 at 14:28
  • \$\begingroup\$ It's actually 300pF which would only make it worse I suppose. Regardless, I tried the multiplexer without the rest of the circuit but it still does not work. I'm guessing it's due to the high source impedance. \$\endgroup\$ – Johnny Jul 9 '18 at 14:31

Your Answer

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

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