I'm trying to multiplex 4 signal pairs using a 74HC4052 Dual 4-channel analog multiplexer/demultiplexer. My signal is a square wave from 0-3.7V. The output of the chip then goes to a circuit that does some measurements. For whatever reason I'm finding that something happens to the signal going through the multiplexer that makes the resulting measurement wrong. I don't have a scope to see whats going on. I know

The only thing I can think of is that the multiplexer circuit resistance changes over the range of input voltage that I'm using. VCC is the same NOT higher than VIH of the signal. I'm finding it difficult to read the datasheet to extract the info I need, but it does seem to imply this will happen under some circumstances.

I'm thinking that something like 4 x Maxim IH5144 SPDT switches connected to a 4-1 demultiplexer might be a better solution. However, the lower Ron the better for my measurement circuit. So perhaps 2 x Maxim MAX4741 might be better. But for circuit layout having single switch circuits would be better.

This is what I *think* I need from an analog switch:
-lower Ron the better
-flat Ron over input voltage range
-ability to switch signal at VIH that is the same or close to VCC without distortion.
-VCC from 3.7V to 5.0V

Any suggestions would be greatly appreciated. Thank you.

 

What is the impedance of the measurement circuit?
How fast are you switching the signal between channels?
Can you set Vcc to 5V?

 

Switching is every 1000ms (1sec)

Technically there's nothing preventing me from using a higher Vcc. Its just that the output high of the measured signals is always a little bit less than than Vcc because the measurement circuit is also doing the excitation of the probe. So I would need 2 power supplies 3.7V and 5.0V. The lower being for the measurement circuit power and the higher for the multiplexer. I was thinking of trying that tomorrow afternoon.

The impedance is the big question. I'm not sure what the actual input impedance of the measurement circuit is, but whats being measured is conductivity. So over the range of conductivity we are interested in, the resistance of the probes will range from 250ohms (4000 uS) to 143Kohm(~7uS) This I think is the real problem. The 74HC4052 analog switch does not seem to like the changing resistance. Doing a 2 point calibration without the multiplexer works. With the multiplexer in the circuit the calibration produces garbage.

The readings using the multiplexer are stable and consistent (although not useful). If anything it appears as if the measured signal is getting clipped or compressed at one end of the range. Not sure which. Hence, why I was wondering if a switch with a lower and stable Ron might be better in the long run than the one I'm currently using. Thanks.

 

Would it be possible to do a 3 or 4 wire measurement?
That would minimize the effect of the mux resistance on the measurement (but requires more mux channels).

Yes a lower mux resistance would help.
The ON resistance of the 74HC4052 is around 200Ω and varies some with signal voltage level.
It should be sufficiently below 250 ohms so it doesn't affect the desired accuracy i.e. ≤2.5Ω for 1% accuracy..

 

The probe is only 2 wires so 3 or 4 wire measurement is out of the question. Unless you meant something else.

I know its asking a lot but could you check the MAX4741 datasheet and tell me if its a *better* switch than what I am currently using. I'm not stuck on the 74HC4052. It was just suggested and its my first attempt. Checking on Digikey the MAX4741 looks to be around Can$5 each and I would need four for each system. That's pretty pricey. The 74HC4052 cost 50 cents!!

The TI TS3A4741 seems identical if not better than the maxim part and at Can$1.42 more affordable. Its SPST but I would just connect the two enable pins together. Thanks.

 

The probe is only 2 wires so 3 or 4 wire measurement is out of the question.

How is the measurement made.
What instruments are involved?

I know its asking a lot but could you check the MAX4741 datasheet and tell me if its a *better* switch than what I am currently using.

A lot better.
It has an ON resistance of 0.8Ω as compared to 200Ω.

he TI TS3A4741 seems identical if not better than the maxim part

It's slightly worse with an ON resistance of 0.9Ω but is still better than needed.

 

This is the actual "instrument". The probe is simply 2 wires connected to platinum tipped ends. Note that we are only interested in conductivity not the other parameters its capable of.

The whole point of multiplexing the front end probes is to reduce the number of measuring circuits we need for each unit. You can see its US$58 each plus if more than one circuit is used together then isolation is required because its a conductivity measurement. So that's another US$24 for the isolator. Its also a final board space issue.

So 4 x 58 + 4 x 24 equals huge bux. We can accept the cost of one measurement circuit for now until we can design a cheaper replacement in-house. Or just get a better OEM deal for the measurement circuit.

Yes the Maxim part is better. Would you recommend getting the TI part to experiment with or is there more to try with the 74HC4015 first? Like increasing Vcc. The TI part description says its rail-to-rail switching which seems like what I want.

 

Ideas:

1. Ask for an equivalent circuit of the probe interface. You might get luck especially if your an OEM. If you could give them a volume per year they might be more receptive.
2. Output into a potentiometer (10 M) Guess and reduce the amplitude to 1/2 what it was initially, no load. Take the POT oout and measure wiper to common. That should be the input Z.
3. OPTOMOS relays. They are basically FET switches with a low RdS ON that are LED operated. Same as low as 1 mA.
4. The range of conductivity you need to measure should also matter.
5. I emailed them about their units of uS and uS. They are not consistent in the their datasheet. See page 1 and Page 7.

 

Yes the Maxim part is better. Would you recommend getting the TI part to experiment with or is there more to try with the 74HC4015 first? Like increasing Vcc.

You could try the 4015 at 5V, but I expect its ON resistance would still be too high.

I would give the TI part a try if raising the voltage of the 4015 is not sufficient.

 

The IXYS optomos relays look very interesting. I found the LCC110. Its a SPDT form C relay. But, its expensive at Can$6.76 each. The 35ohm on resistance is not ideal but better than what I currently have with the 74HC4052. For an extra $2 I can take Ron down to 20ohms with the LCC120. So now that I know what to look for I found this:

Toshiba TLP222A-2. Its a 2 x DPST form A relay with an Ron of 2ohms. Cost is $3.14. And it comes in DIP which is easier for prototyping.

Thoughts on analog switch vs photorelay in my application? Thanks.

 

Now this is ironic. If I go with the TLP222A-2 DPST relay then I need to multiplex/switch 2 inputs at a time for each signal pair. Hmm, sounds like a better, more proper job for the 74HC4052 demultiplexer.

 

http://www.ixysic.com/home/pdfs.nsf/www/LBA716.pdf/$file/LBA716.pdf $6.90 USD at digikey, 1 NO and 1 NC contact 400 milli-ohms

You didn't look hard enough,.

 

The Toshiba TLP222A-2 works like a charm. I'm actually only switching 3 pairs so I opted to forgo the multiplexer to save board space and switch directly using I/O pins. Very low Ron and simplicity win. Its a useful chip so I'm going to buy more for my parts kit.

I'll followup with a schematic, but I just connected the two inputs of each chip together, but with separate current limiting resistors. I daisy chained the load sides of each chip to each other than to the input of the measurement circuit. If I switch the inputs correctly (since I'm not using a multiplexer) the measurement circuit only ever sees on probe. Thanks for the help.