Proton BASIC Community

Hi,
 I am planning to monitor 16 PT-100 temperature sensors using a MAX31865 RTD module connected to a PIC & 16 switching relays. So the design  is based on 16 to 1 relay multiplexing.

Relays occupy space & have a finite switching life. Can there be an elegant FET based switching solution? I am banking on you guys to generate ideas.

Thanks.

What are you switching? Could you not use DC-AC or DC-DC SSRs? If DC-DC SSRs are used, bear in mind about 1.2v voltage drop if I remember correctly.

I interpreted that as multiplexing the inputs using relays?

JOhn

Hi,
You could use these converters
https://docs.rs-online.com/21dc/0900766b81347f5c.pdf

and then a multi channel a to d converter and read in all channels via SPI.
This one has 8 but I am sure there are 16 channel ones
https://www.mouser.co.uk/new/analog-devices/adi-ad411x-24-bit-adcs/

Bob

Quote from: John Lawton on May 07, 2025, 04:03 PMI interpreted that as multiplexing the inputs using relays?
JOhn

I misunderstood the situation. Ignore my earlier post.

My understanding is that he wants to FET's instead of realays.

There are some ideas on the interweb, e.g.

https://electronics.stackexchange.com/questions/376694/multiplexing-pt100-with-cd4051

This could be simplified if you don't need 4 wire operation.

John

Quote from: RGV250 on May 07, 2025, 05:00 PMHi,
You could use these converters
https://docs.rs-online.com/21dc/0900766b81347f5c.pdf

and then a multi channel a to d converter and read in all channels via SPI.
This one has 8 but I am sure there are 16 channel ones
https://www.mouser.co.uk/new/analog-devices/adi-ad411x-24-bit-adcs/

Bob

Thanks Bob. I could have but it would increase my cost & space manifold since these converters won't be cheap.

Quote from: John Lawton on May 07, 2025, 07:07 PMThere are some ideas on the interweb, e.g.

https://electronics.stackexchange.com/questions/376694/multiplexing-pt100-with-cd4051

This could be simplified if you don't need 4 wire operation.

John

Interesting John.
I have used CD4051 extensively for switching analog signals. I have measured the drain source resistance when a switch is ON & found it typically in the order of 100 Ohms or less....this matches quite closely with the Rds value mentioned in the datasheet. So 4051's can never be used to switch RTD's since the value of the RTD (138.5 Ohms at 100 degC) is comparable to the switching resistance.

My only hope are three p-channel MOSFETs which are hard driven to reduce Rds to the level of milliohms & use them as a 3-pole relay. I'll test my idea someday.

If you use a constant current drive and 4 wire operation as the Stack Exchange posting shows, then I don't see that the on-resistance is a critical factor as long as it isn't too high.

John

NXP do a very low ohmic 8-channel analogue switch with balanced ohmic differences per channel in the 0.05 Ohm range HERE.  With typ 0.5Ohm series resistance and 0.05 ch-ch variation, the fixed offset is about 1.3C and ch-ch variation 0.13C.  You will likely find they are extremely well matched part to part from the same batch, thus a general offset cal can be added in the MCU

TI also have some 2x1 sub 300mOhm switches.

The NXP XS3A4051 seems cheap enough, one can always parallel switches to lower the on resistance. How fast do you need to switch and how often? I have used reed relays in the past for a binary weighted resistance substitution device, it worked fine. DIL reed relays can have a life expectancy of hundreds of millions operations and fit in a standard ic socket.

If using 4 wire connections one could use CD4051's use two to switch a constant current supply to the PT100 the voltage drop across the FET switch would be irrelevant. Using another two wires measure the voltage drop across the PT100 by using another two 4051 switches into a hi z differential analog input. PT-1000's are always an option.
Ken K

Talking 4000 series CMO's I have a large collection and recently purchased an IC tester.
https://www.youtube.com/watch?v=66hPg_r4IOw&t=1047s
https://www.youtube.com/watch?v=pUQDl6OzON0
I was helping a friend with a repair, one IC had no readable writing, all the chips were 4000 series, the device named the chip without me going about it the hard way.  BTW I would hate to write the code for the IC tester, I guess some poor sole spent  hours entering data and writing code for someone else to copy and sell.

I just had a thought. If one was to use a 1mA constant current supply connected to 16 PT-100 sensors in series. The total resistance would be 1600 ohms so the voltage drop across the series string would be 1.6V.

Using something like a 4051 and two wires per PT100 to select and read the voltage drop across any individual device in the series arrangement. The read circuit would need to have a differential Hi z input. Hum on the line would to some extent be controlled by the fact the PT100's series power supply could be referenced to ground. Liken it to going around with a voltmeter and reading the voltage drop (temperature) across each PT100 in the series string.

I always like reading your replies, Ken.  You should write a book on all the interesting circuits you have developed over your career, including detailed LT Spice sims. 

Quote from: JonW on Today at 08:58 AMI always like reading your replies, Ken.  You should write a book on all the interesting circuits you have developed over your career, including detailed LT Spice sims.

Thank you for your very kind words.