zero-volt-drop solid-state switch for thermocouple 4

[RyreInc]

I need to switch between 4 thermocouple inputs every few seconds so we can use one thermometer. First thought is some 555-type ICs and some electromechanical relays. I desire to avoid big mechanical parts, so a solid-state solution is preferred, but a SSR, BJT, or FET would have a voltage drop.

1) Is there a solid state relay or opamp that will precisely follow the input voltage, and allow me to switch between 4 outputs?

2) Will a thermometer designed to take J-type thermocouples work with a low-impedance buffered signal from an opamp?

Thanks!

 

[zzzjim]

My simple thoughts , accuracy -vs- complexity .

Could a relay or two be useful for killing unwanted channels , as
crosstalk may be a gathering of unexpected sources of interference !

 

[btrueblood]

Why would you want to sample a thermocouple (best response time, for a bare bead of say, 28 gauge wire, is still 10ths of a second) at anything near the microseconds or whatever that the switch transients are measured in? Just switch, allow 10x settling time, take an A/D sample, then switch to the next TC. Even 16 channels, 0.1 second intervals for each, is only 160 Hz...

 

[IRstuff]

?? 1/(16*0.1s) = 0.6Hz

TTFN
faq731-376

 

[MacGyverS2000]

IR... switching frequency. 16 x 1/0.1


Dan - Owner

http://www.Hi-TecDesigns.com

 

[IRstuff]

According to that post, it supposedly takes 0.1s per channel, so 16 channels take 1.6 s.

Doing 16 channels in 0.1 s requires ~6 ms read time.

TTFN
faq731-376

 

[btrueblood]

Take 16 channels, and assume you want to sample each channel at 0.1 second intervals. 16/0.1 = 160.

I purposely did NOT say "0.1s per channel", I said 0.1 second intervals. I also did not say 16 samples in 0.1 second.

But in any of your cases, Irstuff, is the mux and its settling time a limiting factor?

 

[IRstuff]

I just having problems with the semantics.

160Hz sampling is 16 samples in 0.1 s --> 16 samples / (160 samples/s) = 0.1 s, and therefore, each sample is taking 6ms, assuming there's a single meter being mux'd. If that's the case, then the assumption of 0.1s settling is not valid.

TTFN
faq731-376

 

[btrueblood]

I never calculated or stated settling time, I just said you are switching the mux at 160 Hz. Settling time is whatever the system will settle at, and I would take whatever the predicted settling time is and multiply by 10 before taking the sample. Doing all that within 6 ms sure seems do-able, with the fairly low impedance of a thermocouple. I know I've done it before...but had a pretty good A/D board to work with.