Looking for 12CH MUX

Hi Kurumi,

I have some concerns with this application - please see below for the concerns as well as a potential work around. 

1. you don't need a 12 channel mux - you need a 2 channel 12:1 multiplexer (12 inputs to 1 output and vice versa). We don't have any of those parts - we do have a 16:1 multiplexer but it is one channel and you'd need to use 2. These are parts like the CD74HCT4067 - however it is an older part that has a high leakage current that will degrade measurement accuracy. 

2. When doing differential signal muxing - as with thermocouples it is not good practice to have the differential signals being routed through different IC's as the mismatch could be higher than if using channels on the same pathway. So using 2 channel muxes is good so that both +/- signals are routed through the same IC. Sometimes this isn't always possible - but if that is the case using parts with very flat and low on resistance curves will make mismatch between IC's a lot smaller but the layout for both IC's needs to be as identical as possible to maintain conditions across both switches. 

3. The voltage output of a type K thermocouple is very low (range across ~1700C is less than 70mV). That means any source of error in the signal chain pathway is going to degrade results pretty significantly. Adding a multiplexer to this chain is adding a lump of parasitics that are not constant so its hard to remove that error. If multiplexers are used they should have low leakage current since these can cause offsets larger than the voltage you are trying to measure. Also having a flat and low on resistance curve also allows less attenuation due to resistance and a more "constant" attenuation that can be corrected later on in the signal chain. 

4. The ADC in the reference design is 24bits - this is very high accuracy so the multiplexers errors will show much more here than in a lower resolution ADC. 

5. Will the multiplexer be subjected to the heat measured from the thermocouple? (125C is the max we rate on multiplexers)

6. This may require parts that are not cost optimized. 

This leads us to a couple options:

1. Use the multiplexers in the signal chain such that:

Thermocouple feeds into the multiplexer which feeds into an Amplifier/ADC block.  

2. Or place amplifiers before the mux:

Thermocouple feeds into amplifier which feeds into the multiplexer which feeds into ADC.

Option 1 requires a higher accuracy of the mux due to the small voltage differentials it is reading from - the best option here is precision and since there are 2 channels of 12:1 muxing needed the closest option is to construct a 16:1 multiplexer out of precision parts. To do this 6 different IC's would have to be used to support all the inputs. It is built in 2 stages.

1st stage:

4 x TMUX1108 (precision 8:1 multiplexer). This stage will act as a input with 32 possible inputs. 24 of them will be used with 1 mux going to either the + or - side of the application (as these muxes are one channel - but this will solution is the smallest).

2nd Stage:

1 TMUX1136 (precision 2 channel 2:1 multiplexer). This stage will mux in  2 differential inputs and have one differential output that will continue onto the ADC. 

Pros of this design:

Very low on resistance and low on resistance flatness which lead to less attenuation.

Very low leakage current (pA range) which lead to very small offsets allowing the small voltages to be measured.

Cons of this design:

Expensive 

Bandwidth is degraded for putting multiple switches in series

Option 2

If the gain amplifier is placed right at the thermocouple there will be larger voltage swings between each temperature point making the signal chain after teh amplifier a bit less sensitive so small offsets. This allows some flexibility - but with high resolution ADC's the TMUX11xx family is the best to go.

However if a higher on resistance and less flat response is okay for the system designer the first design can be tweaked a bit.

1st stage:

4x TMUX1308 or TMUX1108. Same as in option 1 however the TMUX1308 is generally more cost effective. This is at the cost of a more resistive switch as well as higher current. But the leakage current should not exceed nA levels of leakage so the offset will be smaller than other switches.

2nd stage

1 TMUX1136 - same implementation as above.

Pros of this design:

Possible cost savings on multiplexer but could lose savings in extra amplifier cost. 

Cons of this design:

May still require precision parts (TMUX11xx family) to work better

Still has more expensive parts in it.

Essentially while muxing this system may be possible - the high accuracy of these systems combined with the fact that voltage output of these devices are so low makes muxing this system challenging and will degrade resolution. If muxing is pursued in this application I highly recommend the more precision  muxes. Due to the nature of the application the multiplexers need to perform very well without much interference to the signal that passes through it and that doesn't usually lead to cheaper devices.

Please let me know if you have any other questions!

Best,

Parker Dodson