better way for adc multiplexing

[hsn93]

hi,

im trying to build industrial analog front end that can be configured by software to accept (4-20mA) or (0-10V / 1-5V / etc) on same connector pin

where accuracy doesnt matter that much


output to be connected with a filtering cap 10nF or so..


for switching i dont like to use relays due to relaiblity issues so
im thinking about SSR like AQY282S (with TVS with Voltage @Ipp < 60v)

also becasue there is 2 ssr per configuration it means 2x resistance =~5 ohm (ie 5% of 4-20mA lost?)

but this would require 4 ssr for 1 channel .. they arent cheap (1.5$ each).
also it will require a not gate for 2/4 ssr.. overall i think there should be simpler way i didnt think about?


so what would be the alternative robust configurable analog front end?



I'm wondering what way being used in PLCs.
and if the above way is good enough or not ?

[pwlps]

If this is for signal multiplexing you don't need any high-current SSR like AQY282S,  for your application I would suggest an SPDT switch like this one or similar (e.g. ADG1633, is cheaper):
https://www.analog.com/en/products/adg1634.html?doc=ADG1633_1634.pdf#

[Brutte]

industrial analog front end that can be configured by software to accept (4-20mA) or (0-10V / 1-5V / etc) on same connector pin

Maybe 74HC4051 and provide 8 different frontend circuits. One for 4-20mA, one for 0-10V one for 1-5V etc. Then you can select which one is fed through in software.

[pwlps]

industrial analog front end that can be configured by software to accept (4-20mA) or (0-10V / 1-5V / etc) on same connector pin

Maybe 74HC4051 and provide 8 different frontend circuits. One for 4-20mA, one for 0-10V one for 1-5V etc. Then you can select which one is fed through in software.

74HC4051 has a >100Ohm on-resistance (vs. 5 Ohm for ADG16xx), so ok if you can afford a voltage drop of 2V at 20mA.   It seems you have to pay much more if you want a very low on resistance.

[Kleinstein]

If one does not have a higher supply at hand, one may not be able to use the simple CMOS switches. These are limited to signals inside there supply. One could still use CMOS switches after the divider.  It should be OK to always have the divider (possibly higher resistors) connected. So one could likely get away without a switch there.  So it would be mainly about connecting the 150 Ohms or similar resistor for the 20 mA signal.

If the signals are only positive, one could use just a single divider chain of some 25 K , 10 K and 150 Ohms and have a N-Channel MOSFET  (e.g. 2N7000) in parallel to the 25 K and 10 K resistors. The control off the FET would however need some 8-10 V to turn it on.  A low R Photomos could be an option here.

[hsn93]

industrial analog front end that can be configured by software to accept (4-20mA) or (0-10V / 1-5V / etc) on same connector pin

Maybe 74HC4051 and provide 8 different frontend circuits. One for 4-20mA, one for 0-10V one for 1-5V etc. Then you can select which one is fed through in software.

74HC4051 has a >100Ohm on-resistance (vs. 5 Ohm for ADG16xx), so ok if you can afford a voltage drop of 2V at 20mA.   It seems you have to pay much more if you want a very low on resistance.

i think another problem is that it swings with temperature or varies from chip to another?

[hsn93]

If one does not have a higher supply at hand, one may not be able to use the simple CMOS switches. These are limited to signals inside there supply. One could still use CMOS switches after the divider.  It should be OK to always have the divider (possibly higher resistors) connected. So one could likely get away without a switch there.  So it would be mainly about connecting the 150 Ohms or similar resistor for the 20 mA signal.

If the signals are only positive, one could use just a single divider chain of some 25 K , 10 K and 150 Ohms and have a N-Channel MOSFET  (e.g. 2N7000) in parallel to the 25 K and 10 K resistors. The control off the FET would however need some 8-10 V to turn it on.  A low R Photomos could be an option here.

i guess this will make it from 4 switches to 3 ?:

[Kleinstein]

Have the 150 Ohms at the low side, and 1 switch is enough.  A 2 nd switch may help to reduce the effect of the switch resistance.

One point to keep in mind is possibly protection for the 150 Ohms - it may not like 10 V.

Depending on the circuit that follows, one could also just switch at the low side. This 150 Ohms resistor would than need differential sensing or a very low R FET to switch.

[Brutte]

74HC4051 has a >100Ohm on-resistance (vs. 5 Ohm for ADG16xx), so ok if you can afford a voltage drop of 2V at 20mA.

Yes, of course you use mux in a feedback of an opamp. Then it does not matter if it is 100R or 110R when feedback resistors are 100k.

I am not sure how many switched ranges OP wants to have. If that is 4 or less then alternatively 74HC4052 can be used and then two channels can be serviced simultaneously.

[Vovk_Z]

I think there is a possibility to use only one switch to connect-disconnect 150 R shunt from circuit if accuracy doesn't matter that much. ("Curent/Voltage mode").
It can consist of several physical electronic switches to decrease their resistance.

[wraper]

hi,

im trying to build industrial analog front end that can be configured by software to accept (4-20mA) or (0-10V / 1-5V / etc) on same connector pin

where accuracy doesnt matter that much

IMHO just connect 150R resistor with MOSFET for 4-20mA. For voltage ranges, simply use maximum range for everything and scale (ADC data) in software only. Or use different VREF voltage for different ranges for better accuracy. SSR makes little sense since ADC won't be isolated anyway.

[wraper]

Oh, BTW even if you want full range of ADC for every voltage range, you don't need switches. Simply make several voltage dividers in parallel and connect them to separate ADC inputs. Again simply select in firmware which ADC input (divider) you want to use instead of switching anything.