Electronics > Projects, Designs, and Technical Stuff

Analog switch for voltage reference? Accuracy?

(1/1)

konfu:
Hi,

I am trying to build a device which should offer a very accurate voltage reference to multiple outlets.

I want to use some ultra high precision reference voltage sources like MAX6126 to generate a very exact voltage. I then need to switch this voltage to several other parts of a circuit.

So instead using a mechanical switch to switch from A to B, to C, ... I thought about using something like a 74HC4066.

Will that 74HC4066 influence the accuracy of the reference voltage? As far as I understand this device is made of a N-channel and P-channel mosfet with about 28Ohms of resistance. So it will influence the voltage if I understand correctly.

Any comments?

grumpydoc:
Why switch it - just connect directly.

But to answer your question... The "on" resistance of the switch will form a voltage divider with the input resistance of the stage which is going to use the reference. If this resistance is high enough then variations in the "on" resistance won't have much effect on the accuracy. Usually voltage references can't source (or sink) much current anyway so you normally have pretty high impedance loads for them so you might be OK. However, at present only you know the values.

Not sure I'd do it that way though.

codeboy2k:
ohms law... as grumpy said.  There will be a voltage divider formed with the switch on resistance and the input impedance of the destination.

To counter this as much as possible, you'll want to buffer your reference output first, with a precision chopper stabilized, zero-offset, zero-drift, low noise op-amp, then send it to the switch, and make sure the destination inputs are also as high-impedance as possible. Usually this means that the inputs of the destination should be another op-amp stage.  If you are feeding multiple ADCs or DACs just remember that you need to consider the input impedance of the destination, and how much current it will draw from the reference source. This current will cause a voltage drop across the switch.  If your destination draws too much current as to cause a voltage drop that adds too much error, you need to buffer the input to that destination.

The MAX44250 is a good buffer for the MAX6126. The best grade 6126 has initial accuracy of 200ppm and the MAX44250 has maximum offset voltage of 6uV which is only 1.5ppm of the 4.096V reference (1.2ppm if you use the 5.000V reference). It's also low drift, at only 19nV/°C so it won't contribute much error over temperature. So if you put it in a box and get a 30°C temperature rise inside the box (from 25°C room temp to 55°C) then it contributes 19nV*30 = 570nV or only 8.8ppm of the 5.000V reference

So... REF-> buffer -> switch -> destination  and keep track of your error budget and you'll be ok. Maybe need to additionally buffer some destination inputs too.


PS: the TI OPA277 (10uV Vos and 100nV/°C drift) and the TI OPA2188 (25uV Vos and 30nV/°C drift) could also be used as a cheaper alternative to the MAX44250.  These 2 op amps are about half the price of the MAXIM op-amp, but are not as good. However, the TI op-amps still might provide adequate accuracy for your needs.  Only you know best what error you can afford.

 

Navigation

[0] Message Index