Can A Si8410ab-d-is Carry An Analog Signal?

[xavier]

I need to isolate two sensors that output a 0-5v analog signal but are attached to the same ADC. Is it possible to use a SI8410AB-D-IS to do this? In other words can you apply a 1v DC signal to the input and get 1v out on the otherside? I first thought not but the video at digikey :
http://search.digikey.com/us/en/products/S...51-5-ND/2170658

The video seems to show a voltmeter changing in value. But the guy references the duty cycle on the input so I'm still kind of thinking it's just digital signals only.

I am aware of isolation amplifiers and figure I can go that route if there isn't another viable option.

[Sch3mat1c]

You need an ADC and DAC. Notice title, *Digital* Isolators. Table 3 indicates V_IH = 2V minimum and V_IL = 0.8V maximum, typical TTL compatible threshold levels. In other words, if you apply a voltage somewhere between 0 and 5V, you are guaranteed to get a "1" (whatever V_OH is specified as) for any input voltage over 2V, and "0" (i.e., V_OL) for < 0.8V. Somewhere in the 0.8-2V range, it will actually transition. It may have some gain (a narrow range where the output changes gradually), or it may exhibit schmitt trigger behavior (hysteresis).

With 150Mbps available, this speedy part could transmit high bandwidth analog if you use a high speed serial ADC/DAC pair, with a clock oscillator and a second isolator to transmit the clock for the data. With minimal control logic (at worst, a CPLD; if you end up having to buffer data, you probably wouldn't want to do it at this speed with discrete logic), it should be possible to interface an ADC and DAC directly to each other in this way. Of course, if the required bandwidth is fairly low, you could use a cheap PIC microcontroller on each side to perform both ADC, serial transmission and DAC output seamlessly.

What is the nature of your signal sources? A cheaper solution is probably possible. Typically, you don't want isolation amplifiers, because they are made for medical interface purposes with reinforced 8kV+ insulation and agency ratings; for very little real electrical value (bandwidth, accuracy and distortion typical of a general purpose op-amp) you might spend $50.

If nothing else, you say the signals are going to an ADC -- it's probably easier to put a lone ADC at the sensor and provide it with isolated power and a digital interface.

In case you're wondering -- there are many analog solutions, but most of the simple methods suck. Optoisolators (LED and phototransistor) have poor repeatability and linearity; they are only suitable for use within a feedback loop (which is why switching supplies get away with them). Transformers are good, but only work with AC, so it basically comes down to having the same limitations as pretty much any radio -- do you use amplitude modulation, frequency, pulse width, etc.? The ultimate isolation, of course, is a full-on radio receiver, whether actually radio frequencies, or light instead (fiber optic links are usually used in the same way an RF carrier is, so analog fiber transceivers can be used in much the same way as regular radio can).

Tim

[Gorgon]

You can use a linear optocoupler like these from Clare to isolate a signal. This is like an isolator with a servo feedback.

TOK

[Sch3mat1c]

Not bad:
http://search.digikey.com/us/en/products/L...CLA113-ND/95983
They exist and are reasonably stocked. You'll need an op-amp on the input side to drive the LED.

Tim

[xavier]

Thank you for the suggestions. The clare LOC110 looks like it's perfect.