Sample and Hold Analog Voltage

Sep 28, 2005 7 Replies
A

I am designing an electronic interface to SCADA. The SCADA typically takes 4-20 mA input for logging data to a central database. So I need to convert my voltage output from my instrument/device into current using a V/I convertor(like AD1B22). My instrument has just one analog out but I need 4 lines to interface to SCADA. For this I am using a 4:1 MUX. Now since these 4 lines should hold valid data at all times I need to latch in voltage values before converting them to current outputs. I was thinking of using sample and hold circuit to latch analog voltage values. When I looked at sample and hold circuit like AD585, the data sheet talks about droop rate(change in output voltage with time due to leakage capacitive current).



So I am thinking that the only way to latch analog voltage is to convert it into digital, then latch the voltage and convert it back to analog. Am I thinking on correct lines or not ? Any other ideas ?



Vote
Twitter Facebook LinkedIn
A

My sample frequency is not even KHz, it is Hz. I need a sample every 5 minutes. So if I use LF398 I dont need to latch data at all. Is that true ?

Vote
Twitter Facebook LinkedIn
A

Ok I got it. So if I sample every 5 min, I need to consider if droop of

25mv is acceptable to me or not. Correct ?
Vote
Twitter Facebook LinkedIn
J

Since latching a digital value will totally eliminate droop and shoot-through, that\'s how I\'d do it.
Vote
Twitter Facebook LinkedIn
P

"aman" schreef in bericht news: snipped-for-privacy@g14g2000cwa.googlegroups.com...

The AD585 is a fast SAHS for fast applications. For you application an LF198 or LF398 is more suitable. This ones have a droop rate of 5mV/min. If you keep your sample frequency at some kHz, this drooprate is neglectable.

petrus bitbyter

Vote
Twitter Facebook LinkedIn
P

"aman" schreef in bericht news: snipped-for-privacy@g44g2000cwa.googlegroups.com...

Yes.

But I don't understand whether you need only one sample every 5 min or if you can get only one sample during that time. That makes a lot of difference. If you can sample several times/minite you can still use that LF398.

If you can get only one sample per 5 minutes you have to consider whether or not 25mV is acceptable. In a practical circuit the result will be better but you can't count on that. You may find SAHSs with better droop rate. Analog Devices offers one they claim to be "droopless", the AD5533. You may want to use that one until you see its pricetag. It's also a little bit overkill as it offers 32 channels. It just does what you suggested: AD-conversion, storing the digital data and DA again for the output. It also claims 0.018% accuracy so I guess they use a 14 bit or better AD converter.

So if you want to go that way you'll have to look what accuracy you need. If your signal is in the 0 to 5V range, an 8 bits AD-converter has a 19mV conversion error already. That's in the same order of magnitude the LF398 offers. If you have to do better, I'd take one good enough AD-converter and Multiplex the inputs using (reed)relay. Store the output in flip flops or registers (the old LS364 comes to mind, but you may be able to do better these days). Then it takes one DA converter per channel.

petrus bitbyter

Vote
Twitter Facebook LinkedIn
J

it's not how often you sample it's how long it takes te preform each sample that determines the suitability of the sample-and-hold circuit.

after you've converted it to digital what the sample-and-hold circuit does is immaterial.

Bye. Jasen

Vote
Twitter Facebook LinkedIn

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required