ADC multiplexing

I can build it for $5000, since i am doing a similar one anyway. I'll throw in Android USB interface for free, so you can look at the stored data with a phone/pda/tablet. We are logging 10 channels at 16 bits, so adding 22 channels is no big deal. Micro SD is included, but bluetooth is extra.

Does a parallel interface indeed waste less time? Have you looked at microcontrollers, and their available peripherals, lately?

What?!? Please tell me that you are a student, not a senior engineer at a company that makes medical equipment.

Yes, if you're sampling each channel at a rate of 250Hz, your overall sample rate will be 8kHz. If you can't work out _why_ this is, and if you're still in school, now may be the time to switch to a business degree. If you can work it out just fine but lack confidence -- buck up, in the real world you'll need to be smarter than your boss, so you need to grow a backbone.

Yes, you need one that can acquire data that fast, and you need one that can store however much data you'll collect -- which means that you need to know how much that is, which means you'll need to know how long you'll be collecting data for.

Can't you reason it out? You seem to be putting a lot of reliance on applying formulas that you've learned, without thinking things through. Has the total number of channels changed? Has the sampling rate per channel changed? If the answers are "no" and "no", then can the overall result change? If so, how?

If you ignore channels 16 through 31, what's your overall sampling rate? If you ignore channels 0 through 15, what's your overall sampling rate? Can you draw one circuit block that can use a 16-bit ADC and that can be used twice to do the job you want?

I would suggest that you work out _on paper_ how the data is going to go. Make a timing diagram that shows when each word (or byte, or whatever) must be moved from where to where. Then make sure that it all looks like it'll work.

Does the Bluetooth and SD card overhead impact your ADC decision or your microcontroller decision? Do you know how many bits you need to twiddle to talk to the SD card? The Bluetooth?

I think (and hope!) that you're a student. In the real world you need to do the same thing as an engineer that you do as a student: get a bunch of data sheets, and start looking at details. Find a microcontroller, figure out _exactly_ how it'll talk to the ADC, figure out _exactly_ how it'll talk to the SD card, etc. You absolutely positively can't be vague here "then magic happens" is only acceptable _before_ you've finalized your microprocessor choice. Afterwards "then magic happens" turns into "then we have to choose a new microcontroller at great expense".

lts

Sound like a jr. engineer just out of school, and she hasn't done any of this stuff before. But that's OK, she can pay you or me to do the first prototype, and learn. 32 channels at 16 bits in mV range will be challenging. I'll do her a big favor and take on the project, if her boss would pay.

You will need a phase shifter/attenuator on each channel that matches all the channels up on both timing and amplitude as path length differences and attenuation differences between channels will cause problems without real controls in place on each channel, typically toward the tail end of the path. Then, a splitter/combiner to mux them all together. That is also where you look to see if they all got synched up right (calibration). Get 'em all within a few femtoseconds of each other and then you data (or its timings) can be relied on.

That's analog side mux.

Is 70 Hz the highest frequency in the channel ?

Are there interference at higher frequencies (such as mains third harmonic at 150/180 Hz).

Does the frequency range go down to DC ? Is DC accuracy needed (offsets and temperature drift) ?

This would typically require more than 40 dB gain before the ADC (either on channel specific amplifiers prior to the mux or a common amplifier between the mux and the ADC).

Is the signal single ended (ground referenced) or is a differential signal available ?

One way of doing this is to use channel specific (possibly differential) amplifiers with more than 40 dB gain and some low pass filtering to cut out frequencies above 70 Hz and use a 32 channel single ended mux into the ADC.

Alternatively, at such low levels and if DC accuracy is needed, use a differential (2x32 channel mux) followed by a 40 dB differential amplifier.

The input frequency range should be limited before the mux or use oversampling at much higher frequency than 250 Hz/channel and do per channel digital low pass filtering in software, before downsampling to

What kind of mains hum levels at 50/60 Hz are expected ?

You may have to chose the sampling frequency with this in mind.

Most likely 32 pre-amps before the mux. Otherwise, cross talk of several mV in the mux will overwhelm the signal. I didn't even bother to suggest the details to her. She needs me or one of us here to do it.

Hi,

Can anyone know about the 32 channel ADC that I can work with?

Thanks Jess

Have you not read the responses?

We've told you several different ways to solve the problem, none of which involve using parts that semiconductor vendors don't bother to make because a competent engineer knows how to solve that problem with less money by using a small collection of parts.

Either you've been assigned this job because you're getting paid to do it, in which case you should expend some effort to _do it_, or you have been assigned this job because you're a student who will learn from completing this job, in which case you should ... expend some effort to _do it_.

But, like Linnix, I, too, have a solution. I'm not so crass as to suggest that I design this for you and take your money. Oh no -- I just happen to know that this is a bog standard part built by Wescott Design Services. It happens to only be available on a 9-month ARO basis, there don't happen to be any pictures (or schematics, or board layouts) available at this time, the first one will cost you about as much as Linnix would charge you do do the design and prototyping, and it'll look like something built in a small shop on a prototype basis. But the _second_ one will only cost as much as parts + assembly + a good healthy profit.

Than you veru much for your emcouraging remarks. Can any one suggest a

jess

Disguised stupident from Bangalore. Male, of course.

There aren't any. You will have to make one.

Actually this part is kind of interesting.

Here's a 64 channel one:

But really, make one from a, say, 4 channel ADC and some analog multiplexers.

okay. Thanks.

Look guys, I know that you guys are professionals and master of electronics. You know if you do not like the posted question or the intelligence of the person who is posting the question than please do not answer the question. Let it go... Why you bother to reply at all? Is it not a waste of your time? Let some body else reply who really do want to help. Your replies might stop other people to reply. I hope you understand . Thanks jess

Welcome to usenet!

On a more serious note: what kind of device are you designing? And what kind of project is it? Providing this info will help to get more sensible answers.

sci.electronics.basics is aimed at beginners, and gets the OP different types of responses. This group is intended for people with a lot of experience to bounce ideas off each other. Posting from Google Groups also means that most people will never see your posts, because of all the spam posted from there.

Jessica Shaw expounded in news: snipped-for-privacy@15g2000vbw.googlegro ups.com:

As Neil Young would croon in Ambulance Blues:

"And there ain't nothin' like a friend Who can tell you you're just pissin' in the wind."

Warren

Get back to us when it is.

-- Rich

Well, the Cypress PSoC 3 might do the trick (CY8C3xxx). It can do

To handle the 10mV range, you may need external amplifiers or route the mux output to an external amplifier. You may need runtime calibration methods and precision external voltage reference according to your accuracy needs.

PSoC 3 comes at below 6USD

Regards

Klaus