16 bit AD converters

Has anyone direct experience with 16 bit ADC's ? In particular we are trying to make a choice between the LTC 1859 and the Texas ADS 8345. We also looked at the Analog Devices AD7686, which is comparable specwise, but in a leadless chip package. Would like something a bit easier to handle like a SOIC package, that we can at least prototype fairly easily.

We need at least 7 channels, 16 bit, 100 Ksps or so. SPI bus to processor is preferred. At that speed, SAR seems the only way to go, which is fine. 3 or 5 volt single supply is ok. We can scale the transducer outputs to anything reasonable inside this range.

The ultimate aim is to produce a full 16 bit real resolution over at least 6 channels, at 100 Hz or so per scan, plus 1 channel where we can accept lower resolution and speed. I figure that by using the speed, we can sample and average to get at least 16 bits of real resolution. Absolute accuracy is not too much of an issue, but resolution is.

I have use 12 bit devices for years, but I know 16 bits is a much harder target.

Suggestions from anyone who has used either or both these devices, or any others with similar performance are welcome.

--
Regards,

Adrian Jansen           adrianjansen at internode dot on dot net
 Click to see the full signature
Reply to
Adrian Jansen
Loading thread data ...

We use AD7699 in a few products. It's an 8-channel mux'd 16 bit ADC that works very well. This is one place we use it:

formatting link

We only used one channel in each ADC to keep the sample rate up, but we use its internal mux in other products. It is a leadless package, as so many things are these days. It is just barely hand solderable.

You could maybe consider using a 24-bit delta-sigma ADC per channel. I think there are some duals around these days, too. Many of them can handle low-level differential inputs, like load cells, directly without any front-end amps.

John

Reply to
John Larkin

I've used some, but I haven't really wrung them out.

TI makes some fast sigma-delta converters that may fit your bill -- check out their web site. You have to clock them pretty fast, and if you want to get 100ksps you have to let them dictate the readout time (or fiddle with their clocks), but they may do what you need.

--
Tim Wescott
Control system and signal processing consulting
 Click to see the full signature
Reply to
Tim

Consider audio ADCs. They are simple and inexpensive. It is relatively easy to get 16 bit resolution with 100KHz simultaneous sampling up to 8 channels in one part. The DC performance could be a problem though.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

formatting link

Reply to
Vladimir Vassilevsky

We have used the LTC1865 16-bit serial ADC and have been quite impressed with it. I have to admit, however, that we are only using the 13 MSB's in this particular application, so if those are the only high-quality bits we wouldn't know.

One thing is above 12 bits, the shortcomings in the analog components really start to show up. We started using OP275's for general power regulation and generating offsets. They were too noisy. After much fooling around we found that the AD706 is REALLY quiet at long time scales, minutes and hours. (Our experimental runs can last a week, and a 1/f step will knock everything out of calibration.)

We also used an AD8132 differential amp on the input of a 14-bit ADC, and found it was noisy enough to be noticeable. We went over to the TI THS4131 and saw a significant reduction in noise. We had to slow down the whole system a bit to allow more settling time, but it was worthwhile.

Jon

Reply to
Jon Elson

Thanks to all for the comments, useful info.

--
Regards,

Adrian Jansen           adrianjansen at internode dot on dot net
 Click to see the full signature
Reply to
Adrian Jansen

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.