I'm about to design some AD7699s (16 bit ADC with 8 ch input multiplexer) into a new design. There's a possibility that one or more input channels might be overdriven. If the converter is being used to sequentially measure successive input channels, will the channel following the overdriven one (in time) be affected by the prior overdrive? More than the usual crosstalk?
I'm hoping to avoid having to power the channel driving op amps with power supplies equal to the voltage reference. I'd rather run the op amps and ADC from the
The datasheet is a tad fuzzy on this. It clearly says that the ESD diodes can handle +-130 mA, but they don't say that analog performance won't be screwed up by pushing current into them.
If you power the driving opamps from 0 and +5, there won't be a problem. But that may not give you the signal swing you need.
I recall doing a test on the ESD diodes, and I think I did see a channel-channel interaction when current was pushed into the ESD diodes. I can't find my notes on that, but I seem to recall it took a few mA to make trouble.
If I were you, I'd make sure that very little current can go into the analog inputs. So don't drive them more than a few tenths of a volt below ground or above +5.
Incidentally, if you use them in pseudo-bipolar mode, against a VMID low-side voltage, make sure VMID is Vref/2 +-0.1 volt. Otherwise weird things will happen.
John
--
John Larkin, President
Highland Technology, Inc
jlarkin at highlandtechnology dot com
http://www.highlandtechnology.com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro acquisition and simulation
The general rule is that you are only allowed to source very little current into the IC. The figure of maximum input current into any pin is the current at which the device is assured not to enter SCR latchup. But, that figure is only related to when the chip will short the supplies. Analog performace is degraded well below this. Our rule of thumb is to keep the current into a pin below 1% of the latchup current.
One device may have less susceptibility whereas another might have more. Specifically, we have seen one mux input that was overdriven affecting another active channel at very low currents
Active clamping is a good way to handle this and even better to limit the supply to the input circuitry (opamp). If you use active clamping be sure that the clamp responds quickly to not let any spikes through. Spikes will normally not trigger the SCR latchup, since the latchup takes time to occur, but can still F*ck up the measurements...
What I recall from testing the 7699 is that ESD current to ground was ok, but pushing milliamps into the upper ESD diode on a channel caused microamps to be injected into other analog inputs. Something like that.
The ADI data sheet is criminally deceptive, claiming that the ESD diodes are good for 130 mA without noting that just a few mA will cause errors. ESD diode charge spraying, and sprurious mosfet turn-on, are studiously ignored on most data sheets.
--
John Larkin, President Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators
Yes. Clamp or otherwise limit the input swing to close to ground and
+5 volts. It's unfortunate that VMID has to be VREF/2 +-0.1, because that makes the pseudo-bipolar inputs swing exactly to ground, and that makes the low-side clamping more difficult. The first time I used this ADC, I missed the +-0.1 volt gotcha and used 2.5 as VMID. Clamping was easy but Bad Stuff happened.
That's
Something like 500 ohms maybe. We did one board where we used a series resistor and then a cap to ground into each input, and had weird nonlinearities from charge injection, so we took the caps out. Charge injection can also mess up an opamp that drives the ADC pins directly.
It is a nuisance to make a 16-bit-accurate front-end that includes clamping.
I *think* the ground side is immune to moderate ESD current injection, which makes the clamping situation much easier, but I can't swear to it. I usually keep notes when I measure stuff like this, but I can't find them in this case.
--
John Larkin, President Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators
hnology.com=A0 jlarkin at highlandtechnology dot com
Yes, I got bit some years ago by a TI ADC that had a similar restriction to their "differential" input. Instead I'm using the prior opamp stage to handle dynamic differential aspects, and using the CPU with its auto/calibration to shift the midpoint where it belongs (it has to fix some system offsets anyway, so midpoint shifting doesn't cost anything more).
Thanks for the input resistor tip - comparatively easy to adjust once we get to building.
My one experience with over driving ADC chips was that as soon as the input voltage exceeded VCC, all channels would be screwed up. The explanation that we got (or made up for ourselves -- it's been a long time) was that as soon as you start spilling carriers into the chip through the protection diodes, they spill _all over_.
The reason it's my _one_ experience is because I Never, Ever, let that happen again.
I don't know what would happen if you exceed the reference but not analog VCC -- I would expect that you'd have a much higher chance of just getting an 0xffff from the ADC without any overt misbehavior, but I would also expect that the effect is going to vary from chip design to chip design.
--
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
Why am I not happy that they have found common ground?
Tim Wescott, Communications, Control, Circuits & Software
http://www.wescottdesign.com
Analogue muxes do the same thing, plus they connect all their inputs together.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net
Many analog mux's also turn on one of the mosfets in common-gate mode if you go a few tenths beyond either rail.
--
John Larkin, President
Highland Technology, Inc
jlarkin at highlandtechnology dot com
http://www.highlandtechnology.com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro acquisition and simulation
I expect that's what was happening--it sure was puzzling at the time (mid 90s).
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net
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.