AD7768 analog input range in PIN mode

Hi John,

I would check that the reference voltage to the parts is as expected at the pins.

It could also be being clipped at the driver amplifier. Are you able to check this and check that you have observed the necessary headroom of the amp? What amp and what supplies do you use to drive the ADC?

Regards

Niall

Hi John,

Firstly, I would say if you are using the ADA4622-2 that this amp may struggle to drive the AD7768-4 directly on it's own. Generally we would need to use an additional driver amplifier between it and the ADC to settle the input to the ADC due to the fast switching of the modulator. Something like the ADA4940 for example. The AD7768-4 analog input precharge buffer can help here to some extent. In terms of performance the ADA4622-2 would only make sense in Low Power mode (Eco) due to the voltage noise density, bandwidth and slew rate. 

Can you confirm that the expected voltages are on the AD7768-4 supplies and RegCaps? How do you apply the common mode voltage to the amplifier? Have you tried a DC input or is it an ac source you are using? If you could send a screenshot of what you are seeing it might help to narrow down the problem.

Regards

Niall

Thanks, Niall.  That was my understanding, and indeed it is what I see with the EVAL board, but my breadboard clips when the differential input exceeds 2Vpp.  My breadboard actually has three AD7768-4 daisy-chained, and I’ve tested channels in each of them (in case one was damaged), they all exhibit the same mystery.

jsh

schematic attached – I’ve added analog input buffers – doesn’t change much

jsh

REF+ = 4.096 at the device inputs.  I’ve tried driving the inputs directly as well as with buffers – not much difference, I’m using ~7V (V_AMP) supply for buffers.  I used AD4622-2 for buffer(s).

The datasheet is a little fuzzy on what happens in PIN mode after /RESET: it says the device is set to a particular configuration, but what I observe is that it reads the DECxx and MODExx pins and sets to that ODR and power mode.  In the schematic I sent you the /RESET and /START are pulsed manually, but more recently I’ve connected to a microcontroller that sets the DEC and MODE bits, then generates short /REST followed by two /START pulses in quick succession.

jsh

Thanks, Niall – I understand about the analog buffer drive issue; my intent is to provide different analog “front ends” for different applications and my first application needs high impedance buffers but will be used at the lowest ODR (1KHz) in Eco mode.

Your request for pin voltages turns out to be key: after carefully probing with (literally) a needle, I discovered that my “decal” for the device had terminals 55 and 56 (FORMAT0, FORMAT1) reversed, so I have GND on FORMAT0 and IOVDD on the pin that is FORMAT1 on the 8-channel part and labeled DGND on the -4 part.  Not sure why it works otherwise, but I’m guessing that the pin reversal has something to do with the issue I have seen.

jsh

Niall, as so often happens I spoke too soon.  I have lifted the FORMATxx pins on one of 3 daisy-chained AD7768-4 on my breadboard and tied them properly (FORMAT0 to IOVDD and FORMAT1/DGND to GND) and run some more tests, no change.  Here are the pin voltages:

It occurred to me that the floating AINs might pose a problem so I tied all the unused ones to Vcom and obtain these results:

actually from a couple of days ago but no change:

not specific to the problem but just information - spectrograms of the first 17 channels in my 3 daisy-chained AD7768-4 “breadboard”.  You may recall in an earlier message I was surprised to find the 4 “missing” channels in each AD7768-4 in the serial stream.  In this case the unused AINxx were just floating

what do you think?  Did I damage the parts with the FORMATxx mixup?

jsh

Hi John,

I don't expect that the FORMATxx mix up would damage the part, especially since it seems to be converting fine, but with added distortion.

The next thing I would be pinpointing is your signal source. We would always use an audio quality source to characterize our ADCs. (I'm not sure they can even go below 10Hz) Is the source known to be good, especially at the low frequencies you seem to be running at? What's the distortion specified as for the source? The FFT data looks similar to what I would see if either a) the amplifier doesn't have required headroom/bandwidth or b) I applied a sine wave from an everyday signal generator.

Regards

Niall

Hi John,

Firstly I must clarify, you must use an amplifier to drive the inputs of the AD7768/AD7768-4. The analog input precharge buffer is not a full buffer but an aid to the external driver amplifier only. This application note details the performance achievable with various different amplifiers driving the AD7768/AD7768-4 in all three power modes. You must include an anti-aliasing filter on the inputs. The R used needs to be sufficiently large so that the driver amplifier is stable and not oscillating. Without this buffer the low output impedance source will not be able to drive the switched capacitor load.

I failed to acknowledge that the input should be ~ -12dB from full-scale for a 2Vpp input in my last post (assuming you have a single ended input). I was focusing on THD. What max output codes do you see? I'm not fully clear, is the graph shown in image010.jpg a plot of the input or a plot of the ADC output? 

Regards

Niall