How to measure a single-ended +/-10V signal with the AD7768-1?

Hello Niall,

I worked through the post you suggested.

The LTC6363 (a quite impressive part) with gain of 0.5 converts my single ended +/-10V input signal int o a +/-5V (differential) signal. The AD7768-1 is principally capable to convert this signal range but it's "on  the edge", and I need +6V and -1V power supplies for the LTC6363.

What about the AD8475? It has a gain of 0.4 and would convert my +/-10V input signal into a +/-4V (differential) signal for the AD7768-1. This can easily converted using a 4.096V reference and a single +5V power supply for OP-Amp and ADC.
Especially the single +5V supply eases my PCB layer stack and allows me to use more ground-layers for signal separation.

Will the AD8475 limit the AD7768-1 performance?

What do you suggest as high impedance input buffer,
a PGA like e.g. the AD8250 or a low noise OP-Amp?

Unfortunately I will not have enough PCB area to place a 4 additional OP-Amps, implementing a 8th order Low-Pass-Filter.
What is your LPF suggestion that requies minimum PCB area and does not degrade AD7768-1 performance?
(AD7768-1 will alwys run at full speed modulator frequency of 8.192MHz)

Regards, Niels

Hi Niels,

If +6V and -1V supplies are a concern could you use a gain of 0.4 using discrete resistors and the LTC6363? The bandwidth of the AD8475 will likely be too low to settle the effects of the switching from the modulator of the AD7768-1. 

I would also look at the harmonic distortion performance of the AD8475 vs the LTC6363. It might be possible to use an amplifer like the ADA4610 followed by the LTC6363/ADA4945 to drive the AD7768-1. 

One thing to point out which may not have come across clearly on the suggested post, an 8th order antialiasing filter in not required. Take a look at the AD7768-1 datasheet, figure 93, for details of the low pass filter required. The digital filter rejects alias tones and the user only needs to take care of images relating to the modulator frequency. A 3rd/4th order filter is sufficient, provided the selected amplifier has enough bandwidth to provide the required rejection. 2* Fmod is the frequency at which the digital filter provides no protection.

Regards

Niall

Hi Niall,

sorry for my late response, but I did not get an e-mail alert for your post.

When using LTC6363 with discrete resistors, what about the (unwanted) effects that acome accross with resistor mismatch?

What resistor tolerance and temperature drift performance would be necessary? Are 0.1% resistors with TK50 sufficient?

Regarding the AD8475 bandwidth: did you consider that the AD7768-1 has input precharge buffers that ease the driving requirements of the external OP-Amp?

Regards, Niels

Hi Niels,

Certainly you will need to consider resistor mismatch and temperature drift. I would recommend resistor tolerance of 0.1% as being the worst case, you may find that your application requires greater precision. For the temperature drift, some factors to consider would be the temperature range of the end application, and whether you can calibrate or temperature compensate in some way?

Even with the precharge buffers enabled, the AD8475 may struggle to settle the AD7768-1 input when the AD7768-1 is running at the fastest Fmod rates. At lower Fmod rates you will find that the AD8475 can drive the AD7768-1 input with no adverse settling effects.

Regards

Niall

Hi Niall,

the end application's temperature range will be from -40°C to +85°C. The product will be calibrated at 25°C in the factory, but nevertheless temperature drift is of concern.
What way of temperature compensation would you recommend? I could add a temperature sensor on-board.

OK, so AD8475 is definitely too slow. What about using the AD6363-0.5 with additional 350 Ohm series resistors at it's input pins? This should result in my required gain of 0.4 with only two external resistors, isn't it?

Do you have a suggestion how to implement the required 3rd/4th order filter with minimum components/OP-Amps?

Regards, Niels

P.S.: I still get no email notification when you write a post in this thread. :-(