TIDA-01471: 8-channel IEPE vibration sensor DAQ

Hello Jorge,

If your signal bandwidth is 15kHz to 20kHz, then you probably need sampling speed above 64kSPS. note that higher sampling rates would improve the SNR and requires more power and higher speed SPI.

At this speed, a multiplexed system might be challenging and I believe would degrade the performance expected from the converter and front-end. we are talking here about 18b of ENOBs, not so many multiplexed systems running at higher speed can achieve such high resolution. the high speed amplifiers required to support switching transients would inject a huge amount of noise to the signal path.

If you are looking at 8 channel implementations, I would suggest the following:

- use 2 x 4 Ch based on the ADS1274 which is 4 channel simultaneous sampling using the same ADC core. this way you have the benefits of integration (power, and footprint) as well as the benefit of separate channels (higher SNR)

- if variable gain is not needed, I suggest to drop the variable gain stage to improve SNR. or if really required, then better to use a low noise VGA.

- I suggest to use a simpler current source instead of the XTR. it can be either fixed current (2 or 4mA), or variable. I'm going to publish an article soon about alternative circuits for high side current source to be used in this circuit. the simple fixed current is shown below.

-  along with the REF6225 that I have in the TIDA-01471, you can also use the new REF3425

The 4 channel circuit would look like this:

The current source would be something like this:

Hello Jorge,

There is an important thing to note if you are going to use sigma-delta converters (like ADS127L01) in multiplexed architectures: Sigma-delta converters have a digital filter integrated, and it's supposed to be used in continuous conversion and the filter assumes all samples belongs to the same channel. If the converter is used with multiplexed channels, the digital filter has to be flushed out between channels and there will be a latency of multiple cycles in-between. for example if ADS127L01 will be used in multiplexed mode, the maximum sampling frequency would be around 100 kSPS instead of 512 kSPS, this might be sufficient for multiplexing 2 channels only given the BW you mentioned. Moreover, the filter used to attain the shortest latency is not the flat-band filter favored by the IEPE sensor applications, which means further digital processing is required after the ADC.

The alternative is to use a SAR ADC (like ADS8900B converting 20b @ 1 MSPS) instead, or opt to the simultaneous sampling which I mentioned earlier.

The TIDA-010045 uses MSP432's internal 14b ADC which, when runs full speed, would achieve about 11b of ENOBs, the TIDA-010045 targets the cost-effective low-performance application. I assume your "DAQ" module would demand higher performance.

Kind Regards,

It seems that MUX is better put after the scaling/HPF as mentioned before, that means we have 3V signal to multiplex. I would say TMUX1108 should be good for 8 channels. but some simulation beforehand is necessary to avoid any surprises. the full TINA-TI circuit model of the signal chain is available on the TIDA-01471 folder.

Hello Goerge,

The proposed current source is quite simple and possibly sufficient for the application.

It's not clear to me what is meant by "common-mode voltage of the current source", is it the output DC level? this is set by the load like in IEPE sensor case and should lie within the compliance range of the current source.

In general the specs for the current source needed for IEPE sensor are:

- 4mA min current, might be higher according to the cable length or lower for some sensors.

- 0-22V compliance range (dependent on the sensor DC bias range and output range)

- 24V min supply

- ~1uV rms noise (to achieve 17 ENOBs with the given front-end in TIDA-01471)

- >300kOhm output impedance at 20kHz

for the IEPE sensor application, absolute accuracy, drift don't play important role in performance as the IEPE sensor output stage DC and gain doesn't depend on the excitation current.

There are many different topologies that can be used for implementing the current source, one of them is mentioned in this ADJ article.

Best Regards,

Ahmed