0

Damaging Analog-in pins Itsy NRF52
Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.

Damaging Analog-in pins Itsy NRF52

by arlenn on Wed Jul 28, 2021 12:50 am

I am using an ItsyNRF52, and have managed to wipe-out a second analog pin. The first one wrote off as me being careless and ESDing it. This second one is definitely not the case. It was working happily, and while it still reads happily the resistance has dropped from the meg-ohm range like the remaining good pins to around 200ohm. This is the same symptom as the first one I seem to have damaged. 200 ohm seems way to low for a mis-configured pull-down.

I am not sure what I am doing to kill these things?

I am getting my input from an Allegro ACS781 isolated current sensor supplied with 3.3v from the Itsy, schematic included below. The voltage is less than 3.3V and with the LPF current is limited 3ma. I don't feel I am being abusive at all. I am not out to blame anybody, these aren't the cheapest things, and I cant keep replacing them.

It itsy is powered from a 12V switching wall-wart though an isolated power-supply module. There doesn't seem to be any other damage to the Itsy other than these analog channels that keep getting wiped out. Being isolated I really doubt I am getting high voltage tracking though the ACS781, but I will look for whiskers or something like that. The measured process has been up to 30-40v, so if there was some stray voltage, I am sure I would have seen it before this.

I would appreciate any thoughts on this. Does extra care need to be taken around the NRF52 analog pins? do I need more protection here?

acs781.png
acs781.png (30.41 KiB) Viewed 98 times

arlenn
 
Posts: 29
Joined: Wed Mar 11, 2015 3:06 pm

Re: Damaging Analog-in pins Itsy NRF52

by arlenn on Sat Aug 14, 2021 5:52 pm

It might be a bit pre-mature to declare victory, but I haven't had any more damage in a while...

Just about every manufacturer of SAR ADCs seems to recommend buffering the input of the ADC. I think this primarily to maintain voltage stability during the reading from the capacitive action of the device. After having a lengthy discussion between a colleague of mine who has worked with SAR ARCs extensively; the only thing we could come up with is the relatively large capacitor, part of the LPF, was causing an inrush dv/dt event with sufficient current to wipe out the ADC mux.

I have reimplemented the circuit by bodging on a buffer amp (OPA340) using the following articles as a design guide. It has been holding up, so far.

https://pdfserv.maximintegrated.com/en/an/AN6034.pdf
https://e2e.ti.com/cfs-file.ashx/__key/ ... -REV-b.pdf

arlenn
 
Posts: 29
Joined: Wed Mar 11, 2015 3:06 pm

Re: Damaging Analog-in pins Itsy NRF52

by adafruit_support_mike on Sun Aug 15, 2021 3:08 pm

Most ADCs have their own sample-and-hold capacitor at the input, so I wouldn't expect inrush from an external capacitor to be a problem.

The drop from high impedance to ~200 Ohms sounds like a mosfet with blown insulation between the gate and the channel. That's a common form of ESD damage, and there are plenty of mosfets in a pin's IO circuit.

I'm inclined to think the second dead pin was damaged at the same time as the first, but took longer to fail.

arlenn wrote:Just about every manufacturer of SAR ADCs seems to recommend buffering the input of the ADC. I think this primarily to maintain voltage stability during the reading from the capacitive action of the device.

Pretty much. The sample-and-hold can also be treated like a resistor whose value is proportional to the size of the capacitor and the switching frequency. It draws current from the input source, and will produce some voltage drop on a passive signal source. An op amp isolates the input signal from s&h loading, and has close to zero series impedance at the output.

The op amp can also filter the input to prevent noise folding. In any sampling system, you can only get accurate readings for frequencies up to half the sampling frequency. Higher frequency signals still get sampled, but at intervals larger than half their wavelength. The resulting samples will look like a frequency less than half the sampling rate, and multiple high-frequency signals will map to the same low frequency.

If we assume a 1kHz sampling rate for convenience, 1100Hz, 2100Hz, 3100Hz, 4100Hz, and so on will all alias to 100Hz. 900Hz, 1900Hz, 2900Hz, and so on will alias to -100Hz, which doesn't make much sense for sine waves, but shows up in movies when the spokes of a wheel seem to be rotating backwards.

An unfiltered signal can have high frequency components many times faster than the sampling rate of a device measuring the signal, and all of them will alias down to apparent frequencies that don't exist in the actual signal. A low-pass antialiasing filter will get rid of the high frequency noise and eliminate the artificial in-band signals.

Op amps are the tool of choice for making filters, so you get a low impedance signal limited to the ADC's sampling rate at the ADC's input.

adafruit_support_mike
 
Posts: 63701
Joined: Thu Feb 11, 2010 2:51 pm

Please be positive and constructive with your questions and comments.