Electronics > Projects, Designs, and Technical Stuff
Needing feedback and advice on refining a product for manufacturing
<< < (3/3)
coppercone2:
for 24 bits EMI.

I noticed a huge difference with just copper shielding on a 16 bit system, built with isolation and all that stuff.

not so much with the DUT (I tried shielded and unshielded cables) but much more so for the actual circuitry.
OM222O:
the top layer will be a solid ground plane (except for the places where there are connectors which will help with shielding. I might also add some metalic tape to the inside of the box if I see it makes a big difference.
coppercone2:
my circuit had a solid ground plane but on the bottom layer, the impedances of the filters and stuff were around 20k and the input impedance was high with a instrumentation amplifier (but leading wires out of the box did not seem to result in high noise, when comparing shielded and unshielded RTDs.

iirc i had something like 14 bits usable without a shield that went to full 16 bits usable with a shield and much less spikes, resolving a 20Hz filtered signal. IIRC something like 0.002C usable fairly stable resolution with I want to say 300C span of 100 ohm rtd biased by like 300uA? i don't remember the details but I used oversampling and some DSP. i could do it better now because I resolved alot of uncertainties about the design over all that time and I can identified some mistakes with capacitor choice, circuit lay out, and I have better equipment to tune the design. still want some specific signal analyzers to get into the heart of whats going on behind the scenes though.

IIRC it was a 16 bit system oversampled to 18+ bits and filtered after over sampling. You can get like 0.001 but it was kinda slow and I did not like how it behaved with the time constants of the measurement piece, It looked stupid without higher speed (i like when it moves around nice not jumping last digits), I like dynamic response, it also got some jumps once in a while which was where I was messing with different box-car filters etc to try to stop that behavior but it was getting really slow and I don't have the energy to deal with trying to make efficient sort algorithms for stuff like median/outlier rejects. We will see eventually. Past the 18 bit oversampling it was janky. I suspect there was RF and other issues. Did not have money at the time to pursue it and i got burned out working on the same thing with little reason.

the whole circuit had a bunch of stuff and it was medium density and filled a 6x6 inch sheet.

with a modern Delta-sigma ADC you should have a much easier time though. I was using a mid speed SAR.

I do suspect though, that aside from the spikes, more then 3 bits of oversampling for resolutions sake is going to end up shitty. I don't think they are made that well.

if you need code I used quite a bit off some arduino controlled beer fridge blog post that I found a while back. at least for more complicated stuff like trying FIR filters etc
OM222O:
Ah! that's your issue then! SAR ADCs usually have bad INL and noise response. If you read the datasheet the ADS1219 is a SigmaDelta ADC which automatically over samples at 1.024Mhz. as I showed in the pictures, I could read downto 10uV accurately (not just resolution) with +-1 LSD. I used a fairly basic RC filter with 100n X7R and 1K resistors (cutoff frequency of about 1.6KHz) I can improve that by using better capacitors to reduce noise,  and maybe add an option to bypass the capacitors for better AC performance (unlikely tho ... I don't wanna add a lot of complexity for no good reason the data acquisition mode isn't that great with 1Ksps anyways!) but I think 5uV accuracy should still be possible with 1uV resolution. that is honestly more than enough for anyone who buys a device like this rather than a 400$+ lab equipment. that's not the market I'm aiming for either, so I'm pretty happy with how it performs so far.
Navigation
Message Index
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod