If anyone has some experience please share. 
Hi nomsmon, this is something I would also like to know.
Unfortunately I don't have the environment or equipment to do real accurate tests, my limit was comparing to the Atlas Scientific circuit but based on how pH is read and the tests I've done I think it's quite accurate.
I believe after a 2 point calibration on mine and Atlas there was a slight difference but close. Unfortunately I didn't document everything but I was happy with it. I did find my circuit was more stable than Atlas, the 3rd digit would bounce more with Atlas but on mine would move 7.001, 7.002 etc. Going to 10,000ths was all over on both so no point.
My circuit uses an ATtiny85 microcontroller, it's always calculating the pH and gets polled for the data. It has a decent opamp that connects to the microcontroller ADC and gives it high resolution so it can read in the 1000th's pretty well. It also has full isolation on power and data lines which stops interference from other probes and interference.
Currently the onboard code can manage a 1 and 2 point calibration and temperature compensation for readings but it's actually capable of much more. Sadly I'm not the best programmer and had to disable some features to use the Atlas driver in reef-pi. It can also do temperature compensation for the buffer solutions, a 3 point calibration and the ability to report the low and high slope of the probe to see health. If any programmers read this and can remove the Arduino Wire library I would love to hear from you.
With all that said you can only read the pH in reef-pi, if you want to calibrate on circuit you need to use a script I have. I also have a script that allows you to change the I2C address, reset the defaults and recall any of the stored information. The circuit is slow compared to Atlas, I think Atlas takes about 1.5 seconds to produce a reading, mine is set to every 3 seconds as I do a lot of sampling to get a better average, I found that made it more stable.
Here's some readings over 12 days I recently captured from my tank using this probe but it's over a year old now and it wasn't calibrated, my tank sits at about 8.2, I had a collection of 30 days in reef-pi but didn't save the chart, it was similar.
This chart shows a rolling average over 20 readings so you get smoother lines, you can see how it goes up and down during a 24 hour period. The temperature of the tank also fluctuated during this period and you can see it in the low/highs. I think Aug 30 and 31 the temps dropped. Aug 26 and 29 I was messing with controller so they got screwed up. So even if it isn't completely accurate you can still spot any changes in your system.
This chart is same data but actual saved values, no averaging.
You can see it's a lot more choppy but for a graph like that it still looks pretty good.
And while I've been typing this I started up a terminal capturing some readings so you can see what kind of data the bottom chart represents, these are every 4 seconds. This is off the controller, during my tests with the standalone version it would stick to a reading and I would think it stopped or locked but it's not quite as stable integrated into the controller, I'm trying to improve this, I want 10,000ths lol. And on occasion I have seen small spikes, it could be 8.632 and then go to something like 8.751 for a reading so I don't like that and trying to improve.
All these usernames are too much to remember for me. I was already interested in buying some stuff from Robo-Tank, but after this response I am absolutely sold
I'll design some tests to run when I get mine; may ping you for input on test parameter design
Makes sense and also explains why isolation is so important on the circuit.
The AT micro controllers are awesome and that's an awesome feature set; I'd be super interested in the temperature compensation. I'm a programmer by trade and hobby. I don't do embedded systems, but I've been looking for a good project to jump into it
That's awesome, thank you so much for sharing. Those numbers look great to me; deviation is relatively small and we do expect some natural fluctuations in pH.
So the controller emits a reading every 3 seconds but it is sampling off the ADC much more frequently than that? Does that mean you're doing some kind of aggregation over the 3 second window to produce the single measurement value?
www.sparkfun.com
can you try a grounding probe?
Robo tank ph circuit implements the open source reef-pi ph circuit, which has galvanic isolation built in. In fact that’s why it’s bit cheaper than equivalent commercial ph setups.
Sorry for taking so long to reply, I had to disappear for a bit.
My pH circuit is very different than the open source version, it has it's own microcontroller and works similar to an Atlas circuit, in reef-pi it uses the ezo-ph driver.
oh, forgot that. Is the schematic opensourced ?
