Temperature probe calibration

[svejk]

Is there a way to calibrate the temperature probe(s)?

I'm planning to equilibrate
- in ice bath @ 32ºF/ 0ºF
- water at room temperature @ 22ºC
- at higher temperature

Out of the box, the two probes are off by about 2ºC (but agree with each other) and I'd like to correct that before proper use.

[hebnern]

Currently we don't support this. I am surprised that they are off by that much. We are using DS18B20 sensors which are supposed to be accurate to +/-0.5C. I guess the manufacturer's rating is a bit optimistic.

Calibration could be a bit tricky to handle. For instance we would have to consider:

• Does the calibration offset/curve apply to the probe itself or the port that it is connected to?
• If it applies to the port, what happens if you accidentally swap the probes at some point?
• If it applies to the probe itself, we would have to remember the unique address of the probe and associate the offset/curve with that, regardless of what port it is connected to.
• What if you have more than two probes? Do we have to remember offsets/curves for all of them? What is the maximum number of probe calibration curves that must be supported?

I think this would be a nice feature to add, but it would take some thought to get it right. What do you think?

[svejk]

The +/- 0.5 definitely looks indeed quite optimistic,
but as the brew bit is a temperature controller, it's affecting the core capabilities.
What's the point in PID controlling, when you are far off.

Now to the calibration:

• If you have unique addresses for each probe, than you can calibrate by probe id.

• Single point calibration (offset) might be a first step to correct the reported temperature.

• If memory is an issue, keep only the profiles for the currently connected probes (or keep the "profiles" as simple as an offset).

I agree that this can be turned into a nice feature, but a basic offset is a must, as the probes seem to be a little bit out of spec.

[hebnern]

Ok, I think a simple offset per probe id should be easily doable.

[hebnern]

We just pushed 1.2.0 which addresses this.

Thanks for the feedback guys. We just pushed version 1.2.0 with the following fixes: FIX: Fix display of negative quantities between 0 and -1. NEW: Add a configurable screen turn-off delay. FIX: Fix static setpoint units/conversion when set to Celsius. FIX: Add hysteresis to PID mode to prevent constant heater/cooler cycling. FIX: Fix hysteresis implementation. NEW: Add probe calibration utility. FIX: Fix rapid relay cycling caused by invalid cycle delay setup.

[tracyj]

I just tried the probe offset, and it doesn't seem to change anything. If I go into the probe offset and change it by -1.7 degrees, and then back out with the left arrow, nothing has changed. The displayed temperature still remains the same. Then if I go back into it, the offset is back at 0.0. I would think that it would have the previous offset set. Am I not doing something wrong? I have also updated to the latest firmware. If I check for updates, it says that it is up to date.

[bmoney]

We just found the bug and the fix will be in the next update. Thanks.

[bmoney]

We just released version 1.3.0 with the update to address this bug if you would like to download it and try it out.

[svejk]

Took me some time to get the software update installed (thanks iphone! the wifi hotspot did the trick to get over the problem to update 1.0)
Probe offset is simple and works.
Thanks guys!

[dan_maslowski]

Just fired this up and immediately subscribed to the unstable update path. Very first thing I did was check the temp probes. Probe 1 was reading 7.7 degrees F high. Backed it off in an ice bath and am very happy. Calibrating probe 2 now.

I agree that an ABV setting/probe would be nice. Thanks gents this is excellent so far.

Edit: Just set probe 2. Off by 4.9 degrees F. I agree with the poster that suggested this. With out accurate temp reading and calibration this would have been far less useful. Obviously a feature to add will be to have a button that says calibrate probes. Push the button, insert probe into ice bath. Record for two minutes and automatically apply the offset. Fairly trivial bit of code I would imagine.

[hebnern]

Yeah, we were thinking of doing an automatic calibration, but figured that inevitably someone would ask for manual control anyway. So we started there instead.