I recently got a lightly used Hottop 2k+ (comes stock with BT probe & Artisan integration). I started with Rao's suggestions for BT/ROR curve settings- ROR was useless because of noise from BT probe. So I added smoothing.

Christopher Feran advised fixing the BT noise problem, not band-aid the real issue with smoothing. Rip it out & put in a new one. Steve K suggested the BT may have electrical interference.

I'm curious if anyone here with experience w/ Hottop or Artisan has input on the BT probe / Artisan settings!

Use the lowest settings you can and make it useable for you. Every machine is different. You can read On Idle Noise on the Artisan blog and decide if you think its worth changing it. Could be the Hottop has inherent noise or it could be a ground loop in your system. The Phidgets Hubs are better than the old meters. And they have a new Hub out that might help some, so you could try that and see if helps, and get a 3MM RTD instead. Only you can decide if this matters to you on the HotTop - I am not familiar with how precise the control is or how quickly it reacts. Some smoothing seems ok in your situation.

Michael; I went back and forth with Taylor on different Artisan settings, effect of other devices/ground loops in the circuit, etc. The TC's run into the HotTop controller, then to Artisan, and I didn't know if it might have its own quirks (e.g. like the Center meter does). Indeed one of the possible remedies might be to bypass it and use a Phidget meter instead. But before spending money experimenting I advised Taylor to post here as the Artisan team might know more about HotTop issues. FYI; The first profile in his post is purposefully with limited smoothing to possibly show some of the underlying noise.

Steve and Taylor if the meter is in the Hottop I think you use smoothing, and first learn to roast with that on the Hottop. Then decide if there is sufficient control present that doing something else matters. I suspect since its not a gas drum with heat retention, learning to anticipate what it does is the best course of action and one set of meters is just going to have a different anticipation factor than another.

The two examples were very helpful. I thought the second one was useable for learning the Hottop.

Maybe another Hottop owner might know more.

A single point only, but I was using a 3s delta BT/ET with smoothing of 6. Smooth curves was set to 7, and drop spikes enabled.

These settings are in Config->Curves->Filters.

The default filtering settings for the hottop were awful IMO. I'm sure that you could improve on my settings, but I found these were a huge improvement from the default.

I should also point out that the periodic ripple in the first set of curves looks like the effect of quantisation noise to me - you can see this in the "staircase" effect on the temperature curves. The way to get around this is to do some filtering of the temperature signal before finding the gradient. This filtering is what introduces the slow wavy variation in the second set of curves.

I think by playing around with the filter parameters you can get far fewer artefacts, and a usable RoR curve. However, the problem is the physical resolution of the temperature measurements, and I don't think this can be fully rectified with signal processing.

Phidgets don't seem to have a problem with quantisation noise, as the minimum change in temperature is lower than the measurement noise (0.05F or something).

Thanks all for the replies. I'll definitely try out the settings you mentioned.

I was thinking the same sort of thing regarding the "staircase" form of the BT curve. It seems to me a lot of the noise in ROR up/down is due to the stepped nature of the BT curve.

I've never messed with filter settings but I'll dig into that on my day off!

The hottop does not deliver any decimals (despite I advised the developer to output those before the machine was on the market) and thus delivers a signal which makes it hard to compute a smooth RoR from it. To avoid duplicate values you should increase the sampling rate to 5sec (and hope that the temperature changed more than 1C within those 5sec;).

Under you should tick "Interpolate Duplicates" with a delta of say 0.2C to avoid that two succeeding equal readings smash of the RoR signal no change in temperature equals a delta of 0. In such a case the duplicate value is suppressed by Artisan and interpolated based on the next and previous reading. This avoids those RoR spikes to 0 on duplicate readings.

Other than that and increase smoothing which adds a delay, there is not much that can be done on the software side to this.

If after trying these suggestions the result still isn't acceptable, I'd think bypassing the TC's into a Phidget would be a possible solution, and we know it works well and doesn't cost a lot. You could still use the USB connection to the HotTop for machine control from Artisan, and just reconfigure Artisan to read the TC's from the Phidget instead. Worth exploring. I did something similar with my Phidgets and Arduino for my roaster, but then I don't mind hacking things either.

Note that you each TC can only connected to one meter. Thus you cannot connect them to both, a Phidget and the Hottop electronic. Most likely the Hottop does not work without its TCs connected as it uses their signal for some safety checks. You most likely will need to place extra TCs (or RTDs) into the machine to be connected to Phidget modules.