Bean Temp Whackiness with Trier Use

[ContralaCorriente]

Wondering if you guys have any thoughts and or best practices around minimizing deltaBT volatility, specifically surrounding  the use of the trier? Seems like every single time I take it out I am getting very funky readings plus or minus, which are almost making the Delta temperature LCD unusable.

I will preface by saying that I have already troubleshot any grounding issues, though am using a phidget 1048 and may consider swapping to a mastech just to rule out the possibility that this is related to interference and not incoming air temp from the trier port.

[SusanJoM]

I'd hazard a guess that your reading instabilities might have to do with the settings you have selected that impact how often Artisan requests information from the device. 

What does your Tools > Extras first/Graph page look like? 

[Charlene]

Hi ContralaCorriente,

What does your plot look like? is it drop-outs going to zero? Is it high magnitude jitter?

Ah... never mind, just noticed your .pdf attachments.

Edit: I took a look at your plot. Don't know if it has been smoothed or not but what is there on the delta bean temp is a classic multi-tone; a combination of many signal frequencies aka tones.

You might consider adding ferrite beads to your thermocouple leads to try and filter those unwanted frequencies out.

No ideas on where it is emanating from.

Second Edit: Took a look at a write-up on Artisan's methods of calculating Delta-BT data and it sheds some light.

https://artisan-roasterscope.blogspot.com/2014/01/sampling-interval-smoothing-and-rate-of.html

From what I am reading, any low amplitude tone riding on the BT signal will be amplified within the Delta-BT data set by virtue of settings for BT sample time and the degree of smoothing applied to the BT data and Delta-BT data.

Are your thermocouple leads shielded?

There is DSP code that could be adapted to the Artisan program to mathematically filter out tones but at the expense of processing time.

I would try ferrite beads on all the thermocouple leads... it's inexpensive, is easy to do and they do work. Then, readjust sample time and smoothing settings and observe the results.

Here is a critical question...

Where is the 1048 located on your setup?

Right at the Huky? ...or further away, say sitting at your table top?

The leads between the thermocouples and the 1048 are somewhat like radio antennas.

[SusanJoM]

I also just noticed the pdf documents (Artisan saved graphs are much easier to read, by the way), and I see no evidence at all of electrical noise.  I think your irregularities are completely regular and are just an issue of smoothing and or sampling interval.

[Charlene]

I also just noticed the pdf documents (Artisan saved graphs are much easier to read, by the way), and I see no evidence at all of electrical noise.  I think your irregularities are completely regular and are just an issue of smoothing and or sampling interval.

Hi Susan, nice to meet you.

It is true that if sampling frequency is too low it will create mathematical noise similar to electrical noise.

It's also possible there are electrical tones as well but without putting a scope probe on the thermocouple lead to see what is there it's hard to say there is none.

[SusanJoM]

It's also possible there are electrical tones as well but without putting a scope probe on the thermocouple lead to see what is there it's hard to say there is none.

Sure.  I only meant that I see no evidence to suggest that what you call 'electrical tones' are lurking in the profiles in question.  That's all.

[Charlene]

Sure.  I only meant that I see no evidence to suggest that what you call 'electrical tones' are lurking in the profiles in question.  That's all.

Ahh... so you are saying this is typical of quantizing noise you have seen before when folks get the sample rate off kilter.

I have not used Artisan but have other programs for plotting analog to digital signal data.

[Charlene]

It does help to know the setup details of the application... In this case Artisan:

http://www.home-barista.com/home-roasting/artisan-problem-with-noise-in-bt-curve-t37349.html

(If posting that link is a no-no just say so and I will delete it. I am new to the forum and feeling my way around.)

[SusanJoM]

Links to useful information are always acceptable.  I link to HB all the time, although it is frowned on at HB to link back to here. 

Since I made a lot of the posts in that thread you linked to, I know the information is good.

For your future inquiries about Artisan, you might note that set up has been quite extensively documented in our downloads section.

[edtbjon]

Wondering if you guys have any thoughts and or best practices around minimizing deltaBT volatility, specifically surrounding  the use of the trier? Seems like every single time I take it out I am getting very funky readings plus or minus, which are almost making the Delta temperature LCD unusable.

I will preface by saying that I have already troubleshot any grounding issues, though am using a phidget 1048 and may consider swapping to a mastech just to rule out the possibility that this is related to interference and not incoming air temp from the trier port.

I very much doubt that using the tryer would cause these fluctuations.
First try adjusting the different settings according to the documents in the ARTISAN RESOURCES which Susan have compiled for us.
Second, check any grounding issues for the computer etc.
I think we can rule out faulty thermocouples, as both share the same nervousness, nor should the Phidget cause these kind of (still) small fluctuations. (We've seen one problem with the Phidget if memory serves me, which caused very extreme flucuations, nothing like this.)

Apart from these problems I'd say your roasts looks good (from a "declining RoR" viewpoint). I would maybe hit the gas a little bit harder in the beginning to get a little bit shorter roast times, but you got a good starting point to relate your further experiements to.
I can imagine it's hard to get any guidance from Artisan as that the RoR curve is all over the place while roasting, before thje smooting is applied. TIP: Right-click on the BT curve at e.g. the DE point and choose DE again, which will apply smoothing to anything recorded until then. I do it now and then just to smooth out the curve while in a roast.

[ContralaCorriente]

Thank you all for your responses. I have not had a chance to run some additional tests or roasts, but to answer a few questions that were asked:

Susan,

Screenshot of Extras attached. "Stock" settings here, as uploaded from Dave's (very helpful) settings file. Oversampling was NOT ticked during those roasts. Sampling interval at 3s. Based off of that and what I've gleaned from the Setup and Tuning Guide, I assume your suggestions would be to check oversampling and begin to increase sampling interval? Anything else?

edtbjon,

Thanks for the feedback! One of the points of these roasts was to see how the beans would react to that initial punch of heat. Curiously on batch 3 I gave it full gas (5.0+kpa) at TP and the beans were still slower to react than I wanted. Was using a 400g charge and roasting in my garage, which was at an ambient temp of 50-55 degrees, so that could explain the sluggishness. I am going to try to decrease the charge to 375g and hit the gas at 60sec instead of 90sec the next batches.

Charlene,

Thank you for the thoughtful replies. After I react to some of the smoothing suggestions, or even regardless, I am going to tinker with grounding each TP into a single plug, which many users on homebarista seem to have squashed nasty spikes with. At this point I am not totally convinced (or perhaps totally hopeful!) that this is interference based.

-M

[edtbjon]

Well, 5.0 kPa is definitely "full blast". The roast(s) seem to be somewhat similar to mine, so it's easy to relate. I usually hit the burner at 70 seconds after charge and I support the burner with just a little fan to provide some oxygen to the flame. My typical charge weight is 400g for which I set the burner to about 4.0kPa for the first three or so minutes. That will usually take me to DE at around 6.00 min.
You don't say if and how you control your fan speed, but that "very low fan speed" is probably where I think the use of a Variac is justified. (Well, anything which you can control good enough to give you just a "whiff of air".)
55F isn't that bad for environment temp, just see to giving the roaster a proper preheat. The most important thing is to avoid any wind.