On Idle Noise

Discuss roast levels and profiles for espresso, equipment for roasting coffee.
User avatar
MaKoMo

Postby MaKoMo » Mar 15, 2019, 8:12 am

I received many complaints on noisy RoR signals over the year. The ongoing work of Scott Rao and a broken TC made me reconsider my own roast logging setup. As I had collected many different meters and probes over the years, mostly for development purpose, I wonder which one I should best install in my machine.

My goal was to minimize both, the signal to noise ratio as well as the introduced signal delay, especially for the RoR. Thus I came up with a setup that would allow me to compare different setups on the amount of idle noise they produce.

I switched from TCs to RTDs for reasons detailed in the linked post and I am quite happy with my new setup,
Marko
★ Helpful

User avatar
danetrainer

Postby danetrainer » Mar 15, 2019, 11:48 am

Hi Marko, thank you for this helpful post and link to the tests you performed. I am interested in trying an RTD on my Roaster and my application benefits from bending the probe for a better position in the coffee beans. Omega has a new "bendable" RTD, no bends within 2" of the tip, otherwise 2x radius of the probe thickness. It's available in 1/8" size by custom order, as all the in-stock size are 1/4". Here is a link Omega PR-26

Also, looking at the 4 wire configuration out from the RTD, I could use two independent to the machines PID and the other two to my meter/laptop?

User avatar
MaKoMo

Postby MaKoMo » replying to danetrainer » Mar 15, 2019, 12:33 pm

No, the most-common 4 -wires configuration increase accuracy over a 3-wires and 2-wires configuration for RTDs. All wires are connected only to one sensor element. There are also dual RTD probes with then 8 wires. Note that an RTD is completely different to a TC and requires a different meter. Your meter has to support the corresponding RTD configuration.

Btw. I bent my non-bendable RTD, but only 10cm away from the tip. That worked as it does with TC if done carefully to not ruin the shield or the insulation.

Note also that my new configuration is not the philosopher's stone. Just from all the options I had, this was the one that made most sense for me w.r.t. my goals and my experimental results.

Madman13

Postby Madman13 » Mar 16, 2019, 8:10 am

MaKoMo wrote:My goal was to minimize both, the signal to noise ratio as well as the introduced signal delay, especially for the RoR. Thus I came up with a setup that would allow me to compare different setups on the amount of idle noise they produce.


Wow, what a great blog post! Thank you for your detailed experiments and write-up.. Makes me wonder why we have seemed to write off RTD probes, seems like they may be far superior to thermocouples for the purpose of BT / delta BT since the delta BT is so sensitive to noise. I will add a RTD probe for BT measurements immediately to test. Do you have an opinion on 100ohm vs 1000ohm probes and class a vs class b for this purpose?

User avatar
JavaMD

Postby JavaMD » Mar 16, 2019, 9:23 am

Thanks Marko for the post.
Interesting ... and a non scientific confirmation ... my Santoker 500 has RTD probes and I get a fair reading with it on the ROR with 1 sec sampling and Delta span of 10 sec. With my Mill City ... North Coffee 1 KG I have to increase my Delta Span to 15 seconds and it is no where near as clean as the Santoker.

This is a bit of Apples / Oranges with 2 different machines but goes with your findings.

Steve Egge

User avatar
MaKoMo

Postby MaKoMo » Mar 16, 2019, 9:38 am

Madman13 wrote:Wow, what a great blog post! Thank you for your detailed experiments and write-up.. Makes me wonder why we have seemed to write off RTD probes, seems like they may be far superior to thermocouples for the purpose of BT / delta BT since the delta BT is so sensitive to noise. I will add a RTD probe for BT measurements immediately to test. Do you have an opinion on 100ohm vs 1000ohm probes and class a vs class b for this purpose?


I do not think that PT100 vs PT1000 will make much of a difference. Also the class might not be too important. Mine is a 1/3 DIN which corresponds to AA, but class A and B should work well too. The class label for RTDs is about accuracy, which is not too important for our application and even the accuracy of a class B variant should be high enough. In any case RTDs do not drift as (especially cheaper) TCs do.

Note that I am not saying that RTDs are per se better for measuring the BT than TCs. Just that in according to my experimental results they delivered better results (less noise) and are fast enough. There so many different TCs (and they can be pretty cheap!) and also different RTDs (usually more expensive and with less choices) that I do not like to generalize my results too much. However, RTDs can be an option in some cases. In my case even a better option.

I am looking forward to learn from your tests.
★ Helpful

User avatar
MaKoMo

Postby MaKoMo » Mar 16, 2019, 9:44 am

JavaMD wrote:Thanks Marko for the post.
Interesting ... and a non scientific confirmation ... my Santoker 500 has RTD probes and I get a fair reading with it on the ROR with 1 sec sampling and Delta span of 10 sec. With my Mill City ... North Coffee 1 KG I have to increase my Delta Span to 15 seconds and it is no where near as clean as the Santoker.

This is a bit of Apples / Oranges with 2 different machines but goes with your findings.

Steve Egge


Thanks Steve for confirmation. Your 10sec delta-span produces at least a 5sec RoR delay @1sec. You seem to have still some noise on your BT. Often the data forwarded by the machine has just a too low resolution of the forwarded signal (I am only aware of one roasting machine, the newer Probats, that delivers 2 decimals, most others forward only 1 or none) or noise induced by some other means. Also a low quality meter might have this effect. Note that machines featuring RTDs have them often connected to cheap PIDs just because those feature a data connection and can measure RTDs. However, many of those PIDs do considerable processing generating on their temperature input and thus a corresponding large delay on the signal (mostly because their p-i-d loop runs better with smoother signals, which is the actual purpose of those devices which are here abused as temperature meters). This delay is already in the BT signal and adds to the delay produced by computing and filtering the RoR.

Moxiechef

Postby Moxiechef » Mar 19, 2019, 8:28 pm

Just tested my Phidget 1048 and Omega 3mm TC for noise....I'd say it's a bit noisy.

At about 1:30, you see the airflow kick up. That's when I turned the machine on.

I'm definitely going to try out your new setup sometime soon.
Image

User avatar
Almico

Postby Almico » Mar 20, 2019, 9:07 am

When checking idle noise it's helpful to have a bit of temperature deviation on the probe. If the probe environment is dead calm you might get a reading with far less noise than would ordinarily be present if the probe was "moving".

I tried mine the other day and fortunately do not have any EMI from any motors or fans. Right now I'm running a P-1048. with 3mm TCs. I do get a bunch of noise with everything, including the overhead lights off, all smoothing at 0 and delta span at 1. It is quieted down by DS=10, but I'd rather have a quicker response. I'm game to try an RTD.

So let me get this straight. I would need this as well a.125" or 3mm 4-wire RTD? What am I losing with a 3-wire RTD?

Image

to run the BT probe. Can I hook my ET TC directly into the VINT Hub?

User avatar
MaKoMo

Postby MaKoMo » Mar 20, 2019, 9:27 am

That setup is needed to read one RTD. You do not loose anything by using a 3-wire RTD as the TMP1200 is not really supporting the 4-wire ones to their full potential. Note that you need to tell Artisan in the Phidgets tab which probe type you connected (as is with the TCs).

You cannot put TCs directly into the VINT hub. You need to add a VINT TC meter like the 4-channel TMP1101 or the 1-channel isolated TMP1100. The VINT HUB0000 can be connected to up 6 VINT devices.

Regarding the Artisan setup with the TMP1200. As this is only a one channel device which you might configure as main device in Artisan using its meter popup, you would need to add the TMP1101 channels (0+1 and 1+2) as extra device channels. If you plan to use on of these as meters, you need to link it via a symbolic formula into the main devices second channel.

Here is my current setup: the TMP1200 as main device feeding the BT, the TMP1101 channel 0+1 as first extra device with a sensor measuring drum temperature and another one measuring air temp from the cooling bin. Some other extra devices. 5th extra device reading ET from TMP1101 channel 2. That ET signal is then mapped via the symbolic formula "Y11" to the ET channel of the main device. Note that those Y variables are assigned in order. First main channel is bound to Y1 (BT) second to Y2 (ET), first extra devices first channel to Y3 (here DRUM),....

Image

Image

Image