Homeroasters - Home Roasting Coffee Community » ALL ABOUT ROASTERS » Other Roasters
Who is here? 1 guest(s)
Page 7 of 16: 1...45678910...16
 Print Thread
Skywalker roaster mods
Dan N
Top  #151 Print Post
Posted on 01/15/2024 6:54 AM

Quote

renatoa wrote:
This is absolutely weird... mine is 5K !!! Shock
Connected with a 30k upper resistor in a divisor, and powered from 5V, this give about 0.7V voltage at ambient on NTC pins.

Mine measured 303k with 3 different meters and all agreed. Did you disconnect it before measuring?

Also the 2k resistor could be switched in parallel with the 30k giving 1875 ohm equivalent. Josh's data seems to show they do that to get the 2 readings.
 
Dan N
Top  #152 Print Post
Posted on 01/15/2024 2:04 PM

Quote

Dan N wrote:

I've done yet more analysis on Josh's thermistor data in post #3 and have learned a great deal about the inner workings. I'm now able to calculate the coefficients for the Steinhart–Hart equation to match the data: a=7.119210e-04, b=1.984603e-04, c=1.054138e-07

I added another point for 200C to my calculations (using 17 points now) and have a new set of coefficients that track better at the high end of the temp range that is most critical: a=7.021927e-04, b=1.998391e-04, c=1.015296e-07

Using the equation shows details that are hidden in the discrete data. For example at the highest temperatures when the ADC reading decreases by 1 the temperature goes up by 0.5F. The relative accuracy should be consistent between roasters but not the absolute accuracy. Who knows if they used thermistors with 0.5% or 10% tolerance.

At some point I'll play with this on the Arduino and replace the polynomial formula with a table based on Steinhart-Hart.
 
Mike_Mathis
Top  #153 Print Post
Posted on 01/16/2024 4:47 PM
Meanwhile, us guys riding in the bed of the truck............I'm getting a fatal error on line 5 of SkyCommand as I try to compile.

Compilation error:TimerInterrupt.h No such file or directory

Any ideas where I messed up?
*Kaffelogic Nano 7
*Skywalker V1
*2023 BC-2
 
HarryDog
Top  #154 Print Post
Posted on 01/16/2024 4:54 PM
Load the latest TimeInterrupt 1.8 from arduino
https://www.ardui...interrupt/

So at the top sketch, then include library, then add library zip.
Load library zip.
 
Robotic Kitten
Top  #155 Print Post
Posted on 01/16/2024 8:11 PM

Quote

Mike_Mathis wrote:
I'm getting a fatal error on line 5 of SkyCommand as I try to compile.

Compilation error:TimerInterrupt.h No such file or directory

Any ideas where I messed up?


You didn't mess up. Just remove that line, I think it is a leftover as Josh is not using timers any longer for sending data to roaster.
 
Mike_Mathis
Top  #156 Print Post
Posted on 01/16/2024 8:16 PM

Quote

HarryDog wrote:

Load the latest TimeInterrupt 1.8 from arduino
https://www.ardui...interrupt/

So at the top sketch, then include library, then add library zip.
Load library zip.


Bingo!

Thanks, HarryDog

Everything is working now.
*Kaffelogic Nano 7
*Skywalker V1
*2023 BC-2
 
HarryDog
Top  #157 Print Post
Posted on 01/16/2024 10:55 PM
I don't know enough about the code myself to say if it's not required, so I loaded the library so it compiled.
Any other code clean ups we should be doing?
 
Mike_Mathis
Top  #158 Print Post
Posted on 01/17/2024 10:30 AM
I am somewhat confused pertaining to the temp reads on the roaster. I wanted to install an independent TC to check the accuracy of the thermistor, so I did. The TC is apples to apples in diameter and length. Location is just a little further down and over toward the center from TM. Here is the rub. The roaster works impressively with the controller/TM. If I use the controller, I’m thinking I need to definitely use the TM readings. If you electrical engineer/coder guys figure out a way to make a more responsive TC work with Artisan, I think the roasting dynamics will change quite a bit. For the better? is the burning question. (pun intended). Does this even make sense?
*Kaffelogic Nano 7
*Skywalker V1
*2023 BC-2
 
HarryDog
Top  #159 Print Post
Posted on 01/17/2024 11:15 AM
If a NTC thermistor's ability to achieve the highest accuracy is within the -50°C to 150°C range, and up to 250°C for glass encapsulated thermistors.

I did not see any markings on mine.
Was going to test a RTD in mine to compare but delayed due to weather conditions for 4 days now. I do suspect they will track differently but will see if it ever gets delivered.

We might be wasting our time if it is a glass encapsulated thermistors.
 
renatoa
Top  #160 Print Post
Posted on 01/17/2024 1:06 PM
If you already have a TC fitted in the roaster, all you need is a MAX adapter to the Arduino board, and the code to read temp from max instead skywalker... and draw a conclusion.

No kidding, is an easy task to adapt code as I wrote above.
Edited by renatoa on 01/17/2024 1:12 PM
 
Mike_Mathis
Top  #161 Print Post
Posted on 01/17/2024 2:59 PM

Quote

You didn't mess up. Just remove that line,


Lol, therein lies the problem. I don't know how to remove it. I find this stuff fascinating and frustrating at the same time. The only way I can do this stuff is to bug the electrical, engineer, coding, programming gurus.

.......and I imagine I'm about to wear thin on you guys.

I have the utmost respect for what you guys can do and my coffee roasting is the benefactor of your knowledge. I am constantly running to YouTube because of the stuff you are all posting.
Edited by Mike_Mathis on 01/22/2024 8:11 AM
*Kaffelogic Nano 7
*Skywalker V1
*2023 BC-2
 
Mike_Mathis
Top  #162 Print Post
Posted on 01/18/2024 10:39 PM
As done in another post in this "mod" thread, I installed a TC in the roaster. I won't go into details as that is evident in the other post.

Today I performed two roasts on the SW. First was in "Auto" mode and Second was in "Assist" mode. I used the controller for both roasts but had the TC I installed and connected to Artisan as a no control, recorder. (I can post the jpegs if anyone wants to see them). This was to do a temp comparison between the TM and the TC. We all know the temp readouts on the controller are lower than what a TC shows. My attempt here is to shed light on how much differential throughout a roast. Here are the unscientific results.

Roast 1 Columbian Aponte Narino (Honey processed)
Time Art SW Diff
.30 128 117 -11
1:00 120 114 -6
1:30 126 119 -7
2:00 133 127 -6
2:30 141 136 -5
3:00 150 142 -8
3:30 157 150 -7
4:00 165 158 -7 DE@ 169.2 4:16
4:30 172 164 -8
5:00 179 170 -9
5:30 186 178 -8
6:00 193 ??? ?
6:30 199 190 -9 FC@ 197.9 6:22
7:00 204 ??? ?
7:26 207.7 ??? ? Drop

Roast 2 Guatemala La Flor Del Cafe (washed)

.30 125 124 -1
1:00 125 118 -7
1:30 134 126 -8
2:00 143 130 -13
2:30 152 138 -14
3:00 159 146 -13
3:30 168 154 -14 DE@ 171.2 3:38
4:00 176 162 -14
4:30 184 169 -15
5:00 191 176 -15
5:30 196 182 -14 FC@ 198.2 5:42
6:00 200 188 -12
6:30 204 192 -12
7:00 207 195 -12
7:30 211 199 -12
7:38 212 ??? Drop

Maybe this info doesn't mean anything. Higher diff in 2nd roast probably because washed -vs- honey process?. The ? is where I missed the window for recording temp. My roasts are going too fast with this roaster. Voltage is 124-125V at plug. I believe the high voltage is causing the beans to roast too fast. Would 400g charge help?
Edited by Mike_Mathis on 01/19/2024 8:48 AM
*Kaffelogic Nano 7
*Skywalker V1
*2023 BC-2
 
renatoa
Top  #163 Print Post
Posted on 01/19/2024 9:22 AM
A motorcycle P45T 45W 6V bulb mounted series on the mains chord would reduce voltage by 6V, (or 5V when roaster is at 70%), thus bringing the voltage back to 115-120V standard range.
 
Mike_Mathis
Top  #164 Print Post
Posted on 01/19/2024 12:47 PM

Quote

renatoa wrote:

A motorcycle P45T 45W 6V bulb mounted series on the mains chord would reduce voltage by 6V, (or 5V when roaster is at 70%), thus bringing the voltage back to 115-120V standard range.


Thanks for the suggestion. The first course of action I will implement is reducing the heat a little. My DE times were not too far off. Looking at the profile I can see where the controller did not cut the heat back at a point when it would have extended the Maillard phase and lowered the RoR from the teens. I'm simply going to have to take control of this roaster. When you stated 1000W was plenty for this roaster, you were without a doubt on the money. I wonder if increasing to 400g would benefit my situation.

On a different note:

Has anyone printed the funnel? What material did you use? How is it doing? I'm having one printed in PLA. I know the heat-resistant properties are not very high for this material, but the exposure time should be minimal. We'll see. The local printer said any other material would be a special order for the material and would be more than the US $14 I'm paying for the PLA one.
*Kaffelogic Nano 7
*Skywalker V1
*2023 BC-2
 
HarryDog
Top  #165 Print Post
Posted on 01/20/2024 1:54 PM
Hello all I added the Renatoa code to allow Temperature change between C & F, but when Artisan is in F mode it reads the probe but not in C mode?

If just switching in artisan should work I definitely did something wrong.
I added the code to the lines in the notes, wrote it to the Uno?
What to check?
 
Dan N
Top  #166 Print Post
Posted on 01/20/2024 5:11 PM

Quote

HarryDog wrote:

Hello all I added the Renatoa code to allow Temperature change between C & F, but when Artisan is in F mode it reads the probe but not in C mode?

If just switching in artisan should work I definitely did something wrong.
I added the code to the lines in the notes, wrote it to the Uno?
What to check?

This shows which lines to change (Robotic Kitten's code change):
https://github.co...620a57a44d

Does your change match that?
 
Dan N
Top  #167 Print Post
Posted on 01/20/2024 5:42 PM
Some thoughts on thermistor accuracy and whether a thermocouple mod is worth the effort:

This thermistor is a close match to what we have, and its datasheet is the most detailed I've ever seen for any thermistor. It shows resistance range and temperature tolerance for every 1C increment from -30C to 300C:
https://www.analo...K_100.html

The temperature tolerance at 200C is +/-3C while a K-type thermocouple tolerance is +/-1.5C. That's not a big enough difference for me to consider a change. All roasters will read a little low or high at any given temperature. Once you've done enough roasts to know the number where FC happens and the number for end/dump that gives the best flavor coffee, it doesn't matter what the absolute number is.

HB is full of threads where people struggle with this same issue, try a lot of changes, and usually just go back to using the original temp probe.
 
HarryDog
Top  #168 Print Post
Posted on 01/20/2024 5:49 PM
If I set artisan to C temperature mode, the sensor reads in F for a few seconds then goes to zero when I think the Uno looks to read the Probe in C? Looks to work if Artisan is set to F temperature mode.

In my edited version I had the comments and no extra spaces, I added those spaces and removed comments but same action when running the code. It compiles and uploads but does not work when set to C in artisan. Could I have a setting wrong in Artisan?
 
Robotic Kitten
Top  #169 Print Post
Posted on 01/20/2024 6:31 PM

Quote

HarryDog wrote:

If I set artisan to C temperature mode, the sensor reads in F for a few seconds then goes to zero when I think the Uno looks to read the Probe in C? Looks to work if Artisan is set to F temperature mode.


UNO always reads in F and then translates to C if Artisan tells it's using C. You could try this sketch https://github.co...ommand.ino


BTW, I've finally disassembled my SkyWalker and there were a lot of screws to remove. But I've disconnected the thermistor and measured it at different temperatures:
  1. Ambient room temp 19.7C, Thermo couple measured 71 and thermistor 313.1-313.5k
  2. Ice water, Thermo couple measured 37 (I don't believe it, as my IR thermometer measured 33F and thermistor 731k-732k
  3. Heating water in a kettle, Thermo couple reported 106F and thermistor 128k
  4. Heating water in a kettle, Thermo couple reported 135F and thermistor 66k
  5. Heating water in a kettle, Thermo couple reported 151F and thermistor 46.7k
  6. Heating water in a kettle, Thermo couple reported 181F and thermistor 25k
  7. boiling, Thermo couple reported 215F and thermistor 13.5k
  8. boiling water ant Thermo couple settled at 215.5F-215.6F and thermistor 13.44k


Also, looks like the board has the logic for zero cross detection, so the question is whether just to remove the MCU and replace it with UNO/blackpill or if it makes sense to build a TC4 based board with the same form factor for a drop-in replacement.
Edited by renatoa on 01/21/2024 3:34 AM
 
Beanz
Top  #170 Print Post
Posted on 01/20/2024 7:00 PM
I suspect the thermistor has been selected as the manufacturer uses these in high volume in other electrical appliances such as water boilers they manufacture. They are not seeking the response times in a water boiler that we see as essential in a roaster but the 1% accuracy at 100C (from Dan N’s data) is ideal for a water boiler.
A thermocouple in volume will cost more than the thermistor. They have designed a low volume coffee roaster and used an off the shelf high volume temperature measurement device already used as an approved part in house. If they had selected a thermocouple it would be at a higher cost and another item in their inventory and supply chain.
It would be interesting to look at Mike_Mathias’s results in post 162 and hold both devices at fixed temperatures long enough for both devices to soak and fully stabilise and then measure the difference. I am wondering if there is a lag in the thermistor getting to temperature and stabilising, it is 4mm in diameter and if you take into account a layer of stainless over the glass which is encapsulating the sensor and the fact that there may not be good thermal contact between the glass and stainless sheath as it is probably filled with epoxy. This may not account for the total difference but the gap may narrow.
If someone had the ability to immerse the thermistor and the thermocouple in boiling water for 5 minutes it would be an interesting experiment. If the temps could be tracked coming up to temp and then cooling we could see if the response times and accuracy vary.
The device is probably Chinese source such as Fenghua
https://www.china...(1)(1).pdf
 
HarryDog
Top  #171 Print Post
Posted on 01/20/2024 8:09 PM
Looks like a simple calculation inversion was the issue.
This if (CorF == 'C') v = 5 / 9 * (v - 32);
to this if ( CorF == 'C' ) v = (v - 32) * 5 / 9;

This line change did not seem to make a difference.
char CorF = 'F';

Does Arduino do something odd with the variable v=5?

Maybe brackets around the 5/9 would work, did not test that?
Edited by HarryDog on 01/20/2024 8:23 PM
 
renatoa
Top  #172 Print Post
Posted on 01/21/2024 2:46 AM

Quote

Robotic Kitten wrote:

Also, looks like the board has the logic for zero cross detection, so the question is whether just to remove the MCU and replace it with UNO/blackpill or if it makes sense to build a TC4 based board with the same form factor for a drop-in replacement.


Wondering what could be the purpose of ZCD, when the lamp is controlled using slow PWM with one second cycle, it's very visible.

I know this is the SKW mods thread, but I think this "enthusiasm" to propose from start the factory electronics demolishing would not help to make this roaster so popular as it deserves...
There are some words in the starting post of the spare parts thread that deserve a little attention, and consideration... ;)

Quote

...they were NOT set up to sell replacement parts.
The rep stated they didn't want people to think the roaster was unreliable.


If we start with the assumption that we paid $400 for the mechanics only, then I would return without regret to the builders subforum, having in my mind "I'll make it better for this amount" Grin
Edited by renatoa on 01/21/2024 6:24 AM
 
renatoa
Top  #173 Print Post
Posted on 01/21/2024 2:51 AM

Quote

Beanz wrote:

A thermocouple in volume will cost more than the thermistor.
...
The device is probably Chinese source such as Fenghua
https://www.china...(1)(1).pdf


Manufacturing the thermocouple itself is close to nothing, the electronics for reading are the main price factor.

...

The datasheet from that link list models of NTC with a range up to 105C degrees only.
 
renatoa
Top  #174 Print Post
Posted on 01/21/2024 3:12 AM

Quote

HarryDog wrote:

Looks like a simple calculation inversion was the issue.
This if (CorF == 'C') v = 5 / 9 * (v - 32);
to this if ( CorF == 'C' ) v = (v - 32) * 5 / 9;

This line change did not seem to make a difference.
char CorF = 'F';

Does Arduino do something odd with the variable v=5?

Maybe brackets around the 5/9 would work, did not test that?


Oups... my bad... should foresee this...
When evaluating an expression the compiler start considering it is of integer type... even if it contains float variables somewhere in the middle of expression...
As a result, from the start 5/9 is computed as... ZERO ! and from there the rest is faulty.
Moving (v - 32) in front forced the expression type evaluator to float, as v is float variable. Thus, 5/9 is computed right, having a float result.

Another way to write this the right way would be ...

v = 5.0f / 9 * (v - 32);

... i.e. write 5 as a forced float number.

Sorry made you bang your heads... :(
 
Robotic Kitten
Top  #175 Print Post
Posted on 01/21/2024 5:23 PM

Quote

renatoa wrote:

Wondering what could be the purpose of ZCD, when the lamp is controlled using slow PWM with one second cycle, it's very visible.


Maybe they are trying to play nice by not turning on the 1kW load during mid cycle?
BTW, the heater is controlled by JT816HA which is an SCR??? Does it mean that the heating element is sized to work only at half of the AC wave?
 
Page 7 of 16: 1...45678910...16
Jump to Forum: