Who is here? 1 guest(s)
 Print Thread
PID'ed Gas-fired fluid bed roaster
tamarian
And here's the video. I tried to get a closer look at the flame sensor. Also a demo of the sample trier, but since I was holding my phone/camera, I couldn't raise the jar to catch all the beans. At the end, you'll see my dump mechanism. May need some adjustment by raising the cooling tray, since my roast chamber is a bit higher than before. Enjoy:


Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
JETROASTER
I am continually amazed at what shows up on HRO!! Nicely done.
Is the trier spring-loaded or gravity?
As you move forward, you'll find that almost any obstruction of your exhaust system during the dump cycle will expedite the process.
Very nice! -Scott
 
tamarian

Quote

freshbeans wrote:

I am continually amazed at what shows up on HRO!! Nicely done.
Is the trier spring-loaded or gravity?
As you move forward, you'll find that almost any obstruction of your exhaust system during the dump cycle will expedite the process.
Very nice! -Scott


Thanks Scott! The trier just uses gravity. Was planning on making it spring loaded, but it worked quite well without.

You'll notice at the end of the dump the blower was reduced. I also use a small plate that I slide into RC to catch the downward heading beans and redirect them out, but was hard to use while taking video.
Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
allenb
This is very impressive! BTW, you've got the first affordable, PID controlled gas fired fluidbed that I know of.

When you've got things where you want them and you're real comfortable with the new setup be sure and shoot another video.

Allen
1/2 lb and 1 lb drum, Siemens Sirocco fluidbed, presspot, chemex, cajun biggin brewer from the backwoods of Louisiana
 
allenb
I forgot to mention that in the video it looks like ET is now hunting a bit. Did you make the changes in max control output and changes to gas pressure you mentioned earlier?

Allen
1/2 lb and 1 lb drum, Siemens Sirocco fluidbed, presspot, chemex, cajun biggin brewer from the backwoods of Louisiana
 
tamarian

Quote

allenb wrote:

This is very impressive! BTW, you've got the first affordable, PID controlled gas fired fluidbed that I know of.

When you've got things where you want them and you're real comfortable with the new setup be


And you've got the first affordable PID controlled gas fired drum that I know of smile
Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
tamarian

Quote

allenb wrote:

I forgot to mention that in the video it looks like ET is now hunting a bit. Did you make the changes in max control output and changes to gas pressure you mentioned earlier?

Allen


Yes, but seems like I need to reduce the pressure further to 1 bar from 1.5. In addition, I need to twist the collar a bit to better position the ET TC threaded into that collar. Once Artisan monitor was shut off, it seem to have smoothed the ET curve a bit, this is with default setting. But BT curve was smooth throughout the roast. Doesn't bother me, as it is not that far from what it did with electric element.
Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
tamarian
Couldn't resist another roast after turning regulator down to 1 bar. This is with 750 grams El Salvador Burbon. Saved profile as a screenshot so no smoothing/filtering takes place. I left it for 4 minutes on preheating at 125 to see how it behaves without beans. Once the beans were added, Fuji was on 25% power or less.

forum.homeroasters.org/forum/attachments/screen.png

Next would be with 0.75 bar, and with decent beans that I may cup. So far all stale beans, which will be used to season a grinder coming my way soon.
tamarian attached the following image:
screen.png

Edited by tamarian on 09/20/2013 2:35 PM
Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
tamarian
Today was quite productive. My planned tuning and pressure adjustment was limited by my 16 bar gauges, which offered very little feedback. So I located 2 bar gauges and combined with the multimeter re-did the whole driver tune up. My assumption that pressure changes needs retuning the driver were wrong. The drive tune up is only current based. A multimeter is essential, as the values found for min and max are way off from the documentation. Calibrating while running the roaster is necessary as well, while looking at the flame and pressure gauges and multimeter at the same time. It was quite tedious since some of the clockwise and counter clockwise instructions were mixed up, so the multimeter and pressure gauges came in handy.

Test roast following the adjustments was flawless. I have not seen this level of control and precision with my PID while on electric element. PV and SV where less than 1 degree apart during the entire roast (except the charge/recovery stage).

First, here it is idling after pre-heating:
forum.homeroasters.org/forum/attachments/idle.png

And the test roast screenshot:
forum.homeroasters.org/forum/attachments/scrgas.png

Same via artisan:
forum.homeroasters.org/forum/attachments/as-tanzaniasample21sep2013small.png

Note that the PID was set for 160C at 4 minutes and 199 at 8 minutes.
tamarian attached the following images:
idle.png scrgas.png as-tanzaniasample21sep2013small.png

Edited by tamarian on 09/21/2013 2:42 PM
Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
tamarian
Next up was the flame sensor. Since the torch and igniter were moved way down from initial build, the flame sensor had a hard time detecting the flame. I've adjusted the value in the Arduino sketch to be more sensitive, but still wasn't reliable. I've remove the glass cup, and replaced it with a shot glass where the bottom of the glass goes inside the pipe by an inch. Then placed a small mirror to reflect the flame and igniter on the flame sensor. This worked like a charm. Finally wrapped the sensor with some cut up baking silicon to protect it from the heat. My plan B, if this did not work, was a periscope. I'm somewhat disappointed that it worked, since I prefer a periscope ;-)
Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
allenb

Quote

Tamarian wrote:

Next up was the flame sensor. Since the torch and igniter were moved way down from initial build, the flame sensor had a hard time detecting the flame. I've adjusted the value in the Arduino sketch to be more sensitive, but still wasn't reliable. I've remove the glass cup, and replaced it with a shot glass where the bottom of the glass goes inside the pipe by an inch. Then placed a small mirror to reflect the flame and igniter on the flame sensor. This worked like a charm. Finally wrapped the sensor with some cut up baking silicon to protect it from the heat. My plan B, if this did not work, was a periscope. I'm somewhat disappointed that it worked, since I prefer a periscope ;-)


I am officially giving you our "most creative and ingenious roaster builder" award! ThumbsUp

I've never seen such ingenuity in overcoming design obstacles as your shot glass/mirror fix!

That's great news on the increased control accuracy. Those profile curves are rock solid. Its also great news that even with the proportional valve's hysteresis issue that perfect control stability is possible using a 0-10 volt control output. My fear was that there would be large swings in temperature as a result of over correcting due to the hysteresis.

Keep the updates coming!

Oh, by the way, out of curiosity, when you've got things totally set and have a few roasts done, let me know if you notice any difference in the cup since converting to gas heat

Allen
1/2 lb and 1 lb drum, Siemens Sirocco fluidbed, presspot, chemex, cajun biggin brewer from the backwoods of Louisiana
 
tamarian
Thanks Allen, I'll wear it with pride :-)

Cupping reports won't take long as I'm ready to roast some of the good beans. Fortunately, most of the beans are ones that are logged previously, used by others who are now waiting for their fix, and will let me know what they think.

Here's the flame sensor alarm/relay sketch. It's very basic, but maybe of use to those new to Arduino:



void setup()
{
  pinMode(12, OUTPUT);      // Control signal, used for alarm and relay
  pinMode(8, INPUT);        // Flame sensor digital reading, 0 or 1
  Serial.begin(9600);
}

void loop()
{
  float AnalogV, DigitalV;
  AnalogV = analogRead(A0);    // Analogue output of flame sesnsor
  DigitalV = digitalRead(8);   
  Serial.print("AnalogV: ");          // useful for debug and tuning
  Serial.print(AnalogV);
  Serial.print(" - DigitalV: ");
  Serial.println(DigitalV);

  if (AnalogV > 499)            // The higher the value, the darker it is
  {
    digitalWrite(12, HIGH);   // Turn on alarm, relay and ignitor
    delay(3000);              // Wait for 3 seconds   
    digitalWrite(12, LOW);    // Turn off alarm, relay and ignitor
  }
}


The main igniter is on full time, from startup of the roaster, until it is cooled down and turned off. The Arduino relay runs a backup igniter.

If the main igniter burns out, and for whatever reason the valve is open and there is no flame, the alarm sensor will detect that in less than a second. It will turn on a buzzer alarm, switch on backup igniter. It will repeat this loop every 3 seconds, until a flame is detected, or roaster turned off.

The igniter takes about 2 seconds to get hot enough, so that is a limitation of surface igniters. May have to replace with a spark type igniter. Would love to put a spark plug in there, but they seem complicated.

Having carefully calibrated min current to keep the valve barely open, and setting PID's low signal (2-5%) to keep the flame on when in ramp/soak mode insures the flame won't go off during a roast, even if the igniter fails mid-roast.

This will very likely reduce the possibility of the alarm going on during a roast. But if it does, my safety steps would be to first check the glass window to see if flame is really off. If so, slow the blower down to help backup igniter work. Close gas supply. There should be enough gas in the hoses to combust for a second or two. If I hear that, or see it in the glass window, I'll re-open the gas supply.
Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
tamarian
Wanted to test the limits by roasting a batch that is small for the roaster, 200 grams, and roast it fast, to see how the PID Fuzzy control copes with it. No problems.

forum.homeroasters.org/forum/attachments/sulawesi22sep2013small.png

One change I should do is lower my charge temperature, as the recovery is very fast and results in overshooting in the early drying stage.
tamarian attached the following image:
sulawesi22sep2013small.png

Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
allenb
Very nice! Down the road I would definitely play around with the continuously pulsing spark ignitors and see if it's a reliable means of ignition.

For whatever reason, when roasting in the 1 bag Sivetz, we always found without exception, a superior cup when starting with the roaster at room temperature versus pre-heated. Have you ever cupped the same coffee from a pre-heated roaster versus non-preheated? If not, you might give it a try. Obviously, if you're in production mode, cooling it down to room temp is probably not an option. If you're thinking this must have been in our head and imagined, you're not alone but it was always the case.

Allen
1/2 lb and 1 lb drum, Siemens Sirocco fluidbed, presspot, chemex, cajun biggin brewer from the backwoods of Louisiana
 
tamarian
Well worth a try, Allen, and easy to test.

Done two roasts today, with beans I'm familiar with (Sulawesi Toraja and Yirg. Konga), where I had their profiles "nailed" in the past and have been repeating the same profile for each. I have duplicated these profiles today with gas, to the same exact degree and same exact profile.

Having not changed anything in the thermometry, the aromas seemed familiar, but the colour is much lighter on gas than it was on electric element, on the exact same dump temperature.
Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
jkoll42
This is a very awesome build - looks amazing! Very excited to see the use of a PID for a gas roaster.

One question though - are you really roasting in your basement with gas and without any exterior venting?
-Jon
Honey badger 1k, Bunn LPG-2E, Technivorm, Cimbali Max Hybrid, ECM Synchronika w/ Flow Control
 
tamarian

Quote

jkoll42 wrote:

This is a very awesome build - looks amazing! Very excited to see the use of a PID for a gas roaster.

One question though - are you really roasting in your basement with gas and without any exterior venting?


Not without exterior venting. The house uses a 1/2HP evaporative/swamp cooler with about 3,000 CFM. Just opening the door above those shallow stairs will push the smoke out. I'm starting to like roasting there, which was only temporary while building it, so I may add an exhaust path, since we don't run the cooler in the winter. Should be simple enough as there's already a tiny exhaust fan just above the roaster. Probably just needs an upgrade to a beefier unit.
Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
tamarian
I just received an Arduino Nano board to replace Arduino Uno. This is to reduce the clutter, and spare the Uno for more demanding projects. Here's what it looks like, mounted on a mini breadboard, with a loud buzzer and bright red LED:

forum.homeroasters.org/forum/attachments/img-20130924-wa0007.jpg

Now the wires coming out go to the flame sensor, instead of another box for the breadboard:

forum.homeroasters.org/forum/attachments/img-20130924-wa0006.jpg

You can see a pic of the old setup with two boxes here: http://forum.home...post_45500

I have removed the igniter relay, until I figure out a more instant igniter. For now, running on two igniters full time during the roast.
tamarian attached the following images:
img-20130924-wa0006.jpg img-20130924-wa0007.jpg

Edited by tamarian on 09/24/2013 12:25 PM
Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
tamarian
Here's the viewing glass/shot glass with the mirrors and flame sensor in action:

forum.homeroasters.org/forum/attachments/img-20130925-wa0000.jpg
tamarian attached the following image:
img-20130925-wa0000.jpg

Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
tamarian
Did two more test roasts. First one was very dark to record/calibrate second crack start and end. I rarely go into 2nd, so thought I might find out when it occurs in my setup, and see how far I can go before 2nd crack:

forum.homeroasters.org/forum/attachments/guatemala2ndcrack25sep2013small.png
forum.homeroasters.org/forum/attachments/img-20130926-wa0000.jpg

Next was testing extremely slow and ultra light roasts, steady RoR, which is how I roast for Arabian/blond coffee. I forgot to turn on Artisan for a minute or two, so the roast was 17 minutes total to 170C, and the first minute was rise to 120C:

forum.homeroasters.org/forum/attachments/kona25sep2013small.png

forum.homeroasters.org/forum/attachments/img-20130926-wa0001.jpg
tamarian attached the following images:
img-20130926-wa0001.jpg img-20130926-wa0000.jpg kona25sep2013small.png guatemala2ndcrack25sep2013small.png

Edited by tamarian on 09/25/2013 10:56 PM
Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
tamarian
By now, the roaster have gone through 12 roasts in total, 8 with sacrificial beans that will be used for grinder seasoning, and 4 roasts for actual use. Espresso roasts still need more rest, but the brewing roast (Yirg. Konga) turned out great. Still debating if it is better than when roasted with electric element, but certainly not worst. It is very clear to me that the colour looks lighter, at the same level of roast. The aroma is intense and sweet, but Konga is always like that.

I need to start weighing my gas tank, to see how many more roasts it can do before a refill, to avoid running out of gas mid -roast.

I think I'm done with this roaster for now. I do have some mods on my mind, but they are not coffee related, just electronics. I'd like make the whole thing Arduino based, without the Fuji PXG4, and without the valve driver. But I have a long learning curve before getting into that, and if I ever get to it, it will post it in another thread.

If there are any details in this build that are not clear, feel free to ask.
Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
tamarian
For those interested in in the proportional valve's performance, I forgot to point out something that was confusing during the valve's tuning process. There is some sort of "lag", but only takes place when you change directions, from opening to closing or from closing to opening. I think this is what they meant by this chart:

forum.homeroasters.org/forum/attachments/untitled_4.jpg

So while tuning by setting the PID to manual mode and setting the output gradually in 1% increments, you get instant feedback via the gauge, flame and sound. But when you decrement the output, you have to decrement by a larger margin until the pressure starts going down and noticing a decrease in flame. The works the other way around too, when it starts decrementing, then you start incrementing.

Fortunately, the impact of this "lag" is minimal if the PID is well tuned, and the input pressure to the valve is calibrated via regulator to not give more pressure than the output max your roaster needs. A well tuned PID will minimize the need to aggressively change directions from opening to closing. And supplying no more pressure than the max needed, increases the granularity of the output control.

This took me a while to figure out via trial and error. I was fine tuning for one direction (up/open), then get not so tuned behaviour.
tamarian attached the following image:
untitled_4.jpg

Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
Dan
Tamarian, I'm not surprised to see this. It's called hysteresis. Just like in a steering wheel, when you turn it the opposite direction you have to take up the slack before the wheel begins to turn. I don't know if a PID has a function to deal with this or not.
1 pound electric sample roaster, 3 pound direct-flame roaster, both handmade; modified Mazzer Mini grinder, LaSpaziale Vivaldi II automatic espresso machine. When the electricity goes out I make vacpot coffee from beans ground on my Zassenhaus hand grinder, and heat the water with a teakettle on the gas range.
 
tamarian
Thanks Dan. The valve's documentation mentions that the valve's hysteresis is a maximum of 10% of max current.. Which explains why it had wide swings when the input pressure was way over what is the max needed for the largest batch. Reducing input from 4 bar to 1 bar reduced that swing.

Clippard has a 2% hysteresis valve, around $160, that requires a bi-polar stepper/chopper driver.

--

On a separate note, those interested in proportional valves and drivers may benefit from this discussion on earlier build thread, starting from this post: http://forum.home...post_43716
Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
tamarian
While our house's evaporative cooling was enough to exhaust the smoke, the old de-chaff contraption was only good for outdoors, as the screen still lets some small chaff dust escape, that may eventually clog under my laptop's keyboard. So I used a large paint can for a cyclon. Nothing fancy inside and no screening, as whatever escapes will go outdoors. The old chaff can needed cleaning every 4 or so roasts. This one can hold a lot more, and with the blower on max maybe "self cleaning" Roflmao
tamarian attached the following image:
img-20130930-wa0001.jpg

Wa'il. 1 Kg PID'ed gas-fired fluid bed roaster, GS/3MPS, K10F
 
Jump to Forum: