Exploring North Gas Roaster Automation: Part 1

Discuss roast levels and profiles for espresso, equipment for roasting coffee.
User avatar
slickrock
Posts: 272
Joined: 13 years ago

#1: Post by slickrock »




Intro:

I've been procrastinating and promising drgary for nearly two years to post on my journey of roaster automation with my 1KG gas roaster. Its been a worthwhile endeavor for me to automate my rig and I've bean doing automated profile roasting for over two years now. (I used to leverage my time on public transportation to write detailed reviews, but my recent business clients require driving on my part, so I've had less time to dedicate to new write-ups). There is really nothing novel here about PID-automating a roaster in these forum circles, but automating a gas roaster is different: there are entailments related to automating gas flow in a drum roaster along with the challenging characteristics of PID tracking on heftier solid drum roasters that exhibit more sluggish power response driving changes in BT.

The roaster in question that I automated happens to be the North TJ-067. I can probably be counted as an early adopter of this roaster, basically before Mill City became the fully established US distributor of North roasters that it is now, and before the TJ-067 became a mainstay in roaster circles over the last few years. Interestingly, for me at least, it was my interest in roaster automation that initially drew me to this model. However, this topic should be of interest and applicable to anyone considering automation for the their own manual gas drum roaster, irrespective of make and model (even down to the Huky and it's equivalents).



Evolution of Interest in Automation:

My thoughts and interest in roaster automation had actually developed well before getting the North; actually all the way back to my Hottop days in the 2000's, where that particular roaster had no automation options at that early time (basically the year it came out). Way back then, the I began to be enchanted with the idea of "recording a roast" and basically replaying it. I had some challenges trying to upgrade me Digital Hottop to a programmable model that I need not get into here, but this ultimately took a back seat to my desire to roast larger batches (near the pound range) than what the Hottop could do; not to mention an even stronger desire to move to gas roasting and away from electrical elements for heat production.

Enter the Huky 500 an my early-adopter escapades shared on this forum. In fully embracing this roaster, I basically put my desires of roast automation on the back burner. Why? For reasons obvious, there are ostensibly no electronics in the Huky - its basically a full manual roaster with you as the prime mover: the only way you can adjust power is with your own hand on the needle-valve gas supply. The only way you can control ventilation is with your hand actuating the damper. And its up to you to monitor your temperature and timings to control your roast profile. (Now I'm sure many of you are saying "well that's exactly the point", but please hold on onto that thought as I'll be covering this aspect in a companion post [LINK TBD]). So in effect, the Huky is perhaps the most unsuited sample roaster to consider for roast automation. Consequently, my automation journey took a back seat for a few of years... and that's probably a good thing, as I really got more in touch with profiling and roast control along the way.

Into the next decade came inklings of a new, bigger, better large capacity Hottop on the horizon. Before it ultimately turned out to be vaporware, the promise of 1kg batch Hottop that would likely provide electronic roast control with hints of automation capability started to rekindle my interest in roaster automation once again, but this time on a higher-capacity roaster. When the projected cost of this roaster turned out to be stratospheric, I began look elsewhere and it was in that journey that I happened upon the electrical version TJ-067 back in early 2013. While this roaster checked off a number features I was looking for, the thing that was most interesting to me a the time was that the electrical version of the roaster was robust and capable at 1kg capacities... and very suitable for roaster automation.

Why? For the simple reason that that this roaster had an onboard high-power Solid-State Relay (SSR) to control the heating elements and that typical power duty controllers are primarily designed to drive them (e.g. TC4). In fact, most electrical heater-element roasters on the forum, including the Quest, Hottop, Behmor, etc., can more readily be converted to roaster automation compared to gas roasters because SSRs (onboard and otherwise) can be used to cycle the heating elements OFF and ON - rapidly at times - without the click and wear of mechanical relays. For it was this reason (ironically in hindsight) why I initially had my sights on the electrical version rather than the gas version of the North roaster.


Example of a TC4 wiring diagram for electric roasters, but which is unsuitable for gas roasters, especially those with AC fans

While I was initially impressed that the gas-version of the North roaster with its on-board PID control and auto-ignition capability, it became clear that the onboard PID did not output variable voltage duty, but rather only supported ON/OFF mode to control the onboard ON/OFF solenoid gas valve. Any attempts to automate to "duty cycle" control the ON/OFF valve on the roaster would be met with the same issues of driving a mechanical relay as stated above (worse actually since the auto-ignitor would be taxed with frequent spark re-ignitions. Because gas roasters don't use SSRs (i.e. only manual needle valves), at this point it seemed to me that roaster automation with a gas roaster was out of reach with any kind of low-cost solution, hence my initial deference to the electrical North, even though deep in my bones I really wanted the gas version, based on my extensive experience with the Huky.

The PID, the Valve, and the Controller:

It was around this time that fellow HB member tamarin, who was also active on the Home Roaster Forum, turned me on to the idea of using a Proportional Valve to electronically control pressure on a gas roaster (i.e fluid bed or drum type).



Basically the proportional valve can smoothly regulate gas pressure without the ON/OFF duty cycling of mechanical relays and re-ignition: as more voltage/current is applied to the valve, the valve opens up in the same proportion to the voltage/current applied. So if say 3.5V is applied to valve that would fully open at 10V, then the valve would immediately open by just 35%. That way, if I could employ a variable voltage PID, I could use it to drive an "analog like" proportional gas valve (in contrast to a duty-cycle PID driving a "digital-like" SSR in an electrical roaster). After some further research, it became clear that I might be able to have my cake and eat it too: to be able to cheaply automate a gas drum roaster! So armed with this information, I went and group-purchased the gas-powered roaster, along with drgary, directly from North Coffee a few years ago.

As a North early adopter, It took months to get my setup gas-integrated, instrumented, seasoned and stabilized for manual roasting with Artisan as well as understand the capabilities (and limitations) of the onboard PID. Automation was a distant thought at this time, though I was able to leverage the onboard PID to target stabilized drop temps for roasting. Over time though, I began to appreciate certain design aspects of the roaster that would be conducive to setting up automated PID control: one is that the North has outboard plumbing, so it would be easy to added additional valving and bypass lines to support the introduction of a proportional valve; second, is ample internal chassis room to hide the control box to drive the valve.



Actually, this would be a good time to talk more about this valve and control design.

A PID-controlled gas roaster requires the following:
  1. Proportional Gas Valve: I chose the Clippard EVP Valve because of its inexpensive price, though others of similar capability could be used. I chose the valve model with the widest stock inlet size to allow the maximum gas flow at low pressure (which is ideal for propane/LP line pressure), and for obvious reasons, the valve is normal off (i.e. shut when no power is applied). I situated the valve controller just outside the wiring harness inlet that is used for the stock shutoff valve for the roaster. The wire that connects to proportional valve is feed through that hole into the under chassis (see below).

  2. Valve Driver Controller: The Clippard valve is actually a "current-controlled" valve and I knew that I would have to develop some more involved electronics to control it via a PID controller, which typically expect "voltage-controlled" devices to drive. Luckily. Clippard also offers a companion valve driver controller to simplify the integration, providing voltage-to-current conversion, scaling, as well as number of bias controls to easily dial-in the sensitivity and high-low limits on the valve.



    Not surprisingly, this controller is more expensive than the valve itself, though you save quite a bit if you order the controller naked without an enclosure. BTW, some other valve manufacturers provide the valve with controller built-in, but those valves can even be more pricey. I housed the valve controller and PID controller (see item 3 below) in a transparent, compact Arduino housing I picked up on Ebay.

  3. Duty/PID controller: Basically, this is the the control system used to regulate the duty of the valve based on the Set Value (SV) temperature of the roaster, which is one of the main values to control with a PID. (I won't spend time here covering the theory and function of if PID control since there is a wealth of it the can be gleaned from other forum threads found here and on the Internet). In my case, I chose a headless Arduino-TC4 Shield combination (i.e. no LCD ) for my PID setup. I chose this versus a typical self-contained PID controller such as a Fuji for number of reasons:
    1. namely because I intend to fully control the PID via Artisan, so there is no need for a display and physical buttons; because the Arduiono-TC4 piggyback setup is rather compact, low power and can be easily stowed inside the roaster;
    2. because I can store as many profiles and set points as memory would allow;
    3. and because I can more easily integrate Bluetooth for wireless operation.

    I won't go into the finer details of Arduino as programmable general purpose device controller, but you can find a wealth of information related to it on the Home Roaster Forum. Traditionally, though, a Arduino-TC4 (or TC4) is primarily used to do two things: one is to measure temperature from thermocouple probes and two is to optionally drive the duty of SSRs to directly control power to electrical heating elements and/or DC fan control (for example to control power and fan levels for a HotTop directly from Artisan via slider controls in the UI). However, a more specialized capability is to use the Arduino-TC4 as the PID controller itself: by compiling and loading a sketch (i.e. source code) into the Arduino board to provide PID processing logic, of which there are a number of available sketch lineages that can be used for this purpose (more on this later).
  4. Roast Profile Program: Although technically optional, this is where Artisan comes in and provides a huge value proposition for roaster automation. Typically, Artisan is used for temperature tracking only so you can guide the roast as it is happening and review the roast profile afterwards. Better still, with the aid of a TC4, Artisan can be use to set the duty of the roaster power and/or fan controls of the roaster directly from the UI, with slider controls, programmable button controls, events, and the like. But the coolest feature is when you can use Artisan to drive the PID, allowing it to track against a roast profile you've provided ahead of time within Artisan. The best way to think of this is that Artisan is continuously changing SV to track to roast curve. This is mode by which my entire setup hinges upon.
  5. Temperature probes: This is an implicit requirement, as the roaster must be setup with appropriate temperature probes to measure Environment Temperature (ET) and Bean Temperature (BT), at a minimum. Often times, roasters have dubious temperature probes placed in suboptimal locations to properly capture ET and BT, and these issues are obviously roaster-specific. For my particular setup with the TJ-067 (pre-Mill City), I settled on placing the BT probe using the existing left-hand hole used for the screw secures the sight window.



    I used a small tube bender to form the the probe to bend up to the left where the bean mass settles when the drum is in medium speed rotation. For ET, I used the bottom right-hand hole of the bean chute for the ET probe. I had to bend the probe just slightly to miss the rather sizable drum fins. Keep in mind tube bending is an art form, so I would purchase multiple temp probes, with the intention that one or more of them are sacrificial to practice. For cable routing, I chose the same cable routing path that the stock roaster has down through the under-chassis of the roaster circuitry.
  6. Plumbing: this is gas roaster after all, so gas plumbing is required. I chose the following design, through there are plenty of other approaches to consider, such as a bypass valve, but this was a less expensive route.



    For my setup, the gas flow is branched in a bypass configuration to the normal gas flow for the proportional valve. For the main line I added a simple cut-off ball valve. When opened, the roaster is manual mode, where the gas flows through the ball valve up through the stock needle valve to that is used to control the roaster by hand. When closed, the roaster is in duty/PID mode, whereby gas can only flow through proportional valve for automated control. In this mode, the stock needle valve needs to be fully open.


    One complication is the need to account for the the differing back pressures of the ball valve and proportional valve, because the inlet of the proportional valve is much smaller then that of the ball valve. For that, I strongly recommend placing an additional needle valve in series with the cutoff ball valve, adjusting the needle valve so that the back pressure is equalized with the proportional valve. If you don't do this, then you will have to adjust the upstream gas regulator pressure each time you switch from manual gas control to PID control. You will notice from the photos that I didn't include this extra needle valve, so I'm plagued with this issue until I get around to installing one. An additional plumbing consequence an issue with branching the gas circuit into different lines with pipe thread connections: you will need at least two swivel fittings in your plumbing so that both lines can be fitted at the same time when connecting the rest of the plumbing up with standard pipe threading. These parts were a little hard to find - here is an example. Alternatively you can use a hose with a swivel on on side. In hindsight, a less complicated approach would be to use copper fittings and solder them together.
TC4/Valve Controller Integration Details: As stated above, the TC4 Is more than just a temperature interface shield for Arduino: it has interfaces that are designed to be integrated with other devices that can drive duty control for power and fan speed. Jim Gallt, the architect of the TC4, made some assumptions about duty cycle, power isolation, and the type of devices it would likely be controlled by the TC4. Namely, the design (along with his sketch lineage) appears to be biased to control SSRs with the intention of primarily controlling heating element roasters that operate with very low frequency on/off duty. Additionally, the design and original sketch lineage assumed DC fan control, including an transistor current amplifier to isolate current draw of any low impedance connected devices. All of this is just dandy if you are connecting to a Hottop or Quest, but for the Clippard valve driver controller, some changes are in order, which can be found in my saga to get these two devices talking to each other on the HomeRoaster forum some years ago. Basically the changes boil down to two aspects. First is to bypass the amplifier/isolator and use Arduino to drive the valve controller directly; not only because it is not needed because it is high impedance, but the impedance isolation circuitry does not work well when operating at a high duty cycle needed for valve driver. Why is a high duty cycle needed for the valve driver? Because the valve driver board needs basically a 2KHz+ duty cycle to simulate a DC voltage input to control the valve in an analog manner (i.e. as opposed to on/off digital manner of an SSR). Accordingly, the second change is to modify the include statements in Jim's TC4 sketch so that the duty1 terminal used to drive the valve controller will output at 2KHz. This is actually simpler than it sounds because Jim provides readable "commented out" include statements that you can "un-comment" to activate them, after which recompile of the sketch will get you going right along.

Fan Duty Control and PID Sketch Concerns: Once you've bought the valve, controllers, plumbing, probes, housing, etc. you will need to load up a sketch into the Arduino for PID control. This is where things can get complicated if you are not Arduino/software inclined, and I won't spend time here on the finer points of Arduino programming and compiling, all of which can be found on the web. That said, matters are even more complicated because there are multiple sketch branch versions to choose from. At the time I embarked on this endeavor (around 3 years ago), there were several sketch lineages to consider, not all of which provided a complete set of capabilities I needed to fully automate the TJ-067:
  1. Jim's sketch lineage included the much needed duty cycle frequency configuration for the valve controller but it could only support DC fan control. My "early" TJ-067 has an AC fan without a VFD, unlike the newer North roasters which sport a VFD. As such, my roaster requires phase controller and zero cross-over detectors to operate the fan with duty control from the TC4. This was the version I ended up going with as I intended to tackle fan control at a later time.
  2. An alternate sketch lineage exists from green_cardigan (the original progenitor of PID control on the TC4) on the HomeRoasters forum the does support full AC Fan control on the TC4, however, it didn't support the capability to easily change the duty cycle frequency needed of the valve controller. (Power control wins over fan control, so I elected at that time to use Jim's sketch for my setup and forgo automated fan control for my setup.) Some good news though: recently, green_cardigan released a new sketch that supposedly allows both duty cycle frequency control plus AC fan control. Although I have not tested it, this opens the door to duty AC fan control as well as PID power control driven from Artisan through the TC4.
  3. Another recent development is that Artisan 1.0 itself supports built-in PID control. This means that Artisan simply outputs duty values to the TC4 to control power, rather than output Set Values (SV) to the TC4 to control its onboard PID. This basically obviates the need of PID control on the TC4 altogether. It also would allow Artisan to theoretically provide PID control on both duty terminals of the TC4 (i.e. you could PID control the fan as well as power). Currently, as far as know, there is no TC4 sketch that provides PID control on both duty terminals. And with built-in Artisan PID control, PID logic in the current sketch lineages can be done away with altogether to save memory space.
Now all the above technical interlude isn't half as interesting as setting up Artisan and developing synthetic roast profiles to track your profile automatically with PID control; all of which I'll be covering in Part II of this review later on. There, I'll be covering conceptual arguments for roast automation as well as particulars specific to PID roasting with Artisan on a larger drum roaster like the North.

07/11/1991, 08/21/2017, 04/08/2024, 08/12/2045

cruiten
Posts: 35
Joined: 15 years ago

#2: Post by cruiten »

Very nice post on a very interesting topic...

I am looking forward to reading part 2.

SJM
Posts: 1819
Joined: 17 years ago

#3: Post by SJM »

Nice work, SlickRock.

John Gahlt ? Are you sure you don't mean Jim Gallt?

User avatar
AssafL
Posts: 2588
Joined: 14 years ago

#4: Post by AssafL »

Awesome work. Very impressive.
Scraping away (slowly) at the tyranny of biases and dogma.

User avatar
JK
Posts: 626
Joined: 12 years ago

#5: Post by JK »

Joel can you rewrite that in Stooge? That's my first language :)

Thanks for sharing, I get most of it...
-----------------------------
I'm on a Mission from God!

grundemp
Posts: 3
Joined: 7 years ago

#6: Post by grundemp »

Hi there,

Thanks for sharing this post. Very interesting and I'm in the process of automating my 2 KG north roaster as well. I've decided to use the same clipart proportional valve but having some problems getting adequate flow and I'll need to increase the inlet pressure. I roast usually with 4.5 KPA pressure. Can you please let me know what pressure you have set your cylinder regulator at to in order to get sufficient flow through the valve?

Also very interested to find-out how you regulate the fan settings and looking forwards to your follow-up post. Once I have my roaster working I'll be happy to share my experiments as well on the forum

Best regards,

Peter

User avatar
woodchuck
Posts: 311
Joined: 18 years ago

#7: Post by woodchuck »

Excellent job Slickrock, thanks for sharing it with us.

Headala
Posts: 917
Joined: 10 years ago

#8: Post by Headala »

Very interesting; you make me want to upgrade to a gas roaster! :D

User avatar
slickrock (original poster)
Posts: 272
Joined: 13 years ago

#9: Post by slickrock (original poster) »

grundemp wrote: Very interesting and I'm in the process of automating my 2 KG north roaster as well...I roast usually with 4.5 KPA pressure. Can you please let me know what pressure you have set your cylinder regulator at to in order to get sufficient flow through the valve?...Also very interested to find-out how you regulate the fan settings and looking forwards to your follow-up post. Once I have my roaster working I'll be happy to share my experiments as well on the forum
Good to know that you and other folks are jumping in on this. Do let us know your progress for your particular setup. Answers to your inquiries:
  1. First, what valve did you order? I ordered a ET-P-05-6025, that has a "0.060" orifice, which I recall is the largest standard-issue one they offer. This one produces the least back pressure, but as mentioned in the original post, this too is still more than the standard needle valve and because it is in series, you will need to adjust for it from the pressure regulator.
  2. I have not measured the outlet pressure from my regulator. I'm using a cheesy single-stage 10PSI variable regular (with external knob) for my setup. I adjust it from time to time to maintain 5.0KPA when the TC4 is set at 100% duty (i.e. valve full open) - this is done, if at all, when warming the roaster. As the tank nears empty, I'll back off the regulator to keep the 5.0KPA target, which is also a good signal to go out an get a tank refill.
  3. For me with my non-VFD North Roaster, I still use manual control as I've yet to automate the fan. I basically have 4 fan settings separated by Drying, Ramp, FC-start, and FC-end phases, which are effectively the only times I touch the roaster panel besides turning it on/off (i.e. everything is pretty much done through Artisan now). As discussed in the original post, during the formative time of my automation quest, no sketches where available for both high-frequency duty control and Zero-CD AC-Fan control, but this has since changed with some of the more recent sketches available. I would be curious if and how you get both Power and Fan going (after all the, TC4 supports 2 duty outputs), though having a VFD Fan should simplify things further.
  4. Your post has given me the motivation to get back into gear and post Part 2. Out shortly.
Good Luck.
07/11/1991, 08/21/2017, 04/08/2024, 08/12/2045

justgrindit
Posts: 84
Joined: 7 years ago

#10: Post by justgrindit »

slickrock wrote: Your post has given me the motivation to get back into gear and post Part 2. Out shortly.
Excellent writing, thanks for taking your time. Looking forward to Part 2 :) .

Post Reply