Homemade Sous Vide Controller
October 18, 2011 7:45 AM Subscribe
I'm interested in setting up a PID and thermocouple to use a slow cooker for sous vide. Most guides are kind of handwavy about this process. What should I look for in a PID? Are parts from Ebay reliable enough, or are they likely to be unreliable (either in terms of holding it near 1-2 Fahrenheit of the target temp or not lasting more than a few projects) or not work? Or what about using an Arduino?
I have a mid-1970's era slow cooker (I think my parents actually got it as a wedding present), and I'm interested in using it for sous vide. I've had good results with the beer cooler method, but now I'm interested in moving on to tricks like 48 hour brisket/pastrami that are less safe and less reliable with constantly adding hot water to keep it at the target temp. Automation is the ideal solution.
I've seen fully assembled PID and thermocouple switches that handle mains power from Sous Vide Magic, but the price is $140-170, and the separate parts I've seen on Ebay and lightobject are much cheaper for building a PID controller for mains power.
If I understand guides correctly, I want to wire up a SSR rated for my slow cooker's wattage with a proper heatsink (mechanical relays tend to fail according to one guide I saw) to the DC output on the PID, wire an extension cord to the SSR insulating and enclosing it well since it is near an appliance with water in it, and also hook up a type J thermocouple to the PID and put it in the bath.
Also, how difficult is it to make/get an enclosure for this that are reasonably splashproof? I'm half thinking of repurposing an old rubbermaid container with the input wires going in, but are there more professional enclosures out there?
Straying a bit from the style of design popular online, how would hooking up my Arduino Uno (I know it's more than is needed for this, but it's not doing anything else ATM) to a Type J thermocouple and a PowerSwitch tail (if I can find one) work? It looks to me like this would be more doable, as the mains power is relatively isolated and the PowerSwitch tail can handle "15amps @ 120vac," which is 1800w (enough for the slow cooker) if I remember Physics right. And I feel like implementing the PID's algorithm for the Arduino would be a fun exercise. To improve usability (read, not hardcoding temperatures and using serial output), I could also add a serial display and some buttons to adjust temperature. I like the idea of this, but it doesn't seem to be as widely done as using PIDs, so I wanted someone more experienced to give their opinion on my idea.
So, what result is best for me, a guy who is more than willing to do some basic electronics and programming, even if the result doesn't end with a clean box?
I have a mid-1970's era slow cooker (I think my parents actually got it as a wedding present), and I'm interested in using it for sous vide. I've had good results with the beer cooler method, but now I'm interested in moving on to tricks like 48 hour brisket/pastrami that are less safe and less reliable with constantly adding hot water to keep it at the target temp. Automation is the ideal solution.
I've seen fully assembled PID and thermocouple switches that handle mains power from Sous Vide Magic, but the price is $140-170, and the separate parts I've seen on Ebay and lightobject are much cheaper for building a PID controller for mains power.
If I understand guides correctly, I want to wire up a SSR rated for my slow cooker's wattage with a proper heatsink (mechanical relays tend to fail according to one guide I saw) to the DC output on the PID, wire an extension cord to the SSR insulating and enclosing it well since it is near an appliance with water in it, and also hook up a type J thermocouple to the PID and put it in the bath.
Also, how difficult is it to make/get an enclosure for this that are reasonably splashproof? I'm half thinking of repurposing an old rubbermaid container with the input wires going in, but are there more professional enclosures out there?
Straying a bit from the style of design popular online, how would hooking up my Arduino Uno (I know it's more than is needed for this, but it's not doing anything else ATM) to a Type J thermocouple and a PowerSwitch tail (if I can find one) work? It looks to me like this would be more doable, as the mains power is relatively isolated and the PowerSwitch tail can handle "15amps @ 120vac," which is 1800w (enough for the slow cooker) if I remember Physics right. And I feel like implementing the PID's algorithm for the Arduino would be a fun exercise. To improve usability (read, not hardcoding temperatures and using serial output), I could also add a serial display and some buttons to adjust temperature. I like the idea of this, but it doesn't seem to be as widely done as using PIDs, so I wanted someone more experienced to give their opinion on my idea.
So, what result is best for me, a guy who is more than willing to do some basic electronics and programming, even if the result doesn't end with a clean box?
I have an eBay PID, SSR & thermocouple setup. It works fine, though I sometimes notice that the calibration drifts a little bit from one use to the next (not a big deal to correct, but it can be a pain). I generally find my setup with a slow cooker results in reported temperature fluctuation in a roughly 0.5C range (and checks with another thermometer seem to confirm this). I don't think that even a 1C variation is a huge deal though, especially as you won't be using a circulator, so your water temperature won't be totally consistent throughout.
I built mine in an old computer UPS - it is about the right shape and size and it already has a power cord and outlets in the back, saving some work and resulting in a safer product (grounded chassis, solid construction).
I also recommend purchasing a simple immersion heater than you can place inside a cooler (e.g. hooked to the rim of a mug) to allow for temperature control of a larger volume container (though mostly I make yoghurt in my cooler).
posted by ssg at 8:03 AM on October 18, 2011
I built mine in an old computer UPS - it is about the right shape and size and it already has a power cord and outlets in the back, saving some work and resulting in a safer product (grounded chassis, solid construction).
I also recommend purchasing a simple immersion heater than you can place inside a cooler (e.g. hooked to the rim of a mug) to allow for temperature control of a larger volume container (though mostly I make yoghurt in my cooler).
posted by ssg at 8:03 AM on October 18, 2011
Best answer: The cheap Chinese and Taiwanese processor controllers from ebay are exactly what you expect from those sorts of tools: they work surprisingly well, there will be one or two quirks, and they'll survive about half their expected lifetime. Which is to say: they do well enough for most cases.
Your design is basically solid; just pay attention to what's isolated from power and what isn't. You're building something that would never seek a UL approval sticker, but it doesn't mean you should build it to be dangerous. You can build a box if you have any way to form thin aluminum, or Google "project box" and you'll find places you can buy any number of them. Install proper strain relief if you can, etc. Also note that there's no reason that your controller can't sit a good ways away from your cooker.
I'm not an arduino guy (all my home projects have moved to ARM cores), but the easiest thing to do is probably buy a single-chip thermocouple solution (it will have the amplifiers, CJC temp sensor, maybe even open thermocouple detection, etc., all in a chip). You can either go with a chip having analog outputs (which you then sample with your A/D) like an AD595 or with a digital-out solution (which you just hook up to a spare SPI or I2C port) like an ADT7302. There are tons of offerings out there for such things.
Likewise, you could either use the PowerSwitch you linked or just connect to the solid state relay yourself. The topic of the control loop is slightly more complicated, but it's nothing you can't learn in bites. Start with the "P" part only and low-frequency PWM the SSR on your output value. Add in the "I" part and learn how fun non-linear controls are. Then realize you don't need the "D" part for something like a temperature-hold oven and save yourself the work. ;)
Both results would work. The first one you mention will get you up and running in maybe a day if you're diligent. The latter one might take weeks, depending on your skill levels with the electronics and code. It's really what you'll enjoy most.
posted by introp at 8:05 AM on October 18, 2011
Your design is basically solid; just pay attention to what's isolated from power and what isn't. You're building something that would never seek a UL approval sticker, but it doesn't mean you should build it to be dangerous. You can build a box if you have any way to form thin aluminum, or Google "project box" and you'll find places you can buy any number of them. Install proper strain relief if you can, etc. Also note that there's no reason that your controller can't sit a good ways away from your cooker.
I'm not an arduino guy (all my home projects have moved to ARM cores), but the easiest thing to do is probably buy a single-chip thermocouple solution (it will have the amplifiers, CJC temp sensor, maybe even open thermocouple detection, etc., all in a chip). You can either go with a chip having analog outputs (which you then sample with your A/D) like an AD595 or with a digital-out solution (which you just hook up to a spare SPI or I2C port) like an ADT7302. There are tons of offerings out there for such things.
Likewise, you could either use the PowerSwitch you linked or just connect to the solid state relay yourself. The topic of the control loop is slightly more complicated, but it's nothing you can't learn in bites. Start with the "P" part only and low-frequency PWM the SSR on your output value. Add in the "I" part and learn how fun non-linear controls are. Then realize you don't need the "D" part for something like a temperature-hold oven and save yourself the work. ;)
Both results would work. The first one you mention will get you up and running in maybe a day if you're diligent. The latter one might take weeks, depending on your skill levels with the electronics and code. It's really what you'll enjoy most.
posted by introp at 8:05 AM on October 18, 2011
Response by poster: exogenous, is the STC-1000's sensor waterproof? If not, could I set it up with another thermocouple? The more I read, the more PID seems like overkill for what I need. I like how it works for both heating and cooling, as I could also rig it up with a fridge as a lagering cabinet, which is something I've been curious to try.
I'm definitely thinking about getting an immersion heater for larger jobs, as I have a 10 gallon water cooler/mash tun, and while it's been okay for beer cooler sous vide, it'd be really great to set it up with a heater so that I could handle cuts too big for the slow cooker/make a lot of something at once.
posted by mccarty.tim at 8:22 AM on October 18, 2011
I'm definitely thinking about getting an immersion heater for larger jobs, as I have a 10 gallon water cooler/mash tun, and while it's been okay for beer cooler sous vide, it'd be really great to set it up with a heater so that I could handle cuts too big for the slow cooker/make a lot of something at once.
posted by mccarty.tim at 8:22 AM on October 18, 2011
Here is a full sous vide cooker on Make Projects, with part numbers, downloadable templates, and build instructions.
posted by no relation at 8:53 AM on October 18, 2011
posted by no relation at 8:53 AM on October 18, 2011
Best answer: Yeah, the sensor itself is waterproof. The comments on the website I linked suggested that it might not be food-grade, but I don't see that as a problem for food in a bag.
posted by exogenous at 11:21 AM on October 18, 2011
posted by exogenous at 11:21 AM on October 18, 2011
Best answer: Last note: if you roll your own control loop some day, google "take back half" for a very clever and simple algorithm for doing temperature control without the typical instability headaches due to sensor lag. (I used it most recently for a reflow toaster oven controller and it has performed wonderfully with none of the usual surprises. The only way I can make the loop unstable is if I crank the gain way, way up. The algorithm is both satisfying (simplicity, ease of matching to physical processes) and disturbing (empirically tuning the loop gain rather than deriving it ab initio).
posted by introp at 11:58 AM on October 18, 2011 [1 favorite]
posted by introp at 11:58 AM on October 18, 2011 [1 favorite]
This doesn't sound like a PID control problem. It sounds like bang-bang (i.e., deadband) territory.
PID achieves one type of optimal response to a step input. If you were adding stuff to the unit as it cooked, then a PID would make sense. If not, then you want to be full on until you reach your deadband and then off. You are not likely to see any overshoot to speak of.
Most of the little PID controllers I have bought on eBay have autotune, anyway. It takes a while, since you have to let it cycle through heating/cooling cycles a few times.
I have bought several little tiny ones with 240 VAC outputs, and others with PID outputs for driving SSRs. The number to choose from is staggering. Prices are all over the place, too. Too many to count in the sub $50 range, last time I looked.
posted by FauxScot at 6:07 PM on October 18, 2011
PID achieves one type of optimal response to a step input. If you were adding stuff to the unit as it cooked, then a PID would make sense. If not, then you want to be full on until you reach your deadband and then off. You are not likely to see any overshoot to speak of.
Most of the little PID controllers I have bought on eBay have autotune, anyway. It takes a while, since you have to let it cycle through heating/cooling cycles a few times.
I have bought several little tiny ones with 240 VAC outputs, and others with PID outputs for driving SSRs. The number to choose from is staggering. Prices are all over the place, too. Too many to count in the sub $50 range, last time I looked.
posted by FauxScot at 6:07 PM on October 18, 2011
You have lots of good answers so I can't really add . I will say that I bought Auber PID controllers for a smoker and sous vide and they were excellent/still going years later/GREAT service and communication.
Unfortunately I do not have the sous vide set up anymore or I would post pics to show the layout.
http://www.auberins.com
posted by plumberonkarst at 2:44 AM on October 19, 2011 [1 favorite]
Unfortunately I do not have the sous vide set up anymore or I would post pics to show the layout.
http://www.auberins.com
posted by plumberonkarst at 2:44 AM on October 19, 2011 [1 favorite]
plumberonkarst and I have similar experience with the auber units. $39 on eBay when I bought them last year. PID, SSR, relay outputs, autotune. I bought SYL-1512A units. They are ridiculously small. Power directly from the line. Way cool.
posted by FauxScot at 5:54 AM on October 19, 2011
posted by FauxScot at 5:54 AM on October 19, 2011
Yeah, I've got an Auber PID in my espresso machine. Cheap, works great, and way less messing around that doing something yourself with a micro-controller (unless that messing around is the fun part for you).
posted by markr at 5:23 PM on October 19, 2011
posted by markr at 5:23 PM on October 19, 2011
Response by poster: The STC-1000 I got was faulty, but I found its sensor works with the Arduino using some public sketches, so I may just wire it up to a remote control power switch to turn on and off the slow cooker.
If the Ebay seller in Hong Kong ships me another like they should, then I have a redundant solution in case I want the Arduino for something else, or say, want to make sous vide food and a lager in a thermal cabinet at the same time. Otherwise, I chargeback.
posted by mccarty.tim at 8:45 AM on October 30, 2011
If the Ebay seller in Hong Kong ships me another like they should, then I have a redundant solution in case I want the Arduino for something else, or say, want to make sous vide food and a lager in a thermal cabinet at the same time. Otherwise, I chargeback.
posted by mccarty.tim at 8:45 AM on October 30, 2011
This thread is closed to new comments.
You could use the same NEMA junction box for an enclosure for a PID-based controller if you want.
posted by exogenous at 7:57 AM on October 18, 2011 [1 favorite]