[Syzygies]

I'm part of a design effort to build an inexpensive analog controller for BBQ pits. I keep Art of Electronics handy, but 'in over my head' is where I generally like to be, and where I probably am right now. If anyone would be amused to comment on this circuit, I'd be grateful.

But first, what's the relevance to this forum?

1. Linear regulators and op amps are our bread-and-butter, so the circuit should be obvious to various people here.

2. I own the BBQ pit in question, and I'm eager to bring arbitrary quantities of smoked brisket, ribs, pulled pork to future Bay Area meets. Here is how you can help.

Arguably the best home BBQ pit is a Kamado, a studly relative of the Big Green Egg. These pits are remarkably stable by themselves, but the temperature can wander a bit during overnight cooks, the fire can even go out; natural convection is mildly unstable in our sense. I wish that the solution were as simple as soldering on a capacitor!

A control circuit, using a fan and a thermocouple, is a convenient way to apply negative feedback, locking in the temperature over 20 hour cooks. Various hobbyists pioneered this using digital circuits and 1,500 lines of code, then a commercial company has lately been cleaning up, The BBQ Guru. These gadgets cost $100 and up, one sees dozens of them at any competition.

So the backyard Weber kettle guy, who doesn't want to drop $150 on a commercial controller or lose a month coding a microprocessor, is out of luck. So are my friends who are broke after buying the Kamado itself.

What this begs for is a $20 analog circuit to play exactly the role the CMoy has played in this community. There's a $10 chip, the AD595, that does all the heavy lifting of converting a thermocouple input to a reference voltage, and opening and closing a switch in comparison to a second, externally provided reference voltage.

What I want to generate is a reference voltage that wanders in a tight (say, 5%) sawtooth pattern around its mean. The effect will be to create a range of temperatures for which the fan oscillates on and off, causing the system to stabilize somewhere in this range. This is called proportional control, more sophisticated than taking care of hysteresis, but simpler than true PID control.

My most recent draft circuit, pieces liberally cribbed from both the CMoy and Art of Electronics, is this:

Dual jellybean op-amp, first half a relaxation oscillator, second half a follower.

This isn't audio, are my tantalum caps good enough? Any other comments?

Edit: This revision avoids mucking with the power supply, instead assuming a clean 5V supply that can be used elsewhere in the controller. I mix down a (1/4, 3/4) sawtooth signal (roughly triangular, but not perfectly so) to the desired amplitude, then mix down the resulting signal to the needed voltages for comparison with the thermocouple:

 

[SnoopyRocks]

Can you post a block diagram of the system? It would make it a bit easier help. Meausuring with a thermocouple makes sense but I don't quite follow how you plan on adjusting the temperature in the pit.

 

[tangent]

As it happens, I recall an article in Circuit Cellar, which I'm pretty sure involved a PIC and a modded Kamado. Look it up. It might get you the benefit of those $150 controllers for a DIY price.

 

[Syzygies]

Quote:

Originally Posted by SnoopyRocks Can you post a block diagram of the system?

I drew you one, but both of my scanners are acting up at the moment.

The AD595 converts the type K thermocouple into a standardized voltage. It also includes a comparator, so one can throw a switch when this standardized voltage exceeds a user-provided voltage. I want to provide that voltage. By providing a slowly oscillating triangle wave, I can get "proportional control" more or less for free.

This switch is amplified to run a fan, like a small computer fan, which feeds air to the fire. More air when the pit's too cool, less air when the pit's too hot.

Quote:

Originally Posted by tangent As it happens, I recall an article in Circuit Cellar, which I'm pretty sure involved a PIC and a modded Kamado.

Yup, this guy is part of the debate over the new design efforts. We've been having a religious debate over digital versus analog, while a third party has actually been soldering. The PIC approach simplifies using a chip like the AD595; that circuit cellar article was 1,750 lines of code.

Reflect for a moment on the effect the CMoy amp has had on our community. It's cheap, it works, it draws people in every day who've never soldered anything more complicated than a lamp cord. I'm thinking a $20 analog BBQ controller could be the same thing. Last summer, I looked at the digital DIY projects out there, freaked out thinking about the potential time sink, and bought the commercial product instead. I have friends who don't want to spend so much, and the project amuses me.

I actually own the commercial BBQ Guru, this isn't about the money. I didn't want to say, thought I'd get more help if people thought that the brisket I offered actually depended on this project!

But, hey, analog is new to me, so I'm rather enamored with it. I've been programming my whole life, the last thing I want to see is a new machine language instruction set. It took me longer to read the Tiny15 datasheet than it did to solder my MINT.

 

[SnoopyRocks]

Quote:

Originally Posted by Syzygies I drew you one, but both of my scanners are acting up at the moment.

thanks
Quote:

Originally Posted by Syzygies The AD595 converts the type K thermocouple into a standardized voltage. It also includes a comparator, so one can throw a switch when this standardized voltage exceeds a user-provided voltage. I want to provide that voltage. By providing a slowly oscillating triangle wave, I can get "proportional control" more or less for free.

I don't see how this is a proportional controller. This sounds like an on-off type to me. ??? Maybe the block diagram will help.
Quote:

Originally Posted by Syzygies This switch is amplified to run a fan, like a small computer fan, which feeds air to the fire. More air when the pit's too cool, less air when the pit's too hot.

Now I get it.
Quote:

Originally Posted by Syzygies Yup, this guy is part of the debate over the new design efforts. We've been having a religious debate over digital versus analog, while a third party has actually been soldering. The PIC approach simplifies using a chip like the AD595; that circuit cellar article was 1,750 lines of code.

Why does it have to be so complicated to necessitate digital control/signal processing here? All that is needed is a simple feedback system with a temperature sensor, opamp and DC fan. 1750 lines of code seems like overkill.
Quote:

Originally Posted by Syzygies But, hey, analog is new to me, so I'm rather enamored with it. I've been programming my whole life, the last thing I want to see is a new machine language instruction set. It took me longer to read the Tiny15 datasheet than it did to solder my MINT.

The world is analog - it's not going away.

I find it so frustrating that people acutally think there are 'digital' amplifiers and speakers.

Even digital signals are really analog.

 

[kevin gilmore]

this might help
http://www.ecircuitcenter.com/Circui...pid/op_pid.htm

I do have some schematics for an all analog watlow temperature
controller around here somewhere. Can't find it at the moment.

There are all sorts of digital pid controllers on ebay for $30 to $50.
I prefer and use almost exclusively the eurotherm controllers which
can be put into continuous auto tune mode and would be perfect
for this application.

 

[Syzygies]

Quote:

Originally Posted by SnoopyRocks thanks I don't see how this is a proportional controller. This sounds like an on-off type to me. ??? Maybe the block diagram will help. Now I get it.

For anyone else who didn't get it, comparing a triangle wave to the probe temperature turns the fan on some of the time if the probe temperature is anywhere within the range of the triangle wave. Near the bottom, the fan is usually on, causing the pit temperature to rise. Near the top, the fan is usually off, causing the pit temperature to fall. The pit temperature will find its equilibrium somewhere within the range of the triangle wave, not necessarily at the midpoint. The effect is proportional control. Using a triangle wave is my idea for implementing this in analog, although I can't imagine how else to do this in analog. These days this is usually done in digital.

A true PID controller notices that the equilibrium is not at the midpoint, and tries to adjust for it. An on-off controller simply stays on below the setpoint, yielding an irregular fan period that works quite well in practice. Nevertheless, I had been thinking that proportional control is the best compromise.

Have I drawn the simplest approximate triangle wave generator, with a long period, and user-adjustable amplitude and mean?

I'm sure I'll learn just fine by trial-and-error, but some things confuse me. This doesn't need Black Gate caps, but it probably needs at least solid tantalum 1uF or so, to be stable in this app?

Quote:

Originally Posted by kevin gilmore this might help

That's impressive. I searched for analog PID control, didn't find anything so clean! I might have to build this.

There's a quirky ramp-and-soak component to BBQ pits, starting the fire is very different from holding it. Manually setting a proportional controller, with a wider interval at first, has a certain "stick-shift" transparency to it that true PID control doesn't have, at least for me. The issue is amping up the fire to the point where it overshoots badly, when you're starting it. But for beating microprocessor code in analog with a few $0.40 op amps, that circuit rocks!

However, it looks to me like it outputs its control as a voltage. This still needs to be converted to "sometimes on, sometimes off" fan control, for the specific application of a BBQ pit. It seems to me that leads to circuits like I've drawn, that will also do just fine by themselves, if they work.

 

[SnoopyRocks]

It still seems like this is more complicated than it needs to be. All that needs to be done is to vary the airflow though a fan. With a DC fan (like the 12V fans in PC) this can be easily done without a triangle wave - more voltage means more air flow. Less voltage - less airflow. Moreover the flow rate is variable between fully on and off, allowing for fine adjstments. Is a special fan required here that necessitates the triangle wave?

As far as setting the DC level and adjusting the amplitude: An inverting opamp circuit with two pots will do this. One pot adjusts the DC level on the positive opamp input, setting the ouput common mode level. The other pot adjusts one of the resistors (either feedback or input) to set the gain and thus the amplitude.

 

[Syzygies]

The original Kamado thread where these designs are being discussed sees a lot of humorous hassling from people making fun of the geek talk. ...Exactly as one would expect watching a BBQ pit, hanging out with this gang.

Then this post included the following schematic, that floored all of us:

 

[Frenchman]

Couldn't you PWM the fan in order to control its speed, as opposed to the on/off system? Perhaps taking the signal from the thermocouple and feeding it into a difference amp with the user set temp as the other input. This produces a voltage proportional to the difference of the preset and the input and then you use that voltage to control a PWM controller which drives the fan. That gives you PD control, an even better solution.

--
Edit, I just realized I said PWM as OPPOSED to an on/off. Haha, not only was it dumb, but I'm a nerd for thinking that was funny

 

[Syzygies]

That's an interesting organization of the circuit. My slow triangle wave is effectively PWM, if you think about it. Would the PWM circuit you have in mind be simpler?

At higher speeds there's an issue with Pulse Width Modulation and fans: Too little average voltage, and they just stop. "Stiction". This has to be taken account of as a nonlinearity in the control algorithm, or else any attempt at proportional control breaks down and becomes on-off control.

 

[blip]

I respect doing this analog... it definitely makes it more accessible and the intellectual challenge is definitely more interesting.

But if you ever wanted to consider using a microprocessor, I'd bet you could do it with a PICAXE for about $8-15 total (depending on how much accuracy you wanted) and maybe 30 lines of code. (All you would need is a simple feedback from the temperature sensor to the chips built in PWM output.) You could do it for even cheaper if you want to use a PIC. See here for a similar application in temperature control during home-brewing of beer: http://users.tpg.com.au/adsl2y58/htm...ontroller.html

Anyway, back on topic. You can somewhat solve the fan stall problem with PWM through careful fan selection. Do a little digging around one of the silent computing forums and I'll bet you can find a 12v one that won't stall out until 4v or so. I would think that that would provide you with enough range... or am I missing something?

 

[Frenchman]

This could also be done with very simple pic, a 8 pin 10 series. You could simply sample the temp sensor with the AD module, and then find the difference between the PWM signal and the temp signal (sampled with 8 bit res), and then add that difference to the PWM signal accounting for overflow conditions on the top and bottom. Me and a buddy implemented this for a throttle controller with the temp sensor input being replaced by a hall effect sensor. The hall effect sensor pulses were captured and converted to RPMs, and that was used as the input signal. It worked extremely well with a bit of tweaking on the gains. I think either way would be technically easy, but only the analog method can be done by someone without the programming hardware. Unless you sold programmed pics to people wanting to make one.