Hi guys,

My roaster is broken at the moment, so I have it apart trying to figure out why it nots working. I think I have (hopefully) narrowed it down to a dodgy contactor which is on order. While its open though, I have long been thinking about PID'ing it for profiling and was hoping for some advice.

It seems as if it could be quite straightforward, but I only say this through what I have learned by osmosis. There is already a temperature controller which outputs to a contactor to turn the electrical heating element on and off when below and above the set temperature. Also fed into the temperature controller is a J type thermocouple.

At present the temperature probe is outside the bean mass, and it is pretty much set to a static temperature. As the beans come up closer to the temperature, it results in the heating element cycle being progressively more OFF than ON. Roast duration is largely set by the initial drop temperature, which you must change significantly depending on the volume roasted. I manually tweak the temperature control up or down to give it a little more or less energy at a particular time if I think it needs it. The element is inside the drum, and when power is off to the heater, it is still providing heat as its still glowing red, just at a reduced rate. Airflow control is fairly limited and any changes made don't effect things much. Drum speed is fixed.

I would like to move the thermocouple into the bean mass, and control the roast via a PID with setpoints to maintain a profile, ideally linked to a computer. Because of the nature of the beast, I don't think a PID running on an environmental temperature input would work that well, since at present it basically remains static.

I am hoping that I can locally source a PID that will be an easy physical swap with the current temperature controller. I can get the FUJI PXR-4 easily and I have heard several people have used this with success. I don't want to rewire the roaster at all, thats above my skill level. Essentially, I want to remove the current controller and replace it with the new one and I am hoping because its already electrical and runs with contactors that this might be fairly compatible. Looking at it, I am hoping it is as simple as that, then obviously having to program the new PID.

So here is my current temperature controller, EMKO ESM-3710,

http://www.crobel.net/emko/English/inse ... %20V00.pdf





And here's the Fuji PXR-4 data

http://docs-asia.electrocomponents.com/ ... 6b88c4.pdf

The contactors which the PID would control are here

http://hongkong01.rs-online.com/web/sea ... &R=0758721



Any advice on compatibility with the current thermocouple/contactors and the feasibility of doing this would be greatly appreciated. I've got a week off work, so a perfect time to play with this.

Cheers, Chris

I have no idea what your roaster is, but for a small drum roaster or any air roaster, a contactor is a disgrace, guaranteeing a crappy roast regardless of control algorithm or sensor placement. You need something that, from the beans point of view, looks like an infinitely regulable, non-oscillating heat source. A solid state relay switching on a 1 second duty cycle is standard, since the heat storage of the roast chamber can filter out the 1 second oscillation. If you are really precise, you will get an SCR that switches the 60 hertz wave form like a dimmer.

The other consideration is that the sensor controlling the PID can be placed close to the heat source, so it monitors the environmental temperature in the roaster. This will give you the highest quality control, but requires the most manual oversight and tweaking to get the right bean temperature profile. Driving the PID on bean temperature is fully automatic, but can be unstable, creating up and down cycles in the environmental temperatures. The most elegant solution is to set up a programmable micro-controller to do PID on the environmental temperature, changing the target based on feedback from the bean temperature (this used to be called dual loop control, but with a computer, it's best not to use PID for the outer loop of bean temperature, but rather a series of if-then style heuristics). I know a few people who've tried to implement this, but haven't read the details.

If you are using a single loop ramp/soak PID like the Fuji PXR, it is best to experiment with different sensor locations while logging multiple temperature channels inside the roaster to see how stable and successful the trial placement is. Inexpensive two and four channel temperature loggers are now made in China and can be bought for $125 to $175. These are invaluable for a coffee hobby (and roasting meats to everyone's favorite degree of doneness)

Its a 1kg Toper Cafemino. Heaps of photos and details on the below review.

http://www.bellabarista.co.uk/pdf/Toper ... ptedv4.pdf

I've actually been quite pleased with it over the last couple of years. Whilst the digital ON/OFF cycle is not ideal, there is so much residual energy in the heavy drum, the air inside the drum (since the airflow is relatively low) and the element (inside the drum, still glowing red hot for a minute or two after power off) that I don't believe its a major issue as it might be in an air roaster. I think in reality that when the element is switched off, the radiant energy it is releasing reduces from 100% at the moment it is turned off, to perhaps something more like 50% by the time it cycles back on... complete guess at the end percentage, but hopefully you get my drift... it might be turned off, but its still adding energy. The measured ET with the standard probe placement seldom varies more than 1 or 2 degrees, other than when the beans go exothermic and thats when I typically bump the set point up a little to prevent the heating element being off too long.

I'm not so sure regarding placing the sensor close to the element, as the direct radiant energy would make it read much higher than the general ET I would have thought. Its presently shielded from the element by the drum's axle.

Different sensor placement trials does sound like a good idea. I can drill a hole in the drum exhaust for a better idea of ET perhaps, as this would not be influenced by the radiant energy.

But my real question initially is, can I do a straight swap for another controller to let me play with some of these variables. I don't want to get into major roaster surgery, or it may end in tears.

Thanks, Chris

1. The Fuji PXR4 can work, but it will only allow you to create temperature patterns using minutes as the time interval. Instead I would recommend a Fuji PXG4, which is more expensive, but allows you to program in seconds.

2. Jim is giving good advice in recommending that you switch to solid state relays (SSRs) instead of mechanical contactors. Not only will they last a lot longer, but, switched on/off at one second intervals, they actually will allow your heating elements to last longer. That's because the elements don't expand and contract significantly at that switching rate.

3. The controller must be ordered with a control output that matches your application. If you stay with mechanical contactors, you must specify "Relay Contact Output" for "Control Output 1."

4. Keep in mind that a PID controller like the Fuji does not like to be in a very hot location. That could lead to inaccurate temperature readings and shortened lifespan.

AndyS wrote:3. The controller must be ordered with a control output that matches your application. If you stay with mechanical contactors, you must specify "Relay Contact Output" for "Control Output 1."

Might it also be ordered with DC voltage output, then add a low-amperage SSR for switching the AC control signal to the contactors? Thinking this could be a first step in the direction of a future conversion to solid state control.

Jim

JimG wrote:Might it also be ordered with DC voltage output, then add a low-amperage SSR for switching the AC control signal to the contactors? Thinking this could be a first step in the direction of a future conversion to solid state control.

That's a very good idea.

Hi guys, thanks for the thoughts so far.

I can get the PXR4 locally, but I am not sure about the PXG4 however I will investigate tomorrow. Other than the minutes vs seconds, are there any other significant differences? I have a week off work now which I could use to play with this stuff, but if its worthwhile then I could wait to order from abroad.

I didn't realise when ordering you had to specify the type of output. That being the case, it would probably be prudent to go for the DC option. There is a fair amount of room in there, the white coloured circuit breaker behind the contactors could be shifted further towards the back of the roaster, creating some additional room towards the front.



Temperature-wise I also think it would be ok. With the heating element in the drum, and the 'roof' of the electronics enclosure also thermally separating it from the drum (and a fan in there, to the right of the stirring arm motors transformer), its probably not too bad. The black tube is just air from the cooling tray, so it doesn't get too hot.

You will need to forgive my ignorance, but does the PID behave differently when triggering a SSR vs a contactor. I understand the longevity issue with rapid switching (and the annoying clicking), but does the frequency of pulses change also, or is this something that you can choose. Just trying to get my head around the 1 second duty cycle, does that mean every second it is pulsed for whatever duration the PID thinks necessary ranging from 0.0000 to 1.0000 seconds?

Now the next thing that I'm thinking is, how many SSRs would I need? The current contactor is 3 poles and a N/O switch. Now in the above picture, I think I would only need to change the contactor labelled 3 (I switched 3 and 4 for problem solving). To me at least, it appears A1/A2 are the control leads (rightmost leads) which make the 3 left most contacts, with nothing connected to the N/O one.

The three remaining contactors (from L-R) are for 1. Drum and exhaust, 2. Stirring arms in the cooling tray, 3. (marked as 4 in vivid) Sends power to the temperature controller (I think) which in turns triggers the last one which would need to be replaced with SSRs. The brown, yellow and blue wires coming out 2T1, 4T2 and 6T3 are directly to the heating element. The first three could stay as contactors as they are not cycled, just ON/OFF switches in essence.

So would I need three SSRs to replace the one contactor? and the PID, and heatsinks?

These guys are based in HK and offer next day delivery. Any recommendations for the above?

http://hongkong01.rs-online.com/web/

The three heating elements I have are 220V-50ohm, 220V-60 ohm and 240V-75ohm if that makes a difference.

Thanks for your time,
Chris

gyro wrote:I can get the PXR4 locally, but I am not sure about the PXG4 however I will investigate tomorrow. Other than the minutes vs seconds, are there any other significant differences?

The PXG4 has many more additional features, but probably the minutes vs seconds issue is the only important one. I've used minutes to profile roasts with a PXR-series controller for years (as have many others here), but if you're going to the trouble to do this, why not give yourself the better degree of control?

gyro wrote:You will need to forgive my ignorance, but does the PID behave differently when triggering a SSR vs a contactor. I understand the longevity issue with rapid switching (and the annoying clicking), but does the frequency of pulses change also, or is this something that you can choose. Just trying to get my head around the 1 second duty cycle, does that mean every second it is pulsed for whatever duration the PID thinks necessary ranging from 0.0000 to 1.0000 seconds?

Sort of. The time interval is up to you, from 1 sec to 150 sec. If you set it for 1 sec, the controller will proportion an on/off duty cycle for each 1 sec interval in order to deliver the proper amount of power. 1 second time proportioning will quickly destroy a mechanical relay; that's why an SSR is recommended when running short intervals (AND, they are quiet).

gyro wrote:Now the next thing that I'm thinking is, how many SSRs would I need? The current contactor is 3 poles and a N/O switch. Now in the above picture, I think I would only need to change the contactor labelled 3 (I switched 3 and 4 for problem solving). To me at least, it appears A1/A2 are the control leads (rightmost leads) which make the 3 left most contacts, with nothing connected to the N/O one.

If you're actually running the roaster on three-phase power, then yes, a three phase SSR would be required. If the roaster is running on single phase power (even though the contactor is a three pole), you could undoubtedly wire it for a single pole SSR (and save money).

Hi Andy,

Thanks for the tip on the PXG. I was thinking of updating my controller, and it would be nice to have 15 or 30 second resolution on occasion.

another_jim wrote:Thanks for the tip on the PXG. I was thinking of updating my controller, and it would be nice to have 15 or 30 second resolution on occasion.

I first got into using a PXG for the espresso profiling pump. As you might imagine, trying to program pressure profiles in minutes for an espresso pump is, um, "non-optimum."