PID Corretto

[robusto]

Re: PID Corretto

Hmm. That WOULD be a good way to utilise the alarm function. Maybe it could also trip a relay to switch on a cooling fan? More digi-tech than reaching over and pushing a button with the finger!

Or, to switch on a tape recorder to play a triumphant flourish. Which would also wake you up.

Ah, the possibilities are boundless.

-Robusto

[Sparky]

Re: PID Corretto



Great to see this finally up and running. I was wondering how the Auber was performing. It actually sounds pretty close to perfect for the job, especially with the ability to store multiple profiles.

Now Ill have to think about taking that next step myself.

Control control.. I want more control....

[JavaB]

Re: PID Corretto

Hi all....

Just an update on the PIDed Corretto....

Whilst it does work very well indeed and is truly "hands free"..... Im not as happy with the results as those obtained using a linear temperature control (moving the heatgun further away.... or my earlier - soon to be current again- high powered triac controller).

After several months of use I have concluded that the espresso produced (well to me anyhow) when using the PIDed roster tastes different.... and have done a few checks.....

The air supply from the heatgun is constant and the PID switches the element on/off every few seconds.... with an even longer off period whilst ramping up slowly to second crack. During the off period the element cools quite quickly (almost no thermal mass to stabilize it) and pumps out relatively cool air.... thus cooling the beans directly under the jet of air..... (When the heater cuts back in, the surface of the bean mass is heated to a much higher temperature before the PID registers the increase also)

The thermocouple.... being buried in the bean mass.... sees this change much later and so reports the temp is still correct whilst the beans on the surface are cooling or being locally overheated...... So you get what appears to be a perfectly smooth increase in temp indicated on the PID.....

But some of the beans are being cooled (those on the surface).... and their temperature can drop by 10 degrees or more before the heater kicks back in....... and that will affect the roasting process and the taste of the resultant espresso - never stall the roast temperature and certainly never let it drop during roasting!

Once aware that this was happening, I tried the thermocouple on the surface....which is a real pain as you need to move it depending on the volume of beans (weight, variety etc).... This provides a somewhat more constant heating of the beans ... but still not ideal.. (and rather impractical)....

So Ive concluded what you need is linear control of the heat output (so the air is always at the same temperature - a ramp and soak PID with linear output (which I dont think exists....)

Failing that a triac control with gerfingerpoken adjustment looks like the best method (or just adjust the height of the gun!)

[reubster]

Re: PID Corretto

Thanks JavaB,

This has "bean" a really interesting post, I was convinced it presented the next advancement in corretto roasting and I was a certain contender to try it out.

Its very generous of you to share that info and Im even more impressed at the humble way you allowed your "inventors-ego" to overulled by your Palate.........a true coffee snobber!


Some food for thought: If you had a computer program that "held" a roast profile and in "real time" sampled the measured bean temp against it, The "raise/lower/do nothing" commands could control your triac via a serial port.
Just to keep you thinking....... http://www.national.com/ms/CN/CN-6.pdf

Cheers
Reuben

[JavaB]

Re: PID Corretto

Some food for thought: If you had a computer program that "held" a roast profile and in "real time" sampled the measured bean temp against it, The "raise/lower/do nothing" commands could control your triac via a serial port.
Just to keep you thinking....... http://www.national.com/ms/CN/CN-6.pdf

Cheers
Reuben

Reuben,

Hi and thanks for the compliment.. :-[ - but as Luca has said - "its whats in the cup that counts...... " and that isnt as good... so back to the drawing board.......

Yep, Ive thought of that and I think it would work very well.... as long as you have phase control of the power rather than on/off only..... it should maintain the HG output within close limits to the desired profile.....

The problem with the PID is that it is (relatively) slow acting..... which isnt a problem with a large and bulky heat source....e.g. a normal element where the thermal mass stabilizes the temperature.... but a thin piece of resistance wire in a HG with air blowing over it looses its heat very quickly and that isnt reflected by the beans - or at least those near the thermocouple.....

So back to the triac control for day to day roasts..... whilst I ponder a better method of automation....

[Bill]

Re: PID Corretto

What if you used 2 heating elements instead of one? One heating element could be on all the time, and the PID just control the other one. That way it would never blow cold air on the beans as one element is always on. Also, the element controlled by the PID would never go cold (from the PID switching it off), because the other heating element is always blowing hot air on to the PID’d element, and would therefore heat up quicker once the PID switched it back on. Does that make sense?

I’ve found with my modified popper for example that I’ve never needed to reduce the power via the triac dimmer to less than 70% at any stage of the roast. Therefore I could have an element supplying 70% power all the time, and use a PID (or other controller) to just control the other 30% power to a separate heat element..

Your thoughts?


Although (as reubster said above) the ultimate setup would be to use a computer program to control the power via a triac dimmer. If anyone here knows how to do this, I’d be very interested.



Bill

[JavaB]

Re: PID Corretto

Yep, I think that would work..... but if heatguns have two elements then one is low power (mine is 60C) and is used as a "dropping resistor" for the fan.... and is of no use....

But if you could get a HG or popper with two elements.... I think that would work well....

However on a similar tack - you could use a triac set at say 70% (or whatever is required to maintain the ramp temperature between cracks) across the SSR output of the PID.... (both in parallel).... so when the PID is on the SSR is closed and you get full power to the HG..... when the PID is off the SSR is open and the triac supplies the 70%(or whatever is required) power to the HG heater....

That way the PID would only control the "marginal" heating.... with the element always getting at least 70% power.... and so never blowing cold air..... might give that a try!

[Bill]

Re: PID Corretto

However on a similar tack - you could use a triac set at say 70% (or whatever is required to maintain the ramp temperature between cracks) across the SSR output of the PID.... (both in parallel).... so when the PID is on the SSR is closed and you get full power to the HG..... when the PID is off the SSR is open and the triac supplies the 70%(or whatever is required) power to the HG heater....

That way the PID would only control the "marginal" heating.... with the element always getting at least 70% power.... and so never blowing cold air..... might give that a try!

Yep, sounds good. Let us know how that works out.


Bill

[Bill]

Re: PID Corretto

…and I think the thermocouple will need to be mounted somewhere where it can react quickly to the changing heat gun temp output. As you said, with it buried in the bean mass it is too slow to react. The cycle time of the PID switching on and off would need to be kept to a fairly small duration to avoid large temp swings. Maybe mounting the thermocouple right on the end of the heat gun???

[Dimal]

Re: PID Corretto

Gday JB,

Do you actually use the Soak function on your controller, other than for the initial dry-out? Other than that, and even thats questionable I would imagine that the most you would want to use of the Ramp/Soak capability would be just a basic multi-tiered Ramp so that the temperature is ALWAYS climbing.

Id also reckon that you would need to use two t/couples, one in the bean bed as you already have to ensure a continuous rising temperature gradient, and another in the heatgun airflow. This would entail the incorporation of a basic cascade control algorithm which Im not sure an Auber controller has. The idea of determining the power output requirement to maintain a minimum continuous rising gradient is a good one though, I believe that would obviate the need for more sophisticated control and complexity thereafter only requiring additional power and control for the Ramp-up to 1st and 2nd Cracks.

I guess a really easy method to accomplish this would be to use a second heatgun (theyre cheap enough) as the base heat source perhaps with your existing Triac controller adjusted to establish a constant Ramp. Of course, a much simpler method all round would be to find a controller that has an Analogue Output (0-10V or 4-20mA) that you could tie into the gain/driver stage of your Triac Controller to enable a fully proportional control of the heatgun output. Have used systems such as this in the past for a myriad of applications which can be made to work very well.

Anyway mate, what ever you decide to do we will all be watching this space with great interest. All the best,

Mal.

[JavaB]

Re: PID Corretto

Hi Mal,

Yep, the soak function is only used to hold the bean mass at 150C to dry them out.... the rest of the profile is a selection of ramps (at different rates depending on the stage of the roast).

The problem appears to be, even though the minimum duty cycle is 1 sec on 1 sec off... it tends to cycle more slowly.... no doubt due to the thermal inertia of the bean mass.... it maintains the core temperature quite well but the hot air temperature (and therefore the temperature at the surface of the bean mass) fluctuates quite a bit - caused by the rapid change in heatgun output...

Nope the Auber is a pretty basic ramp/soak controller which is really designed to control something with long time constants... e.g. a kiln.... it does a pretty good job if a thermocouple is in the direct air stream.... but even then struggles a bit with the very high rate of change....

A PID with an Analogue output would certainly be far better.... as small changes in drive would also make small changes in output temperature... rather than all or nothing every few seconds.... and during the slow ramp between FC and SC it is probably on for three or 4 seconds.... then off for even longer..... as it senses the temperature increasing too quickly.... by which time it has already overshot on the surface...... even though the core temperature is stable...... and cooling down (which is the bigger problem) is even worse......

Ill give the dual control a go.... as I have both on hand.... so that there is a very slow net ramp up (as required FC to SC).... and the PID can accelerate that up to FC (by bypassing the triac during that phase) and supplementing the triac after FC if / when required.....

[Dimal]

Re: PID Corretto

Nope the Auber is a pretty basic ramp/soak controller which is really designed to control something with long time constants... e.g. a kiln.... it does a pretty good job if a thermocouple is in the direct air stream.... but even then struggles a bit with the very high rate of change....

Hmmm,

I hope Im not trying to teach you to suck eggs JB but have you tried running the controller with Derivative disabled or significantly reduced. Sounds like there could be a severe reaction to every shift away from Setpoint, the opposite of what you really want in this situation.

Cheers mate,
Mal.

[JavaB]

Re: PID Corretto

Hi Mal,

Nope, no sucking eggs - any advice is good advice.....

Yeah, tried changing the D value (cant disable it) but I think the PID is doing the best it can..... in a situation where it is getting feedback by "Chinese whispers"...... it doesnt really know what is going on....

With just a bare bowl.... and the thermocouple making direct contact with the bowl.... it works great both with "Auto Tune" and manually setting the PID values..... the bowl has a reasonable thermal mass.... and integrates the temperature changes.... and the effect of changes to the heat output from the gun is both consistent and direct....

But add beans.....and the whole situation changes..... these shield the thermocouple from direct heat....and the ones being heated dont directly contact the probe until they have rotated around the bowl and been "buried" in the mass of beans.... which could take several seconds..... meanwhile all the beans passing directly under the HG O/P are getting hit with ???C air...... finally the beans at the correct temperature hit the probe.... the PID says enough is enough... (or predicts that is to be the case) and cuts of the heat....

The beans that were way too hot (but have now cooled somewhat) continue to pass the probe for several seconds whilst those under the HG O/P are being cooled..... and this continues until, several seconds later, the cooler beans have been buried and contact the probe...... and during this time the hotter and cooler than desired beans are being mixed by the paddle.... with the probe seeing the average.

So we have a heat source responding with massive changes in O/P air temp almost instantly..... and a feedback loop which takes seconds for the effect to be felt by the PID.....

Auto Tune in that situation is a dogs breakfast - a total disaster!!! But with manual tuning you can get very stable temperature right at the point of measurement..... but right under the gun O/P is a different situation.... which improves quite a bit if the probe is at the surface of the bean mass... it becomes almost ideal.... except the bean mass grows during roasting......

If the probe is above the bean mass the temperature rises quite rapidly.... e.g. 200C in the beans at the top of the mass might be 30C hotter 1cm closer to the gun O/P....... (I assume due to mixing of the hot air and cooler air in the bowl).....

So to me the ideal adjustment is a slow and linear change to the HG output.... with changes more closely matching the time it takes for the change to be properly detected and the appropriate feedback applied....

Just like I do with the triac controller..... make a slight change..... wait several seconds to see the effect of the change and then either increase a bit more or reduce the previous increase a tad..... effectively predicting the change and correcting the output on say a 10 second integration period....

The PID can do the integration fine.... it just cant provide the proportional output control I need

P.S. Ive just tried to measure the rate of change of HG O/P from on to off.... and the rate is too fast for a thermocouple - either probe or bead - to measure...... the rate is almost exactly the same as taking the probe out of the air stream and allowing it to cool in ambient air..... but in any case - it is pretty quick

[Dimal]

Re: PID Corretto

The PID can do the integration fine.... it just cant provide the proportional output control I need

P.S. Ive just tried to measure the rate of change of HG O/P from on to off.... and the rate is too fast for a thermocouple - either probe or bead - to measure...... the rate is almost exactly the same as taking the probe out of the air stream and allowing it to cool in ambient air..... but in any case - it is pretty quick

Hmmm,

A bit of a bugger alright JB. Looks as though an analogue output device is the way to go for fully auto control. Most commercial roasters use analogue output in either 4-20mA or 0-10V, but for proportional valve position control, not because of the factors affecting the Corretto setup.

Cheers mate,
Mal.

[Lizzie]

Re: PID Corretto

right... now i know why i decided not to PID the Corretto .... i dont speak the language!!! ;D ;D ;D

L