Too Much Coffee

I design power stations (my wife sells the coffee) for a living - almost any PID controller will do the job. The difference in the outputs are this; most cheap PID controllers have a small relay (switch) as the output which is then used to drive a Solid State Relay (SSR) which is external to the machine. No problem with this the relay will last for millions of cycles providing it is driving a very small load like an SSR. Some more expensive PID's are fitted with a VERY small SSR output (sometimes called a transistor output) these will last indefinately, but you can blow them up very easily if you short circuit them. By the way - if you want cheap - then you can use a secondary mechanical power relay insted of an SSR, really the only difference is that you will hear the relay clicking in and out continously which is a bit annoying.

Almost any SSR will do, it does not matter if it is too big, it does matter if it is too small. RS components have a reasonable range. Pretty much all machine could be run from a 10A SSR, but 25A is OK, only difference is the price and the physical size. Cooling of SSR's should not be a problem as they are not switched on most of the time - even if you are drawing water/steam pretty continuously.

Last thing to consider is the PID parameters, most PID's have an autotune facility and you should try this first. If Autotune does not give the results you want (a nd it need time to learn so give it a chance) then adjust the P (proportional) control only until you get a fairly steady temperature - then you are likely to find that although the temperature is steady it is not what you ask for. This offset can be eliminated by increasing the I (Integral) control by a few seconds. Now you should have absolutely steady temperature and at the set point. BUT when you pull a shot the temperature will vary, so next (and this is often very tricky) set the D (Differential) control to repsond to rates of change of temperature. I expect you will need a very low setting for D otherwise the temperature will oscillate in a nice Sine wave. If in trouble do not use the D control at all and rely just on P & I.

Not sure how this would work on a heat exchange machine like the ISOMAC TEA I have and I think I will wait for the warranty to expire before I try!! But basically this is bound to be better than thermostat controls.

Anyone needing help setting up or wiring PID's I can help.

Ad.

PID's Part 2

You need to decide what temperature sensors you use. Most PID's accept;

Type J - range is 95C to 760C so not very useful but will work.
Type K - range 95C to 1260C - so even less useful.
Type N - range 95C to 1300C - as above
Type T - range -200C to +400C - OK but difficult to get hold of.
Thermocouples will work outside these ranges but loos accuracy.

It is better to use a resistance sensor (RTD) like;

PT100 - range -150C to 400C depending on manufacturer- which is much more useful as the desired temperature is in the middle of the range.
PT1000 - range as above.

Make sure to get a waterproof packaged version of any sensor you use.

Ad.

Interesting. There are a lot of people who use thermocouples when PID espresso machines. In fact I can't remember anybody talking about RTD's.

Could you explain a little why RTD's would be better (or worse) when used to control:
-water temperature in a espressomachine?
-the hot air used to roast green beans in a popcorn popper?

Its fairly simple. There are two things to consdier.

1. Most common TC's have a range that is far to large for the temperatures that we are interested in. So if the PID controller has a low resoultion Analogue to Digital converter (an most cheap one have 10bit) then they will not be able to resolve a change of temerpature of less than about 1C with most TC's.

2. TC's rely on measuring the difference between the current generated by the TC and a reference current (called 'cold junction compensation') which is generated by the electronics of the PID controller. The closer the TC's temperature is to the Cold Junction temperature the higher the noise as a proportion of signal. So TC's are good at measuring high temperatures but no so good at low temps such as boiling water.

If you use an RTD firstly the range is usually lower, so the resolution per degree C is higher, secondly we are typically operating in the middle of the range of the device, and thirdly there is no Cold Junction compensation required. So RTD's are much better for middle range temperature such as heating water.

Of course none of this means you can't use a TC but if you have the choice I would go for the RTD, you can get cheap RTDS for central heating systems very easily.

I am not sure what the temperature of air for roasting is - I guess around 200C to 250C. In this case a TC is likely to be better but only because it is easier to get a TC with a seriously heat resistant cable.

Hope that helps

Ad.

What about response time? Any difference between the two?

Not really - the thermal mass of the boiler is much larger than that of either a TC or RTD. But for bext reuslts make sure that the sensor is fitted with a good contact to the boiler surface, and at a point that really measures the water temperature, rather than the heater element temperature. Also don't get a sensor that is very bulky.

Ad.

aloening wrote:PID's Part 2

You need to decide what temperature sensors you use. Most PID's accept;

Type J - range is 95C to 760C so not very useful but will work.
Type K - range 95C to 1260C - so even less useful.
Type N - range 95C to 1300C - as above
Type T - range -200C to +400C - OK but difficult to get hold of.
Thermocouples will work outside these ranges but loos accuracy.

Interesting, my thermocouple is claimed to work down to -40C. It would be interesting to try an RTD and see if there is an improvement.

The idea is actually to use a TC or RTD that will be in contact with the water, not the surface of the boiler, so is there a difference in responce time and does it matter?

TC's will work both above and below the cold junction temperature, what is being measured is the difference in the current generated. So -40C can be measured. This just depends on the TC type and to some extent the package it is in.

The better contact with the temperature you want to measure the better the control will be - but does that mean you have to mod the boiler to allow the sensor to be submerged?. I have not looked at the boiler on my Isomac (because it is still under warranty) but I doubt it is fitted with a port for a temperature sensor.

Ad.

Maybe mod the boiler, I don't know yet. I have two similiar boilers but different generations and I haven't decided yet which one to PID. I know of one person who have PID an older generation of boiler and I'm waiting for photos.

Ops, forgot to login. I'm the guest above....

I suspect that you don't need to mod the boiler. Just find a good location, near the top of the boiler and attach the sensor firmly (or even with some heat transfer paste) a cover with a little insulation. The heat transfer between the water and the wall of the boiler will be pretty good so the sensor shoudl respond quickly.

Try this first - will be better that having a leaking boiler!.

Ad.

The idea is to use a hole that already exist on the older generation of boiler.