Cheap PID temp controllers any good?

[trevwhite]

Hi. Seen lots of PID temp controllers on Ebay and Amazon like this one.

https://www.amazon.co.uk/gp/aw/d/B07HJ5GQ2L/ref=ox_sc_act_image_1?smid=A2QDVE2OGM9E0I&psc=1

Anybody used such cheap controllers? Industrial grade appear to be x10 the cost. Wondering what you pay the extra for. Are the cheap ones unreliable? Inaccurate? Rubbish?

[jonroger]

They worked fine for me on a couple of applications.   But note that basic PID is an archaic algorithm from analog days and for critical applications, one can always do better.

[Kleinstein]

PID is quite basic, but it is still a very good solution for a general purpose regulator.  Already setting the 3 parameters for a PID regulator right can be demanding.  Other regulators that claim to be better tend to use even more parameters and thus an even more critical setup.

Those regulators a build in quite some quantity and are thus relatively cheap - just like simple DMMs. The cheap ones are not necessary bad - just a certain risk they cut one corner too much. For a not so critical case they can be OK.

[SeanB]

They do work, just do not trust them in extreme environments, and with operating close to the nameplate ratings. Output wise derate a lot, and do not regard the mounted unit in the panel as being in any way water resistant, as it is not. but provided you use it well within the sensor operating range, keep the actual controller cool in the panel, as in not in an unventilated panel with heat generating electrics mounted under it to gently toast it, and regard the output ratings for the relays as being 4 times what they actually can handle reliably long term. Other than that they do tend to autotune themselves with reasonable accuracy, at least for heating elements with large thermal inertia.

Just be aware the particular unit you refer to has no cold junction compensation, other than an offset you program in, you might want to get a different one with cold junction compensation if you want to control things under 600c where a drift of around 30C from a cold day to a hot one will cause problems. Plenty of them though with cold junction compensation, and if your process is under 150C then forget thermocouples and go for Pt100 or Pt1000 sensors instead, the thin film elements are quite cheap, reasonably rugged and a lot more accurate over that -100C to +150C range than any thermocouple, and a lot easier to connect as well with less thermal EMF issues.

[trevwhite]

Thanks for the replies. Did'nt notice the no cold junction compensation, that is important. I am working with temperatures of 250C to 300C.

Trev

[coppercone2]

They do work, just do not trust them in extreme environments, and with operating close to the nameplate ratings. Output wise derate a lot, and do not regard the mounted unit in the panel as being in any way water resistant, as it is not. but provided you use it well within the sensor operating range, keep the actual controller cool in the panel, as in not in an unventilated panel with heat generating electrics mounted under it to gently toast it, and regard the output ratings for the relays as being 4 times what they actually can handle reliably long term. Other than that they do tend to autotune themselves with reasonable accuracy, at least for heating elements with large thermal inertia.

Just be aware the particular unit you refer to has no cold junction compensation, other than an offset you program in, you might want to get a different one with cold junction compensation if you want to control things under 600c where a drift of around 30C from a cold day to a hot one will cause problems. Plenty of them though with cold junction compensation, and if your process is under 150C then forget thermocouples and go for Pt100 or Pt1000 sensors instead, the thin film elements are quite cheap, reasonably rugged and a lot more accurate over that -100C to +150C range than any thermocouple, and a lot easier to connect as well with less thermal EMF issues.

the post is informative but the way it reads is pretty funny

Almost like the side effects in a medication commercial.. I almost wanna hear someone read that out loud at 200% speed as a 'china disclaimer'



Also, maybe customs should print it out in tiny font and make a special sticker next to CE

derate 400%. thats a good one.

well, you can use a standard #2 pencil as a power resistor... .... .... also, be sure to chill the water you are dipping... avoid gas station ice... and do not use for voltages in excess of.... along the lines of ventilation... drift issues...

[ivaylo]

...note that basic PID is an archaic algorithm from analog days and for critical applications, one can always do better

PID is quite basic...

What are names of contemporary analog or digital algorithms which are preferred to PID nowadays? In the 80s all I learned was PID, the little I did with this in the 90s was also only PID, so these posts got me curious where things went since. I am googling right now, but in case kind folks had easy info to share...

[SeanB]

PID works well enough for process controls with limited change in operating conditions, like a heater with appreciable mass, heated to a reasonably large difference from ambient, and with little change other than a step change with you placing a load in there. Things with low thermal mass, like IR heaters, call for different control, especially if you want a very precise control of say heating a PCB in a IR reflow oven, using a small probe placed on the board, but that is a whole other kettle of fish. they tend to fall in the corner of "fuzzy logic" which is often some variant of PID with autotune enabled, adjusting the PID settings on the fly in near real time, instead of having them semi fixed and only adjusted during set up.

[jonroger]

The algorithms I've written to outperform PID characterize the system on an ongoing basis and make use of this data for the future.  Usually dynamically maintained tables,  lookups and little or no P, I or D.  Agreed, for most applications it doesn't matter.

What's the fastest way to safely get a car from point A to a brick wall?   It's a very non-PID like immediate full throttle to a certain point and then full brakes, stopping just before the wall.  The trick is knowing what conditions need which certain points.