This discussion is locked.
You cannot post a reply to this discussion. If you have a question start a new discussion

Best method of controlling a resistive heater

Hey all, 

I'm creating a resistive heating system (I don't have a choice on that, the resistors are already provided), and I'm looking for the best method of being able to control this. 

Resistors are 50W 4 Ohm Panel Mount Units, of which I've got about 32, but I'll likely need more going forward. 

Currently it uses 4 zones, each of which use a custom MOSFET PCB which is PWM pulsed by an Arduino which is all fed from 4 24V DC power supplies. 

This has worked OK thus far, but it's not the most reliable and I was wondering if there's something COTS which would suit this better that provides a better method of control to be able to vary the temperature of the panel, as the current ones have no sense feedback, so they're unable to temperature compensate. 

I was wondering if an inverter could sort something like this? 

I know I'm limited by the supply of the room (which is just 230V, there's no 415V supply etc.), but I'm curious to see what people recommend in terms of whether to look at motor controllers as off the shelf PWM units, current stabilised sources etc. 

  • So the heaters take ~ 15V   and a touch under 4 A  each to generate 50W? Do they need to be in parallel, or is the standard of insulation such that each of your 4 banks or 8 could be wired as  8 in series  to make a 120 Volt AC driven section ? or you could run the heaters on the secondary of a transformer and switch the primary side with a normal thermostat, but I'm not sure how accurate you need to be.

    How hot is it getting, and what level of control do you need - nearest ten degrees, nearest degree, nearest hundred milli-Kelvin?

    There are standard thermistor units that are intended for underfloor heating, and there are controls for laboratory ovens, but none of that that may be really what you need. There are plenty of industrial PLCs that can be made to respond to a thermocouple and open or shut contacts at a preset level, but  that is probably at least as hard, if not harder to customize than sorting the Arduino.

    What aspect is unreliable about the current set-up, and how much protection is needed to guard against or raise alarms for under or over temperature - what sort of thing  is at stake ?

    It may be trivial or very hard depending on the answers.


  • Hey Mike, 

    I've got freedom to wire them however. I opted for screw terminal versions to avoid having permanent soldered connections. 

    I'm aiming for a maximum temp of around 80 degrees. 

    Control wise, anything within a few degrees would be better than what we have now. 

    The existing system was set up by a predecessor, who developed a custom MOSFET PCB which I believe is rather overkill, as the MOSFET units are capable of around 60A per device, and there's 4 units being used for the 4 zones we have. 

    They're being mounted to create thermal shrouds if it helps give a little more context, under vacuum. 

    The existing MOSFET PCBs were designed to use a PWM feedback to check what's being sent from the arduino is actually being seen by the switching device, but this never worked, so was never implemented as things currently sit. 

    The previous revisions of these MOSFET PCBs were just an off the shelf item from amazon, badged as capable of 30A, but the board wasn't designed to dissipate even a little amount of heat, so we had a few burn out fairly quickly.

    I'd been looking into solid state relays and using an Arduino as a PID controller. 

    I'm just struggling with how to lay this out to be honest, I've got capability of putting a max of 16 resistors per panel, so I was going to wire each panel separately to give zoned control, but it's a case of what supply voltage, do I go with a 230V a.c. solid state relay, or a 48, 24, 12V equivalent and use associated power supplies to try and manage the power distribution better. 

    We're using Arcol HS50 4 Ohm resistors currently FYI, looking at the datasheet for them them shows I'm only going to be getting around 30w of power dissipation based on an ambient of 80 deg C.

  • What's the application? Is it something like process heating that might have a pretty constant power requirement or space heating that could vary between next to nothing in summer to flat out in winter? Or something that might have to respond far more quickly?

    I can think of a whole raft of possibilities - from organising the heaters into banks, each bank with a simple individual thermostat set slightly differently from its neighbour, but organised so that the further below target temperature things are the more banks are switched on - to a proper PID controller (e.g. something along the lines of

       - Andy.

  • Hey Andy, it's a thermal shroud within a vacuum chamber, we're given a specification for what the temperature ramp up should be over a number of hrs/days, then how long that temperature should be held for, before following a ramp down profile. 

    I was thinking PID controller myself, we've got a number of arduino boards sat around which should be able to do that pretty easily, it's the routing of the resistors I'm struggling with. 

    I was considering solid state relays and using mains voltage on them (230V) but over a single resistor that's like 13kw, and it only drops to what a mains socket can kick out at around 4-5 series resistors. 

    I didn't know whether this was the right way to do it or whether it's worth having multiple combinations of series and parallel strings for dual redundancy sake, and the question is open on whether to use 230V or something like a 24V dc supply. 

    There's 4 x 24V dc supplies down there at the minute from the old setup but we have a raft more resistors now, 16 positions per panel we have provision for and 4 panels total. 

  • So what I'm thinking is this. 

    Taking a single panel, I have 16 places for resistors. 

    If I wire up a string of 8 in series I have 32 Ohm series resistance, which on a 230V supply equates into 7.2A approx (1.66kW).

    This allows me to wire up an entire string of 8 and have a backup string if it fails during an operation. 

    I could then run this on a 20-30A solid state relay and control this with the digital pins of an arduino, giving precise on off control, I'd just then need to integrate some temperature sensors to give it a PID loop. 

    How's that sound?

    My next question is going to be how much can I put on a single ring before it blows! as the 4 panels would draw 28.8A ( 6.6kW)

  • I would suggest a commercial temperature controller/programmer. This allows you to programme the required ramp rates dwell times etc.

    3500 Advanced Temperature Controller and Programmer | Eurotherm by Schneider Electric  is an example.

    Do you know from previous work approximately what power you need? Being in a vacuum chamber should reduce heat losses (or have I missunderstood?) This should give you some idea of voltage and current requirements.

    Eurotherm also supply matching power controllers which can include load monitoring and alarming.

    An otherwise unloaded ring will supply 30A, although to get this from one socket you will need a plug with two neutral pins (tounge in cheek).

  • Thanks Roger for the info, I'll check out the controller now. 

    Unfortunately no, my predecessor never bothered within defining that. 

    Yeah the vacuum should reduce heat loss. 

    I just recall that the last time we had this working, it required someone to go and manually program each of the 4 panels with a specific PWM, which then made it a bit of a dumb system, as it wouldn't temperature compensate or go to a setpoint, it was just at that frequency until switched off. 

    I want to do this properly and have it setpoint controlled, with the ability to customise the ramp, hold and fall times, and ideally be able to monitor it remotely as we're looking to have this all eventually output into a digital dashboard of the area. 

    I wanted to initially do this all via PLC but it seems getting hold of PLC components at present are challenging, with extensive lead times. 

  • The Eurotherm controllers have a communication option, It was RS232 last time I worked with them I guess it will be ethernet now.

    Lots of components are a problem at the moment. We have two machine projects with 6 month delays waiting for Siemens inverters Frowning2

  • Arcol HS50   resistors are very good as heaters at well beyond the nominal rating for de-rating. I have used them in the past as the basis for a soldering press to attach PCBs to brass plate.

    I'd only consider them single insulated though and I'd be  wary of mains derived voltage on the element without considering how ADS would behave with a mid-element to case short.

    They will have a heating time constant of seconds to minutes so 'PWM' is probably a bit optimistic at anything much beyond the tens of seconds. By the time they are screwed to something solid, even less nimble - which is good, a slow loop is easy to organise - if you want PWM at all it can be software think 'burn for  x out of ten seconds, check temp and repeat' .

    Given what you have, and the fact it sounds more like a lab-like set-up that may change, I'd have a quick play with  ADC input to the arduinos and consider thermistor as being a lot more robust for EMC than thermo couples if you have pumps, valves and things clattering away. Do you also have a need for plasma in the vacuum firing up as well ? EMC can be a many headed monster in some settings.

    A typical  ten K at 25C device and a fixed 2.2k resistor will give you a pretty linear  voltage division from  50-100C the funny curve of the thermistor and the A/A+B voltage division more or less linearise each other, and after few hundred nF  of filter capacitance you should have a voltage that you can reda on the arduino and make on-off or raise lower the duty cycle decisions on. (and with the arduino drive a temperature display as well or at least 'up down ready' LEDs)

    I'd also add a thermal fuse somewhere so that if it ever goes to run-away something drops the mains, or if there are too many amps, then drops a contactor that drops the mains.  Maybe not this exact model but something like it.

    IF you really do need that many KW, how is the facility cooled ?  I'd not like to share a room with a 3kW fan heater in summer let alone 6kW of static heat.  Less heater and some lagging to keep it in may be preferred though  if you have to cool in a specified time too, then you may need fans or even cooling water.


  • Yeah it's really bad at the minute supply wise! Will look into the eurotherm!