In my project I already have all the circuits for a PWM driving some high power MOSFET. Now the hardware guys are coming with this AC heater, and I've been breaking my head around Triacs and phase controllers but that will be an extra design time that we don't have available.

I wonder if I get the AC through a bridge rectifier (240V, 60Hz) and then through my PWMed MOSFET controlled at 1Khz. I know the final wave won't be pretty to look at in an oscilloscope, but it's just a heater (a resistance), it shouldn't matter if the current is really going or not to both sides.

Does anyone see any problems in doing this setup?

Edit: The PWMs already working in the project through high power MOSFETsare isolate through OpAmps that keeps the proper gate charge levels on the MOSFETs. Those already work nicely controlling a compressor and a pump. The difference from those and the one I'm asking about is only a proper stable DC or a mirrored sine AC between drain and source.

Below is the circuity that is already working for the pump and compressor and the main idea I was trying to re-use it a bit for the heaters. I mean, just because the heater manufacturer said the heater is 240Vac it doesn't meant I can put 240Vdc (full rectified but no capacitor) on it. enter image description here

enter image description here

  • 2
    \$\begingroup\$ You can't provide electrical isolation through opamps. Can you show us the schematic? \$\endgroup\$ – stevenvh Aug 2 '12 at 11:11
  • \$\begingroup\$ agree, the term 'isolated' was used too loosely. It isolates from the MOSFET gate sinking or generating over-current on the MCU pins. Maybe the term 'protected'. We're in R&D stage and current implementation is all in veroboards and an proper schematic does not exist yet, but I'll put something together and edit the post in a bit. \$\endgroup\$ – Budius Aug 2 '12 at 11:24
  • \$\begingroup\$ Use my full bridge drive to get AC on the load. Note need for floating supply for your electronics. A circuit that uses half wave DC for large loads is not acceptable commercially - this results in a net DC component in transformer cores, which they (and the regulatory authorities) do not like. \$\endgroup\$ – Russell McMahon Aug 2 '12 at 11:53
  • \$\begingroup\$ What's the heater's power? \$\endgroup\$ – stevenvh Aug 2 '12 at 12:04
  • \$\begingroup\$ 350W. It's 7x 50W heater arranged around a tube. \$\endgroup\$ – Budius Aug 2 '12 at 12:06

You haven't talked about production volumes and cost-optimization, but a solid state relay (SSR) would give you isolation and phase control in a single package, though at a higher recurring cost than a purpose-built solution.

Many SSRs are available with logic-level PWM input. This link might provide some good reading, though there are many manufacturers of similar products.

  • 1
    \$\begingroup\$ After discussions here with the electronics designer and the process guy, we just ditched the in-house solution and will buy a CRYDOM proportional SSR and control it with a 0-5V signal. It will cost £63 per-unit, but it's not a huge production volume, so it will not be a problem. Thanks. \$\endgroup\$ – Budius Aug 7 '12 at 10:22

One possibly unacceptable effect is that the control circuit is directly connected to the mains this way, and that touching any part of it may be lethal. Proper isolation helps, but it's cheaper to restrict the mains voltage to a part of the circuit.

For a heater you don't need the 1 kHz PWM; its inertia is high enough that you can do it at much lower frequency. I would suggest to use an opto-triac instead of the FET. The triac will have a larger voltage drop than a FET, but for the FET you would have two diode drops too, so that won't make much difference. Also the switching of the FET will cause switching losses as the FET goes through its active region 2000 times per second.

enter image description here

A MOC3041 has a zero-crossing detection circuit, which will give you each time half cycles on or off. This reduces peak currents in the heater and EMI. Instead of varying the pulses' duty cycle you can now control the heater by varying the ratio between half cycles on and half cycles off.

edit re target precision mentioned in comment
Milli-degrees precision is impossible, but not because the on-off control would be too coarse. Suppose you have a vessel with 1 kg of water. (You mention heating around a tube, which may suggest a streaming fluid, but then millīdegrees are even impossibler.) Heat capacity of 1 kg of water is 4.2 kJ/K, so 4.2 J will "heat" it by 0.001 °C. Switching on the 350 W heater for half a 60 Hz cycle gives 2.9 J, good for 0.0007 °C. If you would control each of the 5 heaters individually you can even add per 0.00014 °C, or 14 % of the required precision.

So the coarse on-off control isn't the problem. It's distributing the heat evenly in the fluid, not to mention measuring the temperature. How can you get a to 1 milli-degree even temperature in a vessel (let alone a tube with heating wrapped around it)? Even a magnetic stirrer can't guarantee that. (Note that a magnetic stirrer will also raise the temperature!) And at 60 °C 1 milli-degree is 3 ppm, and to control it to that precision you'll have to measure it to better than 1 ppm. How are you going to do that?

Phase control may give you a theoretically higher resolution, but there's no way to measure its effect, and without the required feedback you can't get the control precision either.

  • \$\begingroup\$ hi @stevenvh, I appreciate your comment and concern. I just edited and expanded a bit more the post. It's not a opto-isolation but there's separation between MCU and high currents. I'm aware that a heater inertia won't see much difference on 1Khz, but there are 3 reasons for it: 1) all the other PWM are already running at 1Khz; 2) there's some huge studies, analysis and tests here in the company to make better heaters (but that's with the mech-engs). 3) I'm trying to reduce my engineering time by not having to create a phase angle controller in my circuit. \$\endgroup\$ – Budius Aug 2 '12 at 11:16
  • 1
    \$\begingroup\$ @Budius - about your third point: there's no talk about phase control, like a dimmer would do. It's just on for a number of half cycles, and then off for a number of half cycles. If there's not opto-isolation then how is it done? Transformers? Can you post a schematic? \$\endgroup\$ – stevenvh Aug 2 '12 at 11:25
  • \$\begingroup\$ hi @stevenvh schematics available. I'm really just trying to re-use the circuit that is already there in the heater. \$\endgroup\$ – Budius Aug 2 '12 at 11:42
  • \$\begingroup\$ @Budius - OK, so you're boosting the uC's output voltage to get enough gate voltage. But I need to see more. "Motor/heater" is that connected to 240 V AC? If so the uC is directly connected to the mains. This may be OK, I don't know what the application is, but usually you'll want the low-voltage side isolated. \$\endgroup\$ – stevenvh Aug 2 '12 at 11:48
  • \$\begingroup\$ What is currently running on the project is normal 24Vdc motors (one for a pump and one for a compressor). This 24Vdc comes from a higher current supply, but as we managed to get a smaller compressor it will probably be the same supply as the MCU board supply in the final thing. And for the heater, I was thinking about connecting it from the 240Vac, through a bridge rectifier on the same MOSFET. I understand what you're saying on the "low-voltage side isolated" and I'm just wondering that if 1 thing goes wrong on our machine they'll have to stop anything anyway. But still pondering. \$\endgroup\$ – Budius Aug 2 '12 at 11:57

You can use a bridge rectifier to allow the use of a DC switch to control an AC powered in two ways.

  • The switch and the load are inside the bridge.
    The switch and the load see PWM chopped full wave rectified DC

  • The load is in the AC line to the bridge and the switch is inside the bridge.
    The load sees PWM chopped AC.
    The switch sees PWM chopped full wave rectified DC.

At 1 kHz and with a resistive load the lack of synchronisation of the PWM frames with the mains should be acceptable.

If you drive a bridge the switch inside the bridge "floats" and needs to get its drive voltages from somewhere. IF your power supply is isolated at output then the whole circuity can float but is essentially at mains. BUT you can put FET + bridge on the mains side, drive via a non-zero-crossing opto or pulse transformer or whatever and have rest at whatever voltage you wish. You still need to get the FET gate drive but a resistive feed from mains will be OK - half wave diode charges FET gate drive cap when mains polarity is correct.

As ever - IF there is not formal isolation between your main circuitry and the mains switch then ALL your circuitry should be considered to be at mains voltage at all times. This applies even if you consider the circuit SHOULD be on the "cold" side of mains at all times.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.