3 x 3kW resisitve heater modulation from 3ph AC?

[max_torque]

I need to modulate 3 resistive heater elements, taking power from a 3ph AC supply (400Vrms).

Options are to use a phase controlled set of Triacs at 50Hz, or to rectify the supply to DC, and use a set of Mosfets to PWM the heaters?

Given that i don't need extremely fine power control, and can "correct" for the non linear power draw vs "on time" for the triac phase triggering, does the first option make more sense?   I'm more comfortable with the second approach, having never designed a Triac based power controller, but a quick scoot round 't net suggests it isn't too hard to do?  (and will probably be cheaper in terms of parts)

[CopperCone]

3kw is alot. High end shit will actually use regulated dc to control heaters. I have some big 3 phase supplies like thiz. They can do short circuit detection, send alarm signals and control slew rate.. Good for things like molybdnum heating elements which need careful control.

Well regulated dc power like this is always nicer to work with.

Triac control wil be alot cheaper though. And more noisy. And most likely it will be a pain if your heating element is highly nonlinear

[Benta]

3 kW isn't much, around 13 A at 230 V. Piece of cake for a set of triacs. As you say. non-linearities can be fixed in the gate drive system. I wouldn't expect a heater to be significantly non-linear (like an incandescent bulb is). Sine wave needs to be compensated, of course.

Yes, you'll need a set of LC filters, but you'll need that even more for a PWM system.

[dmills]

If you are putting out 9kW of heat, odds are there is some reasonably serious thermal mass in play?

If so, forgo phase control and go for zero crossing switching with PWM in multiples of whole cycles, little need for filtering, and very, very standard for heating loads.

regards, Dan.

[oldway]

If you are putting out 9kW of heat, odds are there is some reasonably serious thermal mass in play?

If so, forgo phase control and go for zero crossing switching with PWM in multiples of whole cycles, little need for filtering, and very, very standard for heating loads.

regards, Dan.

Indeed, see "operating modes - Full wave switch TAKT" page 15 of the technical documentation of Thyro-P.

https://www.advanced-energy.com/Thyro-P-OM-EN

[max_torque]

Heaters heat a couple of litres of water, in a dynamically controlled heating and cooling loop.  I suspect that phase angle control would not be needed and i could just run with cycle based control as mentioned.

What is the typical upstream (AC supply side) filtering used? Presumably a reasonably large L followed by some C to try to get a decent power factor?

[oldway]

With full wave switch TAKT and resistive load, power factor is 1 and there is no need of filtering.....

[Photon939]

I agree with the whole cycle control unless you need super precise heating.

I work in HVAC controls and I have installed some 3 phase 480vac electric duct heater controllers.

The controller was a beefy heatsink with two solid state relays joined in the middle by a PCB with a PIC on it. Give it power and a 0-10vdc signal and it pulses the SSRs with PWM at around a 1Hz cycle time.

[max_torque]

Would a single hardware zero cross detector  for say phase 1 to N, and a 120deg/240deg software delay be sufficient for phases 2 & 3, in order to coherently lock the Triac trigger to each phase?

[james_s]

Why reinvent the wheel? Just get some off the shelf SSR's that have the ZC switching built in, then drive them all with a PID controller. All that stuff is cheap and readily available now, just beware the cheap Chinese SSRs which most often are nowhere near the specs printed on them.

[max_torque]

Because this project is to remove cost from an existing device that does just that, drives 4 x 10A AC loads with SSRs.  But a decent SSR costs around £25, compared to around £3 for a descrete triac solution, and that switch can go directly onto the control pcb, meaning less wiring / build time etc.  The existing solution is a hacked together mess of wiring, controllers, relays, contactors etc, the idea is to replace the whole thing with one pcb!

[Zero999]

Just use three zero-crossing opto-couplers, such as the MOC3163 and TRIACs or SCRs: simple and cost-effective.
http://pdf.datasheetcatalog.com/datasheet/fairchild/MOC3162-M.pdf

[james_s]

Is this something that's going to be built in large quantities? Otherwise I can't see there being much savings in replacing £100 worth of hardware (3x SSR and a PID controller) when you factor in the engineering effort.

Certainly you could build your own 3 channel SSR if you wanted, but there's no reason an off the shelf solution has to be a convoluted mess. You can have a very tidy setup with 3 SSR's mounted together on a suitable heatsink with a PID controller on the front panel, the whole thing could fit in a fairly compact enclosure with power in/out and a thermocouple lead.