A triac power control circuit for a transformer.

[dr pepper]

I'm putting together a 1.6v 30a power supply for my wifes pyrography pen as I dont want to pay the silly 170 notes for a new one off the shelf.

I need a triac phase angle circuit, aka light dimmer, but with the o/p going to a transformer (240 ac primary) I need something that is reasonably symetrical, as if it isnt the resulting dc is going to fry my trans, which is allready working hard.
I'd like to do this with a simple circuit using discretes, instead of a micro.

There was something on this forum a while back, cant find it now.

 

[schmitt trigger]

I need something that is reasonably symetrical,

My wife also does pyrography, and from having myself worked for a transformer manufacturer, I can tell you that the positive and negative volt-time product has to be perfectly symmetrical under all load and line conditions, otherwise the transformer will suffer from staircase saturation.

The way I solved it, instead of phase control, is with integral cycle control. This is the best way to do it for heating loads, as the thermal constant averages the power delivered to the heating element.
It also reduces EMI.

Are you familiar with the principles of integral cycle control?

 

[crutschow]

A very large capacitor in series with the primary will block any DC current.

But Schmitt's idea of an integral cycle controller is likely the best. It may be better to place that at the transformer output rather than input, to minimize the transformer input surge current.

 

[dr pepper]

I dont work for a transformer manufact, but I get involved with industrial controls, what you mention I think is what in industry is burst mode power control, where you switch a full cycle at a time, and omit cycles to reduce power, this I had thought of, but for me I'd have to use a pic to do that.

A cap might do but it'd need to be big, the secondary current is around 50a, so that would need some need hefty devices.

I managed to find a internal pic of a burnystar pyro machine, there is a pcb in there and a trans, theres only 5 or 6 discretes on the pcb, doesnt look as though theres a load of smd's on the back, so with this in mind I bunged a standard dimmer on mine, and connected the pyro tip and let it soak for an hour, the trans got warm but not too bad, the dimmer was a little warmer but seems ok, so I might go with that.

Now having looked at the dimmer it says on the back, transformer output 250va and theres a diagram of a trans connected pictorially to the terminals, maybe this dimmer was meant for transformer low voltage lighting.

 

[schmitt trigger]

Indeed, burst control.
That can be achieved with a comparator to sync with the zero crossings, a counter, and some gating logic.

But if your dimmer is intended for a transformer load, then its phase control is balanced well enough to prevent a residual DC component.

For a standard dimmer, the issue is the Diac that sets the triggering voltage. Diacs do not have *guaranteed* symmetrical breakovers, as you can see on page 1 of the following datasheet.
The small difference will cause a small phase angle asymmetry.
But again, you might have a lucky diac in your dimmer, where the breakover is symmetrical. No problem then.

**broken link removed**

 

[MrAl]

Hello,

We had a similar problem back in the 80's, although it wasnt with a triac phase control it was with a synthesized AC PWM going to the primary of a transformer as part of a frequency converter.

The solution there could not be an "integral" cycle control because the problem was with each half cycle, where one half cycle could be slightly different than the next, but in a somewhat regular fashion. We had to use instead an "integrate" solution where we integrated the output pattern over several cycles and that would tell us if there was an offset.
This is very simple in principle. If there is a DC offset then when the signal is integrated over a somewhat long time period (10 to 100 cycles lets say) it shows up as an actual DC signal of small amplitude. This is then applied as feedback during the appropriate half cycle depending on the polarity of the DC signal. This causes a change in only one of the two half cycles in any full cycle period, which adjusts the DC offset output and thus corrects the problem.

The main problem we had was that the DC offset caused more audible noise in the (larger) output transformer (typically 1 kilowatt or higher) because it would push the operating point higher up on the BH curve of the core material.

What actually helps the problem too is resistance. The more series resistance the less this affects the transformer because it starts out as a DC offset voltage, and it is that resistance that makes it a current. So a small resistance means large offset current and a larger resistance a smaller offset current. This means adding even 0.1 ohm to the primary could help quite a bit in cases where the offset isnt too bad to begin with. Of course it goes without saying that the application has to be able to put up with 0.1 ohms extra primary resistance without a problem of any kind.

But wow, 30 amps? I hope you dont intend to draw that directly from the wall socket. Remember that a triac by itself does not convert power, it just limits it almost like a linear circuit. That means if you get 30 amps output at ANY voltage, no matter how low, you've got 30 amps at the input too. The only way around this is with a transformer unless your 'heater' can work with a DC voltage and then you can build a buck regulator. I am not familiar with what your heater type is, so i cant say for sure.

I noticed that a lot of things are going to 'integral' or as i like to call it 'integer' cycle control now, like soldering stations, to regulate the temperature. Not a bad idea really.

 

[dr pepper]

25a is on the primary side, and is only 1.6v, so is only 40w or so.
I see the point about dc pushing the cores operation up on the b/h curve, however the dynamic change in magnetic flux will be the same, so the increase in noise must be for another reason as the b/h curve for iron laminate is fairly linear till it gets to the saturation point. Maybe its caused by harmonics present by badly distorting the sine waveform using phase angle control.

 

[schmitt trigger]

Maybe its caused by harmonics present by badly distorting the sine waveform using phase angle control.

That is correct.

Integer control (thanks for correcting me, that is the proper name) is still the option I would pursue. Even with discrete CMOS logic, it is not that complex of a project.

I remembered that back in 1987 I built and published in Audio Amateur magazine, a vacuum tube's heater regulator utilizing integer control (see attached image). Utilizing just 4 ICs (a dual comparator, a binary counter, an opamp and an analog switch) I built the regulator and had circuitry to spare for implementing a soft start (to reduce the heater's thermal shock) and a B+ delay (to allow the cathode to warm up before the plate voltage was applied).

And sorry, but I don't have a better resolution image. This is a scan from a Polaroid photo I took from the magazine.

 

[MrAl]

25a is on the primary side, and is only 1.6v, so is only 40w or so.
I see the point about dc pushing the cores operation up on the b/h curve, however the dynamic change in magnetic flux will be the same, so the increase in noise must be for another reason as the b/h curve for iron laminate is fairly linear till it gets to the saturation point. Maybe its caused by harmonics present by badly distorting the sine waveform using phase angle control.

Hello,

The dynamic change is not the issue, it's the DC bias that comes into play. It doesnt matter that the change is the same, it matters that it moves up along the BH curve so one end moves closer to saturation. When the 'dynamic' change occurs with DC bias the core gets pushed into near saturation at one end of the curve but not at the other. Because it gets near saturation it draws even more current but only then, and when at the other end it doesnt get bothered so it creates more noise than usual. It's the same situation as a buck converter inductor, except the transformer is not made to be able to handle too much DC whereas the inductor in the buck circuit is, and the frequency is much much lower well within the human hearing range. Add to that the fact that the converters were maybe 10 kilowatts, and we got noise complaints from several buyers. Nobody wanted to sit in the same room with those things

 

[dr pepper]

Yes, assuming that the transformer core enters the section of the bh curve that is non linear.

Anyway I have a working circuit, which is surprisingly simple, a pwm circuit operating at 2 or 3 hz, this fires the triac, the heater wire has enough thermal interita to maintain a constant temp at 2 hz, works well, at the moment I'm using a zero cross opto to drive the triac, I'm going to try it non zero cross with just a tranny driving the triac, I think it'll be fine.

 

[MrAl]

Hi,

Sounds cool. Just remember that driving a triac with a low voltage transformer will mean it will not turn on at the zero cross and may actually be far from it, depending on the voltage of the transformer. That's because the turn on gate voltage is not zero. In fact, with a 6v transformer winding and 1v turn on it wont turn on until about 7 degrees, and with 2v turn on it wont turn on until about 14 degrees. If a higher voltage winding is used it is probably better, with a dropping resistor of course. Maybe 12v would be enough.

 

[dr pepper]

That is interesting and worth knowing.
However the triac in this instance is powering the primary at 230v, the secondary is 1.6v, and is loaded by a short piece of wire as the pyro element.
I tried to get this to work without the opto, which I can do till it comes to powering it with a transformerless supply, the gate needs around a watt to make the triac fire, that requires a really big cap and zener for the supply.