Need help desiging a Simple and Cheap AC regulator

[RobF]

Gentlemen,

I am not a power guy and I'm struggling trying to find a solution to my problem. I need a circuit that will allow me to adjust its power to a fixed resistive load. The input MUST be capable of 100-240V AC and the output power must be consistent throughout the input range.

My solutions so far:
I used a very simple SCR for phase angle firing. I can adjust the resistor that feeds the gate to change the total power output of the controller (higher voltage). This works perfect BUT it does not work on both 110V and 220V without changing the resistor values.

Second solution was reverse phase angle firing; Use a P-channel mosfet or IGBT to control the load. The gate would be connected in parallel to a capacitor and resistor connected in series across the AC line. This works because when the voltage is rising the fet is on until it is able to charge the capacitor up enough to turn it off. This also works very well. This is nice because it is on at zero crossing and turns off at a voltage determined by the resistor/cap value. The problem with this is that it is not consistent between 110 and 220V.


This controller heats up a heating element so the quality of the output is not important. It also doesn't matter that the output is only half wave. The biggest challenge is that the heating element must reach the same temperature regardless of its input voltage (given the input is 100-240V)

Have I missed any possible solutions? I need something simple as it is a disposable item and must be inexpensive to manufacture.

Thanks!

 

[VEBill]

Stable power or stable temperature?
What temperature?
What power level?
What thermal time constant for the element?
What time-period requirements for 'stable'?

Your voltage range sounds suspiciously like the worldwide range of consumer voltages - from certain parts of Japan (100VAC) to Europe (240VAC). So is it a variable voltage, or a worldwide market?

If the goal is stable element temperature (and assuming temperature and power are in the right ranges), then aren't there PTC devices that might do the job in one step? But not likely for many reason. But if you're lucky, then one component.

Another approach is to have a small low power circuit that mirrors the high power circuit.

Another approach is to simply regulate the variable voltage down to (say) a stable ~90 volts.

Or allow a temperature sensor to control the gate of a triac or SCR. This should be 'light-dimmer cheap' if your element can average out several AC cycles durations.

 

[RobF]

Power is actually the most important. I am looking for a temperature rise from ambient not necessarily a certain temperature.

The power level is approx 35W. The time constant is a few minutes as its rather large.

My voltage is the worldwide range, I just neglected to mention it for reasons unknown to me.

This is a controller and is mounted a distance away from the element. Additionally a PTC is not an option because I cannot add or modify to the element.

A mirror is a new suggestion however I'm not exactly sure how to apply it to my application or if it would be cost effective. Can you elaborate on this?

Regulating the voltage would definitely work but it would have to be regulated down to about 48V (DC or RMS) to properly power the element. Anything less and the element does not get to temperature in a timely manner. The voltage doesn't have to be perfect however too much more and you'll heat the element too far. The problem I have is how to regulate the voltage down affordably. At 48Vdc, I need up to 1.5A available.

Closed loop control would be great but it is just not practical in this application. I cannot mount anything on the element and it needs to be a dirt cheap solution. I'd love to explain the details but the big guy in the corner wont let me.

Thanks for your suggestions!

 

[Skogsgurra]

Even if you can easily have a circuit that delivers a constant 48 V RMS voltage at all input voltages between 90 and 264 V (remember that there are tolerances that must be considered) using a triac and an eight-leg microcontroller, it will probably be banned by most utilities because your waveform will be terrible. And, at 1.5 A - that is not trivial.

I think that a much better solution would be to use a switcher with 48 V DC output. There are lots of 24 V devices and probably also quite a few 48 V devices. Just google to find them. They tend to be consumer market priced. So economy should not be a problem.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[waross]

Put a piece of nichrome wire in series with the output. Measure the voltage drop between the ends of the wire. That will be proportional to the current. Use this voltage to control the current. You may want to rectify and filter a little, and then use the DC voltage for your process variable.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[electricuwe]

At that power level the cost of the power electronic part is negliable compared to the control and interfaces. Unless I'd need hundreds of it, I'd prefer to buy instead of designing on my own. You will not pay a large penalty, if you take a higher rated unit.

 

[RobF]

I guess I should elaborate more. The solution must be under $15 so that eliminates every commercial 48V supply I've come across. I've discussed high volume pricing with a manufacturer of a nice 48V supply but they just cannot come close to my target.

My quantities are 10k+ so it is affordable to build a custom solution. Even if I go commercial, in those quantities I have significant lead times that approach the lead times I've received from my contract manufacturer on one of the initial 110V only designs.

 

[Skogsgurra]

In that quantity, a triac and a little processor can easily do what you are after at the price you are willing to pay. But you will probably not get a nice box and terminals for that money. Just a PCB with the components and the MCU program. Lead times needn't be more than around six weeks for the null series. But the harmonics contents of the current will not be acceptable in industrialized countries.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[VEBill]

The low power mirror concept is used (for example) in consumer electric blankets - at least the ones I've seen.

The (cheap) controller regulates the temperature of a very small heater in the controller. The big heater is wired in parallel with the small heater. So they tend to track each other. It's not ideal for precision work, but the same technique is used in many consumer appliances.


Power doesn't necessarily equal temperature rise. Variable air flow over the element (for example) will change the temperature of the element.

 

[LionelHutz]

I'm thinking that using a triac and a low ohmic value series resistor (of the correct power rating) combined with a processor would allow you to measure the voltage and current and then control the power level to the heater. Look to the Microchip and there is likely line powered, voltage measuring and light dimmer examples to work with.

If you want to run without a processor then you're still likely looking at some sort of analog feedback circuit phase controlling a triac or 2 SCR's.

 

[Skogsgurra]

Not necessary to measure/control current. The OP said it is good enough to control voltage. Simple.

Still, the PF at 230 V will not be acceptable.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[VEBill]

OP: "The time constant is a few minutes as its rather large."

Gunnar: "Still, the PF at 230 V will not be acceptable."

A cheap solution to #2 is #1; slow things down and take advantage of the thermal time constant.

 

[RobF]

Hi Gunnar,

I am not aware of any PF requirements in the US. We have harmonic current requirements but I believe it is for applications consuming > 75W.

Can you point me to some material I can read regarding EU/CE requirements for PF and harmonic current?


LionelHutz,

You have an excellent idea. I did visit their site and they have some very nice application designs which would be perfect. The problem I see, aside from PF and harmonic current, is that it requires a much higher component count and requires programming an MCU which takes more labor. I would have to call my CM and run a BOM on the design before I could go through with it.


As of now, the best solution I can come up with that meets my cost requirements is to have two separate versions.

Thanks for everyone's input!

 

[Skogsgurra]

Yes VE1BLL, some kind of packet control? That could work. A simple scheme where every second period is forwarded at 100 V, every fourth at 200 V - and probably some other combination. Switching that fast probably avoids the flicker associated with packet control.

I would think deeply about that if I were the OP. Tolerances will not be very good. But I have a feeling that it isn't needed in this case.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Skogsgurra]

sTiZZiT

Right. Less than 75 W doesn't have any limit re PF or harmonics. So, it will probably be OK using a triac. The programming of a microprocessor will hardly be affecting the unit cost dramatically if you are in the 10k+ bracket. It will make smart solutions possible that probably makes the total cost lower at the end of the day.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[RobF]

VE1BLL & Skogsgurra

I knew I posted on here for a reason. A combination of what LionelHutz and VE1BLL said would make for a mean little controller.

With this I could even have say some kind of input (hall effect) to change the total power output for other future products or if the customer requires a different "level" of power. I say hall effect because the board will be potted. Run a magnet across the magic spot with an LED for visual feedback and you'd have a inexpensive control over temp. The reference design i saw on Microchip's site actually had an IR receiver so you could use a remote to set the brightness (was for an electronic dimmer). This would be even cooler but means I'd have to design or source a IR remote.

If they give me enough time, I will go with this solution.

Thanks again everyone. I'm very glad I posted this thread!

 

[Compositepro]

As has been suggested, electric blanket controllers do almost exactly what you trying to do. They control power to a heater without feedback from the blanket. They used to be a simple bimetalic switch wrapped with a heater wire. The heater wire "mirrors" the heater you are trying to control and thus will compensate for input voltage variation and ambient temperature.
I'm sure you can find a Chinese manufacturer that can supply what you need

 

[LionelHutz]

If you need real simple then you can do a voltage measurement with a processor, look up a phase angle in a table and then just adjust the output to that phase angle.

If you want fancy then you could measure the power (volts and amps) and run through a PID type control to regulate the output. If you want even fancier then do zero-crossing switching (basically a 60hz PWM) and eliminate much of the harmonics.