using pic to reduce wattage by changing duty cycle

[jimg]

I have a simple circuit-

120V AC ---- Diode ---- 6800 ohm resistor ---- 48V dc relay ---- Mosfet as switch --- AC neutral

The diodes and mosfet are rated at 250V. There is also a diode and 10uf cap across the relay.

I am turning on the mosfet now with a 12f629 pic.

By my calculation, the resistor is dissipating about .4 watts and is generating more heat than I would like in the enclosed space alloted.

Would it be possible to eliminate the resistor entirely by feeding the mosfet a square wave from the pic? If so, what would be the approximate on/off ratio and what problems can I expect? I'm a little confused since it's switching a 60hz half wave but it shouldn't make any difference if I chop uniformly through the waves and the area between.

 

[Nigel Goodwin]

You would need to smooth the 60Hz to give reasonable DC - but the scheme doesn't sound a good idea to me?, it looks like the original idea was flawed?.

Why not use a 120V AC relay, and switch it with a triac? - seems a lot simpler?.

For that matter?, could you use a solid state relay?, you don't mention what you are switching.

 

[jimg]

Well, most of the design criteria were heavily influenced by cost and size, since I want to make about 50 of these. The relay will be switching up to 10A of 120v AC.
The js1-48V relay I am using costs $1.57 where a usable pcb 10A 120v ac coil relay is $4.56, and the dc relay with the 10uf cap seems quite happy with no chatter, etc. but there may be problems I'm not aware of. Heat and size are problems, so by replacing the relay, the heat sink required by triacs, etc were prohibitive.

 

[Nigel Goodwin]

You could try it and see how it works? - I would suggest trying something like 25% duty cycle at a reasonably high frequency.

 

[jimg]

Thanks Nigel. If you're willing to even consider the idea, it can't be totally insane. By a reasonably high frequency, would you think about 1Khz? I have no experience in these things and I don't want to generate radio noise, etc.

 

[Nigel Goodwin]

Thanks Nigel. If you're willing to even consider the idea, it can't be totally insane. By a reasonably high frequency, would you think about 1Khz? I have no experience in these things and I don't want to generate radio noise, etc.

I would suggest trying between 1KHz and 10KHz, and see what happens - I would also suggest STRONGLY that you include some type of resistor in series with it - probably a smallish carbon one, so that if something 'nasty' happens (like the FET goes S/C, or the PWM locks permanently ON) the resistor will go O/C before anything really frys!. You can specific safety resistors that go O/C when overloaded, without big clouds of smoke.

 

[jimg]

Ok! Now to convert from a 12f629 to a 12f683 and learn all about PWM

 

[Hero999]

I would put the 10:mu:F capacitor in parallel with the transistor and relay and the reverse flywheel diode only accross the relay. The idea is the inductance of the relay coild keeps the current flowing when the power is turned off.

Nigel is right but I'd even consider adding a small fuse for extra protection.

 

[jimg]

So-

120V ac -- diode -----  relay  ---  mosfet  ----  neutral
                    |           |             |
                    |---diode---|             |
                    |                         |
                    |------------ cap --------|

something like that. I don't understand why, but I'll try it.

Wait- If I do that, the cap will be charged all the time at about 160 volts regardless of the state of the mosfet.

 

[Sceadwian]

Acording to the manual for that relay the switching time at 100% coil voltage is somewhere around 7ms's and the release time is about 2ms's. You're not going to be able to have a switching frequency higher than about 100 hertz or the relay contact is basically going to hover somewhere in between close and open states. Also according to the PDF at it's 10^7 switching lifetime is rated at only 180 cycles per minute, it's probably significantly lower at 50 per second which is the highest frequency you're going to want to switch a relay at. Asuming that the rated lifetime is halved at 50hz the relay is going to fail after about 280 hours.

 

[kchriste]

Why not go to a fullwave rectifier and put a small capacitor between the Hot lead and the AC input to the bridge? Choose the capacitor so it's capacitive reactance drops enough voltage to give you 48V for your relay. Or does the neutral lead need to be in common with the FET? Then something like this may work also:

120V AC ---- 100ohm resistor ---small cap---->|---- 48Vdc relay ---- Mosfet as switch --- AC neutral
                                           |
                                            ------------------|<------- AC neutral

The -->|--- is a diode and the small cap needs to be able to handle the peak AC mains voltage.

 

[jimg]

Sceadwian-

Acording to the manual for that relay the switching time at 100% coil voltage is somewhere around 7ms's and the release time is about 2ms's. You're not going to be able to have a switching frequency higher than about 100 hertz or the relay contact is basically going to hover somewhere in between close and open states. Also according to the PDF at it's 10^7 switching lifetime is rated at only 180 cycles per minute, it's probably significantly lower at 50 per second which is the highest frequency you're going to want to switch a relay at. Asuming that the rated lifetime is halved at 50hz the relay is going to fail after about 280 hours.

I'm not quite understanding what you're saying here. I don't actually want the relay to switch at those rates, I want it to switch and stay switched until I stop sending it pulses, the chopping is to drop the effective voltage through the cap/relay from 160V to 48V. The 7ms switch time and release time of 2ms is considerably lengthened by the cap, hopefully greater than 1/60 of a second to stop the chatter. This is not a problem in my application, anything up to half a second delay would probably be ok.

kchriste-
I don't use a full bridge because the circuit needs to be relative to and tied to neutral, and that would just be twice the voltage I had to throw away. My experiments in the past make me think the "small cap" as the voltage limiter is actually a physically large non-polarized cap and I would still need a cap across the relay to stop the chatter. I will give it a try to see how it works out however.

 

[eblc1388]

Would it be possible to eliminate the resistor entirely by feeding the mosfet a square wave from the pic?

Rather than to go into all that trouble of switching the control element, why not try to eliminate the resistor by replacing it with a capacitor.

 

[jimg]

If you are talking about a standard transformerless power supply type of thing, in order to supply 48 volts at 8ma, you need a very large capacitor, no?

 

[eblc1388]

If you are talking about a standard transformerless power supply type of thing, in order to supply 48 volts at 8ma, you need a very large capacitor, no?

Depends on what large means, size or capacitance. The capacitor need to drop 72V @8mA so it is about 9K equivalent.

The capacitor value that gives 9K at 60Hz is 0.3uF from my calculation.

 

[jimg]

Well I game to try most anything, but I couldn't get it to work. This is using my original circuit but with the resistor replaced by a capacitor, right? Using a 1uF cap, I'm only getting some very short spikes of a few volts out of the cap, so I must be misunderstanding the circuit.

120V ac -- diode ---1uF cap---|--relay  ----|----  neutral
                              |             | 
                              |-10uf  cap --|

 

[eblc1388]

Capacitor works on AC only. If you give it a rectified voltage, then the capacitor would simply charged up so you get no more current.

Try this:

120V ac --1uF cap---+--diode-----+---relay----+----  neutral
                    |            |            | 
                    |            |-10uf cap --+
                    |                         | 
                    |-----reversed diode------|

Instead of wasting half of the AC energy, you can place the relay and 10uF smoothing capacitor at the +/- outputs of a bridge rectifier(BR), while one end of the BR AC input connects to the 1uF capacitor, another end of the BR AC input goes to MOSFET for switching the relay ON/Off.


Hope I have explained it correctly.

 

[jimg]

OK, that works. .47uF cap gave overall current draw of 20ma and voltage at the relay of 50V. The only non-polarized cap I had of high enough voltage was bigger than the relay, so I'll have to shop around for something smaller! What would be the voltage rating of the capacitor required? 50V or 120V or 250V ??
Metal Polyester the smallest?

We're dropping 2 watts somewhere, bigger than my original loss with the resistor, so I have to try the bridge method.

 

[3v0]

Well, most of the design criteria were heavily influenced by cost and size, since I want to make about 50 of these. The relay will be switching up to 10A of 120v AC.
The js1-48V relay I am using costs $1.57 where a usable pcb 10A 120v ac coil relay is $4.56...

Have you looked at Electronics Goldmine for relays.

https://www.goldmine-elec-products.com/departments.asp?dept=1114

They have a lot of relays in your pricerange. One of them and a 20 cent NPN might solve your problem.

3v0

 

[jimg]

Thanks for the link 3v0.