24v 4 channel AC Chaser (not a dimmer)

Go To Last Post
16 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hey there fellow freaks.

Recently a client came to me and was wondering if my company would be interested in manufacturing some simple 4 channel light chasers 24V / 200VA total load. I said I would get back to him.

Anyway, I've looked into it a little but am a little confused by the vast array of Triacs that are out there these days. Seems every Triac manufacturer is claiming their triacs are the best.

I was thinking i could manufacture this chaser using a ATMEGA8 solution that is non-isolated using 'logic level' triacs and using a simple zero crossing detector using the ATMEGA8s comparator, that way i could keep the component count low. Ideally this board would be mass manufactured so a nice SMT based solution is what i envisage, if the board fails it would be a throw away afair.

I just want to throw this idea out here to get some advice/ tips / feedback before I commit to an order of a handfull of triacs, etc.

I searched the forum here for threads dealing with Logic Level triacs and didn't find any, so perhaps this can be a good starting point for some discussion.

Look foward to some feedback.

thx
oddbudman

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

To go non-isolated could be seen as the act of a money chaser who has no regard for the end user.

Suggest you have a good think about the implications of manufacturing a unit that could result in death if it goes faulty, when that risk could be reduced at the design stage.

Keep it simple it will not bite as hard

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If you are geting the 24v from a transformer, no prob. Just run the 24vac thru a bridge rectifier, no filter cap... run the fw rectified ac to the lights, use logic level mosfets, drop the 24v to 5 with a regulator for the avr

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Mike, I'm a little confused about your response (good pun though :) ). Perhaps i am just being ignorant, but isn't 24vac SELV. I'm interested in finding out how I can somehow make this device so deadly I can make it kill - If for my own safety in the design phase.

If your worried about overvoltage protection I was planning on a overvoltage clamp circuit with a fuse to isolate the input if some bozo does wack 240V in by mistake.

Bob was correct with the 24V transformer call. Yes it will be running off a 24V AC transformer.

Bob, why the mosfet solution? Seems interesting, but not sure why you would suggest it over a triac. Could it be more efficient than a triac? (the losses in the bridge + mosfet Vs losses in the triac).

I suppose the mosfet circuit could also be used as a LED PWM dimmer aswell if a DC powersupply input was used instead of the AC supply. Furthermore perhaps the mosfet could allow for more control over turn on/ turn off times.

Anyway, thanks for the responses thus far. I look foward to getting some more.

oddbudman

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

My reasoning was: you need to have a diode, small cap and small 5v regulator for the cpu, but the lights dont need regulation... they can run right off the fw rectified 24vac... just thought tieing everything to a comman ground and using 5V logic level mosfets might be simpler than triacs... do you pull the gate down to turn triacs on? Do you need one of those wierd diac thingies? See, I know knothing about them. "If the only tool you have is a hammer, then everything looks like a nail..."

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The mosfets could run cooler than triacs because a triac when in conduction
has over a volt drop across it. I usually just estimate 1 watt per amp of load
current with triacs. Using some of the new fets with very low on resistance
might work pretty nice for this application. The "down" side is that the bridge
rectifier must now be larger and it will now dissipate over 1 watt per amp
(unless you use Schottkys) of load current. Perhaps use fet pairs as AC
switches to avoid running the lamp load through the bridge.

Tom Pappano
Tulsa, Oklahoma

Tom Pappano
Tulsa, Oklahoma

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

See next statement

Last Edited: Wed. Oct 6, 2004 - 07:51 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If you still want use AC for the Target and not DC, just do following.

Use a bridge recifier between the load and the transformer.
In the middle of the bridge use a mosfet as switch (also possible for dimming with zero cross detection).

The supply is generated with a extra bridge recifier direct connected to the transformer.
The bridge for the Mosfet is used to solve the parasitve recifier problem of the mosfet.
with this circuit you have a current flow always from top to down and you can use a simple n-FET.
The only Problem is the Ground level for Mosfet Gate.
By the second recifier the supplyground of avr should be near the same level than the Mosfet Drain pin.

Other solution is to use 2 mosfet antiparallel in sequence.
But for tis solution you should have a transfomer with a additional supply coil for the avr to solve the ground problem between the mosfet and the AVR

Regards

Max

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
Other solution is to use 2 mosfet antiparallel in sequence.
But for tis solution you should have a transfomer with a additional supply coil for the avr to solve the ground problem between the mosfet and the AVR

Or you could use a photovoltaic optoisolator to drive the gates, since this is not
a high speed switching situation.

Tom Pappano
Tulsa, Oklahoma

Tom Pappano
Tulsa, Oklahoma

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

MOSFETs that I'm familiar with, have internal diodes across the D-S. Therefore 2 MOSFEETs antiparallel would look like 2 diodes antiparallel, conducting all of the time.

TRIACs are triggered by current, not voltage. So, you need to know the current required to trigger the TRIAC. Many can be triggered with the gate either positive or negative. (With respect to MT1). Most that I've used like negative gate best. Look at the data sheet carefully, since the trigger current is dependent on the MT2 voltage, and you want to trigger at a reasonably low voltage to limit your noise output.

When you trigger a TRIAC, beware that while it's conducting, a significant current can flow out of the gate. If you're driving directly from an output port, I suggest a pair of antiparallel diodes in series with your current limiting resistor.

Finally, good layout is a must. A cold incandescant load draws a lot of current while heating, and you don't want this current to be transformed into a noise voltage in your logic.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thought I'd chime in on the safety issue.

Under "ideal" conditions, 24 volts can still kill - especially if it is 50 or 60 cycle AC. The heart is much more sensitive to an AC waveform in that frequency range than it is to DC. So if you can develop around 5 mA across the heart at 50-60 cycles, you can kill.

Couple of stories related to us in an electrical safety class - one guy that was taking the idea of touching the 9-volt battery to the tongue for the tingle but wasn't thinking and was doing it with a twelve volt car battery and jumper cables. His tongue literally exploded. A car battery has a much lower internal resistance (obviously) and so current through the guy's tongue was huge.

Another that had figured out you could do similar tingling with a 12 volt car battery and two buckets of salt water - one for each hand. It was a lethal experiment.

Those are both extreme examples, but one was lethal at only 12 volts. All you need is enough conduction through the skin and you can kill - especially with AC. Isolation is not something to be taken lightly. Inside the body, the saltwater conducts very well and if the current is across the heart, possible death.

Since you say you have a 24 volt/200 watt supply, you can push slightly over 8 amps at 24 volts. If everything is dry and you know that nobody would ever be exposed to the internal voltages, you might be OK but I have to go with Sutton and say isolate or have a real good faith in good luck or divine intervention just in case your item gets used in ways you don't anticipate.

Please note - this post may not present all information available on a subject.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
If everything is dry and you know that nobody would ever be exposed to the internal voltages

Although technically correct, I would not go with "nobody would ever be exposed".

A lamp can break, and this might cut the skin. If you then probe the wound with the exposed filament wires, it bypasses the skin resistance and you can get a lot more current. If you break two lamps and put one into a wound in each hand, :evil:

You have to draw the line somewhere, and UL/CSA/EU standards and regulations can help a responsible engineer to find that line. In most countries, that line has been drawn at 25 volts AC for most unprotected wiring.

This doesn't mean that I suggest being reckless just because it's "only 24V". But, opto isolators??? You don't leave bare wires laying around, and you cover terminals sensibly. A cover over a driver board (and any hot things) is another sensible thing...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks for all the replies people, keep them rolling in. Especially if you want to drop any further advice regarding using Triacs (or even IGBTs).

I've just got some leftover parts together and plan on trying the AC Rectified / Mosfet switching approach over the next few days.
I will be using some ST ”OMNIFET” FULLY AUTOPROTECTED POWER MOSFETS (come with overload cutout and error feedback). All tied together with a ATMEGA8. We'll see how things go.
I'll endevour to keep this thread updated as I go along.

Anyway, for interests sake, here is a picture of the board which I am looking to supercede.

The ICs are a 555 timer, a CD4017 decade counter, MOC3041 400V Zero Crossing Triac Driver Output Optocoupler and the Triacs being used are BTA12 600BW. There is a 12V reg on side of this pic too. You can all probably guess how this all works :)

thx,
oddbudman

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I thionk a lot of these comments are over kill.

firstly there is no reason to worry about zero crossing detection. this is only a problem
of transients being sent back through the 240V power system and you do not have this problem (the transfer is one great big filter). The only other transient issue is heat disapation, but as you are not switching exevery cycle (as qa dimmer would) this issue is inconsequential.

any triac will do (the cheaper the better) as long as it takes 10Amps (200VA at 24V) and definetly the esiest way to drive it is via a triac optocoupler. This results in just two resistors, a triac and an optocoupler per channel. Note ther are very cheap quad SMD DC to AC optocouplers eg Toshiba that sell for around USD $3-4. you do not need any filtering on the AC side becasue you are using low speed switching. you should be able to make them for around USD15 easy and you nshould be able to fit them in something slightly larger than a match box.

not that heat disapation is not a big issue with the triacs as you are using slow speed switching so all you need to worry about are I2R losses. As an example if you are swithcing 10A and te internal resistionace of the TRIAC is 0.01 ohm and the duty cycle is 50% then you will only be disapating 0.5W per triac, therfor no heat sinks should be requried.

Why use Mega8 what is wrong with TINYAVR and cutUSD4 of the cost of each unit.

Lachlan

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks for your insight Lachlan (for interests sake I'm in Australia too- brisvegas). Much Appreciated. I'll look out for one of those quad optos tomorrow.

Currently we are using ATMEGA8s in many of our other projects and use them in high quantities so sticking with them makes sense ATM.

We currently are only paying $2.80 AUD each for ATMEGA8s in PQFP.

I was also thinking about using the ATMEGA8s ADC for speed control and a few other stray I/O lines with a DIP switch for chase pattern control. So although there will still be spare I/O, for the moment it's not -complete- overkill.

I'm looking to use SMT compontents throughout the board, that way I can handle the heatsinking on the PCB, and keep the unit nicely self-contained in a Transformer like enclosure.

I'm starting to think this venture may be worthwhile....

thx
oddbudman

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
As an example if you are swithcing 10A and te internal resistionace of the TRIAC is 0.01 ohm

Unfortunately TRIACs have an on-voltage that is almost constant, and for most of them it is 1 volt. I think the MOSFET design is a good way to go, assuming you can find a bridge rectifier in the form factor that you need.