Help figure out how to Dim LED MR-16 on either AC or DC

[tconley]

I was looking for some help reguarding dimming of led based mr-16 lights. They are capable of both running ac or dc at 12 volts. I was planning on using them for some lightning effects and other color wash effects. I was looking at having between 8 and 16 lights for halloween. I need help on figuring out the best way.


After some discussion in the chat we went over the pro and cons of several ideas.

1. Using a 16 channel renard with builting ssr to control them on 12v ac.
Con - This would work but the dimming would not smooth and there might be problem using 12volts
Pros- I already have a board ready to build just need to order the bits and peices from mouser

2. Using the renard c in conjunction with the 595 and a 12v dc power supply from a computer. Using it without the ssr and the mr-16 directly driven.
Cons - have to order the renard C and board chained to board to the equation
Pros - I already have an unbuilt 595

3. Using a striat renard design to handle the dimming.
Cons - No board yet designed and I am not capable of doing the design.
Pros - Single board solution designed to specifically address the issue.

Does anyone have any other ideas that might work.

Reference Information
LED Mr-16
http://www.besthongkong.com/index.php?cPath=9_29

[tconley]

I have heard about the d-light DCx16 - Gold Edition my problem is it isnt compatible with vixen. How hard would it be to make something similar based on a renard type design for controlling 12v dc.

http://www.d-light.us/index.php?page=shop.product_details&flypage=shop.flypage&product_id=48&category_id=11&manufacturer_id=0&option=com_virtuemart&Itemid=26

[Macrosill]

T,
In reference to the 595 driving the led floods directly at 12v dc. Would the ULN2803's be able to supply the current needed to supply all the floods with full power at once if this statement is correct, "you can have all channels driving 350mA X 8 at an appropriate duty cycle", I read that somewhere. I looked at the 2803 data sheet and it says the peak current can be as high as 600ma with a continuous current of 500ma. But the continuous collector current max is 500ma. Does that mean you can only sink 500ma total at one given time with the 2803? I am not good at reading the data sheets.

I was thinking going the same route and have wondered this for about a week now.

[P. Short]

The data sheets are a bit ambiguous. Each collector has a separate pin, so I interpret that 500 mA number as a per/output spec. But the power dissipation spec for the total package also has to be met. This means that you can't drive all 8 outputs at 500 mA (since then the power dissipation would be 8 * 500 mA * 2V = 8W, where the 2V is my guess for VCE(sat) at 500 mA).

The spec for the particular MR16 that TC selected, though, says that the typical power output is 2.4W, which translates to 200 mA at 12V. At 200 mA, however, you could drive all eight outputs of the ULN at the same time, although it would get warm, and some sort of heatsink might be a good idea.

--

Phil

[Macrosill]

I also read that you can parallel the outputs of the 2803 to get a higher current capability. This would take 2 channels to drive 1 lamp but that would be acceptable if I could get away without using a ssr.

Any thoughts on this?



edit: just did the math and the led floods I am looking at are 2 watts each at 12v. So they use 166ma. I guess I should be okay driving 8 of them directly from the 2803.

Another point, each location of floods will have 1 red, blue and green, only using 6 of the 8 outputs, 3 from each rj45 so I only have to run one cable and no splitting. This will also allow for future expansion. I would probably not drive each color at the same time. So then I would only be using a quarter of the outputs at once, max.

[P. Short]

That would work, and would be the thing to do if you were running into the 500mA/output limitation. It doesn't help when you are hitting the power dissipation limits for the package.

I think that it may possibly work well enough by just modifying an SSR to run from 12VAC. Most of this is just my musings on how to fix things if it doesn't work well.

--

Phil

[FireGod]

Deleted, my bad

[P. Short]

Editted...

The issue with using 12VAC is that the SSRs aren't guaranteed to turn on unless the voltage is above 3V, or so. This is based on the sum of the triac gate voltage (1V), the opto saturation voltage (1V), and the drop across the gate resistor (1V if the resistor is 22 ohm and the triac sensitivity is 50 mA). This is a high percentage of the peak AC voltage (17V peak), so I am concerned about how well the PWM operates at the lowest dimmer levels (perhaps below 10%).

--

Phil

[FireGod]

Would using a sensitive gate triac help?

If the dimming below 10% is not so great, I would just not go less that 10%.

Could a transistor be used? I am really rusty on transistor operation but if I recall correctly it might work.

[P. Short]

Yes, both of the first two suggestions would help. And remember, that is a worst case scenario, when the parts are at the edge of their specs and at the environmental limits. In real life, the triacs and optos turn on with less than the worst case voltage, and require less than the worst case current.

And the third suggestion would also work if DC was used instead of AC (although the current SSR board cannot be used for that).

--

Phil

[FireGod]

Do you think that using these would work in the current SSR?

782-CNY17F-3 Optocoupler Phototransistor (http://www.vishay.com/docs/83607/83607.pdf)

Pins 5 and 6 would need to be jumpered on the optocoupler.

511-TIP127 Power Bipolar Transistor PNP (http://www.st.com/stonline/books/pdf/docs/4128.pdf)

The transistor would need to be turned around when installed.

[P. Short]

That's a good idea - I'd been overlooking the obvious idea of using a common collector circuit (mostly because I rarely use them, because it is often inconvenient to provide good base drive).

However, the implementation depends on how the SSR is wired up. The one that I'm looking at is here (http://computerchristmas.com/ForumBoard/read.php?f=4&i=7285&t=7285), and that circuit needs an NPN transistor (TIP122, or similar). The heavy common rail on that SSR needs to be connected to +12V, and the (cathode) terminal of the MR16 needs to go to ground.

I think that the method that you propose would work with SSRs where the pins 4 and 6 of the optocoupler are switched around from the picture in the link above. In this case the heavy common rail of the SSR would be ground, and the other end of the MR16 leds would go to +12 VDC.

An alternative optocoupler would be to use a 4N35 or 4N37, which I prefer because it has more sources. But the CNY17F-3 would work just fine.
--

Phil

[ErnieHorning]

You’ll have to hack the board because the Opto-TRIAC and Opto-Transistors have a different pin out.

With the FireGod method pins 5 & 6 could just be jumpered.

[FireGod]

Phil, you got the part number right and I got it wrong. I was actually proposing a TIP-122 not a TIP-127.

Edit: So using this method you would drive the LED with DC voltage and use a PWM signal to the optocoupler.

[P. Short]

Ernie,

I don't understand your first sentence.

--

Phil

[P. Short]

A small detail (or two).

1) Firegod didn't state it explicitly, but the base of the optocoupler must be left open. This is done by either selecting an opto without a base, or by cutting the base pin so that it doesn't make contact.

2) These parts work fine for lower current levels, but using a darlington increases the Vce-sat, and hence the power dissipation on the transistor. For higher currents (above 2A, or so) it might be better to chose a regular power transistor (not a darlington), and give it more base drive by selecting a higher-gain opto and/or give the opto more input drive as well.

--

Phil

[P. Short]

Another factor affecting power dissipation is that the darlington may never fully enter saturation. The collector of the TIP12x is always going to be roughly 3V above its emitter, determined by its own VBE and the opto's VCE.

This should not be an issue for using this circuit with the MR16 LED lights, though. It really only applies when you are trying to push more than 1A through the transistor.

--

Phil

[wjohn]

I can switch DC loads and have tested the circuit using a 1W MR16 LED, however dimming is not practical using the RENARD,

for simple ON/OFF switching of a DC load, I used a H11AA1 opto to isolate the 595 from the load, (overkill) and a BD139 NPN transistor connect common collector to the OPTO with the LED MR16 connected to the Emitter and down to Gnd.

595 switching worked fine.

[P. Short]

What happens when you try to dim with Renard (assuming that you are using the PWM firmware)?

--

Phil

[wjohn]

i did not use PWM, but I will.

Those really short pulses in the standard code just aint right for DC ;-)