Another DC SSR

[P. Short]

Here is another possible design for a DC SSR intended for use with micro-based controllers (with appropriate new firmware). As currently designed it uses current-control rather than PWM to control the power delivered to the load.

It can be easily modified to do PWM by adding some jumpers to the board, though. In this mode it can be use with either MOSFET or bipolar transistors. With BJTs it will provide a constant current to the load for use with some of the MR16 or other LEDs lights.

It uses a shunt regulator (in place of the usual series regulator), so the capabilities of the regulator no longer determines how much voltage the controller can handle (just the breakdown voltage of the transistors).

The design has provisions for a bridge rectifier on it, so it can be powered with either DC or AC, although the output is DC only. It is limited to about 2A total current, based on the size of the traces and of the bridge rectifier when the board is power with AC).

I'm working on this design because I'm not totally satisfied with the existing SSR designs. This was the only type of SSR that I've tested so far that provides smooth dimming of the MR16 LEDs over a wide range (this design allows a much more linear-appearing ramp to the dimming than the PWM techniques, no matter whether AC or DC is used). Second, I'm hoping that this will provide a testbed (and hopefully the solution), to LED flickering (when dimming) that some people have noticed.

The design needs a little bit more work (primarily changing the pinout of the optos to be the same as the one that John uses on the coop DC SSR board).

--

Phil

[NogginBoink]

I'm not sure this will work.

It looks like I'll need to send two commands to set a channel output: one to the DAC and then one to the MUX, right?

But then as soon as I send the next command to the DAC, the MUX will forward that new value to the previously selected (wrong) channel on the MUX, and as soon as I send a command to the MUX to switch to a new channel, the first channel has no input from the DAC and goes to zero.

Uh... right?

It appears that this SSR will only be able to drive one output at a time. Granted, the uP could scan them at a high rate, but still...

Wouldn't it make more sense to do away with the MUX and use four DACs?

Or am I missing something?

[P. Short]

The cap on the input to the op-amp holds the value while the mux is disabled (or another channel is selected). The mux and caps are a lot cheaper than 4 DACs (or a 4-channel DAC).

It's basically a sample-and-hold circuit.

--

Phil

[Trepidati0n]

P. Short,

I know there are a bunch of design challenges using digital so close to AC. Not hard to get around, but it adds a few parts but makes the system a bit more rugged.

The largest advantage of this type of control is that you are now somewhat immune to the source (whether it is DC or rectified AC). But this is where problem begins and ended for me when I thought about this...who eats the heat from the delta voltage? While this design issue is not a show stopper, it might be well beyond the average DIY’er to maintain. It also might require heatsinking of the FET’s.

Therefore I have gone to a system where I just use rectified DC and just do a PWM. The approach I now use is that all loads connected must be able to survive 100% and just duty cycle down.

I was about to finish my design and make a PWB...but then I saw the Lynx system here. Hopefully the designer will be a bit more forth coming in the future about his design so I can see whether or not I want just use his or still strike out on my own.

[P. Short]

1) This is a low-voltage design (12V, give or take).

2) Those are BJTs, not FETs.

3) The transistors are the components that dissipate any excess power. As this is a low voltage system, and the LEDs tend to maintain the same voltage across them over a fairly wide current range, there isn't quite the dissipation problem that you might think. If the LED current is set to about 200 mA (max) the power dissipation in each transistor would be under 1W for the blue and green LEDs and about 1.2W for the red lamps. These are not a problem for TO220 transistors.

4) The MR16's in question do not do well with PWM, as they have a huge amount of capacitance in them that tries to counteract any attempt at PWM control.

In any case, the issue is moot, as the issue of controlling those particular MR16 LED lamps has disappeared off the screen. As for RJ's Lynx design, I rather suspect that it is just using SCRs (rather than triacs) for PWM control of pulsing DC (full-wave rectified AC power line).

--

Phil

[Trepidati0n]

I apologize about the BJT/FET thing. I rarely use BJT's anymore in the designs I do. The varience in Vbe is often way to large for time/life calculations. Therefore FET is on my brain.

As for the capacitance issue (I didn't know it ws that bad with 30 in series), has anyone considered puting a very low capacitance diode in series with the chain? Also, if the built in resistors that are in some LED strings were remove this should also reduce turn on time since the FET/BJT would be on as hard as possible.

Happy Holidays...

[P. Short]

The lights that I'm talking about are not strings, but rather multi-LED modules that fit in standard MR16 sockets like those used for spot track lighting. There was a group buy by members of the planetchristmas.com forum of 32-lamp modules from easy-led.com some while back. There is a thread about 1 page back in the LED section of the planetchristmas.mywowbb.com forum about the problems that were encountered. Those types of lights sometimes have various combinations of diodes, caps and resistors in them, perhaps to reduce flicker and to make them work in both 12VAC and 12VDC sockets.

PC group buy lamps (http://www.easy-led.com/big_01.php?mod=LR0101S&name=Spot&name2=&id=4)

--

Phil

[Trepidati0n]

I now understand where the design was coming from.

The point of view in which I looked at your design was assuming you had a random length chain of LED's or lamps (or any type of load really).

This last year I was experimenting with hot glue and silicone (to act as a diffuser), Red, Green, and Blue LEDs in a clear pastic tube to try and create nice full color light bars of custom length. However, there was still too much hot-spotting for my liking.

[joneslights]

Would this design work for me to power a 12vdc strip light? Here is a link to the product. http://www.acolyteled.com/products/stl25dc.html

[P. Short]

No. It takes new firmware in the (Renard) controller, which hasn't been written. That was largely just a concept for discussion.

--

Phil

[joneslights]

No. It takes new firmware in the (Renard) controller, which hasn't been written. That was largely just a concept for discussion.

--

Phil


Thanks Phil,
Nice work you do BTW

Greg

[omzig]

Phil, how far did you get with this? Did you ever get it to work?

[P. Short]

I had some PCBs build, and did some testing, but haven't pursued it.