Advice needed for dual driver setup.

[Mash]

Hi guys, here is the scenario:
Continuing project for bedroom LED lighting. Will use 12-16 crees on a mains dimmable driver for general purpose/mood lighting, and plan to use 3-4 Bridgleux or mulitled spots for high illumination (only when needed), on another mains dimmable driver.
However in order to simplify and eliminate several switches/dimmers etc I want to do the setup so that one switch/dimmer controls the whole setup. After lots of thinking, the ideal way would be that the dimmer on its first 75-80% of rotation operates the Crees, then when turned higher. kicks in the Bridelux/spots. I thought of doing this by way of a relay. which reads the output voltage of the first driver, which is feeding the crees, and adjusted so it kicks in the AC to the second driver at a user set threshold level.
Now the questions:
1- Is there an easier/better way of doing this? Any products that do this already?
2- Are relays adjustable so I can get the switch point where I want it?
3- Have never used a relay, so can a knowledgeable member point me to the right one (in farnells, eg)? Spec as this: UK mains, 230V; the LED circuit which will be controlling the relay: max 32V, 350mA.
Thanks in advance!

 

[evilc66]

A rotary switch might be an easier option. Just set each pole up to have a different resistance/voltage for the different brightness levels, and have the last few contacts hooked up to the Bridgelux driver(s).

 

[kethd]

I think your idea is novel/intriguing/feasible... but I worry that your understanding of circuits/electronics is not adequate to pull it off.

I'd be surprised if there were a standard product to do what you want. Such a circuit might make use of a relay, but it must be driven by some additional circuitry because the turn on/off points of a bare relay are too crude/sloppy/unpredictable.

I think you need someone who knows electronics to build you the relay with a driver circuit that has a trigger point that you can adjust. The circuit needs to be designed with a little hysteresis so that when the dimmer is right at the trigger point the hi-power mode does not flicker ON/OFF.

If you were very lucky, you could achieve your goal with just a relay and just the right series resistance in the relay coil circuit, but I don't recommend you try this.

You don't actually need a relay at all; there are many ways to do your switching with power semiconductors. Either approach is OK, mostly a matter of the preferences of the person designing/building the circuit. (Relays always break eventually, since they are physical. Semiconductors can last forever, but are more vulnerable to glitches and stresses.)

Sadly, the bottom line is that this idea of yours is great in theory, but for you will make this project more complex instead of simpler...

 

[Mash]

Thanks for your answer guys!
The problem with the rotary switch option is that I think I will need two dimming switches (one by bed, one by entrance), thats why I dont think the rotary switch would be sufficient
And thanks for your advice kethd, exactly the kind of info I wanted. I am sure it can be done by an electronic circuit, but I dont have enough knowledge to build one, or access to a friend who could! I thought of the relay eventually as something that I could manage and would do the job required, but seems not!
And as per your advice re power semiconductors, I know absolutely nothing about them, so would have no idea how to build/implement it! Unless osme kind soul shows me the exact way to do it!
Please do post more if you think of more, thanks!

 

[kethd]

Technical question for driver geeks:

Do most dimmable constant-current drivers put out a variable DC current, or a variable-duty-cycle waveform?

 

[Mash]

KTEHD, here is a link to a mains dimmable driver:
http://lumotech.com/en/led-drivers/...1-12va-minidim-mains-dimmable-led-driver.html
There are datasheets there, hopefully they will answer yoru question.

This not the model I want to use, since it works only with trailing edge dimmers, but is one of the few mains dimmable drivers out there.

PS re the power semiconductors: Since you know more about these than me, can you reccomend me a design off the net, simple to build and understand? I wouldn't even know where to begin looking!

 

[kethd]

There are many different possible design approaches to your situation.

I love the idea of using just one combined driver. For example, imagine one dimmable constant-current driver with two warm white Bridgelux for background lighting and one cool white Bridgelux with a focusing reflector for bright spot lighting. The three LED are in series. They would ordinarily all be on, but then we add a special circuit to short out the cool white spot at low levels, release the short at high levels. That would be fun! Could be scaled up to serve your desires, and would be most efficient I think.

But I think you would be happier with an entirely different approach, as you originally asked. What you need is to get from me or someone a box that connects your 230VAC dimmer output, that has a trigger adjustment level and a switched output that you could do whatever you want with. There could be such a general-purpose product, but I don't know of one...

Which directions would you like to discuss further?

background resources:

http://www.evecto.com/dimmers_1.html
DIMMTRONIC Leading Edge and Trailing Edge Dimmers

http://sound.westhost.com/lamps/dimmers.html
dimmer circuits and waveforms

www.iwatt.com/downloads/Reference%20Design%20EBC864.pdf
Dimmable Offline LED Driver with iW3610
complete schematic

 

[evilc66]

Technical question for driver geeks:

Do most dimmable constant-current drivers put out a variable DC current, or a variable-duty-cycle waveform?

Depends on the driver, but most switched inductor buck drivers try to emulate a DC voltage. There is a little ripple on the output, and that will be defined by the filtering and the drive frequency of the driver IC, but typically it's considered just DC. The duty cycle of the drive frequency from the IC will set the current level, with the inductor and filter components averaging and smoothing the output. At least, that's what I've seen in my experience with that type of driver.

 

[evilc66]

You know what might be the easiest overall? Use a small, simple micro controller like an Arduino as the interface. That way you can have multiple control sources, as well as multiple drivers hooked up, and you can make it do whatever you like. The coding for it isn't terribly hard, and the interfacing electronics are pretty straight forward. Plus, you can find them under $20 if you don't mind soldering all the parts yourself.

 

[Mash]

Thats a great idea, hadnt thought of that! Looked at arduino before for other purposes, but never got into it!

On another note, what happens if I use a relay on the AC input side? ie the ac input will be controlled by the dimmer, so could i set up a relay to switch on the current to the second driver according to the AC voltage change? his way I would cut out the problems of detection on the DC output side, but would it be problem free?
KETHD, I would love it if you can show me the plans for a ready or easy to make power semiconductor setup as mentioned by yourself! Just a simple cicuit to switch on the AC side of the second dimmer at a preset level. This way the 2nd driver is not running all the time, and also dont have to think about the max load on the circuit when mixing and matching different multiple LEDs, as you mention above.

 

[kethd]

Google Solid State Relay.

Let's focus on SSRs. You need a modified-input version of SSR. You can easily find many SSR circuits online. If you had electronics background, we could go the route of doing something bespoke from scratch. But in your case I think we should start with a commercial SSR and add an input circuit matched to your needs.

You can readily get SSRs from Amazon, eBay, Jameco, DrillSpot etc. You want something that can comfortably handle outputs up to 230VAC plus, and has sensitive low level DC inputs in the 5-24VDC range. Pick out a couple SSRs that look attractive to you, and we'll discuss them, then work on an input circuit.

(PS: What current level are you thinking of running your Crees at?)

 

[Mash]

I think 24v would be too low, since drivers supply max of 32V on each leg, unless we do a circuit (which I think we will need, since the SSRs are not adjustable, ie ANY voltage in the 5-24 range would open them). Current level would be 350mA, or less likely 700mA.
Did you happen to see my previous post re controlling the second driver from the AC side? Any relay or other adjustable "threshold" activated device out there? Something that eg, block the output if its below eg 180V, but as you turn up the dimmer it allows it to go through until the max 230V (would ZENER DIODE rated at eg 180V work)?

HAd a look at a supplier, and found something like these as examples:
http://www.rapidonline.com/Electron...ys-PCBs-24V-SSR13/200035/kw/Solid+State+Relay

http://www.rapidonline.com/Electron...s-with-LEDs-SSR02/200032/kw/Solid+State+Relay


If Im reading the spec correct, seems good.

 

[kethd]

We must add a circuit to the SSR input, so we probably want a low rated input, around 5V. SSR inputs are naturally low-volt, since we want an opto-isolated type, so really we are basically driving an LED inside the SSR.

The plan now is to try to control the second driver from the AC side. I had been hoping to control it independently of the Cree driver. But now I am thinking that working with your very high peak line voltage (almost 400Vdc) is very inconvenient, so it would be better for our SSR input circuit to be connected to the Crees. We can use a current-sensiving resistor of about 1 or 2 ohms, and separately leech off enough current to power the circuit, hopefully only around 10mA.

The physical package of the PCBs 24V SSR13 SSRs you found worry me a little because you need to keep the inputs and outputs very separate -- otherwise, your Crees will be connected to your line voltage and there will be a shock hazard.

The SSR02 range of solid state relays look better - nice big separate screw terminals. Nice range of DC input and nice range of AC output. The drawbacks are that they are rated for immensely more load current than you need. This could be good (because it implies very low ON insertion loss at your light load) but it could be bad because there might be OFF leakage currents that would be significant in your usage. I am also worried that the output is AC-only, just crude SCR which means no ability to turn off output except when load current passes through zero. Probably would work, but since both your AC from the dimmer and your second driver input load might be "odd", I might be more comfortable with an SSR with hi-volt AC/DC output based on MOSFET or something else that is able to turn off in a more definite way. (Or, maybe it is safer to only turn off when the current goes through zero, if your load is inductive.) I'll look around.

If you got a mechanical relay with 230VAC inputs and 230VAC outputs, it MIGHT do exactly what you want, with no modifications! Mechanical relays have inherent on/off hysteresis, which is convenient for you, though it could be too much. The trigger point would be repeatable enough for you. If it was too sensitive, you could add series resistance. (Using a zener instead of a resistor would be trickier.) But if it is not sensitive enough, you would have to get one that runs from a lower input voltage. You could get a relay with 120VAC input and add lots of series resistance, then reduce the series resistance until you get the trigger point you want. If you screw up, you'll burn out the input. But relay inputs are inductive, which could be a problem for your dimmer. The odd nature of your dimmer source and LED driver load (inductive?) could be a problem for the relay output contacts. There could be audible hum noise, click-clacks etc. You might be lucky, and love this low-tech approach. But from a design perspective, it does not allow the control of the trigger point and trigger hysteresis we would want.

You can't take this crude approach to SSR etc non-mechanical relays because they do not have built-in hysteresis. Their trigger point is not just unpredictable, but actually a danger zone. They can end up flickering or partly-on, stressing the load and the dimmer and the SSR itself -- parts could overheat, burn out, go up in flames, etc.

The SSR06 might be OK:
Panel Mount SSR 4-32VDC, 0-380VAC 10A load with LED, Zero Cross 60-2052 £13.80

The cheap SSR04 is tempting:
Solid State Relay, 9.6-14.4VDC, Random, input current 10mA 60-2066 £3.16

It's tempting to just start with the SSR09 (but perhaps unsafe/unwise):
PCB Mount SSR, DIP Pin-out, Load Voltage 400VAC, Control current 5mA-20mA 60-2064 £2.84

 

[kethd]

circuit design

Input from a 1 or 2 ohm current-sensing resistor in series with the Cree LEDs.

Power supply tapped from two or three Crees, about 9-10V, will only steal about 10mA.

Circuit based on LM393 comparator with adjustable trigger point and a little hysteresis.

Output to 5Vdc-in SSR.

I'll try to sketch out a circuit and find some way to post it...

 

[kethd]

Here is an attempt to sketch a circuit in ASCII:

SSR TRIGGERED DRIVER CIRCUIT - 21 OCT 2010

CREE Driver #1
POS
 |
 VCREE
 - LED
 |----------->|----------o-----o-----------o----------o
 |         1N4001        |     |           |          |
 VCREE                   |    .-.          |         .-.
 - LED                   |    | |          |         | |
 |                       |    | |100K      |         | |1K
 |                       |+   '-'          |         '-'
 VCREE            100uF ===    |           |          |
 - LED             16V  /-\    |           |  ___     | .------.
 |                       |    .-.    /-------|1M_|-\  \ |      |
 |                       |    | |<--\      |       |   \|      |o
 VCREE                   |    | |10K \  |\ |       |    | SSR  |
 - LED                   |    '-'     \-|+ \       |    |      |
 |          ___          |     |        |LM393>----o----|      |o
 |---------|10K|--o---------------------|- /            |      |
.-.               |+     |     |        |/ |            '------'
| |CSR           ===     |     |           |          from DIMMER
| |1-2 OHM       /-\     |     |           |         to DRIVER #2
'-'               |      |     |           |            BRIDGELUX
 |----------------o------o-----o-----------o
NEG              10uF
                  3V

(created by AACircuit v1.28.6 beta 04/19/05 www.tech-chat.de)

If it seems all garbled, you'll have to copy it into a fixed-font display mode.

The good news is, it is not that many parts, and should not be too sensitive to construction details. The parts are all common and cheap, aside from the SSR -- and even that could be quite cheap.

The bad news is, if you are not comfortable building a circuit that includes an integrated circuit, this could be quite a challenge!

 

[Mash]

Thank you very much sir for taking the time to explain and draw the plans, it is really appreciated!
I understand excatly your point about the hysterisis of the trigger point, where if you are just on the tirgger limit, one instant you could be above by a small margin and one instant below, therefore the system keeps switching states very rapidly which is not a good thing!
Your circuit design looks great, and from looking at it, I think the capacitors allow for a little variation in the voltage to combat the hysterisis, and the 10K resistor would be a pot for adjusting the trigger point, am I reading it correctly?
I am quite comfortably soldering kits etc, however the usual problem I have is transferring the circuit design to a physical board, and the way to build the board (etching etc)!
Again thank you very much for your info!

 

[kethd]

I am encouraged by your response! I didn't know how far you'd be able to stick with this...

Yes, you are keeping me awake at night -- but I am happy to try to help, hoping that maybe others might find this exercise helpful someday.

The capacitors actually are not related to the hysteresis. The 10uF is to ensure we have a steady dimmer-level signal, regardless of ripple etc in the driver waveform. The 100uF is to ensure a steady power supply voltage. They are both probably larger than absolutely necessary -- this is just a draft circuit.

If we wanted an absolutely rock solid trigger point we would further stabilize the reference voltage, but I think this is adequate for your needs, and I am trying to keep the parts count down (I am a perfectionist and I have a tendency to get carried away).

Yes, the 10K adjusts the trigger point. Ideally it would be ten-turn, but you could get away with single-turn. Wire everthing up, turn everything on, set the dimmer where you want the trigger point to be, then adjust the pot (carefully) to put the trigger point right there.

The 1M is positive feedback which controls the amount of hysteresis. If it is more than you want, make this resistor bigger up to 10M. Or you could make it smaller down to perhaps 100K.

Almost the same circuit could be used to either turn the second driver off or on at the trigger point -- or there could be two SSRs, one that turned on and one that turned off...

Note also that you do not need the Bridgelux driver to be dimmable. The second driver could just go straight to full on if you wire the SSR output from mains instead of from the dimmer; perhaps you could use a cheaper driver?

With a circuit this simple there is no need at all for an etched board. You could just plug the parts into a protoboard and they would work fine. You could put them in a bare perf board and run wires. If you want to be very neat and fancy you could use those little radio shack boards that have a place to solder an IC with modular copper traces over the perf holes to arrange your connecting circuit.

The heart of building this is the LM393 8-pin DIP. Build and mount it close to the SSR. The whole core circuit should only take up about a square inch or two. The wires from the Cree chain should be able to be long, several feet - the two caps should take care of things...

Most circuits like this are happier if they are more compact. But don't focus on that. If you have problems translating the circuit to a physical layout, it is OK to try to make the physical layout a lot like the circuit drawing, let it spread out some so that you can easily see and understand what is what.

If you have built kits before, with integrated circuits (and they eventually worked), you can do this -- even if it doesn't work the first time... you can always post pictures and we can make suggestions...

 

[Mash]

kethd! You said you are encouraged by my respose!
:grin2:
Buddy, I wouldnt be asking you for info, and your time, thinking and writing here, if I didnt value your input; but thanks for the comment! As alway your time and knowledge is very much appreciated!


By coincidence, today I was in a light and electrical shop, and was talking to the guy who is pretty knowledgable, re dimmable drivers, double dimmers etc.
And he said he will contact someone he knows, who builds specialist drivers etc and enquire about this situation we are discussing here. Then I come home and see your great contribution!
Now, I had an idea and wanted to ask your permission first.
I want to give your circuit plan to this guy to build for me, but given the nature of his business, he might in the future build them for his own use/sale as well, ie use your idea.
But I wanted to ask you first, and will only do the above if you are ok with it. Otherwise will just try to build the one for myself!
Now all this is happening in the UK, and the chances of you maybe finding out one day etc etc were very slim, BUT I have my principles and I like sticking to them, especially if somone like you puts in the effort to helping me!
Thanks again!

 

[kethd]

It would be more practical for you to have someone build this for you. There are always details that come up along the way that need fine tuning. For example, I was thinking last night that there should be another resistor from the output of the LM393 up to the positive voltage rail (maybe about 100K), because the open collector output will otherwise be floating and poorly defined when the output is off, due to the SSR input being an unpredictable load...

This would be a good project for a technical class learning electronics design. (And very timely with the green-eco connection.)

I greatly appreciate your consideration and principles. I'd love to be able to profit from my knowledge and ideas, but I don't think that is practical in this case. My goal is to be helpful in general, and help promote LED lighting in particular. I give these ideas and designs in the spirit of "open hardware design" Creative Commons copyleft. I would ask that anyone building on my ideas likewise openly publish their refinements, and not assert copyright/patent exclusive ownership. I have no problem with anyone profiting from actually building anything -- I encourage it! (As long as everyone remains free to copy the designs for themselves, or to sell in turn...)

It would be fine if someone turns a circuit like this into a little PCB and sells it as a bare board or kit or populated, with or without the inclusion of the SSR. It would be pretty easy to make it pretty adaptable, to all situations where one wanted a secondary load triggered on/off by the state of an LED driver...

However, I fear that the market for such a specialized product would be incredibly tiny -- probably too small to be worth pursuing.

I think that my previous embodiment would be a much more commercial product. Imagine a black box. It accepts AC mains and AC dimmed inputs. It outputs AC mains and/or AC dimmed. Each of these can be on all the time or switched ON/OFF before or after the trigger points. It could allow all kinds of combinations. It would be quite easy to design such a multi-purpose modular product. It might have many uses that we have not thought of yet. You could use it for your purposes, and there might be an actual market for it.

There might as well be a version that actually includes the dimmer itself -- then there is just the mains input and various outputs, maybe with some switches that control the modes. It could be a finished ordinary box that could be exposed, or hidden with the wiring, or a compact module that would fit in the same space/usage as an ordinary in-wall dimmer switch. Of course, at some point you get into safety standards, official testing etc and then the whole thing gets so bogged down that it is not worth doing, since the market is probably not really that big... Someone in the far east could do it, sell on eBay, avoid the regulations?

This is a much less "efficient" approach than the circuit I drew for you, because of the challenge of deriving the low-voltage power supply that the circuit needs from the full line voltage. But it can be done, and done well, and because it makes a much better general-purpose product I think it is the way to go, commerically.

In any event, keep us updated on developments -- and no matter what, post good pictures of the innards and outards of whatever you end up doing!

 

[Mash]

Nice to see you being so generous with yoru work!
If you have the time, do you want to do a finalised version of your circuit, with all your improvements and additions, and I will check to see if the guy is willing to play!