DIY LED driver for reef lighting

[jm82792]

Check your library. If they are not real busy they might help. Then maybe your local copy shop.

Good idea!
I'll go to my library,
it's really small and empty so is should be set.
I could probably print the thing off without even bothering to ask,
other than paying the $.10 a page they require.
For the transfer paper doesn't glossy and non recycled magazine paper work decently?
I got all my parts yesterday and will first for fun do my RTC.
I tried to breadboard it out but it didnt work,
I think it's because I tried it without the 2.7K pullups and it being so late I wasn't thinking...

 

[der_wille_zur_macht]

FWIW On an Arduino, the internal pullup resistors are automatically activated when you use the Wire library. USUALLY that's enough to get I2C working though IMHO they are too weak to rely on.

 

[TheFishMan65]

dwzm, I have bad news. You wanted efficient drivers. Well from an electricity usage the analog dimming is better than PWM :sad1: according to a seminar I watch yesterday. Me too, oh well.

 

[der_wille_zur_macht]

Do you have any data from the seminar to qualify that? What type of driver was it focusing on?

With my revised lighting plan for the 360g I'm really not AS worried about efficiency, since I'll only have about 300w of LEDs anyways. But still it would be interesting to learn more.

 

[TheFishMan65]

Let's say you want 50% light.

So with PWM you are running a 50% cycle. So assume XP-G and IIRC that is 3.3 volts at 1 amp and about 340 lumen. So power would be 3.3 volt * 1 amp * 50% = 1.65 watts.

Now for to get 170 lumen by changing the current we only need about 450 ma so the voltage drop is 3.1 volts or 1.395 watts.

Few other points
Analog
- Quiet no switching noise
- Better efficiency as current decreases
- Wavelength varies due to band filling and quantum-confined start effect
Digital
- Noisier input needs filtering
- Lower efficiency
- color changes from junction temperature

DWZM check your email

 

[der_wille_zur_macht]

Ah ok, I get it. The efficiency difference is due to the inherent efficiency/current slope. That makes sense, though it's probably not as bad as it might seem, since in the real world, some of the efficiency loss at higher current is due to higher temps, which wouldn't be as significant with PWM.

At any rate, this should reinforce a standard practice: use analog methods (and/or driver design, i.e. sense resistor) to get a "base" current that you'll be running at most of the time, then use whatever means is convenient to dim up and down for dawn/dusk effects, keeping in mind that PWM isn't QUITE as good as analog.

Though it's probably still arguable that people need to be careful when applying this to commercially available drivers, or "black box" drivers, because I wouldn't be surprised if some of them are doing some sort of signal conversion internally.

 

[TheFishMan65]

I am not sure it can get much better the PWM is about 85% as efficient as the the current controlled. This was using the charts (yeah I ended up looking it up, because there were to many unknowns) for a junction temperature of 25C. I would think the PWM circuit would run hotter than the current controlled version and could only get worse for efficiency.

The design goals seem correct.

 

[terahz]

If the driver is configured to run on 100% PWM all the time and pwm is used just for dimming, I doubt the difference in efficiency will be noticeable, but what you say makes sense. I guess the other option is to put a pot on rsense and use that for dimming

 

[TheFishMan65]

terahz, I am a little confused by your wording. I think you are getting that if 100% PWM is considered your daylights hours then there is the difference in efficiency then. The dawn and dusk would be controlled by PWM and some efficiency would be gained with a current controlled process.

 

[Obi-dad]

That makes sense on the difference in efficiency at the LED - what about the efficiency of the analog driver vs the digital driver (considering only internal efficiency of the driver circuit)? For example, I think some of the analog voltage convertor ICs get very hot when throttling them back. Or, if one is talking about using a variable resistor to throttle back voltage on a power supply, that will produce heat (less efficiency).

So if the internal analog driver efficiency is less than the CAT4101 driver efficiency, then maybe the whole setup (driver efficiency + LED efficiency) is a wash.

Anyone know how efficient analog drivers are internally?

 

[va7zt]

OK, stupid question time! I'm sure this has been answered but I don't recall. With the 4101 driver, do you need the 5V to test it? I would like to test them as I build them but my 5V source is in my light hood. At present I don't care about the PWM, just want to make sure the drivers are working.

 

[TheFishMan65]

Yes you need 5 volt to power the CAT4101 and something must drive the PWM. Constant 5 volt to the PWM is ok.

They did not mention driver efficiency that I remember - put I only caught part of that seminar. Listen to it yurself the link is on the DIY LED thread.

 

[terahz]

terahz, I am a little confused by your wording. I think you are getting that if 100% PWM is considered your daylights hours then there is the difference in efficiency then. The dawn and dusk would be controlled by PWM and some efficiency would be gained with a current controlled process.

My point was that dimming with PWM is really a problem if the driver is configured to run at say 1A, but it is PWMed all the time at 80% to get equivalent of 800mA without changing the resistor. In that case, yes you are much less efficient.

But if you have the driver configured at 800mA and stay on 100% PWM (which is basically no PWM) all the time except dusk and dawn it is probably not too bad to use PWM for dusk and dawn.

Maybe a digital pot instead of fixed Rset will be best, but I don't know if it is worth the complexity.

 

[TheFishMan65]

I started to go that way. But wanted to get the lights done before I started the controller part so a true pot.

We have no said the same thing 3 different ways hopefully everyone else will understand us now

 

[der_wille_zur_macht]

analog driver vs the digital driver

If you mean linear vs switching then in general the switching reg will be more efficient over a wider operating range. But since our operating ranges are essentially known ahead of time, we can design out that typical difference in most circumstances. See all the recent discussion related to matching the power supply's voltage to the minimum requirement for the CAT4101. If you do that, it'll be as efficient or more so than a switching reg like the NCP3066.

On the other hand, if you were trying to design a driver for something where voltage was all over the map (say, a battery powered flashlight, or automotive lighting system, where voltage can vary by several volts), then the switching reg would probably win out on average, or you might even be FORCED to use a switching reg (in SEPIC topology) if the input voltage potentially varied both above and below the desired output voltage.

 

[Obi-dad]

If you mean linear vs switching then in general the switching reg will be more efficient over a wider operating range. But since our operating ranges are essentially known ahead of time, we can design out that typical difference in most circumstances. See all the recent discussion related to matching the power supply's voltage to the minimum requirement for the CAT4101. If you do that, it'll be as efficient or more so than a switching reg like the NCP3066.

On the other hand, if you were trying to design a driver for something where voltage was all over the map (say, a battery powered flashlight, or automotive lighting system, where voltage can vary by several volts), then the switching reg would probably win out on average, or you might even be FORCED to use a switching reg (in SEPIC topology) if the input voltage potentially varied both above and below the desired output voltage.

I was a little confused by the terminology - I was thinking that the CAT4101 was PWM. Never mind, that is only the control signal to the CAT4101

 

[der_wille_zur_macht]

I was a little confused by the terminology - I was thinking that the CAT4101 was PWM. Never mind, that is only the control signal to the CAT4101

Yep, you're right. It's a linear driver with respect to the output. It takes a PWM signal, or an analog signal, for dimming PWM via the PWM pin, analog via the sense resistor pin. By design, nearly any of these IC's can take an analog signal, since you can just trim your analog signal down to some safe voltage range and add it to the sense resistor pin on the IC.

 

[jm82792]

FWIW On an Arduino, the internal pullup resistors are automatically activated when you use the Wire library. USUALLY that's enough to get I2C working though IMHO they are too weak to rely on.

Okay I see.
When I get the board for the RTC etched and get it working with the internals(it does now) I'll use my own, I'm not sure how but I'll figure it out.
I'm waiting for my doubled sided clads still,
when I get them then I can give all of the etching and such a shot.
I bought extra CAT4101's because I know I'll mess up at least once.

 

[der_wille_zur_macht]

I wouldn't even bother trying to get it working with the internal pullups. Just put the externals on and forget the internals are there. They're really weak. For external pullups, you just need a resistor from each of the two I2C busses to +5v. You can even splice them into a wiring harness if you're using one, or solder them anywhere on any board on the I2C network, stick them in headers, etc.

 

[jm82792]

Okay I see, take 5V from arduino,
then through the resistor(s) and give it to the 2 IC pins

What value should I use?
2.7K?
I've seen a few different values on a few different arduino related RTC(same chip) websites.
I'll buy them and integrate them into the PCB that I've yet to finish because I need to get to Kona(hour away) for a drill.
I don't plan on using any other IC2 devices,
for the temperature readings I'm going to use some thermistors..

Thanks for starting this thread,
I've enjoyed the DIY journey