8+ bit LED fixtures on DMX

[deranged-angel]

Hi,

I've scouted the internet and can't seem to find what I'm after so I'm hoping you lovely people will be able to help.

 

I want to use an LED fixture but it needs to be more than 8bit. It's not easy to get good data sheets on these things to determine whether the fixtures have 16bit for each colour or 16 bit overall, I.e divided by RGB. Has anyone used LED fixtures with more than 8bit? 10 bit would do, really but 16 would be nice. Also, how are they controlled when they are via DMX? Obviously, a DMX cable can only handle 8bit so is there some way of using a couple of channels per colour?

 

Hope this makes sense.

 

Thanks

Emma.

[daemon]

Generally most fixtures run 8-bit per colour, giving 24 overall and 16.7million possible colours, which is also more than your "16-bit overall i.e. divided by RGB".

[csg]

why do you need more than 8 bit per channel control?

 

If the argument is concerning fine and very precise control of colour, then you will be better looking at fixtures which use rgb + white and amber. these fixtures still use 8 bit control, but are much more capable of providing subtle colours.

 

ive never seen a fixture capable of 16 bit operation, but the principle would remain the same as moving light 16 bit movent control - ie ganging 2 channels together with a course and fine channel combining to give the finer resolution control. this would require suitable software for whatever control platform you use

[niclights]

I can understand the demand for more precise control but I don't know of any, which is rather useless.

 

But for implementation the console/soft must be capable. As said above it is done with two channels where software allows definitions of the two associated 16-bit channels and then handles them automatically as one where, for each single increment of the 'course' channel the 'fine' channel will go through its whole range. This will often be represented as decimal points in percentage notation and for numeric input of values. Common on P/T now for the majority of moving fixtures.

[peter]

EvenLED is 16-bit LED dimming and uses 2 DMX channels as described by other posters. Our control desks all provide 16-bit RGB control where possible on the fixture.

[deranged-angel]

hmmmm thanks for the info there.

The reason I need more than 8 bit per colour is for colour matching. Some of you may have answered other posts I've put on here regarding my MA project which is to match the colour of Shakespearean stage lighting using LED. I have used an LED Par that had 8 bit per colour and when trying to colour match, if I added, say 1 more unit of red (out of 255) then the colour became just too red. I need that higher bit rate to fine tune the colour.

I've looked at using white and amber with it but not closely yet. Will have a look, thanks!

I kind of wanted to make this project something that I could give a solution for so that people with a couple of LED pars would be able to pick up and use, should they want to make their stage look like it was lit like in Shakespeare's time. I'm playing with an Arduino at the mo - some programmoing software to implement PWM control but I kind of wanted to have an element of using technology that was "to hand" as it were, so that people don't have to go out any buy a load of LEDs and some complicated kit.

 

What I don't understand is how the fine tuning works. If you have one channel on your desk that is your normal 0-255, which is the max a DMX cable can carry, how can a fine tune work?! How can this have more steps in it? Niclights, you say that it works in decimal points but I dont see how this is possible with PWM control which most desks run off. Or am I wrong?

 

Sorry for all the questions. It's all a bit jumbled in my head!

Thanks

Emma.

[Tomo]

What I don't understand is how the fine tuning works. If you have one channel on your desk that is your normal 0-255, which is the max a DMX cable can carry, how can a fine tune work?! How can this have more steps in it? Niclights, you say that it works in decimal points but I dont see how this is possible with PWM control which most desks run off. Or am I wrong?

You've misunderstood.

Consoles don't use PWM* - that's purely a dimming technique that's used by some designs of dimmer.

 

To properly understand DMX control outside of dimmers you need to stop thinking of the DMX stream as a series of channels - they're not.

DMX Consoles actually output up to 512 bytes per universe.

 

How the receiving devices interpret those bytes is entirely up to them.

What's a 16-bit number? It's two 8-bit numbers stuck together.

 

Many consoles that understand 16-bit attributes can display this by showing 16-bit attributes as 0-65535, and 8-bit attributes as 0-255.

 

*Not strictly true - some analogue output consoles use PWM to generate the analogue voltages.

[deranged-angel]

Thanks Tomo,

Time to get on to the blue-room wiki me thinks and properly understand DMX! I knew it was DMX 512 but somehow 8bits 0-255 got wedged in my brain somewhere alongside it. Will do some further reading then may get back to you on this subject! Thanks all for your help.

Emma.

[smalljoshua]

The DMX specification allows 512 8bit bytes per frame.

 

A 8bit byte allows a range from 0-255. There are a maximum of 256 possible values that the single byte can have.

A 16bit byte allows a range from 0-65535. There are a maximum of 65536 values that the 16bit byte can take. This is because 256*256=65536.

 

To get a 16bit byte you use 2 8bit bytes from each DMX frame. If you imagine that the Lighting Desk works in a percentage with an infinite number of decimals.

 

To work out a 8bit byte the desk divides the percentage by 100. It then multiplies it by 256, because that is the number of values that an 8bit byte can have.

 

To work out a 16bit byte the desk divides the percentage by 100, as with an a bit value. It then multiplies this by 65536, because that is the number of values that an 8bit byte can have. The desk then has the 16 bit value. The best way to explain the next bit is with an example. If you have a 16 bit value of 12755 you Davide it by 256 then round it down you get 49. Multiply this by 256 to get 12544. Subtract that from 12755 to get 211. These are then your 2 8bit values. The course channel would have a value of 49 and the fine channel would have a value of 211.

 

HTH

 

Josh

[paulears]

I'm confused, you can see a single step on the red channel? I can imagine a 1 unit increase on a 1 to 10 scale being seen, but 1 on a 1 to 100% dimmer channel is tiny - using 16 bit resolution is surely totally pointless. Or is it the brightness curve on the fixture is seriously uneven, and maybe only has a fixed number of steps between off and full? As in nothing happens until the DMX level gets to the next step?

 

On paper 16 million colours is an awful lot - far more than we can discriminate - so something odd is happening here - isn't it?

[deranged-angel]

Paulears - yes, because I'm matching the colour of a candle, the LED is only 1 foot away from the white screen that it is being compared on to. That is why it seems to obvious. There is actually a massive difference. This isn't down to dimmer curve as I have already done experiments with a lux meter and dimmer and the level is was on is in a very steady section of the graph.

 

I suppose once it's in the rig, you might not notice it so much but I don't think they'll give me an MA degree for saying that! Haha. It has to be precise and it's part of looking at the use and suitability of LEDs too.

 

Now then....back to DMX. Me, oh my, my little brain is turning to slush.

 

I've got some conflicting suggestions from here and with wiki. Wiki says:

 

Using these types of devices on older lighting controllers would result in two adjacent channel controls being used to adjust a single movement axis. One would be referred to as the coarse and the other as fine, indicating the relative amount of movement control each channel provided. The coarse channel would allow values in multiples of 256, such as 0, 256, 512, 1024, all the way up to 65280. The fine channel allows the addressing of all in-between values, by adding between 0 and 255 to the value obtained by the coarse channel. Thus the fixture's movement can be controlled more accurately.

 

So, I've got my 16bit LED Parcan. As DMX can only carry 8 bit (knew I'd not plucked it out of the air! Wikipedia tells me this is so), This needs to be broken down in to 2 channels - a coarse channel and a fine channel.

The way I read it from the wiki page is like this:

Lets say I want to initially put my red LED at 50%. This would be a value on the coarse channel of 32640 but realistically, it would be 126 on the 8bit signal.

Now, I want to tweak it ever so slightly. I then go to my fine tuning channel and start to move it from 0-255. This adds on to the value already on the coarse channel.

 

Is that right? Or is small Joshua right? Or did he actually say the above but I just didnt get it?

 

Oh mushy mushy brain please solidify so I can use you again.

Sorry if I'm being thick - it's just not sinking in at all.

 

Emma.

[gyro_gearloose]

Just to be pedantic for a moment, all bytes are 8 bits. If its 16 bits its called a word and if its four bits its called a nibble.

 

PWM on anArduino is simple. They have 3 or 6 channels of PWM dimming built in depending on which CPU you are using. Its very easy to get them to receive DMX as well. I've written some code which receives DMX and controls two LED driver chips. This give me 32 channels of dimmable LEDs. I've hooked this up to 10 Piranha RGB LEDs to make a simple row of pixels. I'm controlling them with MagicQ and I can get smooth fades between colours.

 

 

EDIT : You are almost right. 50% would be 127 on the coarse channel and 255 on the fine channel.

[smalljoshua]

Lets say I want to initially put my red LED at 50%. This would be a value on the coarse channel of 32640 but realistically, it would be 126 on the 8bit signal.

Now, I want to tweak it ever so slightly. I then go to my fine tuning channel and start to move it from 0-255. This adds on to the value already on the coarse channel.

That is basically what I said but you put it differently.

 

Josh

[boatman]

.

.

Is that right? Or is small Joshua right? Or did he actually say the above but I just didnt get it?

.

.

 

Yes, Yes and Yes.

 

I'm with Paulears on this one and find it very unlikely that you would notice a 1-bit change in a true 8-bit PWM control. I suspect that you will find the particular LED fixture you are using doesn't actually have an 8-bit control channel but only 5 or 6 bits; receiving 8-bits from the DMX line but ignoring some of the least significant bits in the control circuit.

[gyro_gearloose]

The way I worked it out was this : 16-bit numbers range from 0 to 65535, so 50% is 65535/2=32767.5. Since you can't have fractions this gets rounded down to 32767. I then divided this by 256 to get 127.99609375. We keep the 127 as the coarse byte and multiply 0.99609375 by 256 to get 255. This gives us the fine byte.

 

 

Or you can cheat and use the Hex function of a calculator....