Senior Member
Ok so we are up to:
(Updated at full speed)
X - 12 bit
Y - 12 bit
RGB - 8 bit (can change the B to a V in software)
(Updated not as fast as 60 kpps)
1 Shutter Output
7 TTL Outputs for Misc.
60 kpps constant? Can adjust in software by doing multiples of points or add software control point (point pulling algorithms) to reduce the speed seen by the scanners.
No frame buffer.
$100 target cost.
How is that for a spec list?
Senior Member
With a FT232R the limit is 3 Mbps... 60 would not be possible with that. You would need around 4.32 Mbps. 30 kpps is possible for that 2.4 Mbps.
Bit Rate = ((12 * 2) + (8 * 3) + 8) * PPS
If we go with a straight USB micro ... we leave out a lot of hobby folks with the programming and understanding of it. It also makes the boot loader more difficult and such too.
Does anyone have any really big objections to a 30kpps limit?
Creaky Old Award Winning Bastard Technologist
Dont use maxims unobtainium parts and NO PWM dacing.Originally Posted by cfavreau
Steve
Qui habet Christos, habet Vitam!
I should have rented the space under my name for advertising.
When I still could have...
Senior Member
Good advice.
The design so far is a spec list. We can add no maxim parts and no PWM for analog out to that list too.
I was not thinking of using either.
Senior Member
Never mind about the 30 kpps limits of the USB chip.
What is everyone's opinion on 30k vs 60k pps. Keep in mind Budget System. Also keep in mind that maybe once it is up and running and we can play with it it maybe able to do 60 kpps if the right parts and pieces are picked.
The point output rate adjustability is dertermined by the architecture of the micro and USB interface.
Senior Member
Fail. We need both a B and a V. Why not just use the same DACs for X, Y, R, G, B and V? That way those suitably inclined can use the spare channel if they have one as a Z or whatever.
We should also have an accessible I2C output from the micro that people can hook beam table actuators to. Simple PCF chip or two.
The limits of the USB chip need not be prohibitive if we use a little compression (like, for example, RLL or even a simple Huffman thing) on the link. This need not stop people from programming it themselves if we provide our code as a library (preferably with source, so they can modify and improve it).
If we use something like an ATmega644 that gets us 4 kbytes of RAM, which is plenty to have a small framebuffer.
-J.
Last edited by heroic; 05-12-2009 at 16:52.
Senior Member
Why the Violet channel ... does the standard projector have RGB and V or just RGB or RGV? I only have 3 lasers (PCAOM's excluded of course). I have seen quite a few others that only use 3. Steve is probably right about the Quad DAC being the same price... Just trying to keep it simple though. The more data you have per point, the more memory you need and the more bandwidth you need to transfer it. Just trying to keep that stuff in check too.
An ATMega644 is definitely high on the list for a choice. 20 MHz ~ 20 MIPS. Lots of pins and 4k of RAM. If a FT245R is used instead we have up to 8 Mbps... which should be Ok even for RGBV. Although right now it is best to keep it to a specs list. I bet someone else already has something that will fit the bill designed.
Not sure about the I2C... Anyone else want to chime in? I think good old TTL will work well and fit most people's projector. Good idea though... worth exploring.
Senior Member
An increasing number of people are going RGBV. It gets you a much wider gamut than RGB and far better colours than RGV.
There is a difference between data and information; a simple entropy coding scheme will reduce the data while keeping the information.The more data you have per point, the more memory you need and the more bandwidth you need to transfer it. Just trying to keep that stuff in check too.
It's called SanguinoAn ATMega644 is definitely high on the list for a choice. 20 MHz ~ 20 MIPS. Lots of pins and 4k of RAM. If a FT245R is used instead we have up to 8 Mbps... which should be Ok even for RGBV. Although right now it is best to keep it to a specs list. I bet someone else already has something that will fit the bill designed.
I want both :-)Not sure about the I2C... Anyone else want to chime in? I think good old TTL will work well and fit most people's projector. Good idea though... worth exploring.
Senior Member
No on the 60 (or max) kpps constant. While it's possible to get close to a target point rate by doing point repeats (such as I do with the audio DAC), why do it if you don't have to? The mcu can clock the points out at almost any arbitrary rate, so let's take advantage of that.
Speaking of clocking out points at a fixed rate, consider that the point output timer interrupt should have highest priority for low jitter but at the same time we don't want its timing requirements to interfere with USB communications. The ftdi chip would probably allow this, but it's not clear to me if internal usb control would.
Senior Member
60 kpps max output rate - variable (from 1kpps to 60kpps)?
So these specs pretty much limit us to the FT245R (the parallel version). Max output rate 60kpps. Internal USB maybe... would have to do experiments to see if it can work.
Does any one have any experience with working with micros (specifically the AT90USB family) to know if we can do USB and have a 60,000 time per second interrupt go off with no interruptions?
We have exactly 16.66666666 microseconds to do the point output and do other stuff too. Obviously other stuff can be split up between points if need be... but we essentially have to transfer and output 8 bytes in 16 us (round down for overhead). At 20 MHz instruction clock rate we have 320 instructions max to do so.
Not to worry about the hardware yet though... have to nail down the specs.
So summary:
USB Interface
60 kpps max output rate (1 to 60 kpps variable)
12 bit X/Y
8 bit RGB(V to be determined)
8 bit TTL (including shutter)
If we don't use the serial port we can throw in DMX too...
Posting Permissions
Reply With Quote