So the 595 will be source while ULN2803 is sink? What is your idea for the matrix control? Since normally a channel receives a single byte of dimming info will the dimming value be used to address the matrix (4bit->col/4bit->row). Just trying to wrap my head around how the DMX data will be interpreted... Will this just be on/off?
edit - too slow I guess - so dimming will be supported...
First 8 channels are Column control and will always to be set to 100% in Vixen
the next 8 channels will be the Row control and will be dimable.
However only ONE Column is to be selected in Vixen at a time.
Matt_Edwards wrote:First 8 channels are Column control and will always to be set to 100% in Vixen
the next 8 channels will be the Row control and will be dimable.
However only ONE Column is to be selected in Vixen at a time.
Really? Why is that? I understand the current demands would increase (might need to double up the ULN2803 outputs?). How would I turn on all bulbs in the matrix at the same time?
Sorry - I think I'm missing something -- I'll just sit back and perhaps it will make more sense as it progresses...
budude wrote:Really? Why is that? I understand the current demands would increase (might need to double up the ULN2803 outputs?). How would I turn on all bulbs in the matrix at the same time?Sorry - I think I'm missing something -- I'll just sit back and perhaps it will make more sense as it progresses...
Two resoons
the Row are driven by 595s and are capable of 35mA MAX.
the Series resistors are calculated (guessimated) based on one LED on only.
Matt_Edwards wrote:the Series resistors are calculated (guessimated) based on one LED on only.
Perhaps the limiting resistors should be off-loaded and have a separate resistor for each LED? This would be like what Frank did with his Super Strips. Yes - a lot more resistors but it makes it more flexible to different LEDs. I would have to hard-code in the colors for the LEDs otherwise and/or a row or column would have to all be the same type (or color) of LED?
budude wrote:Perhaps the limiting resistors should be off-loaded and have a separate resistor for each LED? This would be like what Frank did with his Super Strips. Yes - a lot more resistors but it makes it more flexible to different LEDs. I would have to hard-code in the colors for the LEDs otherwise and/or a row or column would have to all be the same type (or color) of LED?
This can be considered after the first Prototype is built and tested. I guess I should have called it a POC.
Personally I see only using 4 DMX channels: one for each colour and the 4th for sequence routines. It means adding more grunt into the uC. However, DMC wants to keep it as close to the original Thorpe concept and use Vixen for most of the control.
OK, I'm having trouble visualizing how that hardware design can adequately support dimming (at least with a large number of dimming levels).
Usually I think that the refresh rate should be at least 120 Hz, to minimize flicker and other effects related to a low refresh rate. With 256 levels of dimming (an assumption on my part) that would mean that all 60 (or so) LEDs would have to change state at a 30 Khz rate, or once every 33 μS. Since there are eight rows, each row would be on for about 4 μS (give or take a dead-time to prevent ghosting). While I'm not sure of the response time of an LED, I'll ignore that issue for now...
WIthin that 4 μS the micro has to determine which of the 8 LEDs in that row are on and clock out 16 bits of data to the shift registers. This is going to take a fair number of instructions, and I'm not sure if even the Atmel part is capable of doing this. Is there code or an example that shows that the part can do this? If not, I'd suggest some hardware changes. The first level of change would be to give the two shift registers separate clocks (and maybe separate serial input data pins). This allows one of the shift registers to be clocked at 1/8 the rate of the other one, if necessary. The other possible change would be to use a part that has more I/O pins and eliminate the shift registers altogether...this would allow the micro to avoid all of the clocking operations, and perform simple 8-bit IO writes.
I have been watching this project and like where it is going.
Even with its limitations it has a lot of potential.
Mouser lists ATtiny2313-20SI in the schematic as obsolete.
What is the replacment?
ATtiny2313-20PU is not expected until February.
ATtiny2313-20SU is soic, may mave to give it a try.
JOel
May I suggest wait a tick. the 20SI version is what my CAD library has. a Real BOM will be made available.
DMC is building prototype over the next couple of days. then we will see how it goes.
Cheers
Do you want to use a different pin for driving RCK? If you use the on-chip SPI peripheral (that will let you write the shift registers with two instructions, rather than bit-banging it), MISO gets set as an input, and can't be overridden (at least that's how it is on the ATMegas...)
