All three controllers up and running ...
6804, Parallax/Wiznet, Parallax/ENC28J60 running 36 count strings of GECE's as Univ 1, 2 and 3 respectivly...
At .05 and .10 LOR timings, they ALL appear to working equally... I choose that phrase because the all have realitively the same bahavior with the GECE's.
Sometime they will drop a pixel, or a color on one pixel on one pass will be wrong .. but overall.. I could video them running and you couldnt tell me which was which .. (if you can .. go by a Lotto ticket because you guess REALLY well...)
The only difference in setup via LOR is the ENC28J60 board is running UNICAST.. the other two are multicast ..
Bob
xLights setup
3 strings of 36 GECE's, One on each controller
Custom model as 3 vertical strings (looks good laying side by side on the floor)
Bars Effect, Garland Effect, Fire, On
Only goof is that I havent been able to make the RGB to BRG translation take place on the WizNet board.
Bob
1 string on each controller...
Sounds good, at least we know the enc28j60 can keep up with a 20hz refresh rate. Do you have any other pixel types to test with? I'd like to see if the enc28j60 can keep up with a 40hz (0.025 frame) timing.
I went back to all of the test sequences I have ran since the project started including the original LOR Chase sequence where it chased from Pixel 1 to Pixel 36 every second.
This was my original metronome test... and ALL three boards are in perfect sync.
Remember this is the one that where on/off events crossed timing marks and I was chasing Red then green, blue then white.
It works perfecty on all 3 boards ..
I do have some 2811's but I would have to do some digging to get to them...
as for your rate question .. I set LOR up with .02 timing on the ENC board .. sporadic at best .. hit and miss.. mostly miss (of course this was with the GECE's)
Bob
I went back to all of the test sequences I have ran since the project started including the original LOR Chase sequence where it chased from Pixel 1 to Pixel 36 every second.
This was my original metronome test... and ALL three boards are in perfect sync.
Remember this is the one that where on/off events crossed timing marks and I was chasing Red then green, blue then white.
It works perfecty on all 3 boards ..
I do have some 2811's but I would have to do some digging to get to them...
as for your rate question .. I set LOR up with .02 timing on the ENC board .. sporadic at best .. hit and miss.. mostly miss (of course this was with the GECE's)
Bob
Bob Moody wrote:I went back to all of the test sequences I have ran since the project started including the original LOR Chase sequence where it chased from Pixel 1 to Pixel 36 every second.
This was my original metronome test... and ALL three boards are in perfect sync.Remember this is the one that where on/off events crossed timing marks and I was chasing Red then green, blue then white.
It works perfecty on all 3 boards ..I do have some 2811's but I would have to do some digging to get to them...
as for your rate question .. I set LOR up with .02 timing on the ENC board .. sporadic at best .. hit and miss.. mostly miss (of course this was with the GECE's)Bob
With a string length of 36 geces, the pixels won't refresh fast enough for 40hz operation, regardless of the controller. That's why I was asking about other pixel types.
I do recall that conversation ...
All in all... it looks like it is testing very well ...
Oddly(?) enough when all three boards are run in xLights or Nutcracker .. they are perfect. Not seeing dropped pixels or color issues...
In LOR .. there are the occasional dropped pixel and the occasional color issue ..
If xLights was a 100, then LOR is a 97..
More than likely the drop or color issue would never be seen in a full display ..
Discounting hardware at this point.. I think all 3 are working fine with the GECE's (would be interesting to see how they function with the 2811's ) ..
I'm inclined to think that the remaining issues (although minor) are embedded in LOR and the way it sends packets vs xLights ...
I have Vixen here someplace just not installed, setup and configured .. so I really cant see if the final resolution lies in the sequencer itself.
Bob
Bob Moody wrote:I went back to all of the test sequences I have ran since the project started including the original LOR Chase sequence where it chased from Pixel 1 to Pixel 36 every second.
This was my original metronome test... and ALL three boards are in perfect sync.Remember this is the one that where on/off events crossed timing marks and I was chasing Red then green, blue then white.
It works perfecty on all 3 boards ..I do have some 2811's but I would have to do some digging to get to them...
as for your rate question .. I set LOR up with .02 timing on the ENC board .. sporadic at best .. hit and miss.. mostly miss (of course this was with the GECE's)Bob
Bob, would you mind running the same test with the clock at 80 mhz? I'm curious if the enc28j60 can keep up with the slower clock.
I ran it as you asked and while it performed okay .. it didnt seem to be as robust ...
To clarify, it just didnt seem as stable in that it dropped more pixels and the color issues seemed more frequent... I put it back and it ran like a charm ...
On a seperate note ... a couple of things ...
I think I may have a test bench issue in that I am powering the Parallax boards from the laptops USB ports and the the 6804 and the power for the strings are coming from a bench top power supply (Actually a converted ATX supply)
I'm wondering if some of my enumeration issues arent coming from basically a lack of common ground for the data output pins on the Parallax boards. I'm assuming that the 6804 is taking care of that internally since there isnt another connection option.
I ran in to this before with 2811 pixels when I first started. I had to bond all the grounds between the controller and PS so that they were all at the same reference point ..
I had somewhat of an accident today in that with 3 controllers on my desk and two of them held togther by thier jumpers (parallax to ethernet boards) that it was bound to happen that I bumped one in to the other .. I managed to blow a fuse on the 6804. Once I figured out what happened (add a few choice words for effect) and repaired it, It took me an hour to get everything to sync back up again... I had a string on one board chase down 10 or so pixels then jump to the pixels on another board ... This confused the heck out of me .. how was this happening? Eventually I managed to get them all to enumerate properly and everything went back to good (good being everyone in sync.. on the same bulb at the same time)...
Any thoughts on the grnd theory?
I cant imagine the sync of these 3 strings being that delicate.. if they are, then there is no way I would use them in a display .. but .. I know there are a ton of folks out there using them.. occasional issues but for the most part .. if they were that flakey ... then I think the community would have abandoned them already .. Its not like we are getting them cheap compared to some of the other options that are available ..
Thoughts... comments?
Bob
Jim discovered that some of the gece's are sensitive to noise while they're in addressing mode before the controller starts sending data. His solution was a pull down resistor on the data pins to hold them low until the prop started sending data. I've never seen this problem on my GECEs when I hd them. And it seems like the problem didn't start happening until this year and only on 682 boards. So I'm not sure if it's a manufacturing variation in the newer GECE pixels or something on the 682 or its newer firmware. Bottom line is that GECEs need to be power cycled with the controller, and the data needs to be silent until the data is valid. Also the off rest time required to reset them can be very significant at times, I've seen up to 30 seconds. Grounding the data pin during power cycle seemed to shorten that reset time.
--Jon Chuchla--
Sent from my iPhone using Tapatalk
