Weird AlphaPix16 pr...
 
Notifications
Clear all

Weird AlphaPix16 problem - need HELP!

38 Posts
5 Users
0 Reactions
6,451 Views
OptykMuse
Posts: 181
Topic starter
(@optykmuse)
Estimable Member
Joined: 15 years ago

The universe 12 channels prior to 487 are mapped to another prop as are universes 1 thru 11. I have 8 smaller different style trees with 254 pixels on each one, none are currently connected though (they are all wifi ESP8266 D1 mini controllers).
So string 1 pixel 1 of the megatree does start at 12-487. This matches what I have programmed in the alphapix (and the sACNview transmit test). I just set the sACNview to transmit to only the last few pixels of universe 12 since that is were the megatree alphapix controller starts, and those pixels do whatever effect I select in the Fade Range tab.


Reply
MartinMueller2003
Posts: 7614
(@martinmueller2003)
Illustrious Member
Joined: 14 years ago

I would have to look to see if Vixen would send the same universe to two different addresses. That would be the next place to look. I personally would align the port 1 chan1 to universe 13 chan 1


Reply
jeffu231
Posts: 2457
(@jeffu231)
Famed Member
Joined: 15 years ago
MartinMueller2003 wrote:I would have to look to see if Vixen would send the same universe to two different addresses. That would be the next place to look. I personally would align the port 1 chan1 to universe 13 chan 1

Vixen can send the same universe to multiple controllers as long as they have different unicast addresses. For multicast, the same universe should be setup on one controller in Vixen.


Reply
OptykMuse
Posts: 181
Topic starter
(@optykmuse)
Estimable Member
Joined: 15 years ago

I've had to change priorities since it's now November and I still have not gotten this megatree working, but still thinking about it (during these late night hours) as I work on the rest of the Christmas show. I've been doing some re-reading of a few other threads and came across one I read not long ago about universe channel packing:
https://doityourselfchristmas.com/forums/showthread.php?55719-Packing-universes-reducing-channels-per-universe-to-minimize-network-traffic&highlight=packing+universes
This thread was the reason I decided to pack the channels the way I did, although I'm not sure if this is causing me problems or not. Even with the thread reading I don't think I have a good enough understanding of unicast vs. multicast yet, and it looks like multicast requires extra hoops to jump through and possibly only certain AP's work with it.

Jeff, I'm not sure if I understand your last post regarding multicast... are you saying for a given prop that the universes that support it should not be used in other props? For example, my new small trees I have been working on are 762 channels (254 pixels) each so require about 1.5 universes. Are you saying if I use multicast I need to start the second tree (each tree has its own controller) on a new universe instead of using the last half of the universe from the previous tree, so each tree would use universes not used in any other prop (hopefully I'm making sense with my questions)? I need to know if I'm making mistakes trying to implement multicast. These trees (8 total) are not the megatree I've been trying to get working in this thread but will be used in the show also (only these will be wifi as I've mentioned before).

I've read a number of posts in the forum regarding lag - since I only have 8 wifi controllers do I need to worry about lag using unicast? I have a (so far) unused channel set for use in my AP so shouldn't have any interference problems on the wifi end of things. I'm also positioning the AP very close to the 8 wifi controllers for best signal reception (These boards only have the PCB trace wifi antennas).

Another question... if I do go the unicast route with the wifi controllers do I assume correctly that I need to set up a separate 'destination' for each of the controller's IP addresses in the E1.31 controller settings?

This stuff is REALLY confusing but I think I'm slowly beginning to get it...


Reply
MartinMueller2003
Posts: 7614
(@martinmueller2003)
Illustrious Member
Joined: 14 years ago

I see there are a few concepts missing that are causing you confusion. Lets start at a higher level and work our way through the issues. The below is a very lounge-in-cheek explanation meant to be slightly amusing while presenting information.

In the beginning, there was a software program (sequencer) that needed to deliver its data to a set of semi intelligent lights called pixels. To do this the program took all of the channels of data and placed them in a big block of memory. Then the program had a problem: How to deliver all of that data to all of those lights. That problem is made more complicated because the lights have all sorts of their own rule and requirements for how they want the data presented to them.

Rather than pollute the program with all of the vagaries of the lights, there was a great idea: Lets put a controller between the program and the lights that can convert the block of data from the format the program liked into a format the lights used. Alas, these controllers made the mistake of imposing some of the requirements from the lights onto the data they received from the program (ie RGB, RGBW, RBG etc.... However, that was kept to a minimum and soon programs like xLights and Vixen hid most of the color, port and string nonsense imposed by the lights and elevated those concepts to props and locations.

So were does this leave us? We still somehow have to get the block of channel data to the controllers. Oops, it turns out that sending the entire block of data to each controller is wasteful to the media used to send the data and the controllers would not be able to hold that much data anyway. Controller's only want the block of data that is relevant to the lights they have control over. Back to the program and tell it: You need to organize that block of data in chunks that are of interest to a controller and then send that data to the individual controllers.

Suddenly the media that needs to transport this data decided it wanted to have some say in how the data was moved and imposed its own rules. This started with serial interfaces (DMX) and progressed through Ethernet (E1.31, DDP) to WiFi (less capable Ethernet). While Ethernet allowed bigger sub chunks of data to be sent, the size was not infinite. In reality, you could send about a 1200 byte chunk.

So how much nonsense can we come up with? DMX worked great but there were still a lot of cables around. Hey lets put lots of those cables into one Ethernet connection (E1.31) at the lighting desk and break them out to chains of controllers at the stage. We get to use the same DMX equipment and get smaller wire bundles to the stage.

Along come these holiday lighting guys and we want to use something well established so we jump in and adapt the e1.31 to our use. But it is an alien concept. You need to break that pretty block of data in smaller chunks to fit into the artificial constraint called a universe that holds only a sub block of 512 channels of data. So controllers learned to handle multiple universes so that they can put the smaller blocks of data back into a bigger block (internally) and then transform the data into something the lights can use. OK This works. Ethernet has a lot of bandwidth to waste so the fact that most packets are less than half full is no big deal.

At this point, WiFi raises its voice and says: I want some attention. I have a few limitations I want to impose: First, My design did not care much about real time needs so if you try to send something directly to s station connected to the AP, there are going to be additional delays as I wake up the connection and get ready to send. Well all of those delays began to add up and the lighting guys were not happy about the lag they were seeing on the WiFi network. So they looked at not sending the data block to a specific controller and instead said lets broadcast (no connection establishment etc...) the data and let the controllers pick some sub block out of the big data stream. This ran into another limitation. WiFi broadcast data goes out after all unicast data is sent and it goes out at a lower data rate than unicast data. Another set of rules to be handled and described in other threads. Sigh.

This is already too long so I am leaving out DDP and how it helps a bit with much of this. Just say DDP is almost twice (debatable but...) as efficient as E1.31.

E1.31 nonsense rules that pop up in all of these environments:

Unicast:
the block of data gets broken up into 512 byte chunks (universes)
Each universe is sent directly to the controller that needs it. The controller reassembles the blocks and sends to the lights.
The universe numbers are unique to each controller and can be reused at each controller. This means every controller can use universe numbers starting at 1.

Broadcast / Multicast
The entire world is treated pretty much as a single controller
Controllers pick their data out of the stream of data being sent at a starting point (Start universe + start channel in the universe) through a number of configured channels.
All controllers receive all of the data and throw most of it away.

I know this does not directly answer your questions, but I hope it helps you understand the problems being solved by those weird terms and settings and why none of them seem straightforward and there is no single answer.


Reply
jeffu231
Posts: 2457
(@jeffu231)
Famed Member
Joined: 15 years ago
OptykMuse wrote:I've had to change priorities since it's now November and I still have not gotten this megatree working, but still thinking about it (during these late night hours) as I work on the rest of the Christmas show. I've been doing some re-reading of a few other threads and came across one I read not long ago about universe channel packing:
https://doityourselfchristmas.com/forums/showthread.php?55719-Packing-universes-reducing-channels-per-universe-to-minimize-network-traffic&highlight=packing+universes
This thread was the reason I decided to pack the channels the way I did, although I'm not sure if this is causing me problems or not. Even with the thread reading I don't think I have a good enough understanding of unicast vs. multicast yet, and it looks like multicast requires extra hoops to jump through and possibly only certain AP's work with it.

Jeff, I'm not sure if I understand your last post regarding multicast... are you saying for a given prop that the universes that support it should not be used in other props? For example, my new small trees I have been working on are 762 channels (254 pixels) each so require about 1.5 universes. Are you saying if I use multicast I need to start the second tree (each tree has its own controller) on a new universe instead of using the last half of the universe from the previous tree, so each tree would use universes not used in any other prop (hopefully I'm making sense with my questions)? I need to know if I'm making mistakes trying to implement multicast. These trees (8 total) are not the megatree I've been trying to get working in this thread but will be used in the show also (only these will be wifi as I've mentioned before).

I've read a number of posts in the forum regarding lag - since I only have 8 wifi controllers do I need to worry about lag using unicast? I have a (so far) unused channel set for use in my AP so shouldn't have any interference problems on the wifi end of things. I'm also positioning the AP very close to the 8 wifi controllers for best signal reception (These boards only have the PCB trace wifi antennas).

Another question... if I do go the unicast route with the wifi controllers do I assume correctly that I need to set up a separate 'destination' for each of the controller's IP addresses in the E1.31 controller settings?

This stuff is REALLY confusing but I think I'm slowly beginning to get it...

For multicast, you do not need to make a controller representation for each hardware prop controller you have, but one that represents all the universes or segments of universes that make organizational sense. Then you patch your props to that controller. You only want to have any given universe represented once in Vixen when multicasting. Otherwise you will end up with duplicate universe packets on the wire. Then in your hardware controllers, you set them up for Multicast and configure them for the universes they care about and map the channel segments within those universes to the outputs. This allows you to do as much universe packing as you want. You can even have multiple props listening to the same universe segment that always do the same thing so you do not need duplicate channel data. Remember Multicast is a broadcast mechanism. It does not go to any specific endpoint. Controllers that are in multicast mode listen for those packets and react to the ones that have the universes they are interested in. If you have a controller that is only interested in 100 channels 50-150 on universe 10, it will grab the universe 10 packet, and offset into the packet 50 channels and use the 100 it needs. It throws the rest away.

For Unicast, you have a specific endpoint that the universes are destined for. Only the controller at that endpoint will get that universe. In this case you create a controller in Vixen for each specific unicast controller with the specified destination ip/host. You can use the same universes on multiple unicast controllers, but there is not really any good reason to. It causes confusion when configuring the controllers and there are plenty of universe numbers that each controller can have its own range.

Hope that helps.


Reply
OptykMuse
Posts: 181
Topic starter
(@optykmuse)
Estimable Member
Joined: 15 years ago

Jeff, your last paragraph clicked, light bulbs flashed above my head, and I finally understand now that I needed to have separate controllers defined for each unicast hardware controller. I was trying my best to understand the multicast approach but was not having any luck getting the alphapix to do its thing, so shifted gears, deleted the entire sACN controller I had been trying to get working, and then built a separate E1.31 controller for each hardware controller. Once I completed all this (using unique universe numbers for each at your suggestion) I had controllers for all 8 of the small pixel tree I've built and another for the megatree. Each had the applicable IP addresses, pixel counts and universes set up at this point. I patched each of the already defined elements to the matching controllers, went into the web UI for the alphapix and re-did the ports to match the new universe & channel assignments for each string, then held my breath as I fired up Vixen to run the test sequence...
I now have a working mega tree hardware controller!! The two strings I have plugged in flashed in time with the sequence preview so I'm confident the rest of the strings will work once I have them connected. I still need to do the web UI config for each of he wifi controllers but at least know what I need to do once I get them connected to power.
I'm really grateful for all the help on this. I know i could not have figured this out on my own, at least not in time for this year's fun.
Martin, great explanation of the progression of how all of this came about, I enjoyed reading that and it helps to explain why it's so complex, and I still have quite a bit to learn.


Reply
rayhjr
Posts: 972
(@rayhjr)
Noble Member
Joined: 12 years ago

Glad you got it working! It's a great feeling to get things going Blinky.

The people on this forum are amazing!


Reply
Page 4 / 4
Share: