Combating WiFi Lag - ESPixelstick and Nanostation settings?

[paddler]

Looking for ideas. For my daughters show choir, I'm running 37 Espixelsticks - each one is on a costume and controlling just over 200 seed pixels sewn on to each costume. The lights are powered by a 5v power bank. The problem is that I've been battling lag on a few of the costumes - I have done some improvements over the last month but nothing seems to help. I'm looking for suggestions.

Here's what I have: I'm running xlights in jukebox mode for this live show. I've hard connected my laptop to the network switch...which is connected to a Nanostation M2 access point. Each Espixelstick is set to run in multicast and has a static IP. This is a stand alone show network, not connected to the internet. I'm also connected to a Kulp K16 that controls lights on the set. This is hardwired and I'm not experiencing any problems with it.

Attached are screenshots of the setup.

The Espixelsicks are running 3.2 firmware. I would upgrade to 4.0 beta 5 if someone thought it would improve/fix the problem.

As these are costumes, hardwiring them is not an option.

Thanks for any suggestions. If you want to see a video of the show - here's a link. The costume lights kick in at 7:40.
Link

 

[MartinMueller2003]

The problem is that you are sending DATA over the WiFi. Stop using E1.31 or DDP or SACN and put fseq files on the sticks. Then use xLights to send FPP sync packets. One packet arrives at all sticks at the same time and they stay in sync. You will need lates V4 dated January 2024 on the sticks.

 

[critic2]

Like Martin said, you will want to send just sync packets to cut down on the WiFi traffic. However, it will require you to use a PixelStick V3 hardware that has the SD card option (or equivalent) and use the V4 firmware as referenced above.

 

[paddler]

The problem is that you are sending DATA over the WiFi. Stop using E1.31 or DDP or SACN and put fseq files on the sticks. Then use xLights to send FPP sync packets. One packet arrives at all sticks at the same time and they stay in sync. You will need lates V4 dated January 2024 on the sticks.

This is a live show. Will this work with Jukebox?

 

[MikeKrebs]

At 210 pixels, the vests will require two packets for each one. Is there a config where you could send only 170 pixels so it could be half as many?

Are you packing the channels into the universes or using 2 universes for each?

Does 3.2 have DDP support?

Are any of the performers doing the same thing so you could just make them "set A" (all this group are listening for the same channnels)?

And unlikely that someone would hack your network but you did expose the current SSID and passphrase. Probably want to change that up.

 

[kockie]

as you are only using one color, you could also make a group of 210, then every vest will only use 3 channels, so you could put 170 vests in one universe

 

[paddler]

At 210 pixels, the vests will require two packets for each one. Is there a config where you could send only 170 pixels so it could be half as many?

Are you packing the channels into the universes or using 2 universes for each?

Does 3.2 have DDP support?

Are any of the performers doing the same thing so you could just make them "set A" (all this group are listening for the same channnels)?

And unlikely that someone would hack your network but you did expose the current SSID and passphrase. Probably want to change that up.

I am running it in DDP. No, I'll need to use all 210 lights.

I do have the performers in 2 or 3 groups in xlights (depending on the song). Does this put them on the same channel? I don't see any improvements in this part of the sequence.

(I'll change the SSID and passphrase. Didn't intend to put that out there.)

 

[paddler]

Would adding a second AP help?

 

[dirknerkle]

Do a site survey and see if there's a channel that's not as busy and conflicting with other wireless devices. You're using channel 11, possibly the most common for wireless.

A second AP may help. I had 86 ESP's in my show and it ran noticeably better after splitting it over two AP's and two channels. You're pumping a lot of data out there....

 

[paddler]

Do a site survey and see if there's a channel that's not as busy and conflicting with other wireless devices. You're using channel 11, possibly the most common for wireless.

A second AP may help. I had 86 ESP's in my show and it ran noticeably better after splitting it over two AP's and two channels. You're pumping a lot of data out there....

Thanks. (I followed your other posts when setting up the Nanostation. - thanks again)
Ordered a second AP. I'm assuming that I set these up with a separate SSID and passphrase and move half of the ESP's to the new network?

 

[MartinMueller2003]

You will continue to have Bandwidth issues as long as you are sending data over the WiFi.

 

[dirknerkle]

Thanks. (I followed your other posts when setting up the Nanostation. - thanks again)
Ordered a second AP. I'm assuming that I set these up with a separate SSID and passphrase and move half of the ESP's to the new network?

Yes. The second should help, but it's unrealistic to expect no lag ever. One of the problems with WiFi will be all the phones that people will bring in their pockets to the performance. It used to be that I'd have 35-50 wifi networks when I did a site survey from my driveway (in the midst of all my props) but last time I ran a survey, I had more than double that. You could conceivably run a test an hour before the show and really be happy with the network's performance. Then suddenly, after all the patrons are in their seats and there are another 200 wifi networks in the house, right in the middle of your network, things may bog down again. Ugh. Don't know how to fix that other than changing to a wireless system that's outside of any other interference. And that'll be really tough to find.

 

[P. Short]

Also, is there free wifi provided by the venue that becomes heavily used during the show?

 

[MartinMueller2003]

Or limit the packets sent by using fpp sync.

 

[P. Short]

Do you think that anybody besides yourself noticed any lag? The color was a bit off for the costumes of a few of the singers, but that was all that I noticed.

 

[dirknerkle]

I just watched your original video, and thought it was just delightful. You should be very proud of the girls and how you've helped them do something quite special. Don't despair!!!

A thought I had originally was if the costumes used IR to activate the vests. The down side is that IR is hard to sequence individual props, as clearly that is what you set out to do. But for turning everything on/off and likely without delays, it could work very well. You might send a PM to Lightman (Brian Lincoln) as he has considerable experience using IR for not only his Christmas show ("Glow with the Show") but in theatrical prop management. He's a wealth of information!

 

[paddler]

You will continue to have Bandwidth issues as long as you are sending data over the WiFi.

Understood. This being a live show with a live band, I have very little options. The sequence does not follow an audio track. I manually kick off sequences using the jukebox function. I can put the fseq file on the controller but how does the xlights jukebox function work with this?

 

[paddler]

Yes. The second should help, but it's unrealistic to expect no lag ever. One of the problems with WiFi will be all the phones that people will bring in their pockets to the performance. It used to be that I'd have 35-50 wifi networks when I did a site survey from my driveway (in the midst of all my props) but last time I ran a survey, I had more than double that. You could conceivably run a test an hour before the show and really be happy with the network's performance. Then suddenly, after all the patrons are in their seats and there are another 200 wifi networks in the house, right in the middle of your network, things may bog down again. Ugh. Don't know how to fix that other than changing to a wireless system that's outside of any other interference. And that'll be really tough to find.

I'm ok with a little lag. Right now I have some costumes that are sometimes 4-5 seconds behind. Then they catch up. This is what I'm trying to eliminate.

 

[MartinMueller2003]

Understood. This being a live show with a live band, I have very little options. The sequence does not follow an audio track. I manually kick off sequences using the jukebox function. I can put the fseq file on the controller but how does the xlights jukebox function work with this?

There is no difference. Instead of sending ddp packets the show player sends sync packets. The sticks do what the sync packet tells them to do. xLights knows how to send the sync packets as well as an FPP player. If your board can trigger a show player to output sync pkts then you are good to go. FYI: FPP has this ability.

 

[paddler]

Also, is there free wifi provided by the venue that becomes heavily used during the show?

Maybe. Sometimes. These competitions are at a different high school each weekend. No two are alike. Some are in Gyms and some on stage at a theater. Lots of variables.