• You've been granted Beta access to this site, allowing you to explore some of the new features while they're still under construction. More information can be found in the Beta forum.

Action latency on SA UPB rocker switches?

PeterW

Active Member
I've got some SA 1120W UBP rocker switches to give them a test run. I'm using them to replace the buggy smarthome switchlinc-relay switches (don't get me started about the firmware bugs in those!). I'm using them with with some instant-on electronic ballast fluroescent lights.

With the switchlinc-relay switches, they did indeed turn on instantly. With the 1120W's (in snap-on mode!), there is nearly 1 second latency between the lights turning on or off in response to the rocker being pressed.

I've triple checked with UPStart that they are programmed into snap-on mode. If they are in fade mode, you can see the difference (the lights "brown out" and flicker etc).

Also, the indicator light changes almost exactly at the same time that the controlled lights turn on or off.

Have I missed something or are they supposed to be this slow to respond to the rocker? I could understand delays in response to a command being issued, but the actual rocker switch?
 

electron

Administrator
Staff member
That definitely doesn't sound right, I hope you have more luck contacting SA than I do, maybe they can help you out with this one. Please let us know if you figured out the issue. Also, how much faster do they turn when using a remote (compared with X10 that is)?
 

PeterW

Active Member
My subjective opinion at this point is that they respond to remote commands at the same speed as they respond to the rocker.

I'm wondering if the switch operates in two halves, with the rocker button half communicating with the dimmer/switch part via UPB, with no fast/direct link.

Even the smarthome switches aren't silly enough to do that..

The other possibility that I'm wondering about is if the firmware is waiting for the double-tap delay to pass in order to distinguish between a single or double tap. That would be silly in this case because single or double taps are both set to go "on" to 100%.

I had other problems.. the switch bodies are physically longer than the smarthome switches and wouldn't fit in the old style bakelite j-boxes in this house. I had to destroy the old j-box and replace it with a plastic rework j-box just to get the switches to physically fit in the opening. The smarthome switches are deeper than the SA upb switches, but the SA upb ones are taller.

My wife is alarmed by the audible noise that the switches make when transmitting upb commands. Both the computer interface and the two switches make the same "BZZzzzt!" noise when sending upb packets.

So far, I'm rather underwhelmed. I had much higher hopes.

On the plus side, getting rid of those switchlinc-relay devices from that high traffic area has finally stopped the X10 crash that they were causing.

Did you know that if you cause two or three switchlinc-relay devices to transmit at identical times, they will scramble your X10 network until you power cycle the circuit?

For example, if you have three devices, at D2, D3 and D4, and issue a grouped command, eg: 'D2 D3 D4 DSTATUS_REQUEST', they will all simultaniously transmit, and will detect the collision, and will all back off in an identical fashion and retry, forever. The cycle will never break. I have to cut the power at the breaker box, or disconnect the power from two of the storming devices.

You can achieve this by pressing the toggles at the same time. This is a real problem if you have two or three next to each other and other family members get into the habit of pressing them at once.

Another problem.. Switchlinc-relay devices transmit a stray ON or OFF event when powered up. If you are unlucky and they power up at exactly the same time, that will cause a storm too. This is harder to reproduce though, but not that difficult.
 

electron

Administrator
Staff member
I have heard about this, that's pretty bad! Once you have UPB up and running without any issues, you'll have to do a review or comparison with X10, many people are waiting for a document like that :)
 

PeterW

Active Member
Hmm. There seems to be a firmware quirk on those UPB switches. I still haven't quite figured out what is going on, but I've managed to change the local latency somehow.

When I originally set them up with UPStart, they were set to a fade rate of 3 seconds. I changed the dimmer to 'disabled'. I had the controller plugged into a laptop, running upstart 4.0 build 40.

I later installed a newer version of upstart (4.1 build 50) on the machine that I run HCA 6.0 on (upstairs) and tweaked the link settings etc as I went.

Today, just to see what happened, I reenabled the dimmer, and changed the fade rate to 0.8 seconds (the lowest value), and programmed the switch. And then I changed it back to disabled and 'snap on!' mode and reprogrammed them again.

And now the latency seems to be about 1/3 to 1/2 of a second. It still isn't instant, but it is a lot faster than before.

I wonder if the switches are somehow still evaluating the fade timer, even though the dimmer is "disabled"? Perhaps when the dim level is < 50%, it clamps it to off, and when it is >50% it clamps it to full on. Maybe it is doing 'fade' over 0.8 seconds, which is causing the turnon to appear at about the 0.4 second mark as it crosses the 50% dim level.

And in that scenario, the 3 second initial fade rate could explain the roughly 1 second latency - it might have been 1.5 seconds as the 3-second ramp-up passed the 50% mark.

As far as UPB signal reliability goes, so far I'm rather impressed. I have the switches at the opposite end of the house to the computer, both some distance from the breaker box. I have 40-something X10 transmitters in the house (and for some reason I have X10 problems, I wonder why??). So far, HCA 6 has logged zero retransmits - every command has been acked on the first try.

One thing I've been trying to do is use HCA to bridge my X10 and UPB networks together, using triggers on both sites to relay state change messages and commands. I've been able to get my smarthome keypadlinc keypads to turn on/off the UPB switches - *with status acks*! The keypadlinc's send an on, then a status request. I proxy the on command, turn the x10 status request into a upb status request, and then have HCA transmit back a status-on or off. I also broadcast status-on/off in response to the upb state change messages so that the indicator lights get updated. Suprisingly, it actually seems to work. In other words, I've given my ubp switches a virtual X10 address that works well enough that the other smarthome devices work with it.

HCA has been frustrating though, but that isn't too suprising given that 6.0 is their first cut at upb support. The program could do a heck of a lot more to help. If its going to integrate UPB and X10 networks, they really need to come up with a configurable built-in UPB<->X10 proxy of some sort. A decent, easy to use proxy would really help with gradual or retrofit upb deployments.

PS: has anybody noticed how easy it is to transpose upb <-> ubp ? :)
 

electron

Administrator
Staff member
Have you taken a look at HAL 2000? They support UPB now, so it might be worth taking a look at the trial. Have you been able to time yet to see if it is faster than X-10 when trying to turn a light on through the computer? Thanks!
 

PeterW

Active Member
Yes, it is DEFINATELY faster. Computer commands to devices are really quick! Not instant, but quick.

I control my aquarium lights by computer. I just replaced a bunch of smarthome appliancelinc's with the SA UPB appliance modules, and the improved responsiveness was wonderful!

Sadly, my main bank of aquairums is on a different phase in the house and I'm getting very poor cross-phase signal transfer, so I'm going to have to invest in a phase coupler. Anyway, for now, the lights are on a small extension cord.. ouch, gotta watch out for having two phases in the same rats-nest of cords!, especially near water!

Anyway, so far, the signal reliability has been excellent. It is noticably faster than X10, but not *that* much faster. I just did a quick test and I can send 4 seperate ACKNOWLEDGED commands to 4 devices within about 1 second, and that is including any delays added by HCA6. I'm sure I could use link commands to make it all simultanious, but it is more than good enough for me.

I use appliance modules for controlling water changes in the tanks.. But the fill rate is faster than the drain rate in a few tanks, so I was terrified of getting an X10 system lockup while the water was turned on (which would cause an overflow) I'd been meaning to use my rain8 (with its max run time limit settings) to control it, but decided to give a upb module a try instead. So far, it has been nice to be only needing to send a command to it once instead of 5 times like I was doing with X10 (to be sure it was received).

Anyway, my misgivings about the UPB switches don't apply to the modules and the computer interfaces. Those are really quick. I think the problem with the switches is that they're sitting there and waiting to see if we were going to tap it again to make it a double tap command.

I'm going away next week so I needed to get this stuff sorted out before I left.. floods or lighting problems are not something that I needed. And the damn smarthome switchlinc-relay devices kept on destroying my X10 system, so I had to do something.
 

electron

Administrator
Staff member
You need a phase coupler already? Supposedly most houses wouldn't need one, but it's understandable I guess. Did you ever get in touch with someone at SA? I definitely would like to know why there are delays as well, since Lightolier switches (and others as well I assume) don't have this delay, and support double tapping as well.
 
Top