 lutful... of ideasPremium join:2005-06-16 Ottawa, ON Reviews:
·TekSavvy DSL
| reply to gunther_01
Re: ARGH! Ubnt radios freezing! said by gunther_01:I have heard in the past that as ambient temps drop, devices require more power (current) to operate. I can't recall why, or if it's even true. It is true. Almost all ICs draw more current (and hence power) as the temperature drops. I wrote about this when people were having issues with some Realtek-based radios during 2006/7 winter.
Anyway IC engineers do what is called "timing analysis" at the required extreme temperatures and make sure that all internal circuits operate correctly. Those extremes are usually -20C/70C for extended ; -40C/85C for industrial and finally -55C/125C for a few miltary and satellite ICs.
The temperature is for the "die" inside the IC which depends on both ambient temperature and thermal specs of the IC packaging.
For occasional WISP radio failures in winter, the main culprits are the oscillator and the CPU. Most of them are tested/rated at -20C and we are actually lucky to have them running below that temperature. |
|
 WHT join:2010-03-26 Rosston, TX kudos:5 | reply to Rhaas said by Rhaas:How about climb up and whisper sweet nothings in its ethernet port?? Been reading geek p0rn, I see. 
/spelling |
|
|
|
 RhaasPremium join:2005-12-19 Bernie, MO | reply to lutful ^^^^^  Thats a name I havent seen in a loooong time. -- I survived Hale-Bopp! |
|
 lutful... of ideasPremium join:2005-06-16 Ottawa, ON Reviews:
·TekSavvy DSL
| said by Rhaas:^^^^^  Thats a name I havent seen in a loooong time. I actually hang out at the electronics forum these days. »Haven't seen Lutful posting lately
P.S. I really could not make any critical comments about UBNT stuff back in 2010 without getting attacked by fanboys. Hopefully that has changed.  |
|
 WHT join:2010-03-26 Rosston, TX kudos:5 | Respectively replying, anyone that hasn't extensively deployed UBNT gear shouldn't make critical comments. |
|
 lutful... of ideasPremium join:2005-06-16 Ottawa, ON Reviews:
·TekSavvy DSL
| said by WHT:Respectively replying, anyone that hasn't extensively deployed UBNT gear shouldn't make critical comments.
Obviously there are exceptions like ( »Noisy UBNT gear appearing in our area ) by a veteran WISP. My only comment in that 5 page thread was about adding UNIX "nice" feature in a TDMA radio to co-exist with other WISP gear. And my response to a specific question in this UBNT thread relates to IC temperature ratings. Both are stuff I know a lot about because I personally designed ICs and TDMA radio PCBs.
Hopefully you UBNT guys will allow that tiny bit of freedom of relevant expression by non-UBNT users. After all there are very few of us left. Nothing to fear - you guys already won. |
|
 RhaasPremium join:2005-12-19 Bernie, MO | reply to lutful said by lutful:said by Rhaas:^^^^^  Thats a name I havent seen in a loooong time. P.S. I really could not make any critical comments about UBNT stuff back in 2010 without getting attacked by fanboys. Hopefully that has changed. Even the fan boys got burned  I am using their 2.4G equipment with some success - as long as it is fairly quiet.
I am not happy about the 7kft of tough cable I have deployed that has or is failing.. -- I survived Hale-Bopp! |
|
 jcremin join:2009-12-22 Siren, WI kudos:2 | said by Rhaas:I am not happy about the 7kft of tough cable I have deployed that has or is failing.. Same here... The UV protected stuff that wasn't... I was just at a customer's house yesterday for something totally unrelated and happened to look at the cable and saw a bunch of cracks all over the place. Hadn't totally failed yet, but I'm sure it's just a matter of time.
I haven't participated on the UBNT forums for quite awhile, but does anyone know if they are offering any sort of compensation for the crappy cable that is only lasting a year or two? I have other cable that's been installed for 6+ years and is in almost as good of condition as the day we installed it. Even some cheap non-UV unshielded indoor cable that was intended to be temporary is still in great condition compared to the UBNt stuff. |
|
 | No Compensation yet, they will send you new boxes for the stuff you used... But that doesnt quite cover man hours/ lost revenue.
We had a bit of a rush when a competitor went out of business last year.... Close to 300 installs in 2 months (thats a lot for us).... We just got killed this summer with service calls, left a bit of a bad taste in my mouth..... |
|
 bburley join:2010-04-30 Cold Lake, AB | reply to voxframe I have a nanobridge M5 failing this morning at -29C.
The radio link is still good, but ping times are all over the place if they get through at all. I think the highest response was just over 4000 ms.
Rebooting fixed the problem but only for a few minutes. It has now warmed up a couple of degrees and it is returning to normal.
Now I just have to figure out what to replace the damn things with. |
|
 | Just out of curiosity.... Do you have it set to 23db output power? |
|
 bburley join:2010-04-30 Cold Lake, AB | That's funny, I remember reading about lowering the power level for cold weather, but I couldn't remember to do it at the time.
I will try that and wait for the next cold day to test. |
|
 | I set mine all to 20db and have never had the freezing issue.... Perhaps a more knowledgeable member can enlighten me on why the hell that would make a difference :P |
|
 InssomniakThe GlitchPremium join:2005-04-06 Cayuga, ON kudos:1 | Lutful has made a bit of a comeback and probably has the answer. ubnt told me to do that some time ago and it also worked for me. It was only radios exposed to the wind really for me. Although I'm far from being in the coldest Canadian region. -- OptionsDSL Wireless Internet »www.optionsdsl.ca |
|
 lutful... of ideasPremium join:2005-06-16 Ottawa, ON Reviews:
·TekSavvy DSL
| This is general info for all radios: higher amplifier power levels always have higher possibility of communication errors. Similarly oscillators running outside temperature specs may introduce errors in CPU/RAM and Ethernet operation.
These failure mechanisms start appearing (randomly) when IC internal temperature falls outside spec. Windchill just brings the internal temperature down more quickly. |
|
 | reply to voxframe Strange, I am in Ottawa and have deployed in St Donat Qc.
Look that up on a map: and radios where on top of a mountain. »www.skigarceau.com/
Never had a radio freeze up, of course that's only the 100 I have out there. |
|
 bburley join:2010-04-30 Cold Lake, AB | reply to lutful In my case the radio link was still running. It was unable to return pings properly or pass data. It could have been the oscillator/clock, but then why should I expect that lowering the power level would help in that situation? |
|
 lutful... of ideasPremium join:2005-06-16 Ottawa, ON Reviews:
·TekSavvy DSL
| said by bburley:link was still running ... unable to return pings properly or pass data. It could have been the oscillator/clock If you are sure that your IP network was fine, then it could have been the main oscillator (randomly) going out of spec and affecting multiple clocks on the PCB.
said by bburley: but then why should I expect that lowering the power level would help in that situation? As temperature goes down, power amplifier draws more current and becomes a bit more non-linear ... which causes a bit more bit errors. If you lower the PA dBm level, the situation always improves as long as the link margin is good.  |
|
 WHT join:2010-03-26 Rosston, TX kudos:5 | reply to voxframe UBNT has been recommending reducing power when the radios lock up, well...after they warm up first. |
|
 bburley join:2010-04-30 Cold Lake, AB | I didn't have to wait long for the next cold day. It is -28C and the Nanobridge M5's crapped out again.
It turns out it was the RF link. The GUI shows normal signals/indications that all is well, but obviously not!
I was able to restore the link by setting both to 21 dBm. 21 works, 22 doesn't. The shot is not perfect so my signal strength/quality is getting poor and I am reluctant to reduce the power further.
If it gets even colder, I suspect the problem will return. The long term solution is probably to get rid of them. |
|