You bring up a few fair, and known, points. I'll respond to at least a few of them:
In relation to the water ingress, we've seen enough of these to know it's at least an occasional issue. Especially in hotter climates, one thing we've seen is the gasket cracking in the enclosure. I'm also not completely convinced this isn't a water condensation/surface moisture issue in damper climates. Our enclosure manufacture just switched the gasketing material, perhaps as a result of our whining, to something different. I've also looked at various alternatives for enclosures but haven't found the right one yet. I thought I had one which would have been perfect, until I got the $50 per unit quote in quantity. Some days I miss the pipes, but it was time for them to go from a mostly marketing perspective. I've also looked at the conformal coating in the past, and the challenge has been that the GPS module manufacture has told us this is a no-no since it apparently changes the tuning of the GPS patch antenna. I probably need to re-evaluate this now that there's a different patch antenna on the new modules - but I expect a similar answer. For the past couple of years, I've been trying to move to a module flat on the board and then either using a PCB antenna like is used in devices like cell phones, where the antenna's pattern is such that mounting it flat on vertically oriented PCB results in a vertical pattern like the patch has. This should solve the water getting on the antenna issue, as there won't be anything directly below the seam to leak on. The holdup has been me trying to possibly move to a module with a different GPS chipset at the same time, hoping that this would be better than our existing module. With all the warts (some of which will be described below), I'm finding that the modules/chipset that we're using is actually not that bad in comparison to some others. I even have had an eval of a 'timing grade' gps receiver here, which had more failures than the non-timing-grade ones we're using. I have a few more to qualify, but look for this change in coming months assuming I can find a suitable module and antenna which works. In relation to the random timing loss of lock, I am not going to disagree with you at all. I suspect some of this might be leftovers from the GPS+GLONASS issue from the end of the year which seems to have resolved itself for the most part, but I know enough about that bug that it wouldn't shock me to find that there are still lingering issues. The problem here is that although it 'feels' like there might be more issues, I don't have quantifiable numbers. With all of that in mind, I'm currently working on in-field upgrade procedures for upgrading the firmware for these modules to get them the GLONASS fix. This seems to be a more troublesome problem than it should be, since it's just an issue of getting the right firmware - the problem being that the company who built the firmware for these got gobbled and the successor company tends to be more difficult to work with. I have firmware which does work on the modules, it just isn't an official build by the manufacturer. Around the end of the year, we did switch our basics to a newer module, based on the same chipset. The Basics shipped since then default to GPS+GALILEO. Recently we've been using this module in Aux Port and Deluxes, but with GPS+GLONASS+Fixed-Glonass firmware since the Cambium firmware doesn't understand (yet) the Galileo sentences. So anything you get from us today has this latest chipset in it, and has the GLONASS fix even if it isn't enabled. I will say that testing and in-field reports indicates this is even more stable, not sure how much of it is because of the increased antenna gain, and how much of it is due to the updated firmware, or how much is just a side effect of having a lot more of the older units in the field having customers report problems on. I know at least some of it is measurable on the bench here, so it isn't all based on field reports. The only downside so far is that we do see an uptick in DoA's (but still well under 1%). Frustratingly, the DoA's we've gotten back have somehow resurrected themselves between the field and here, but thankfully there doesn't seem to be any increased in-field failures that we can see other than the DoA's. To Eric's point about the holdover timer.. I understand the CTM2's had this functionality built in. Nothing else I'm aware of has had that except for some our early GPS modules which just produced a pulse no matter what, and when it could align it it would. This was good in the early days, not so much nowadays. I'm looking at a couple options to implement a holdover. First of all, assuming the docs are correct, I can tell the GPS modules to produce sync all the time, or only when they have a 2D lock, or when they have a 3d lock. Sync all the time can be bad. Especially if you're not monitoring lock status, since it means that a GPS can be out of lock but producing sync pulses which are wildly wrong, causing all sorts of issues. 2d lock can be bad too since it shares similar issues. As a result, the modules default to '3d lock'. I've been toying with doing something dynamic in the rackinjector where it is able to dynamically changes the mode, so it waits until you get 3d lock, and then since it should have a good position, switches to '2d lock' mode, or maybe even freerun. At the bare minimum, I probably will allow customers to change the setting statically if they're willing to deal with the ramifications. I have some other ideas I'm cooking up as well, just don't want to say too much until I'm actually a bit farther down the path. On Mon, Jul 13, 2020 at 5:57 AM Mark Radabaugh <m...@amplex.net> wrote: > Forrest, > > As usual there are probably multiple issues going on. We have seen an > increase in the number of DFS hits recently, and we also have a lot of > Packetflux timing equipment on the network. I have noticed that DFS hits > tend to be worse in ‘unusual’ hot weather conditions. And we have > certainly seen a pretty unusual heat wave over the last few weeks. I’m not > sure if this is just heat changing sensitivity of the DFS detections > (temperature is involved in the RF calibration), if there is more > reflection off the ionosphere in these weather conditions, or if the > weather radar systems are just really jacking up power looking for storms. > > As far as timing - I do think there is some timing instability going on > but I can’t pin it down to anything specific. We continue to struggle > with RackInjectors losing the GPS timing signal from the Syncbox Basic > during or after storm events. Typical symptom in the RackInjector fails > to see sync from the syncbox and the AP’s go into freerun. Sat’s in view, > etc. all look normal, just no pulses. Sometimes a power cycle from the > RackInjector will fix it, sometimes physically unplugging it will fix it, > and sometimes you just have to wait. I have instructed the field crews > multiple times to make absolutely sure they screw in every screw tight on > the syncbox but I’m not 100% sure they are doing that. I have seen at > least one come back to the shop with evidence of water damage to the GPS > board at the top. I would really like to see the extra step of conformal > coating on the boards if there isn’t a reliable way of keeping water off of > them. > > We have also been seeing an unusual number of LBT issues with the 3.65 > gear which I believe are related to other AP’s drifting or briefly going > off timing. > > Due to the number of times we were seeing loss of sync we had to enable > sync + freerun in order to avoid session resets. I’m not convinced that > we are not still seeing timing jumps due to the sequence of loss of sync, > into freerun, then an abrupt change in framing when sync comes back. Any > time something like that happens it tends to cause a wave of DFS and LBT > events across the network. I can’t necessarily show anything specific at > this point though. We do get a lot of archived and searchable logging > from our Sumologic syslog server. I’m going to ask the NOC to put > together a report of AP’s reporting timing recovery and any correlation > with DFS or LBT events within a 60 second window and see what we get. > > Mark > > On Jul 13, 2020, at 3:19 AM, Forrest Christian (List Account) < > li...@packetflux.com> wrote: > > I need to be a bit clearer in that I'm not really sure what version this > customer is running. The question about 15.x /16.x came from a couple of > oldish threads which indicated that something broke early in 15, and that > it still wasn't fixed in 16. But I found that unlikely to still be the > case another year or two on. In these year-and-a-bit old threads, the > report was that one had to go back to very early in 15.x to "fix" this > issue. But like I've said before in this paragraph - I find this unlikely > to still be the case - I just was hoping to verify that this wasn't a > common knowledge issue that DFS was broken on 16.x. > > I know this customer has been in contact with Cambium. Based on our > conversations with the customer so far, I get the impression that for some > reason they've decided this is a sync issue. I don't know if this is a > customer determination or if Cambium has told them this. I like your word > dubious as I'm skeptical as well, but I'm also not one to dismiss a > possible cause until I fully rule it out, as they could be 100% correct. > > I could see where if you have an AP with sync broken intermittently > (especially if you have freerun on), you might end up with a DFS event as a > result of things just not being in sync. But I have reason to believe > this isn't the case with any of their AP's - at least not the ones I have > seen the GPS status screen on the RackInjector for. > > I could also see where a stray pulse or two may be misinterpreted by the > AP to be the correct alignment and have the same effect with causing AP to > transmit out of sync as well. But generally, the radios should ignore > these as they're very rare (and exist in both the PacketFlux and official > Cambium gear, so if it's a problem with mine, it should be a problem with > the official gear as well). > > And I agree with you 100% about a dislike for DFS. I have a feeling that > this customer isn't going to help me change my opinion. > > > On Sun, Jul 12, 2020 at 8:23 PM Ken Hohhof <af...@kwisp.com> wrote: > >> I am unaware of any correlation between DFS events and either Packetflux >> or 15.x FW. >> >> >> >> I don’t use a lot of DFS because honestly it seems fussy no matter what. >> But I have a tower with 10 sectors in 5 GHz (8 x 450i and 2 x 450m). They >> are all synced from a Packetflux Rackinjector using Cambium Sync. 4 of the >> 450i sectors are in 5.4 DFS, and I’m embarrassed to find they are still on >> 15.2 FW. Uptime of about 6 months and no DFS events. So I’m dubious about >> all of this. >> >> >> >> The latest production FW is 16.2.1 and it also has a lot of fixes so I’m >> not sure why you would be running something so far behind. As I said, I’m >> embarrassed to find I still have radios on 15.2. >> >> >> >> Has he opened a case with Cambium support? There are some best practices >> with DFS. For sure you don’t want to configure the AP to think the antenna >> gain is lower than it is (not possible with 450m or integrated 450i). You >> don’t want to set the SM Receive Target Level higher than necessary on >> other sectors. Then there’s choosing the alternate frequencies. And I >> suppose a poor sync configuration could cause false DFS detections, where >> an AP sees the signal from an adjacent AP. >> >> >> >> But who knows what causes these events? Somebody’s Linksys reflected off >> a bird? A competitor aiming a new radio? I used to have a 5.4 GHz PTP500 >> backhaul and the end pointed in the general direction of Chicago would have >> DFS events when there were storms. I thought ducting was causing it to see >> distant signals, but it could also have been tripped by lightning. DFS is >> fussy. I don’t like it. If I could swap out all the SMs on those DFS >> sectors for 450b, I would probably move them to U-NII-1. >> >> >> >> >> >> *From:* AF <af-boun...@af.afmug.com> *On Behalf Of *Forrest Christian >> (List Account) >> *Sent:* Sunday, July 12, 2020 7:56 PM >> *To:* AnimalFarm Microwave Users Group <af@af.afmug.com> >> *Subject:* Re: [AFMUG] 450i/450m DFS false detect problem solved in >> later firmware? >> >> >> >> I read the 16.0.1 release notes, nothing really specific about DFS other >> than it being on when it shouldn't be. However, I agree there is lots of >> stuff fixed in there, some of which could have repercussions for DFS. >> >> >> >> Are you saying that mid to late 15.x was generally broken for DFS and >> this is largely fixed in 16.x? I guess my real question should have been >> 'What is the state of DFS in the 450 platform and how fussy is it'? >> >> >> >> I'm still gathering information from this customer but it sounds like >> they're still trying to track down the root cause. Sometime in the past >> week or so they figured out that there was some correlation between the DFS >> events adding a fair bit of PacketFlux gear, so this correlation is now the >> leading root cause in their minds. So now I get to try to resolve their >> problem for them. >> >> >> >> >> >> >> >> On Sun, Jul 12, 2020 at 3:00 PM Dave <dmilho...@wletc.com> wrote: >> >> If they are not running 16.0.1 nuthing can help them from some weird >> issues with the DFS bands. >> >> Lots of things corrected in 15.2 and later for EIRP and SNR related >> calculations the help with H/V misreads and A/B channel alignments. >> >> Read the release notes in 16.0.1 for further info. >> >> >> >> On 7/11/2020 3:12 AM, Forrest Christian (List Account) wrote: >> >> I'm working with a customer that is having problems with DFS false hits >> who is convinced this is a PacketFlux sync issue. I'm never one to say it >> definitively isn't my problem, but I'm skeptical in this case. >> >> >> >> I know that at some point in the past that anything beyond 15.0.2 was >> known to have fairly common DFS issues by some customers. I thought this >> was resolved in later releases, but I also don't see any mention of said >> issue being resolved in any release notes post 15.0.02. >> >> >> >> I was wondering if anyone knew the current status? I.E. if they had been >> seeing the problem previously, and then discovered it was fixed. Or have >> tried recent releases and discovered the problem still exists, etc... >> >> >> >> -- >> >> - Forrest >> >> >> >> -- >> AF mailing list >> AF@af.afmug.com >> http://af.afmug.com/mailman/listinfo/af_af.afmug.com >> >> >> >> >> -- >> >> - Forrest >> -- >> AF mailing list >> AF@af.afmug.com >> http://af.afmug.com/mailman/listinfo/af_af.afmug.com >> > > > -- > - Forrest > -- > AF mailing list > AF@af.afmug.com > http://af.afmug.com/mailman/listinfo/af_af.afmug.com > > > -- > AF mailing list > AF@af.afmug.com > http://af.afmug.com/mailman/listinfo/af_af.afmug.com > -- - Forrest
-- AF mailing list AF@af.afmug.com http://af.afmug.com/mailman/listinfo/af_af.afmug.com
