From a post on the AFMUG list a few months ago by Forrest:
On the cambium ask the synchronization experts thing I'm doing someone pointed out the section in the ePMP manual talking about frequency reuse. I had to go look since I had somehow missed that section. In case you haven't seen it, there's a deep discussion about the proper settings when you have a "mixed" network of internal GPS, CMM3 and CMM4. (See Page 94 in the current release). I admit my ignorance and I hadn't seen this yet, but.....
All I have to say is this: WTF Cambium?
Ok, I know the dirty little secret that the CMM3 and CMM4 produce different timing pulses. Obviously different enough that it matters in the ePMP world. But the difference is easy enough to fix... add a fixed amount of delay which varies based on the timing source. They already have to delay the GPS timing signal to match the CMM4. Why not delay the two sources which are 'early' to match the 'late' source? Or even better, fix the CMM4 so it matches every other sync source on the planet - I'm shocked this hasn't already been done. Ok, I get it, it might be a hardware limitation which prevents one or both of the solutions. But come on, really?
For the record the syncinjectors currently produce "CMM3" timing, not "CMM4". Since it's obvious that the difference matters now, I'm considering adding a software witch which allows this to be configured so you don't have to deal with a 'mixed' timing source since it seems like a CMM4 and internal GPS are identical. This should permit better operation when you have to switch between an internal GPS and a SyncInjector.
The inductance of the cabling distorts a square wave, and the different hardware designs to generate that square wave react differently to the attached inductance resulting in variations between the CMM3, CMM4, CTM, CTM2, and the various Packetflux devices. None of this was very critical on FSK radios but it's getting more serious with tighter timing requirements.
Cambium - can you indicate where 1588v2 is in the software map? Is this a couple months out, or a couple years? I can kludge together timing short term, but would much prefer to move away from the current solutions as soon as possible.
Mark - IEEE 1588v2 support is not in the next few months. We are working on implementation of this, but I don't see it getting into the software until late 2016. There are several features prioritized ahead of this as of now. As George mentions above, there are some revisions to the timing structure (and there will be additional changes moving forward) to help "sync" the various sources of sync...
Sorry Chris, we won't be posting roadmaps in a public forum... we will try to answer as many questions as we can about them, but posting future development work in a public place is asking for trouble. Hope you understand.
5 months later. We are a little late to the 450i party and are looking around at our options for Power + Sync. Rack mount solutions aren't really our thing. We come from CMM 2 and CMM Micro land where we have a midspan sync/power box 1/2 way up the tower. CMM 4 is still on the expensive side and has some shortcomings. While developing the latest and greatest AP's, a CMM 5 sure seems like a good thing to have right about now.
Since there isn't a CMM 5, we are forced to go with the "other solution". How far away is a new Cambium sync/power solution?
I know this is Cambium's forum, but I have to vote for PacketFlux. Forrest is working on a 12-port DIN rail universal SyncInjector. He's also planning a 16 or 18 port rack mount. I believe he said both of these will have two voltage rails (for mixed +24 and +48 volt sites). We use PacketFlux gear at nearly every single site.
As for other Power + Sync solutions, there's Last Mile Gear as well.
Just for curiosity, what is wrong with the CMM4? We exclusively use the CMM4 rackmount now. After too many years of worrying about damaged equipment up towers from lightning strikes, I have moved away from tower mounted gear as much as I can. Not to mention far more switching options, gigabit to each radio, eliminating that single point of failure, etc. I also really like the dual power and now capability to power the 450i gear without issue. I love them.
We are certainly not a startup WiSP, and 8-ports is fine for us. Although I do have a few sites that have 2 CMM4's in a rack.
External switch - this is the only way to go. Anything they could build would be limited. I suppose a built in switch would be okay for startup WiSPs, but anyone that has some substantial networking to do would be using something external.
And I'm super happy that the dongle allows us to support the 450i hardware that uses a different power scheme without requiring a different piece of hardware. I think Cambium did a great job in adding that compatability. If it weren't for that dongle, or having to qire differently, people would add to their complaints that the CMM4 doesn't work with the 450i series.
Price? Meh. If you're saying that 8-ports isn't enough, then the price of the CMM4 is peanuts compared the rest of the hardware you're pluging in. The price may be too high for startup WiSPs, but not growing or well situated ones.
Two power supplies are not expensive - the new ones Cambium is marketing (Meanwell) are quite inexpensive. And there are alternatives and other manufacturers. I am happy to be able to power both the 24 VDC and 48 VDC equipment with these.
Dongle is not even needed. You can follow the pin out in the manual should you choose to. We use dongles because all the cables are uniform so if tested by other people who are not informed it will test properly.
We also use external switches and I agree it's the way to go. Even sites in the past that have the outdoor CMM4s still use an external switch.
As for 8 ports if you have more than 8 ports worth of APs then I imagine that site is paying for its self a few times over each month.
All of our sites are -48v DC systems and I love having all the redundancy built into the CMM4. Dual power supplies, multipul ways to get GPS sync injected and they are a surge arrestors. The Meanwell injectors we also buy redundant 1 for each port.
Don't get me wrong with the muMIMO gets released I imagine we will need gigbit ports but then again an AP which supported 50 subs now in theory can support 150 subs.