First impressions of the C5x


#1

My sample (PtP) pair of C5x’s came last week, and I’ve been having fun testing them out. They’re just indoors, running across the room, but I’ve learned a lot.
I usually like a “good, bad, and ugly” type review, but there’s nothing ugly in the C5x. A few years ago I tried out a pair of B5Lites, which of course are C5s, and I wasn’t impressed. The package was very plasticky, and the antenna didn’t have great front:back. We put it in licensed mode and ran it on 4950, donated to a fire department, and one radio (up on a high rise roof) failed within a couple of months. Worst, though, was the single 20 dB choice. A C5c cost more and then it needed an extra antenna, making it an expensive choice. So I welcome the new C5x; it really takes many steps ahead.

First, the packaging. What a game changer! No plastic here. The base unit is heavy cast aluminum. Very, very solid. It still wants a separate ground, but I doubt it’s as necessary (if the STP is built right) as with the old C5. The competition in this price range is plastic.

Then, the modular antennas. This is great. By itself, the radio looks a little like an old Canopy, which had a really lame antenna built in. But the Canopy didn’t have the screw-threads surrounding it. My test pair has the 20 dB dishes, but I’m only using one dish and one “plain” indoors for now. It only took a few minutes to assemble the dish. Three nuts attach the Cassegrain inner reflector to the dish, and that just screws on to the base unit. A set screw locks it in.

[Aside: A Cassegrain antenna is a dual-reflector parabolic, inspired by the Cassegrain reflector telescope. A basic parabolic dish puts the feedhorn at the focus. A Cassegrain puts a small reflector at the focus, with the driven element behind the dish. That’s how come a dish can screw on to the base unit. The driven element is on the PC board, no feedline, and the Cassegrain antenna amplifies it rather like any other parabolic. Metal, with good F:B, not a round panel like the old C5.)

The C5x deserves an award for its design. And because the antenna is separate, you can go to a site with a box of antennas, and use the one that is required for that path loss. The 25 dB dish puts the C5x in the same range as the bigger clients from brands C and U.

Sorta good, sorta bad: The radio does not come with a PoE. It is very tolerant, and can take any passive PoE from 24 to 56 volts. I got mine running with both a Ubiquiti and Cambium power supply. But those were from 100 Mbps radios, so they didn’t pass GigE. Who makes passive 4-pair GigE? Most GigE is not passive! Since this is a client radio, a box like Netonix won’t usually be there. So you’ll need the Mimosa power supplies. They’re cheap enough, though. Downside: No cord, just wall wart. That’s a pain. It ties up too much space on a small plug strip, compared to a cord. You might need to keep some extension cords handy.

Not ugly: I thought I had found an ugly when I pulled the FCC test report. (US tests are tough.) It showed the frequency range as 5190-5825, not even able to go down to the lowest standard Wi-Fi channel (5180), let alone the “clean radio only” channels down to 5160. And it only shows low power level approval. And when I turned on just the first one, my cell phone’s Wi-Fi didn’t pick it up below 5180 or on channels wider than 20 MHz. Or on “odd” channels in between the common Wi-Fi ones.

BUT that turned out to be a false alarm, a problem in the cell phone. The US radio is Good! It tunes the legal channels from 5160 to 5840 (in 20 MHz mode). It allows EIRP up to, I’d estimate, 44 dBm on U-NII-1, and maybe more on U-NII-3. It does 40 and 80 MHz wide. It reduces power when needed, and if you enter coordinates while “on net”, it phones home to correctly set the TDWR exclusion. And it changes channels without the reboot delay in some competitor radios. The FCC test report does show that it’s a pretty clean radio too, which is what limits the allowed power in most cases. It’s not “FPGA clean”, but clean for a chip radio. Better than the Wi-Fi standard emission mask (which not everything meets).

Not right yet: It doesn’t tell you the power when it reduces it for regulatory reasons. If you set it to +40 EIRP and use a DFS channel, it will lower the power, but the display will show the +40. And while it properly does the DFS Wait period, the dashboard shows the radio as “on” when it’s really not. Mimosa should have the dashboard show the actual, post-regulatory-fix power and status.

With the two radios on the Mimosa power supplies, it shows GigE and the radio PHY goes above 800 Mbps, at least in “aggressive” “EVM” mode. This will be interesting to try outdoors on a long path.

BUT here’s a sorta Bad: Even with the radio displaying a PHY of >700 and a MAC speed >300, actual Ethernet throughput is not great. With the Ubiquiti power supply (100 Mbps Ethernet), I got a speed test result of about 35/12. (My cable modem connection is 400/12, and I’m going through a MikroTik RB951G switch. I get the full 400 plugged straight into the switch.) I assume I’d get much better results with Mimosa GigE supplies at both ends, but that only raised the result to around 70/12. Better but only a fraction of the reported PHY. So it looks to me like there is something keeping the Ethernet port from getting its full potential. To be sure, that’s still excellent for a CLIENT radio, especially in this price category, but as a PtP backhaul, it is not a baby B5.

In sum, the C5x looks like it’s going to be a real winner. Whole buncha stars, and kudos to the team behind it.


#2

Thanks for the write up. Very well written. I was waiting a couple months after our first installs to give my thoughts, but you have covered more then I would have.

Could the radio be causing self interference when you are doing the testing? Being indoors and and so close even with just 8 db antennas you could be saturating the antennas…


#3

It’s not self-interference on the radio side. The reported PHY is good. When I point them near enough to each other than the RSSI gets above -40 or so, the PHY does go down. But I turned the power down to -1 dBm out and the PHY was really happy. It just didn’t all show up on the Ethernet side.

Your point does remind me though of a small nit, that the power control might be better. Right now you set the power on the AP side and the subservient side is set to the same power. I presume the A5/C5 combo has the client power automatically controlled to give the ideal RSSI (if possible) at the AP, doesn’t it? (I don’t have any A5s.) If it could do closed-loop control on a PtP link that could be better than just matching a fixed power level. Path loss may be uniform in both directions but noise levels may not be, given how noisy 5 GHz has become.


#4

Ahh, well figures. Dunno at that point. I may try to duplicate the test with iPerf sometime then and see if I can get the figures to Mimosa Support… When I have the time…

In PTMP Mode the C5# can have Auto Power turned off in the Channel & Power Menu. Dunno about PTP mode, though would figure the same deal.


#5

Auto power is not a setting in PtP mode. It would probably be nice. Auto Everything doesn’t seem to affect power, just frequency. And I always like to keep control of frequency.
I had a thought about the speed, though. What if my PtP key doesn’t actually unlike gigabit mode? It’s capping right around what a 100 Mbps Ethernet would support, under ideal circumstances. I checked the key server and the “new” key it generated is the same one I got a day or so before they changed their policy… Did PtP mode ever have a 100 Mbps lock, or was that only PtMP mode? Or could it be a bug?


#6

Mystery solved!
I found one bad jumper cable feeding one of the C5xs. A fresh jumper got the speed test up to 350 Mbps, close to my cable limit.
This box rocks!
I don’t know if there would have been errors showing on the Ethernet, if the C5x reported them. My experience is that 100 Mbps Ethernet is pretty bulletproof, but gigabit on copper is always touchy. I debugged a nasty situation on a competitor’s backhaul radio via the reported SNR on each pair of the GigE. It was just an insanely noisy (power equipment, motors, etc.) location and the STP couldn’t handle it. The Mimosa radios have MUCH better RF diagnostics but not on the Ethernet.