Hello,
This post seeks your choices and opinions regarding:
which do you use - the u.fl or the mmcx connector?
do you solder the pigtail connector to the minipci card in an effort to increase the mechanical sbability?
?2 comes as a result of observing (not using 
a Proxim MP.11a.
Kind regards,
Larry Vaden
I like the mmcx much better. Just more rugged feeling. On a card like the cm9 which only has u.fl, I snap on the pigtail then glue it so it doesn’t hit anything or come loose.
According to manufacturer’s datasheets, MMCX is good for far more insertions/removals, has --slightly-- better electrical specs. By hand, feel is obviously way more rugged.
U.FL removal is supposed to be done with a tool sold by Hirose. No such requirement for MMCX.
U.FL connectors are intended to go in laptops, PDAs and other places where space is a prime consideration. Hirose’s proud claim: “worlds smallest microwave connector”. Too bad we’re not building the world’s smallest infrastucture gear, or they’d be just dandy. They’re a misapplication for any situation where periodic, routine servicing is likely. I suspect the only reason we see 'em in gear such as Mikrotik is because the market for cards is dominated by fidgety miniaturized applictions.
Even better is SMA. I notice that RFLinx has recently dumped MMCX on their OEM boards in favor of SMA, an excellent choice.
I use I-PEX connectors, seems to be a compromise between u.Fl and MMCX. Although I haven’t used many u.Fl connectors. MMCX is the norm with Cisco gear.
I just saw a web site that has the opption of SMA. I have an ass load of those old pigtails from way back when I used 2meg zcom radios. Not that I would want to use the old pigtail but they were a whole lot better the all these new connectors.
I like ufl pigtails perfectly fine, and they seem to work fine after several change-outs. Also, they do not pop off too easily. MMCX is definitly more heavy-duty, and I would use that if I were soldering.
I get really fed upwith small RF connectors. Too much of a pain to troubleshoot and replace.
I think all equipment should be “N connector” only!
On UFL you will have about 5dB better signal than on MMCX
Really, over twice the signal manages to squeeze through a U.FL? That’s quite a performance boost. I missed that in the data sheets; I’d have thought Hirose would tout such a remarkable advance even more than the miniaturized size.
Actual manufacturer information says the insertion loss of an MMCX connector is 0.2dB, w/VSWR figures marginally better than U.FL at 5ghz.
If you’ve found MMCX connectors w/an insertion loss of 5dB it sounds as though you need to get your money back.
no no no - mmcx is good on 5GHz, but on SR-5 mmcx’s are no as good as Ufl. We tested this on about 10 “different” SR-5 (from other distributors, not a same time) and the result was that…
UFL is better than MMCX on SR-5 about 5dB!!
Of course we have our own UFL - Nf (or Nm or RSMA) with special ultra low loss degin:
I think that is hardware diversity problem for UBNT card, because even if i will change antenna-mode to MMCX in MT some signal is transmited on ufl - I know that’s normal, but on UBNT this signal was too high comparing to other atheros and their diversity.
and we need to use ufl…
as You can see the motherboard on top have 2x SR-5 (1 and 3 slot)
this is more detailed image:
Ah, I think I understand. You’re not saying U.FL are better than MMCX, instead that Ubiquiti’s implementation is borked. That’s quite plausible, especially as it sounds as though you’ve made quite a study of it. Please accept my apologies for my snark!
Have you tried multiple implementations of the MMCX-NF cables, from different sources? If so and you’re seeing the same 5db deficit that would seem to clinch it.
marksx:
What kind of boards are you using in the picture, and is that box indoor or outdoors?
Ubiquiti has or had a post on their web page about the mmcx not working correctly. http://www.ubnt.com/updates/index.htm
It appears that since Aug 30th they have corrected it.
John
Have you tried multiple implementations of the MMCX-NF cables, from different sources? If so and you’re seeing the same 5db deficit that would seem to clinch it.
Yes, we have tested 3 (as i remember) different
What kind of boards are you using in the picture,
Siemens D1218 + P3 800MHz (max 22W power consumption, v. high performance)
4 PCI(or mPCI) + integrated Intel 10/100 Ethernet without IRQ conflicts
[marksx@PEC_1] system resource irq> pri
IRQ OWNER
1 keyboard
2 APIC
3 [a5212_pci]
5 ether1
8 [rtc]
9 [a5212_pci]
10 [a5212_pci]
11 [a5212_pci]
14 IDE 1
and is that box indoor or outdoors?
“indoors” but in 105m above ground tower, and 1m from “window”
This is all really good information to have. We have at least four of the older boards in use.
Now they say “Both the SR2/SR5 now have improved MMCX connector performance. It is recommended to use MMCX connector cables for optimum results. Any questions, e-mail support@ubnt.com”
marksx:
What length has antenna cables in this picture?
Great installation.
Best regards
Kiket
depends on distance from antenna to MT, from 4m to 14m
for example main link (backbone) is 28dBi Grid + 6-7m + SR-5 through ufl-N - signal on 18km is -65dBm (with turbo + Nstreme) about 25-28Mbps full duplex performance (thanks to compression) on other side is same antenna and radio
All cabling is based on CNT-400 from Andrew and on Andrew connectors.
Hi marksx,
Thanks for your explanation, but… 14m of cable?use CNT-400 from Andrew in this length too?I think that this is similar to lmr400, it’s right?.Do not attenuation too much 14m of cable?
I’m very interested in your installation.
Un saludo
Kiket
CNT 400 is smillar to LMR 400
14m @ 2,4GHz - give us about 3-4dB attenuation which is acceptable
5GHz antennas are connected with cabling below 10m, but if you don’t need to make very long connection (for example few kilometers), cable legendth isn’t so important, all you need to calculate budget with Radio Mobile Deluxe.
look that was simulation on this link:
we don’t know how long cables will be (we cutted cable and soldered all conectors 105m above ground ) so this signal minus about 3-5dB is real, when we setting up antennas we had about -59dBm (18km link) so Radio Mobile is very accurate