MikroTik RouterOS • View topic

vspace · **Posted:** Fri Mar 27, 2009 3:59 pm

Hi all.
I have a huge problem with rb433ah'a. I have bought several of them and unfortunately every one has got the same problem.
Configuration on every base station is the same.
Three mpci cards (cm9, cm10, sparklans, ub5 tested), every card working at 5Ghz 20Mhz band. Frequency gaps between evry card frequency are more than 40mhz.
hardware - aluminium sealed IP66 box, long - 30cm pigtails, andrew N clamp connectors, andrew cnt400 cables, at the ends grid antennas or sector antennas (good brand antennas)

Problem shows up while all three cards are enabled or transmitting as small bandwidth speed on links. The CCQ drops down to ~10% or even less.
When I enable nstreme protocol on one card - it works. Great nstreme

But when all cards have nstreme enabled, it is even worse than working without it. Long response times, even lower bandwidth speed on every card, lost of packages.

While tested every link separately (cards disabled or not inserted to rb) it works great.

I assume that the problem is connected with small space between cards/slots.
Does anyone have right solution to my problem? Do I have to shield additionally every card with copper or aluminium tapes/sheets?

nest · **Posted:** Sat Mar 28, 2009 1:55 am

The cards are interfering with each other, directly from one PCB to the next. The shielding on each card and between the cards is not sufficient for operating all three on the same band. You would have to design your own earthed shielding, or better still, use a separate box for each card/RB pair. You could be able to possibly get away with two cards in one RB, but I have found you still will get degraded performance. It all depends on how much degrading of the signal your link can withstand.

RK · **Posted:** Sat Mar 28, 2009 7:18 am

You basically need to run one card at 5200 MHz and the other card at 5800 MHz to get good results.

ste · **Posted:** Sat Mar 28, 2009 8:20 am

vspace wrote:

Hi all.
I have a huge problem with rb433ah'a. I have bought several of them and unfortunately every one has got the same problem.
Configuration on every base station is the same.
Three mpci cards (cm9, cm10, sparklans, ub5 tested), every card working at 5Ghz 20Mhz band. Frequency gaps between evry card frequency are more than 40mhz.
hardware - aluminium sealed IP66 box, long - 30cm pigtails, andrew N clamp connectors, andrew cnt400 cables, at the ends grid antennas or sector antennas (good brand antennas)

Problem shows up while all three cards are enabled or transmitting as small bandwidth speed on links. The CCQ drops down to ~10% or even less.
When I enable nstreme protocol on one card - it works. Great nstreme

But when all cards have nstreme enabled, it is even worse than working without it. Long response times, even lower bandwidth speed on every card, lost of packages.

While tested every link separately (cards disabled or not inserted to rb) it works great.

I assume that the problem is connected with small space between cards/slots.
Does anyone have right solution to my problem? Do I have to shield additionally every card with copper or aluminium tapes/sheets?

Does it help to lower tx-power?
Have u tried to disable csma/cd?
Have u tried to shield the pigtails?

Stefan

wireless12 · **Posted:** Sat Mar 28, 2009 10:05 am

well, its always better a RB with single minipci slot i mean its always better to use 411AH instead of 433AH (exceptions may be there)
because of the following reasons:

1. if u are having only one 433ah and three mini pci crad then u must be using a single POE and adapotor and again only one Cat5 to the router box. if there is nay isue in both of them the entire RB will be like a dead body.. all the network behind it will get affected

2. may be sometime the combination of all three mini pci card on 433Ah show issues like hanging reebot of routerboard.

3. IF one card is damged or dead u hav to climb for it and bring the entire BOX down and then replace it with new one. till that time ur entire network would be agian down behind that RB433AH

so using 411Ah is better in place of 433Ah if there is no special compulsions.

hope this helps u..this is my personal opinion.
hav a great time.

adron · **Posted:** Sat Mar 28, 2009 12:22 pm

vspace wrote:

Hi all.
I have a huge problem with rb433ah'a. I have bought several of them and unfortunately every one has got the same problem.
Configuration on every base station is the same.
Three mpci cards (cm9, cm10, sparklans, ub5 tested), every card working at 5Ghz 20Mhz band. Frequency gaps between evry card frequency are more than 40mhz.
hardware - aluminium sealed IP66 box, long - 30cm pigtails, andrew N clamp connectors, andrew cnt400 cables, at the ends grid antennas or sector antennas (good brand antennas)

Problem shows up while all three cards are enabled or transmitting as small bandwidth speed on links. The CCQ drops down to ~10% or even less.
When I enable nstreme protocol on one card - it works. Great nstreme

But when all cards have nstreme enabled, it is even worse than working without it. Long response times, even lower bandwidth speed on every card, lost of packages.

While tested every link separately (cards disabled or not inserted to rb) it works great.

I assume that the problem is connected with small space between cards/slots.
Does anyone have right solution to my problem? Do I have to shield additionally every card with copper or aluminium tapes/sheets?

What is the distance between your antennas?

ste · **Posted:** Sat Mar 28, 2009 12:38 pm

wireless12 wrote:

well, its always better a RB with single minipci slot i mean its always better to use 411AH instead of 433AH (exceptions may be there)
because of the following reasons:

1. if u are having only one 433ah and three mini pci crad then u must be using a single POE and adapotor and again only one Cat5 to the router box. if there is nay isue in both of them the entire RB will be like a dead body.. all the network behind it will get affected

2. may be sometime the combination of all three mini pci card on 433Ah show issues like hanging reebot of routerboard.

3. IF one card is damged or dead u hav to climb for it and bring the entire BOX down and then replace it with new one. till that time ur entire network would be agian down behind that RB433AH

so using 411Ah is better in place of 433Ah if there is no special compulsions.

hope this helps u..this is my personal opinion.
hav a great time.

Problem is when your network grows. E.g. a site with uplink two connected towers and 3 sectors. If you cant plug more than one card per board you have a lot of boards and cableruns. Operating a wireless network I've learned to reduce equipment as far as possible. Each device has its lifetime and outages. More equipment=more alarms at night, more updates more work.
Stefan

Aug · **Posted:** Sat Mar 28, 2009 3:02 pm

Had a similar problem with high power mpci cards on rb433ah.
Ended up installing one rb433ah and two rb411ah in one box with metal plates seperating them.
Used one cat5 run. 24vdc 2A poe injector. Split the power at the box to power all three via dc jacks and plugged ethernet only into ether1 on the rb433ah.
Used short cat5 jumpers from each rb411ah to ether2 and ether3 on the rb433ah and bridged the ethernet ports on the rb433ah.
This was on a rooftop mounted box.
If it were on a tower, I think I'd run two more cat5 to each unit.

vspace · **Posted:** Sat Mar 28, 2009 4:09 pm

hi.
thanx for many suggestions.

But ... I make my living from wireless networks for some years. What all of you had written is not the problem (or it was checked and marked out in first place).
Using separate boards is an idea, byt the 433 target is totally burned.
When there is any hardware with "possibilities" such as operating 3 mpci cards at once it should work. Nobody said that the 433 series can't operate all cards at the same band.

the concept of setting 2 cards at 5180 and 5700 gives some quality back.. but not the same as used in previous versions of multi mpci rb's. And still only 2 cards working right. so the $ given for hi performance 3mpci board is wasted ...

alu box, grounded.. so there can't be doubt that reason is inside the box. (of course the rb is also grounded)
antennas at same mast - 3 antennas at 2m length. grid and sector types, so there is no back lobs.

now running 2 UB5 cards (as U know, they have shielding all over the surface) and one CM10 cards.
Tx powers @ default, all running 36-54mbits so the lowest power from defaults table ~15-18db

pigtails are made from rg-178 cable, so it is capable of transmitting well over 6ghz. shielding the cables? haven't tried..

I'll try with self made shields between cards... if there won't be any improvement gonna sell this $hity boards and replace with multi card PC's.

These boards were stored as backup hardware... because once we have tried to use them in config of replacing the big station - 6 cards.
They where tested in steel box. grounded. two rb433ah side by side, all slots used.... the result? we have removed them faster than lightning... nothing could compare to earlier PC base station....

Anyone has got working station similar to my config?
Maybe there are some mpci cards that have got "ideal" shielding?

dipson · **Posted:** Thu Apr 02, 2009 3:03 pm

Vspace,

This may not appear like what you have over there bcos I am not into WISP.

I have a box containing RB433AH. It has 3 R52 mpci, 3 pigtail and 3 secor antennas. This is my base radio and it operates on 5GHz freq spectrum.

AIM:
Backhaul link to 3 nodes in about 1km distance from the base to forward video traffic from those nodes.

Each node has a crossroad as an AP for 4 802.11b/g compliant wireless cameras. Also at each node, I have one RB433AH with one R52 mpci in an integrated antenna. What a performance!!!!

Though the installation is just about 2 weeks now, I have not noticed any interference. I hope I am not off topic?

Thank you

Aug · **Posted:** Thu Apr 02, 2009 3:54 pm

I wish the voting button came with comments box.
I've only experienced interference with the 433's using high power ubiquity cards. The ones with R52H's work fine for me.

lcrhea25 · **Posted:** Fri Apr 24, 2009 4:06 pm

I've seen some receiving issues on one of my main towers. We upgraded to the 433 boards. One AP has 1 XR2 card in it and the other has 2 XR2's in it. Each card going to a 120 degree sector. Both APs appear to be having receive issues to some customers and some are fine. Some that are having this issue have good signals. Has it been determined there is a problem with the 433 board and if so is there a fix being worked on for it?

InoX · **Posted:** Fri Apr 24, 2009 6:44 pm

I have 2 cards CM9+R5H on 5570 and 5800Mhz...no problems.

ropebih · **Posted:** Fri Apr 24, 2009 9:59 pm

I have same problems. you can find on this forum several my posts about this issue. I allso notice that more people are having the same problems. I just don't understand why are mikrotik stuff ignoring this problem.

normis · **Posted:** Mon Apr 27, 2009 11:13 am

ropebih wrote:

I have same problems. you can find on this forum several my posts about this issue. I allso notice that more people are having the same problems. I just don't understand why are mikrotik stuff ignoring this problem.

Ignoring what exactly? What problem do you have, you have not written

InoX · **Posted:** Mon Apr 27, 2009 12:36 pm

It may be interference between antennas not between cards...

ropebih · **Posted:** Mon Apr 27, 2009 3:42 pm

normis wrote:

Ignoring what exactly? What problem do you have, you have not written

I can not find right now a themes on the forum. There's 5-6 of them.It's about interferences between the cards as already stated.
Here's example: a RB 433 with two mini PCI (CM9), so two links in direction of 2 towns. One card on 5240, the second on 5560, Neither one of them does not work ok until one of them is disabled. So when both cards are enabled ping is 20-40 ms and max troughput per link is 2 Mbit.
when I disable any of these two interfaces, ping on the one which is enabled goes to 1 ms and troughput to 30 Mbit. These problems I had at a few locations. I solved some of them by replacing the rb433 with alix. Only to note that there is no interferences from other links, Thre's also another link to the 5800 which was Backhaul in second RB.

If someone from mikrotik staff want me to give them access to routerboard with that problem, I can do that. Just let me know.

similar topics:
viewtopic.php?f=7&t=29872
viewtopic.php?f=7&t=25736&p=147995&hilit=ping+unstable#p147995

Quote:

It may be interference between antennas not between cards...

No,there's no interferences between the antennas, because previously the rb133 was working there and everything was ok!!!

ropebih · **Posted:** Mon Apr 27, 2009 5:36 pm

here's two screenshot that you can see what's this about.

wlan1 ENABLE
wlan2 ENABLE

wlan1 DISABLE
wlan2 ENABLE

lcrhea25 · **Posted:** Mon Apr 27, 2009 8:03 pm

I agree that something has to be going on with the 433 board an these ubiquity cards. Like I said earlier. I have 3 sectors 2 are on 1 AP and 1 is on another AP. I having this problem even on the one with just 1 ubiquity card in it and was not seeing this issue untill the board was upgraded from a 532 board to a 433 board.

snowpro · **Posted:** Tue Apr 28, 2009 5:13 am

I have two XR5 on a RB532 running for a few months with no problem.

I also have on 433 one XR5 and one XR2 with no problems

I have two XR5 testing on Alix right now. They are side by side on this board.

tbyte · **Posted:** Wed Aug 12, 2009 11:38 pm

I have THE SAME problem, but with two cards at 2.4Ghz. If I disable a card, the other one works fine!

radocicala · **Posted:** Wed Dec 23, 2009 10:59 pm

Same problem here, a lot of people have this problem. The only thing I know is to use xr cards(they dont have this issue). It is sad that mikrotik is not take care of his customers(support their needs) and a lot of imporant problems stays unsolved. And I think that we should get here explanation how to solve it, what is the problem, what cards are witnesses of these problems. I had rb532 no problems, after changing for 433 big problems. I dont say that it is bad router, but how was said, it is for 3 cards so it should work(and there is not so much minipci wifi cards in market). Alix boards are great, but are more expensive, but next time there are my only choice. I would rather pay 30euros more for something what is totaly stable like something problemous.

JJOliver998 · **Posted:** Wed Dec 23, 2009 11:25 pm

I'm running a 433AH with a R52h and a Kingston 1000mw radio next to each other (due to the heat sink of the kingston it has to go on the topmost miniPCI slot and the R52h on the lowest). I'm running a 1km back haul on the R52h non-line of sight with about 20 trees in the way, whilst the signal strength is low because of the trees I'm not getting interference on this. Both are running on the 2.4ghz without a massive spread in channels!

Just thought I'd add my thumbs up, whilst I appreciate it doesn't help anyone in this specific setting.

colebert · **Posted:** Thu Dec 24, 2009 9:39 am

I have been dealing with this issue lately as well. I wanted to use on RB600A w/ 3 XR5s on a daughter board. Running the XR5s in the same box creates large issues. Each card can "see" the other card w/ -40 to -50 signal strength. Not good. Ended up separating to 1:1 ratio of XR5 to Routerboard and the problem was greatly reduced (though not eliminated.)

Seems to be very difficult to run high power cards @ similar frequencies side by side.

radocicala · **Posted:** Thu Dec 24, 2009 11:54 am

but it is problem mainly of low power cards. Why this problem is not in alix, rb532a? just in new mikrotik boards

rodneal · **Posted:** Thu Dec 24, 2009 6:16 pm

I had the problem initially because the cards are so close that they touch - heat transference! The the slots need to spaced further apart.

I used tooth picks to seperate them and create an air space - I stopped having problems then.

Although if you run 3 UB cards (XR#) you need to boost the amps - to solve this I ran electric to the top of all my towers and grounded and plugged directly - no POE.

Rod

PS - I still need to know what replaces nStreme in v4.4 for all my BH - I caint upgrade beyond v3.30 without loosing all connections - anybody want to start a new thread?

nickb · **Posted:** Wed Jan 06, 2010 2:10 am

I think this problem is best explained as poor design in the RB433. Many people didn't have such problems with RB532 because one card on each side and they were not directly behind each other. With RB433, cards are stacked VERY VERY close and this increases chance for interference.

I still maintain that the RB532 was the best designed MT board. It may not have had the fast CPUs that the newer boards have, but I feel that the RB532 had the best form factor, best power options, best reliability and best physical design layout. I was extremely disappointed when it was discontinued.

changeip · **Posted:** Thu Jan 07, 2010 8:32 pm

iam also having this problem... with 2 xr5 cards in rb433ah units and plenty of power. The cards only fit in the 1st and 3rd slots since they are thick. As soon as both cards associate the other will drop to about half speed. When you force one registration off the speed picks back up after about 4-5 seconds and then will drop again as soon as you let it reconnect.

I have 2 of these on the way to test to see if it helps to move one card away from the other:
http://www.adexelec.com/pci32.htm#MPCIFLEX

Sam

nickb · **Posted:** Thu Jan 07, 2010 8:43 pm

Also make sure you are running at _least_ 80Mhz of channel separation for running cards in the same board. I find that adjacent channels don't work at all, and one or two channels away still gets some interference depending on the transceivers.

changeip · **Posted:** Thu Jan 07, 2010 8:54 pm

one at 5200 and the other at 5850 or so. should be plenty of separation. after a little experimenting it seems like if i turn off nstreme it works better, and possibly rolling back to 3.30 fixes that. i will do more testing and post later.

staslabs · **Posted:** Fri Jan 08, 2010 1:44 am

ropebih wrote:

Quote:

It may be interference between antennas not between cards...

No,there's no interferences between the antennas, because previously the rb133 was working there and everything was ok!!!

We have same experience - rb433ah interference between mpci cards.
But no in RB600 with 2 XR5 cards.

changeip · **Posted:** Wed Jan 13, 2010 7:55 pm

i have in hand two of these:

http://www.adexelec.com/pci32.htm#MPCIFLEX

I don't have a quick way to test the theory of moving the radios further apart, but am willing to ship these to someone who can test them. If you are interested, and are in the USA, email me sales at changeip.com and I will mail them to you to test. I would hope you could test and send them back within a week so I can decide if I should return them (evals) or buy a bunch at a discount.

Sam

changeip · **Posted:** Thu Jan 28, 2010 8:59 am

okay everyone, here is a flash video of this as its happening.

http://h1x.com/jing/dualcards.swf

here is the setup:

rb433ah's
3.30 w/ dual xr5 cards.
32db dual polarity pacwireless hdd5w dishes
nstreme on each wlan individually
pacwireless 24i PoE

I just have the btest going between two identical routers on a 22km link. I am somewhat amazed I am getting 50+ mbps thru a single radio : )

I can kind of understand interference ... but in this video why would disassociating the remote side fix the issue? The second card / AP is still transmitting. I can even run an align / scan and it doesnt happen, but the 2nd radio is still transmitting correct? Are we sure this isn't some other problem? Please take the time to watch the video and help us all figure this out.

nickb · **Posted:** Thu Jan 28, 2010 7:18 pm

Looks like CSMA in action to me.

changeip · **Posted:** Thu Jan 28, 2010 8:03 pm

im still learning here ... should I try turning on and off the CSMA setting?

I tried 'disable CSMA' on the nstreme tab on one and the other and then both and it didn't change the outcome.

nickb · **Posted:** Thu Jan 28, 2010 8:26 pm

Then it probably wasn't CSMA.

Maybe just having the carrier active increases your BER, thus reducing the modulation rates (which is in turn reducing your TCP throughput).

talo1019 · **Posted:** Fri Jan 29, 2010 12:26 pm

I think everyone makes some valid points especially nest & valnet. Also someone is using a R52 and 1000mW kingston, very interesting. The point is with 2 Hi-Power card such as XR5, Kingston, F50-PRO stacked on top of each other, you will get inteference. The best analogy I could think of is putting 2 cell phones on top of each other and expecting great reception versus pointing in different directions or side by side.

Of course, you might get it to work with no problems at different frequencies, and other ways, my opinion is the rb532 was better because of the separation of cards.

For example, If you put a R52H or R5H or cm9 with XR5 or F50-PRO, you shouldn't have any problems....cause one is 350mW or less and one is is 5-600mW. Yes the R52H is called high powered, but the actual mW is only 350mW.

Looking at an analyzer, 1dB to 5dB is huge. That's 5 times the power! 1dB to 2dB is same, that's 2x the power, directly on top of each other.

I'll do some tests more in depth tests if people want.

changeip · **Posted:** Fri Jan 29, 2010 7:03 pm

I guess I have to test that theory by using those minipci flex extenders... i will install one on each side to move one of the XR5 cards to the other side of the case and see what happens.

netrat · **Posted:** Fri Jan 29, 2010 7:37 pm

talo1019 wrote:

Looking at an analyzer, 1dB to 5dB is huge. That's 5 times the power! 1dB to 2dB is same, that's 2x the power, directly on top of each other.

I'll do some tests more in depth tests if people want.

1dB to 2dB is 1.23x the power. 1 to 5dB is 2.46x the power. Decibel is logarithmic.

talo1019 · **Posted:** Sat Jan 30, 2010 7:16 am

Yes, I stand corrected. I probably shouldn't have described it in that fashion. Obviously, 400mW at 26dBm to 800dBm 27dBm is double based on mW. Yes, it's logarithmic in terms of actual power.

I was thinking and describing in a more linear fashion for easier understanding, but...I shouldn't and I'll more careful next time.

ne0031 · **Posted:** Tue Feb 02, 2010 10:36 pm

I have had similar issues with poor quality components, other than that radios and boards, such as pigtails specifically. Normally, when I have these issues, it is from working on someone else's equipment. I've ran 4 xr5s on a single 600, as well as 3 on a 433ah with no issues.

I've even posted in the forums before regarding running 2 radios on the same frequency, on the same 433ah. viewtopic.php?f=7&t=36230

I hadn't noted a response to a question on antenna spacing, could you answer that question?

changeip · **Posted:** Tue Feb 02, 2010 10:37 pm

in my case it's a dual polarity 32db pacwireless dish on each end.

you can only fit 2 xr5s into the rb433ah, there isnt enough room for 3...

ne0031 · **Posted:** Tue Feb 02, 2010 10:52 pm

changeip wrote:

in my case it's a dual polarity 32db pacwireless dish on each end.

you can only fit 2 xr5s into the rb433ah, there isnt enough room for 3...

We use a flexible extender when we have 3 in a single box. http://www.adexelec.com/pci32.htm and look for the flexible MPCI-FLEX towards the bottom.

talo1019 · **Posted:** Wed Feb 03, 2010 12:49 pm

Yeah, we bought 2 of the MPCI-FLEX from http://www.adexelec.com as well, and then realized that they overcharged us when it's almost half the price here in Taiwan

. I admit, it's not the easiest thing to find, probably why adexelec charges so much.

In any case, the MPCI-FLEX type extender is convenient

changeip · **Posted:** Tue Apr 13, 2010 8:28 am

okay just an update on this. maybe i have defective rb433 boards or something.

been running with the second xr5 card disabled on a dual polarity setup. anytime both xr5 cards are used bandwidth drops in half. i have tried adding those mini pci flex cables to both sides, no difference. tonight i just placed the dual radios into nstreme dual mode. linked up, but same issue, half the bandwidth of a single link.

i dont believe its interference, because the 'half the bandwidth' problem does happen if im using the other cards in active alignment mode. thats transmitting at full power correct?

wirelesswaves · **Posted:** Tue Apr 13, 2010 11:21 am

You offered to do some more tests right.

What I suggest is to connect a length of leeky feedy to 2 RB units. Each end of Leeky feeder terminated into "PERFECT" 50 ohm loads.

Couple the 2 units together, being careful not to expose either radio card to levels in excess of -50dbm

Do the exact same of other router board setup.

Use frequencies suggested.

Run the same tests again.

Then replace pigtails with skinny feeder. Turn units off, replace one load with 3 x 50 ohm resistors in series.

Thus substituting load for a poor VSWR.. VSWR now = 3:1
Return loss high and reflected power around 25%.

Repeat test.

The test is trying to prove whether in circumstances of high VSWR. That the return currents and the phase differences are now finding inappropriate paths in the radio cards ground plane or the RB itself and especially whether the skinny type pigtails are re-radiating.

And whether the RF stage is intollerant and producing Harmonics when presented with poor VSWR.

Its the only conclusion that I can come up with, but no one has done any form of test like it yet as far as I am aware of.

At least if this test is done it could probably rule out this issue.

changeip · **Posted:** Tue Apr 13, 2010 9:33 pm

i can try someting like you mentioned. are you saying that the antenna and cabling themselves have bad vswr and therefore causing this problem?

I always come back to this - if I can run active alignment mode on the second cards, which is transmitting and receiving full power, why doesnt it affect the other card? Only when its associated...

0ldman · **Posted:** Wed Apr 14, 2010 12:33 am

I've got R52H's in a 600, one on 5280 and one on 5825, both interfere horribly if the frequencies are closer, and both still are locked at 18mb to keep CCQ around 90%.

I've got a couple of 433's I'm working on with an R52 or R52H as the 5GHz backhaul, R52H as a 2.4GHz AP and an XR5 as a 5GHz AP (likely all to be swapped for N equip before I'm done).

The XR5 and R52 don't cause any drop in CCQ when on adjacent channels, though my signal is in the -60's on the R52 and -70's on the XR5. Both are using the same antenna as the tower with the RB600, Pac Wireless 5GHz grids, just in the testing case, no grid, just the feedhorn. In this case I did place the 2.4GHz AP card in between the XR5 and R52.

The XR5's see the R52's @ -40 or so, the R52's see the XR5 and each other in the -20s.

wirelesswaves · **Posted:** Wed Apr 14, 2010 1:17 pm

Hi Change IP

I dont know whether its VSWR related.

My experience used to be in the design of VHF and MF equipment. I have no practical experience in microwave design, however the theories remain the same whether its SW, MW UHF or Microwave.

I also don't have access to 10Ghz spectrum analyser, else I would already have done the test myself.

So without the aid of proper test gear I can only relate on past experiences in RF design.

It may be very inappropriate for me to suggest the following, but some manufacturers may have omitted some fundamental tests.

a: The performance of the RF stage when the power level is decreased considerably, eg: intermods, distortion, harmonics etc.
In RF stage design, the Impedance of the final devices are dependant on both drive level and current/voltage. So a card may behave as expected and in accordance to the original engineers design at say 26dBm, but things could go horribly wrong when the device is set at lower powers. (equally so at higher powers too). What is not known, is how the gain of the final stages are controlled. But in any case I would expect impedance changes throughout the final stages according to how the devices are driven.

Suffice to say that, usually the impedance looking into the port is around 50ohm and maintained that way in order to transfer maximum power. However, it is unlikely that the final amplifier stages are single ended. They are more likely to be a complex arrangement of push-pull or parallel devices in some form of substrate. What is not known is by how much these internal complex impedances change between stages as the power levels change.

Note that the transport concept has some form of feedback/control loop, self adjusting in order to meet the best CCQ. ie power levels change in accordance to throughput. Thus complex impedances are likely to change too.

So it is not that surprising that some folk are finding more stable throuputs and higher CCQ's when the levels are fixed... In other words some of the complex control/feedback mechanism has been disabled... And then I guess, the intermediate power amp stages within the RF amp are now more stable, since their interstage impedances are now stable.

2: What tests have the manufacturers done when the RF amp is subjected to impedances beyond 50ohm.

Its quite possible that the Rf amp in the card could produce spurious emission, harmonics and various nasties, that from a few metres away could be quite undetectable. But how many manufacturers designed the fundamental operations of the card based around the theory that 2 cards were going to be virtually plonked on top of one another enclosed within a metal case where circulating currents would bread!

Forcing a card to operate into anything else other than its design impedance of 50ohm would I am certain create higher than expected VSWR. Some of which is quite likely to find itself on the outer skin of the shield rather than on the inside surface area of the shield. If the RF currents are allowed to propagate on the outside skin of the shield, then who knows where they will end up.

The return VSWR will propagate back and forth and at varying phases. Adding and subtracting to power levels too.

So it may be a long shot! But its a simple enough test and I wouldnt mind betting that there may be more than one or two manufacturers out there who just didn't consider such implications as I bet most had visions of their cards being used on a 1 per board basis and certainly not stacked a few millimetres apart!

I know there are some other RF engineers out there, maybe they have some feedback on this, since I am an HF/VHF man and would naturally bow to the experiences of any microwave RF engineers on the forum.

aaa · **Posted:** Wed Apr 14, 2010 3:28 pm

Actually HF and UHF theory is not so far. Typically wireless card have baseband radio chip with software controlled exchangeable gain preamplifier and analog feed back input for wide band correction and additional two or three stage power amplifier. Circuit between baseband radio and PA most sensitive from external electromagnetic field. Other wireless card transmission can conducted to this circuit can be reason for output signal degradation. As higher power wireless card you have as higher signal degradation you acquire.

Who is online