View Full Version : M5 Signal Performance
cameron
09-04-2009, 05:47 PM
Hi,
I recieved my Bullet M5 the other week and have been doing some testing, But so far I have had no luck with the M5.
I am testing over 11km's, I have a 12db Waveguide Omni and a 15db Waveguide Sector 130mtrs up a tower 11km's away.
The Current Setup:
Bullet5 (27dbi Dish) Connected to the Sector -61, 99% CCQ (It can see both the Omni and Sector)
New setup:
BulletM5 replaced the Bullet5 no change in antenna or position, -63, 2.5% CCQ (only see's the sector)
So I thought maybe the M5 doesnt like this site so I took it to another site, 14.8km from the tower 100% LOS and no fresnel interference. I did a site survey and it saw the omni at -81/-79 and nothing else, weird I thought as I should of still been in the coverage over the sector.
I unscrewed the M5 and installed the B5 and bingo all 3 of our AP's on the tower showed up (omni, sector, backhaul panel) the panel should not have shown up and it was facing the other direction completely.
It appears the M5 is nowhere near as sensitive, and I cannot get my CCQ above 3% where on the same antenna the B5 sits on 99% CCQ and sits happily at 108mbps on a 40mhz channel and the M5 barley gets 6mbps in one direction and 0mbps in the other.
I am not sure what I am doing wrong hopefully somsone can help me out.
Cheers
Cameron
bobbyc
09-04-2009, 06:10 PM
auto ack being used on the m5? If so it's not quite working and results in exactly what you're seeing (low ccq and low rates), so disable and manually set the distance.
backwards compatibility with 802.11a might not be quite there yet (or perhaps not really intended) so that might be why it's not showing everything on your scans.
Bob C
cameron
09-04-2009, 06:46 PM
Bob C,
thanks for your reply, I tried both on and off. I set the tower on both the omni/sector to a/n using the Mikrotik R52N
drwho17
09-04-2009, 07:44 PM
I'm seeing similar results with my Bullet2M's thus far. Same antenna/same AP's, Bullet2HP's are working better.
rconaway
09-04-2009, 08:02 PM
The original version of firmware had that ACK bug. I haven't tested the RC version yet but I would upgrade first and retest.
cameron
09-04-2009, 08:11 PM
Yea I am using the latest version, My B5 is not the HP and performs alot better. the CCQ sits at 2.5% and wont change and the connected speed wont go any higher than 0/6mbps
rconaway
09-04-2009, 09:31 PM
Turn off autoACK and set the distance manually. That should fix the problem.
UBNT-Robert
09-04-2009, 09:43 PM
Cameron --
If you are using Bullet M5+27dBi grid vs. Nano M5 (16dBi built-in):
Your RX signal strength theoretically should be ~11dB different.
I could not quite follow your post, but it appears you are seeing much more than 11dB difference?
Robert
he is seeing 2.5% CCQ vs. 99% CCQ, thats the important thing
rconaway
09-05-2009, 05:15 AM
Robert, that was a bug in the first firmware that we saw also on a 2 mile link. With autoACK set, the units link up at a higher rate and then slowly degrade. When you look at the main page, it calculates distance at .2 miles regardless of what you set it at originally. Turn off autoACK, set the distance, and I got 100% CCQ and 74% of bandwidth. Problem solved. I haven't retested with the RC version but we will do that Tuesday.
bobbyc
09-05-2009, 08:06 AM
The auto ack problem still remains in RC.
Yea I am using the latest version, My B5 is not the HP and performs alot better. the CCQ sits at 2.5% and wont change and the connected speed wont go any higher than 0/6mbps
One problem is easily fixed, one might be a bullet killer.
If you can establish a connection, you can make it work better by setting the ACK Timeout manually. The AutoACK is totally broken when connecting to most 802.11 APs.
As for not being able to see the APs, that's a whole other problem. I just hope UBNT can fix it with software and it's not a hardware design issue.
cameron
09-05-2009, 05:33 PM
Hi,
I have tried it with & without AutoACK It made no difference, I see there is another thread with the same issue I am having.
Overview:
I have a link over 11km's that I get 38mbps TCP through using all Mikrotik gear, On the client end I removed the Mikrotik setup and install a M5 Bullet, After doing this I could barley see the AP on the other end. The original signal was -61 and 99% CCQ, after installing the Bullet the signal strength was -62 and 2.5%CCQ.
I tried this on a 15km link and it saw the same AP at -81 and similar CCQ, I removed this Bullet M5 and installed my spare Bullet5 not the HP and signal was -79 and 85% CCQ.
In conclusion the Bullet M5 is not performing at all.
Help...
Regards
Cameron
rconaway
09-05-2009, 05:38 PM
Okay, you are combining two different issues. I assumed you had Bullet 5M's on both sides. Is that correct?
cameron
09-05-2009, 05:41 PM
Hi,
Incorrect, On the AP end is running a Mikrotik with a XR5 and 27dbi Panel and the Sector is a Mikrotik with R52N and a 15dbi Waveguide sector running in a/n
Regards
Cameron
rconaway
09-05-2009, 05:54 PM
I think there is a N compatibility issue which would leave you with 802.11a mode which is what you are testing. I haven't tested that mode and compatibility yet which might be your second problem and that may be related to firmware. The only way to run really run your test is with Bullet's at both ends and then wait for the firmware to catch up. However, I could be wrong so someone else jump in here if I am.
cameron
09-05-2009, 05:56 PM
Yea,
I spoke to mike before I tried it and he said they work fine on 802.11a AP's, It also appears this fellow is having the same issue: http://www.ubnt.com/forum/showthread.php?t=13342
Cameron.
rconaway
09-05-2009, 05:58 PM
Then I step back. I'm not using the 802.11a mode in anything with the M's.
cameron
09-05-2009, 06:03 PM
Yea I wasn't going to I was just going to get it working in 802.11a then switch on N and away it should go but but it didnt..
Hopefully Mike can comment when hes around.
Cameron.
Hi,
I have tried it with & without AutoACK It made no difference, I see there is another thread with the same issue I am having.
Before you turned off AutoACK, did you set the correct manual ACK?
This is CRITICAL.
Hint: Use the ACK RouterOS auto detects and add 5%.
cameron
09-05-2009, 11:49 PM
Yea, I set it to what RouterOS said for the link. eg. link was 10km I worked it for 10.5km
lncommunications
09-06-2009, 04:41 AM
Try turning Auto Ack off on your AP (set to 5~10% extra distance) and then set Auto ACK on all Stations, make sure you have AirMax turned off.
...and make sure you have the latest firmware!
cameron
09-06-2009, 07:14 AM
Hi,
I have AutoACK off on my AP's, I have spent alot of time tuning the links for the network.
I also have the Latest RC firmware running on the Bullet M5 :( And Airmax is off.
Cameron.
lncommunications
09-06-2009, 07:23 AM
Hi Cameron, what issues are you seeing? and what are you using to test?
cameron
09-06-2009, 03:45 PM
Hi,
As mentionedo n the previous pages, CCQ will not go above 2.5%, and the radio sensitivity is usless compared to my Bullet5.
Doing a scan with my bullet 5 shows all of my AP's 10km away, doing a scan with the bulletM5 only shows 1 of them.
rconaway
09-06-2009, 07:13 PM
That is the bug with AutoACK. Are you sure you have it turned off?
cameron
09-06-2009, 10:09 PM
100% positive, Checked multiple times while trying to get it working.
UBNT-Mike.Ford
09-08-2009, 10:12 AM
Hello Cameron.
Also two things:
1. Turn off AutoACK and set it to 125% of the distance.
2. Check the center pin of the N Type connector. Apply a small amount of pressure and see if it spins freely.
If number two is the cause, please email me at mike@ubnt.com
Thanks,
radio sensitivity is usless compared to my Bullet5.
http://www.ubnt.com/forum/showthread.php?p=47993
Can you test your Bullets on the bench. I tested a dozen of mine over the weekend and half of them had a bad RF section that added an additional 10 to 14 dB loss.
I suspected for several months there was a problem with some the Bullet M5 radios, but it was only last week that I finally had the chance to lay them out on the bench and narrow it down. I haven't reported my results to Mike yet, so I may be the only one that knows of this.