Power out differences on RB333

galaxynet · Wed Oct 31, 2007 2:44 pm

Now here is an interesting anomoly....

I have 6 RB333s running backup links w/ ROS 3.06rc on two and 3.09rc on the remaining four.
Using XR5 cards in all units.
One set running Nstreme dual, the remaining sets running 'standard' AP/Station modes.
On each and everyone of these RB333 the 'second' XR5 card has a lower output power, by about 10db. In the TX window with the AP/Station mode it shows the first card with power out starting at -28db gradually going to -22 at 54mbps - this is normal....
The second card however starts out at -27 and goes down to -21. Not a big difference - I could live with that if that's all it was.....

However, the reporting from either side actually shows a 10db or slightly more difference in the signals. Same antennas, same cabling, everything is the same. You can swap the MMCX connectors to try and eliminate the antenna/cable - everything stays the same. The second card is just a flat 10db difference.....

I tried replacing the cards, no difference. Tried a 2.4ghz card as the seconday (XR2) - same exact issue - 10db down from the first card.

Now this is strange -

MT - how about it?

Anyone else seeing this same thing?

mstead · Wed Oct 31, 2007 9:02 pm

Hi,

Just a thought but I remember some MiniPCI cards would be 10dB down if you were using the wrong antenna - ie A instead of B in advanced settings. Worth a look though probably not the problem here.

Malcolm

galaxynet · Wed Oct 31, 2007 9:20 pm

mstead -
You are correct in it not being the problem here - and typically if you are on the wrong antenna port (the older SR2 and SR5 cards) then it's more like -20db or worse.

With the XR5 there is only one port so.....

Gerard · Wed Oct 31, 2007 10:48 pm

I just upgraded a tower today from 2.9.48 to 3.0rc9 and I noticed a lost several dB.. After looking at the TX power it seems it is choosing the wrong txpower by default..

SR5 current-tx-powers: 6Mbps:26(19),9Mbps:26(19),12Mbps:26(19),18Mbps:26(19),24Mbps:26(19),36Mbps:24(17),48Mbps:21(14),54Mbps:20(13)

SR2 current-tx-powers: 1Mbps:26(16),2Mbps:26(16),5.5Mbps:26(16),11Mbps:26(16),6Mbps:25(15),9Mbps:25(15),12Mbps:25(15),18Mbps:25(15),24Mbps:25(15),
36Mbps:24(14),48Mbps:22(12),54Mbps:21(11)

I just checked another tower that has 2 XR2's and they are working correctly(28 (18)).. But another one with SR2's in it is showing 26(16).. Which is incorrect..

-Gerard

Edit, It's not just RB333's that are having the issue. It is something with 3.0rc9. (maybe earlier releases too, I haven't tested though)

Thu Nov 01, 2007 10:36 am

I just upgraded a tower today from 2.9.48 to 3.0rc9 and I noticed a lost several dB.. After looking at the TX power it seems it is choosing the wrong txpower by default..

SR5 current-tx-powers: 6Mbps:26(19),9Mbps:26(19),12Mbps:26(19),18Mbps:26(19),24Mbps:26(19),36Mbps:24(17),48Mbps:21(14),54Mbps:20(13)

SR2 current-tx-powers: 1Mbps:26(16),2Mbps:26(16),5.5Mbps:26(16),11Mbps:26(16),6Mbps:25(15),9Mbps:25(15),12Mbps:25(15),18Mbps:25(15),24Mbps:25(15),
36Mbps:24(14),48Mbps:22(12),54Mbps:21(11)

I just checked another tower that has 2 XR2's and they are working correctly(28 (18)).. But another one with SR2's in it is showing 26(16).. Which is incorrect..

-Gerard

Edit, It's not just RB333's that are having the issue. It is something with 3.0rc9. (maybe earlier releases too, I haven't tested though)

I don't see the problem.
XR2 reports 28(18) - correct
SR2 reports 26(16) - correct

http://www.ubnt.com/downloads/ubi_mtik_power.pdf

Thu Nov 01, 2007 10:42 am

mstead -
You are correct in it not being the problem here - and typically if you are on the wrong antenna port (the older SR2 and SR5 cards) then it's more like -20db or worse.

With the XR5 there is only one port so.....

try to set individual nstreme1 links and compare the signals.

galaxynet · Thu Nov 01, 2007 1:55 pm

As I stated in the beginning of this post, power out from the second card is lower than the first card - all modes all the time - AP/Station, Nstreme, and Nstreme-Dual.

Setting the card values manually to their rated power out does not change the power out actually seen - it is still at least -10db from the first card.

Letting ROS set the card power out value automatically results in the cards being set 1db lower than rated but again the power out of the XR5 card is still -10db or more down from the first card.

I don't know how much clearer I can make this. I will be sending a supout file later today to support [at] mikrotik.

Thu Nov 01, 2007 3:42 pm

As I stated in the beginning of this post, power out from the second card is lower than the first card - all modes all the time - AP/Station, Nstreme, and Nstreme-Dual.

Setting the card values manually to their rated power out does not change the power out actually seen - it is still at least -10db from the first card.

Letting ROS set the card power out value automatically results in the cards being set 1db lower than rated but again the power out of the XR5 card is still -10db or more down from the first card.

I don't know how much clearer I can make this. I will be sending a supout file later today to support [at] mikrotik.

so to test it we need 2 pieces of XR5 cards in a router and then comprat the output power at a specific data rate and on the same freqeuncy. And in this cas the first card will have -10db lower tx power ?

galaxynet · Thu Nov 01, 2007 3:56 pm

uldis - thanks for the reply.

It will actually be the second card that will show the -10db or more difference.

As to data rates - it does not matter which data rate you choose. I have disabled all rates except for 6mbps and the pwer difference is still the same. I enabled the other data rates one at a time all the way to 54mbps and turbo modes, power is always at least -10db from card number one.

My XR5 cards are currently set for frequency - card one = 5805mhz (or 5.805Ghz) whichever you want to call it. Card two = 5220Mhz (5.220ghz).

As I also said in the beginning - I have six RB333s. Two units have 3.0rc6 and the remaining 4 units have 3.0rc9. The problem is in both rc's.... I see exactly the same issue on all 6 units.

Thu Nov 01, 2007 4:25 pm

uldis - thanks for the reply.

It will actually be the second card that will show the -10db or more difference.

As to data rates - it does not matter which data rate you choose. I have disabled all rates except for 6mbps and the pwer difference is still the same. I enabled the other data rates one at a time all the way to 54mbps and turbo modes, power is always at least -10db from card number one.

My XR5 cards are currently set for frequency - card one = 5805mhz (or 5.805Ghz) whichever you want to call it. Card two = 5220Mhz (5.220ghz).

As I also said in the beginning - I have six RB333s. Two units have 3.0rc6 and the remaining 4 units have 3.0rc9. The problem is in both rc's.... I see exactly the same issue on all 6 units.

We will try to reproduce it. But note that the output power for each freqeuency is different. Try setting both cards at the same freqeuency.

galaxynet · Fri Nov 02, 2007 12:24 am

uldis -
I understand the power / frequency relationship - I have been a microwave engineer for 30+ years now.....

Since they are in the same enclosure I have set them like this; 5805 and 5825Mhz with exactly the same results. I have 'flipped' the frequencies - card one having 5220 and card two having 5805 and vice versa, - it is ALWAYS the same, card two is at least -10db down from card one. If it were -1db or -2db I'd just chaulk that up to either differences in frequency response or differences in the cards themselves. But not when I have six operating units that are all the same and all have the exact same symptoms. And I have changed all the cards for new, out of the box cards (XR5) so that only leaves one thing - the RB333 and ROS 3.0Xrc

Gerard · Fri Nov 02, 2007 5:09 am

I just upgraded a tower today from 2.9.48 to 3.0rc9 and I noticed a lost several dB.. After looking at the TX power it seems it is choosing the wrong txpower by default..

I just checked another tower that has 2 XR2's and they are working correctly(28 (18)).. But another one with SR2's in it is showing 26(16).. Which is incorrect..
I don't see the problem.
XR2 reports 28(18) - correct
SR2 reports 26(16) - correct

http://www.ubnt.com/downloads/ubi_mtik_power.pdf

uldis,
I apologize, You are correct about the default tx power on the SR's.

However there is still an issue with signals dropping after upgrading to 3.0rcX. I've upgraded most of my point to point links to 3.0rc9 and signals aren't as good after the upgrades. For example I have a 5.8 p2p link that was rock solid with a -80 signal using 2.9.48, After upgrading both sides I'm having trouble keeping the link connected, and I'm seeing -89-92 signals. After running 3.0rc9 for a couple days until I could drive out to the remote tower I downgraded both sides back to 2.9.48 and the link returned to -80-81. Both sides were using rb532's. One side had two XR5's and the other had a XR5 and XR2.

-Gerard

pdf · Sat Nov 03, 2007 6:50 pm

uldis,
I apologize, You are correct about the default tx power on the SR's.

However there is still an issue with signals dropping after upgrading to 3.0rcX. I've upgraded most of my point to point links to 3.0rc9 and signals aren't as good after the upgrades. For example I have a 5.8 p2p link that was rock solid with a -80 signal using 2.9.48, After upgrading both sides I'm having trouble keeping the link connected, and I'm seeing -89-92 signals. After running 3.0rc9 for a couple days until I could drive out to the remote tower I downgraded both sides back to 2.9.48 and the link returned to -80-81. Both sides were using rb532's. One side had two XR5's and the other had a XR5 and XR2.

-Gerard

I would like to add our experience. We have a production link and we were wondering if everything was ok or not installing the new RB333 to our production network, so we investigated on a real link (not lab) to see if everything was working as expected.
TX Power set to 8, and the link is operated by 2.9.4x version of ROS (I do not remember if the hardware is RB532 or some intel based with SR5). Signal (TX/RX) is around -72, so it's good enough.
We installed the new RB333 with XR5, and ROS 3RC9 on one side of the link the other side was always operated by a ROS 2.9.42. After installing it, we noticed that the link does not go up (i.e. the client does connect to the AP) and we decided to explore more. If we higher the power the link goes up and the funny thing is that the TX/RX powers are "correct", i.e. as expected. We do not see a shift of powers or wrong power, we simply see (this is our feeling) that under some power the link is not established. It's like if there is a threshold under which the link is not done., when established we see it works only when we are higher the power to +10dB than the normal values. We changed the boxes with ros 3RC8, the whole box (new RB333 with ROSRC8 and new XR5). Same result. Putting back the old box with ros 2.9.42 everything worked as 1 hour before.

Obviously we don't know if the powers reported by ros are correct or not, but we have tested and we are sure that everything has done with the right procedues. Can somebody do a test with the powers and see what's happening? Please Fix it!!!

Thank you.

galaxynet · Sun Nov 04, 2007 3:36 pm

pdf -
Not to flame you but you're off topic here...what you are describing is that ROC 3.09rc correctly interrputs the manual power out for the SR / XR cards. Meaning that if you put in 8db that is what the card puts out. Not like previous releases that you had to subtract 10 from the power level you actually wanted, i.e. you want -20dbm, so you -10dbm in and 'magically' the card puts out -20dbm.

What we are talking about in this post is that RB333 and ROS 3.06rc - 3.09rc, a second SR / XR card will have it's power out level altered, even if you set it manually the power out from the card is at least -10dbm, sometimes as much as -20dbm, down from what it should be. And it is not because of the antenna selection - on the XR there is only one antenna port...A...and you can't select the other port.

From the 'mail' I have gotten and some more independent research it looks like there is a driver issue. I have to send MT another supout file so they can take another look at this issue.

pdf · Fri Nov 09, 2007 7:31 pm

pdf -
Not to flame you but you're off topic here...what you are describing is that ROC 3.09rc correctly interrputs the manual power out for the SR / XR cards. Meaning that if you put in 8db that is what the card puts out. Not like previous releases that you had to subtract 10 from the power level you actually wanted, i.e. you want -20dbm, so you -10dbm in and 'magically' the card puts out -20dbm.

What we are talking about in this post is that RB333 and ROS 3.06rc - 3.09rc, a second SR / XR card will have it's power out level altered, even if you set it manually the power out from the card is at least -10dbm, sometimes as much as -20dbm, down from what it should be. And it is not because of the antenna selection - on the XR there is only one antenna port...A...and you can't select the other port.

From the 'mail' I have gotten and some more independent research it looks like there is a driver issue. I have to send MT another supout file so they can take another look at this issue.

Probably I was not clear, I meant that it looks like it does not register when it should. I am not complaining about the output power level.

I think the following table can be helpful to understand the problem. In our case we are connecting two RB333 via an attenuator (I guess it was 70dB attenuation between the two devices).

Device - Routeros - Wireless - TX Power - Registration
Wrap	2.9.42	SR5	Default	-59
Wrap	2.9.42	SR5	17      -59
Wrap	2.9.42	SR5	7       -69
Wrap	2.9.42	SR5	0       -71
Wrap	2.9.42	XR5	Default -75
Wrap	2.9.42	XR5	17      -75
Wrap	2.9.42	XR5	7       -85
Wrap	2.9.42	XR5	0       NO
Wrap	3.0rc4	SR5	Default -70
Wrap	3.0rc4	SR5	17      -76
Wrap	3.0rc4	SR5	7       -80
Wrap	3.0rc4	SR5	0       -80
Wrap	3.0rc4	XR5	Default -49
Wrap	3.0rc4	XR5	17      -59
Wrap	3.0rc4	XR5	11      -66
Wrap	3.0rc4	XR5	<11     NO
RB333	3.0rc4	SR5	Default -70
RB333	3.0rc4	SR5	17      -76
RB333	3.0rc4	SR5	7       -83
RB333	3.0rc4	SR5	0       -83
RB333	3.0rc4	XR5	Default -59
RB333	3.0rc4	XR5	17      -69
RB333	3.0rc4	XR5	11      -74
RB333	3.0rc4	XR5	<11     NO

now as you can see when using RB333 with 3.0rc4 XR5 there should be no reason why it should not register at 10dBm (<11). But... if we go under 11 it does not register. The same happens for WRAP. Interesting to note that 3.0rc4 does not work the same for intel and powerpc.

I hope this helps.

Thank you.

gmsmstr · Sat Nov 10, 2007 4:56 am

Just a FYI, this is the same on x86 systems using v3rc9. Rolling back to rc6 appears to fix this. Second radio card in nstream dual link is about 10db less.!

galaxynet · Sat Nov 10, 2007 4:48 pm

gmsmstr -
Thanks for the feedback ON the forum - while I have received about 10 emails expressing the same issue none of these folks were 'willing' to post their results on the forum...

pdf - nice charts - great work - they don't have anything to do with what this post was originally intended to bring to light to the MT user world. I suggest you start another thread describing your issue and what you have found so far.

gmsmstr · Sat Nov 10, 2007 8:50 pm

I have a link that is doing quite well, I will see if I can post some information laster today/tomorrow.

Power out differences on RB333

Power out differences on RB333

Re: Power out differences on RB333

Re: Power out differences on RB333

Re: Power out differences on RB333

Re: Power out differences on RB333

Re: Power out differences on RB333

Re: Power out differences on RB333

Re: Power out differences on RB333

Re: Power out differences on RB333

Re: Power out differences on RB333

Re: Power out differences on RB333

Re: Power out differences on RB333

Re: Power out differences on RB333

Re: Power out differences on RB333

Re: Power out differences on RB333

Re: Power out differences on RB333

Re: Power out differences on RB333

Re: Power out differences on RB333

Who is online