Ohhh nice and what about SR2 that stopped working after a month using the default value? You mentioned in other posts to use "default" for better perfomance. How do I know that all SR2 that stopped working are not affected by this? G-mode, default tx power.

When using default power it is putting too much power, that's why it was getting better signal, but after a while it could overheat and start to work unstable.
To make this card work more stable you can adjust the power to use such table:

I have found out and posted on this forum before that best TX power values are between 21 and 23 everything else is making latency.

Cheers...

How about the SR5??? the same problem?

Also are there any tx-power settings for CM9? Or default it's the best?

I have found CM9 working the best at 30 !!!
Try it and you'll see signal strength improvement. I have 8 AP's with cm9 and no problem for more than 1 month, latency is not afected, everything works fine and stable.
I wish somebody else try this and post results in here.

Cheers...

We have discovered that the SR2 card is putting too much TX-power and that's why it could get hot and also work unstable.
You should adjust the TX-power value for G mode to 21 so it will work at 26db(400mW). For B mode you should set the tx-power to 26 which will work as 26db(400mW).

So what you are saying is not that the SR2 card is putting out to much power, but that MT is tweaking the power levels. The default for the card is 26dB for B - why would MT try to push more power out of the card?

no, MT is not pushing it. the card gets overheated because it tries to work at too much power (more than it can do) and we suggest that you force it to work at correct maximum (400mW) by using the settings as in the original post.

Will future releases correct this issue? It would be nice to have the settings in MT (26db) actually mean 26db out on the card. This would eliminate all sorts of issues as we can confidently set the power level to the factory spec. Right now if the power is set manually or to default even, we may be running to "hot" or just as bad be under powered!

If tx power works like that, why from 0 to 17 I the signal doesn't change? Above 18-19-20-21.... there is some difference. I have equipment to measure the signal. Setting it 1 it's not 1 db but a lof more. Same problem with CM9

tip for the driver: maybe you should change tx power to set mW value instead of db?

I don't know about too much power. I could see the card overheating and failing or signal loss if being overdriven. I feel the SR2 doesn't put out enough. Using 26dbm on anntenna b and b/g with a 5dbi omni antenna my Linksys with Seavsoft has better range @ 200mw.

Also a 100mw 3com 8750 has about the same range.
Honestly, I'm disapointed.

What gives? I paid $30 for the Linksys and $400 for the RB / RouterOS Wisp 500.

I don't know about too much power. I could see the card overheating and failing or signal loss if being overdriven. I feel the SR2 doesn't put out enough. Using 26dbm on anntenna b and b/g with a 5dbi omni antenna my Linksys with Seavsoft has better range @ 200mw.

Also a 100mw 3com 8750 has about the same range.
Honestly, I'm disapointed.

What gives? I paid $30 for the Linksys and $400 for the RB / RouterOS Wisp 500.

And the linksys and Seavsoft devices were also using 5dBi Omni's? When you're comparing devices it's best to compare apples to apples, not apples to oranges.

I've seen a definative increase in range after replacing my SOHO grade linksys with a RB530 w/ SR2 just in my home. For use in the WISP world these devices have made a noticeable difference over the previous CM9's and Prism cards we used before..

ESPECIALLY when all client cards are Atheros based.

In addition your 30$ linksys doesn't come close to offering you the features of RouterOS, sure you could mod out the firmware and run some flavor of linux on it, but when it comes down to it your still running your network on the same cheap piece of hardware that has an incredibly high failure rate.

I don't know about you, but I'd take the quality of the RB500 over the Linksys WAP anyday.

anyway, my 2 cents...

And the linksys and Seavsoft devices were also using 5dBi Omni's? When you're comparing devices it's best to compare apples to apples, not apples to oranges.[/quote]

Yes the linksys had 5dbi rubber duckies. The RB532 showed a major decrease in range in the same enviroment. I know the Linksys is a POS. And I doubt it could handle many users. But my point was "What's wrong with this setup?" Like I said even the 3COM 8750AP @ 100mw was comparable in range in the same enviroment. Of course this is indoors. But we are only getting about 50-75ft range. Lot's of walls and doors. The RB532 is for a commercial application. The RB already needs my attention do to some wierd bandwidth problem. It fails after about 6 hours and requires reboot, but the watchdog doesn't reboot it. It's not lockedup, just becomes extremely slow passing packets to ethernet1(WAN).

Well, I'm not an expert on these things. My wireless experience is from being a HAM and I've been programing since the TRS80. Maybe I'm overlooking something.

So I guess I'm going to spend my saturday in the Hotel bar trying to fix it.

If tx power works like that, why from 0 to 17 I the signal doesn't change? Above 18-19-20-21.... there is some difference. I have equipment to measure the signal. Setting it 1 it's not 1 db but a lof more. Same problem with CM9

tip for the driver: maybe you should change tx power to set mW value instead of db?

The radio cards have a hardcoded minimum TX power level in many cases. In the case of the SR cards the minimul level is 13dB so if you set anything less then 13 then the radio will still output 13dB because it do not have not been tuned and adjusted for anything lower then 13dB. So this is not the radio card you want to use for low power output operation. The same goes with the CM9 it can not be abjusted lower then 11dB I want to recall.

/ Eje

Under Linux you can limit the output power to 1 db, this is not correct.

Under Linux you can limit the output power to 1 db, this is not correct.

On a CM9 possibly because that card I'm not sure. But this is NOT the case on the SuperRange cards because the manufacture don't have that supported in the hardware. They have a 10dB offset so lowest you possible could go is 11dB but the cards have not been calibrated lower then 13dB.

Is MT working on the problems with the SR cards, or just proposing the fix they have. I have to turn the power down to 21 to keep the pings stable. From some of the posts I have read, I am not the only one.

I was told by someone that the output power on the SR2 was not adjustable, it simply put out 400mW regardless of what was set in RouterOS. I gather this is not true?

It is not true!

Excellent, thank you! I am setting up an installation that needs to have the card's output wound back slightly so that is very useful to know

I know that this is not a place for official MT answers, but since this is the place that they gave us the "fix" for this problem, I would like to see if they are going to answer the question I had earlier in this thread. Is Mikrotik working to resolve this issue? MT could work directly with Ubiquiti, I know for a fact they are very interested in getting this solved. Why isn't MT being more forthcoming with a fix? I know I may be stepping on some toes here, but I have built a network based on MT and SR cards, and this problem is affecting my customers.

To answer maxxgrafix observation. Linksys devices (Not sure about Sveasoft), like practically all indoor devices, use antenna diversity. This improves indoor performance significantly.

Audio engineers have first-hand experience in this, the next time you head to a concert or similar event, spend some time looking at the radio mic receivers, these normally have LEDs telling you which antenna is being used.... they are constantly switching as the performer moves about. This shows how important diversity is in mobile applications.

Low-end routers/APs (Linksys, D-Link) and enterprise APs (Cisco, Proxim, 3com) are aimed at mobile applications (laptops, PDAs) so they all use diversity, this explains their use of 2 antennas. MT is not aimed at mobile applications. It's main use is for fixed wireless.

Having said that, if MT implements antenna diversity, I'd be able to put MTs in many more places.

The RB already needs my attention do to some wierd bandwidth problem. It fails after about 6 hours and requires reboot, but the watchdog doesn't reboot it. It's not lockedup, just becomes extremely slow passing packets to ethernet1(WAN).

Just tip: Try to disable connection tracking ...

winbox- manual settings

So I should set all the g settings (6mb, 12 mb and up) to what setting?
and the b tx settings (1, 2, 5.5 and 11mbps ) to another setting? [/img]

Leave everything default. Just play with TX-power cause installations, locations etc. are different.
I can understand that MT can't fix this cause SR2, CM9 etc. have same chipset AR5213. So leaving default means that card is using it's stated TX power, but it does not mean that it's always the best.
With SR2 just go down until you se no latency!

Cheers...

What should be mention the Tx Power as if we use 2.4Ghz-10Mhz mode?

So leaving default means that card is using it's stated TX power, but it does not mean that it's always the best.

What in the hell are you saying? I designed my network based on published specifications from a manufacturer. I did not expect MT to come along and make my network illegal. How is it not the best to use the published specs? I can think of no reason that MT should try and tweak the cards without the end user knowing.

That being said is MT any closer to resolving this issue?

I think what djape means is that running the card at its maximum power will not suit every installation; as with any type of gear you should tune your equipment settings so that it runs at its peak performance, not just leave everything as default.

Regarding legal issues, MT cannot make your network illegal as such - it is simply up to you as the installer to make sure that your equipment specs fall within your local legal boundaries. This may require reducing the output of your cards, if you use reputable brands of card then what you set is what you will get.

If you look back at the study I posted on this a few months ago, you will see that the power adjustment is NOT a 1 to 1. Since the cards don't put out 30 dBm this can't be. Stepping down the power in what seem to be 1 dB increments do not step the power down in 1 dB increments, but rather more than 1dB. Eje's post about 13 dBm minimum is about right at least in my testing. It is true that the card will put out 26 dBm, but when running on the 30 setting, the signal gets extrememly "noisy" on the spectrum analyser. This might explain the latency issues when running like this and the MT frimware may be trying to force the card to operate beyond its limits, thus the overheating. While I can't give details, I can assure you that Ubiquity is working on a solution to make these more compatible with MT. They are VERY concerened about this.

I know that Ubiquiti is very concerned about this - as they should be. I am happy to know that a vendor is trying to make things right. That is awesome. To many vendors in today's market would just say it's somebody elses problem and go on. As to the the legality comment:

Regarding legal issues, MT cannot make your network illegal as such - it is simply up to you as the installer to make sure that your equipment specs fall within your local legal boundaries. This may require reducing the output of your cards, if you use reputable brands of card then what you set is what you will get.

You are nuts. I and many others design their networks based on the mfg published specs. When MT pushes a card up to 30 dB when the spec is 26, that will definitely make you illegal. The same goes with the CM9 cards, they are defaulted to very low powers, and people design with that spec in mind. MT pushing as much power as they can get out of the chipset is tinkering with the spec. I didn't buy MT to have them tinker with the spec, and I bet most people did not know that it happened. I bought MT because of it's functionality, not to have their engineers think they could improve on somebody's hardware drivers.

My bad, thank you for the explanation.

After making this change, I lost about 4-5 db across all my CPE. Is this to be expected with this change to make the card run more stable? I previously was set to "All rates fixed: 30"

What about SR5 ? what should be the right config ?

If I go to set manually it puts 17 db on all rates. Should I put 30 ? it doesnt seem so logical...

I think you should leave SR5 default, but then again you can play a bit

Cheers...

After making this change, I lost about 4-5 db across all my CPE. Is this to be expected with this change to make the card run more stable? I previously was set to "All rates fixed: 30"

I assume no response means this should be expected?

Does this problem (and fix) apply across all version of RouterOS?

When using default power it is putting too much power, that's why it was getting better signal, but after a while it could overheat and start to work unstable.
To make this card work more stable you can adjust the power to use such table:

Code: Select all

/interface wireless set wlan1 tx-power-mode=manual-table
/interface wireless manual-tx-power-table set wlan1 manual-tx-powers=1Mbps:26,2Mbps:26,5.5Mbps:26,11Mbps:26,6Mbps:21,9Mbps:21,12Mbps:21,18Mbps:21,24Mbps:21,36Mbps:17,48Mbps:15,54Mbps:15

With this setting i have 50kbytes/s on my 6km link <rotfl>

we have fixed the power reading in the wireless-test package - please upgrade to that and set the tx-power-mode to default.
This applies to v2.9.11 and above.
Then check again after that.

I have tried many settings.

The best performance on my 6km link (mt 2.9.6) is with "tx-power" set to "default" and "txrate" set to "24,36,48mbit" on both sides.

I wonder why I never get better than 560kbytes/s upload, and 1250kbytes/s download (ftp test)...

There are no other ap's, and no noise. Signal is -69db (no really difference in up/down speed if there is -69 or -78)

(2 bridged together Ovislink's WL-5460 with txrate set to 18mbit has no problem with 1200kbytes/s upload and download)

I didn't experience latency with SR2 running on default tx-power, but ping gains. For some registered clients, ping gain in some cases was higher than 1000%, meaning I send one ping and receive 100. After I decreased the power, everything is fine. And this is the same with two SR2 cards I bought recently and the older one I have works perfectly on full power. This seems to depend of SR2 series, older ones seem to be better. And yes, I am using all 3 ones in exactly the same boxes, same antennas (15dB omni) and stuff.

I didn't experience latency with SR2 running on default tx-power, but ping gains. For some registered clients, ping gain in some cases was higher than 1000%, meaning I send one ping and receive 100. After I decreased the power, everything is fine. And this is the same with two SR2 cards I bought recently and the older one I have works perfectly on full power. This seems to depend of SR2 series, older ones seem to be better. And yes, I am using all 3 ones in exactly the same boxes, same antennas (15dB omni) and stuff.

With 2.9.12?

Regards

No, 2.9.6. I don't think upgrade will help unless power is decreased by default. Do you have some other experience?

No, 2.9.6. I don't think upgrade will help unless power is decreased by default. Do you have some other experience?

The power issue have been fixed in 2.9.12

Regards

Do you mean default power rates are different in 2.9.12 or SR2 is working good in 2.9.12 using the same power rates as default in pre-2.9.12?

http://www.mikrotik.com/download/CHANGELOG_full

In 2.9.11 there is "fixed tx-power selection for SR2 cards in wireless-test", meaning this is the same for 2.9.12, but this means that just values are changed, you can't really make SR2 working at producer declared power without side effects. Or am I wrong? Obviously, there is newer Atheros driver in 2.9.11 and 2.9.12, but I am not sure what does it really improve.

If I just put SR2 to regulatory domain, everything becomes fine even if I don't enter antenna gain. And really I lose maybe 1dB in output.

we have fixed the power reading in the wireless-test package - please upgrade to that and set the tx-power-mode to default.
This applies to v2.9.11 and above.
Then check again after that.

Do you mean default power rates are different in 2.9.12 or SR2 is working good in 2.9.12 using the same power rates as default in pre-2.9.12?

http://www.mikrotik.com/download/CHANGELOG_full

In 2.9.11 there is "fixed tx-power selection for SR2 cards in wireless-test", meaning this is the same for 2.9.12, but this means that just values are changed, you can't really make SR2 working at producer declared power without side effects. Or am I wrong? Obviously, there is newer Atheros driver in 2.9.11 and 2.9.12, but I am not sure what does it really improve.

If I just put SR2 to regulatory domain, everything becomes fine even if I don't enter antenna gain. And really I lose maybe 1dB in output.

I have several AP with 2.9.12 working at TX Power: Default.. and it´s working fine to me... you should try 2.9.12... the driver has been improved.

Regards

The changelog says nothing about any improvements to the atheros driver or SR2 fixes. Where are you getting this information? 2.9.11 says the issue was addressed in the wireless-test package have those fixes been applied the stable wireless package in 2.9.12?

The changelog says nothing about any improvements to the atheros driver or SR2 fixes. Where are you getting this information? 2.9.11 says the issue was addressed in the wireless-test package have those fixes been applied the stable wireless package in 2.9.12?

Check this topic http://forum.mikrotik.com//viewtopic.php?t=6667
and this http://www.mikrotik.com/download.html#v2

Regards

It seems cibernet has right, 2.9.11 change log says:

"fixed tx-power selection for SR2 cards in wireless-test"

http://www.mikrotik.com/download/CHANGELOG_full

But I am wondering is this just a fix of default values for tx-power or really driver improvement so SR2 can work with full power as default. I believe the first is true because this problem is maybe not just software issue, maybe hardware can't really do it. However, SR2 works just fine at lower power and there is really no need to push it to the limits.

Can someone tell us what are the default values for tx-power in 2.9.11 or 2.9.12 for SR2 card?

/interface wireless set wlan1 tx-power-mode=manual-table
/interface wireless manual-tx-power-table set wlan1 manual-tx-powers=1Mbps:26,2Mbps:26,5.5Mbps:26,11Mbps:26,6Mbps:21,9Mbps:21,12Mbps:21,18Mbps:21,24Mbps:21,36Mbps:17,48Mbps:15,54Mbps:15

This is recommendation from some other thread in this forum.

I have found CM9 working the best at 30 !!!
Try it and you'll see signal strength improvement. I have 8 AP's with cm9 and no problem for more than 1 month, latency is not afected, everything works fine and stable.
I wish somebody else try this and post results in here.

Cheers...

Can you really drive CM9's to 30db (1000mw) like this? Isn't the 802.11b default around 18db? Does it have major overheating? Anyway, sifting through some older posts, I found this interesting.

Eric

Well, I don't think card is going to 1000mW but for sure stations are getting stronger signal. What we expirience is 3-10 db gain on all our AP's.

Cheers...

Very interesting, djape. I'll try it out.

Eric

Well, I don't think card is going to 1000mW but for sure stations are getting stronger signal. What we expirience is 3-10 db gain on all our AP's.

Cheers...

You have 30db and Frequency Mode: manual txpower , Country: no_country_set and Antenna Gain: 0

Regards!
Alessio