While we've been converting a lot of rb800's to rb912's to gain lots of extra signal, even with higher data rates, we can't get anything much more than 90-100mb even using NV2 and rates such as 270/240. I assume its because of the slow 600mhz CPU. We saw higher speeds with lower signal and air rates using rb800 + SR71 cards.

Does anyone know of a comparable PCI card that has the super-high TX power as the integrated chipset on the rb912? Or will Mikrotik plan on making a card similar? Sucks we have to loose out on high throughput links with rb800.

DBII F50N Pro.

28dB output, though only 20 @ higher data rates.

DBII F50N Pro.

28dB output, though only 20 @ higher data rates.

Yea that's what we've been using. Nothing beats 26db at the highest MCS rates! Sucks we have to sacrifice throughout for higher power. That rb912 would be soo great if it had higher CPU for fast back haul applications. Why don't mikroik make a rb912-AH that has 800mhz CPU for the ultimate ptp solution? I really think they should also have a special mode setting strictly for ptp that optimizes and focuses primarily on high throughput and latency stability.

why not try overclocking to 720mhz.

why not try overclocking to 720mhz.

Is that safe??? What are the risks??

While we've been converting a lot of rb800's to rb912's to gain lots of extra signal, even with higher data rates, we can't get anything much more than 90-100mb even using NV2 and rates such as 270/240. I assume its because of the slow 600mhz CPU.

go to wireless advanced setting for both master and slave

adjust noise floor threshold to -128

adjust hw retries to 14

the signal strenght & ccq rx/tx will increase

note / im using this method just with ptp ....master+ slave

While we've been converting a lot of rb800's to rb912's to gain lots of extra signal, even with higher data rates, we can't get anything much more than 90-100mb even using NV2 and rates such as 270/240. I assume its because of the slow 600mhz CPU. We saw higher speeds with lower signal and air rates using rb800 + SR71 cards.

Does anyone know of a comparable PCI card that has the super-high TX power as the integrated chipset on the rb912? Or will Mikrotik plan on making a card similar? Sucks we have to loose out on high throughput links with rb800.

What is the CPU usage when you are sending traffic through the board?

While we've been converting a lot of rb800's to rb912's to gain lots of extra signal, even with higher data rates, we can't get anything much more than 90-100mb even using NV2 and rates such as 270/240. I assume its because of the slow 600mhz CPU. We saw higher speeds with lower signal and air rates using rb800 + SR71 cards.

Does anyone know of a comparable PCI card that has the super-high TX power as the integrated chipset on the rb912? Or will Mikrotik plan on making a card similar? Sucks we have to loose out on high throughput links with rb800.

What is the CPU usage when you are sending traffic through the board?

45-50%

While we've been converting a lot of rb800's to rb912's to gain lots of extra signal, even with higher data rates, we can't get anything much more than 90-100mb even using NV2 and rates such as 270/240. I assume its because of the slow 600mhz CPU.

go to wireless advanced setting for both master and slave

adjust noise floor threshold to -128

adjust hw retries to 14

the signal strenght & ccq rx/tx will increase

note / im using this method just with ptp ....master+ slave

HOLY COW!!! My signals on this 30 mile link are now -29db / -35db, what did this do??? is this real signal levels or am I going o burn out the transmitter by running it too hot?

Also it didn't effect the throughput. Data rates are 270/270 and I still only see 97mb throughput. Nothing will be the same as rb800

is this real signal levels or am I going o burn out the transmitter by running it too hot?

yes ...this a real signal reading .......dont worry nothing will happen to your transmitter .....

dnot forget to give me a karam

is this real signal levels or am I going o burn out the transmitter by running it too hot?

yes ...this a real signal reading .......dont worry nothing will happen to your transmitter .....

dnot forget to give me a karam

Didn't help throughput, I still only can push ~110 or so. With rb800 I was seeing ~140mb. How good do the signals have to be??? I got -38db on all chains. I find it hard to believe that is the true signal though. What did changing the settings do to effect it that much? Is it transmitting full TX power at the higher MCS rates??

Is it safe to change the CPU to 720mhz??? What do I risk, overheating?

Seems the latency spikes always happen when the data rates change. REALLY wish I could limit the MCS rates to only go to a certain level and not higher. It keeps wanting to pop up to 300, then high latency, then back down to 270 or 240. With UBNT you can set the max MCS rate, then "Auto" checked. Like a automatic transmission, it will be automatic all the way up until you set the threshold.

in RouterOS you can also lower the max data-rate, use the rat-set=configured and uncheck the last mcs data rates.

The signal reading is changing because you manually specified the noise-floor using noise-floor-threshold. We would not recommend to do so as that could effect on the performance as well.

Please contact support@mikrotik.com about the problem getting the max throughput compared to RB800.

in RouterOS you can also lower the max data-rate, use the rat-set=configured and uncheck the last mcs data rates.

The signal reading is changing because you manually specified the noise-floor using noise-floor-threshold. We would not recommend to do so as that could effect on the performance as well.

Please contact support@mikrotik.com about the problem getting the max throughput compared to RB800.

I never figured out how to unlock the MCS rates. You can unlock the regular rates, but its only for b/g. The MCS tab has always been grayed-out, and there are no buttons on that page like the b/g page to manually change it.

The signal reading is changing because you manually specified the noise-floor using noise-floor-threshold. We would not recommend to do so as that could effect on the performance as well.

i do this step (( adjust noise floor threshold to -128 ))....manually with master and salve in ptp connection ...nothing changes in the performance to the bad .on the contrary ,the perfomance increased....

What exactly does adjusting the Hardware Retries do? I noticed another PtP link on older firmware was set to 15, current was 7, changed to 14. Can anyone shed any light on the effect?

There's just SOO many settings you can change that effect throughput and latency stability. I've always just left it as defaults. Bridge on one side, Station WDS on the other, NV2 protocal, and that's it.

If there's any other things to adjust to help improve things, please shed some light!

..........................

The signal reading is changing because you manually specified the noise-floor using noise-floor-threshold. We would not recommend to do so as that could effect on the performance as well.................

I have to somewhat disagree with "Uldis" on this as I have noted a increase of upto 9dBm of signal and upto 16dB SNR + increase in CCQ, if these figures were in any way misleading it would I expect to result in a drop throughput or link to be dropping out but they are solid and one PTP has increased by of over 2Mb, not much maybe but +2 is better than -2,

..........................

The signal reading is changing because you manually specified the noise-floor using noise-floor-threshold. We would not recommend to do so as that could effect on the performance as well.................

I have to somewhat disagree with "Uldis" on this as I have noted a increase of upto 9dBm of signal and upto 16dB SNR + increase in CCQ, if these figures were in any way misleading it would I expect to result in a drop throughput or link to be dropping out but they are solid and one PTP has increased by of over 2Mb, not much maybe but +2 is better than -2,

I personally believe, from all my PtP's I tried this on, would say that its a inaccurate reading of the signal. I had some go from the upper -50's, all the way to -30's. Why do I believe the true signal hasn't changed? Because the data rates didn't change AT ALL. One would think, that having up to a true 20db increase in signal would have SOME effect on the data rates, yet the rates and throughput are exactly the same.

...............................

I personally believe, from all my PtP's I tried this on, would say that its a inaccurate reading of the signal. I had some go from the upper -50's, all the way to -30's. Why do I believe the true signal hasn't changed? Because the data rates didn't change AT ALL. One would think, that having up to a true 20db increase in signal would have SOME effect on the data rates, yet the rates and throughput are exactly the same.

I will try soon on a AP which has one CPE in the -85++ range see if this setting improves signal and reduces dropouts which are currently occurring, as stated if this setting had an impact on performance, throughput should be reduced.
Also

setting is working for that only legacy wireless card based on 5211 chipset and for newer N wireless chipset cards.

which i discovered when i tried this on PTP using XR5 cards, no WLAN registration on the link

You can try to increase throughput by using of bridge mode and disable encryption, however I have no experiences with the RB912...

Leo.