SRS good performance?


#21

I believe I don’t understood what happened here.

Can you hint what right configuration it need to have on A5 and C5’s ?

I want to say that considering clean channel of course, with low PER and highest possible SNR.


#22

I’m also not clear on what the exact changes were, but I’m glad they worked. Tx power and AGC adjustments were both needed. You already pointed out the AGC tuning procedure (thank you), and Enrique found the Tx power override button on the C5 (great). A combination of those adjustments appears to have made the difference.


#23

Well, after follow the suggestion of let AGC AUTO on C5, the problem start again and I decided try with jperf to see the max performance over-the-air on the link between A5c-C5:

Download:

Jperf - Download -

Server:

Client:

Upload:

Jperf - Upload -

Server:

Client:

So I think the link have enought broadband, i have seen yesterday night 72 Mbps. in both ways, and about 60 Mbps. of TCP traffic both ways too, both with Jperf tool.

But like always the speed test over C5 is not enought good, even if i make some downloads ISO Linux files, drivers, etc

Same files downloaded from the PC on the same switch where A5c is connected are fine.

As always all comments are welcome :slight_smile:


#24

AGC is on Auto by default in SRS (GPS-Sync TDMA) mode. You cannot make manual changes or turn it off. When you click the Power Override button on the C5, it allows manual adjustment of Tx Power only. This does not affect AGC in any way.

Turning off power override results in a power drop from the weakest client. Looking at the packet error rate (PER) over time in the graphs, we can see that there is PER up to ~6% from the client. This could be causing fluctuations in modulation.

On a 20 MHz channel, MCS8 is the highest modulation available. With SRS, airtime is split 50/50 between the AP and C5s. If you start out with 173 Mbps PHY rate, and account for the traffic split and protocol overhead, 50-60 Mbps is expected at the client.


#25

Hi

I believe I am having a similar issue. I have two B5 Lite over a 2.1Km link. AT one end is a VDSL link that Syncs at 67Mb/19Mb. If I test the connection at the Modem it is find with speeds test generally always showing close to the sync speeds. However if I test over the link it only shows around 40Mbs. This has been going on for a long time and I have tested a number of things but come back to the fact the link is doing something to the traffic. I currently have the link configured on Auto 20Mhz and get 172PHY however I have tryed different power and frequency along with channels and find if anything it makes it worse. As J.Enrique has stated I can test over the linke and have pushed a 3Gb file between two PC on either side of the link and have seen 500MBs so can’t see why I cant reach the 67Mbs of the VDSL circuit. I have traffic split on Auto and have changed a fair bit and it doesn’t seem to make a difference? This has been ongoing for a year and I found this article which is the same. I can’t check the link with Fibre as not available where I am.

Really appreciate some help as this is frustrating. on average I am getting less than half the VDSL Sync speed over the Link…

Stephen


#26

Usualy turning ON flow control helps in this situation. Whenever you have transition between port speeds 100>1000 or vice versa you MUST enable flow control. Suppose your modem is 10/100 - link between modem and the router or switch should have flow control on, because B5-lites are gigabit.


#27

Hi Stephen3, thanks for share your experience.

The funny of that is, if a put something to download throw any torrent client, I can see about 40-45 Mbps on A5-5C, so that is what you can expect from an 45-50 Mbps. VDSL line. If you try to do an speedtest from any webbrowser, OS, different computers ( from the side behind C5 ) is always the same 20-25 or 30 Mbps. maximum download speed from that speedtest.

So for me now is not a big problem, because if I make a direct dowload, for exemple game app from Origin store, I see the right speed on the download tool from Origin. But, what happend if I wan to start an WISP provider? I can’t offer for exemple 50 Mbps. because the customer will see 30 Mbps. on his computer and that is not very good for our bussiness.

Another issue that I find is when I choose on the A5c more than 20 Mhz. the speedtest is even lower, for exemple:

20 Mhz. maximum speedtest 30 Mbps. MCS8-173 Mbps. PHY
40 Mhz. maximum speedtest 10-15 Mbps. MCS7-300 Mbps. PHY


#28

Hi Nikolay,

believe me I tried a lot of things, even talk with my provider for replace the modem-router and the result was exactly the same. I have one computer connected directly in to the router having the right speed
( before A5c ) the A5c is connected directly on same router that the PC. The PC and modem-router has also gigabyte ports.

Also I tried to make traffic shaping plans, the same result.

I feel the “problem” is how A5c manage that traffic.

Regards.


#29

So I did some further testing and have found if I turn Traffic Slit to Manual and move it to 75/25 and the key is changing the TDMA window if I move this to 2ms then I see the full seed of the VDSL?? on 4ms it gives poor results. My link is 2km so not sure what this means and when reading the B5-Lite Documentation it suggest on Auto it should optimse the TDMA window but it does’t seem to and is causing the bottleneck on the connection. In regards to the flow control I have a gigabyte router and switch so not sure if this is relevant. I dont know enough about what TDMA does someone else might be able to comment?


#30

On my A5c i cant do that, the traffic is 50/50 and timing also is locked. I have one B5 lite kit, but is on the box now, I dont need to install in a short time.


#31

I have posted a question over in the B5 group to see if anyone can explain this? One thing I have is the VDSL Modem is connected to the Remote (Clinet) unit and the other end my home is connected the Local (AP). I feel like this might be the wrong way around and could be causing a problem with the Auto feature. When set to manual I can stipulate the 25/75 split however I wonder if when set to Auto the configuration is trying to push the other way 75/25. Although in saying that I don’t know how that would affect the TDMA Windows


#32

Hi again, Mimosa´s staff, any advice for that problem?

Thanks.


#33

Thats the same problem I was having, it would get slower. The actual speed through the link would increase but the speed results in SpeedTest would be worse. I am pretty sure this pertains to the way DSL works however I would of thought someone here could explain…


#34

@J.Enrique and @Stephen3 make sure that you are running the latest GA firmware for the A5/C5, which version 2.3.3

Assuming clean spectrum, wider channels will provide higher PHY rates and thus higher through-put.


#35

Yes, right now 2.4, same issue with the speed test. But the most strange is if I move from 20 Mhz. to 40 Mhz. with better PHY Mbps rates the speed test is nearly half than before, that is still happening from the first time that I install the AP A5c


#36

J. Enrique, we made some enhancements in 2.4.0 which improves the through-put on SRS mode as the client count increases. If you are still seeing issues with Speedtest while running 2.4.0, please contact us at http://support.mimsoa.co if you haven’t already done so.