PMP450i 900 SA Accuracy

Hello Everyone,

Just wanted to get everyone opinion on the spectrum analyzer accuracy on the PMP450i 900mhz AP. I just recently replace an or Trango 900 AP with the PMP450i and I am currently having issues with high latency on my wireless subs. When I first put it up everything looked good except on the linktest the uplink percentage was around 75% for every sub. I started at 10mhz at 907 and finally had to drop it down to 905 with a 5mhz channel just to get any type of stability. Could the AP perhaps pick up noise that wouldn't appear on the SA?

At first glance, that noise floor looks pretty good for 900! Is the latency/jitter consistant, or is it intermittent to all or just some of the clients? If you look at your frame utilization when latency is high, is your uplink frame utilization being saturated? ALSO, if your client(s) are hitting their max upload MIR, this will cause lots of latency and jitter, and dropped packets.

What happens when you run a timed SA on both the AP and the SM's?

Sometimes you'll have to run an SA for awhile to see intermittent noise.

Has all the Trango equipment at the site been turned down or removed?

There is another Tragno AP on the tower but I've turned it off for now to verify its not causing the interference. I'll look to run a SA on the client side later today. Frame utilization is low for both download and upload on the AP. The latency/jitter does flucuate quite a bit but it doesn't appear to be related to usage. I only have 8 clients connected to this AP and they are very low users, which is why they were still on Trango gear,  but I'll go back and check QOS settings on their radios. The AP could be hitting its max upload, but with the noise level at the AP I figured I could get more than 5mbps on the upload. Thanks for the insight.

Current Ping Test from AP to Client with best signal on tower.

Client connection recv -56dBm 8x/6x SNR 33V/33H db


Sent to x.x.x.x: bytes=64 seq=0
Reply from x.x.x.x: bytes=64 seq=0 time=480ms
Sent to x.x.x.x: bytes=64 seq=1
Reply from x.x.x.x: bytes=64 seq=1 time=110ms
Sent to x.x.x.x: bytes=64 seq=2
Reply from x.x.x.x: bytes=64 seq=2 time=110ms
Ping statistics for x.x.x.x:
Packets: Sent = 3, Received = 3, Lost = 0 (0% loss)
Approximate round trip times in milli-seconds: Minimum = 110ms, Maximum = 480ms, Average = 233ms

Are your SM's IP's on the same network as your APs IP's? I'd be curious to see what the direct 1 hop AP to SM ping time is without any other BH's or routers involved. There's a ping tool you can use directly on the radios in the 'Tools' section of the radio GUI.

The signal level, modulation and SNR you've posted for that client is excellent. I would not expect that much latency or jitter.

With the AP noise floor looking the way it does, and the SnR you posted for that client, I'd assume that using a 10MHz channel would be possible. I know this is going to sound strange, but can you try a 7 or 10mhz channel width?

I was on a 10mhz first then dropped to 7mhz and finally 5mhz just to get the ping latency down and under control. The ping test I posted was from the AP directly to the client. Client is in simple bridge mode, AP and SU both on same IP subnet.

I'll be playing with this AP again tonight during our maitnance window to see what combo will work for us.

Yeah, I'm not sure what to tell you... I can show you a couple AP's that have way way worse noise then you and our ping times are averaging around 21ms with pretty low jitter. This is our worst connection on an AP with 6 other clients. It's a 7.5mi shot, non-LOS, connected at -69, 15dB SnR, @ 4X-mimo-B. They use VoIP on this connection frequently during business hours with no issue.

Oh an we're running a 7MHz channel width, and 2.5ms frame size.. ALSO... what firmware are you running? I'm using 15.1.

OH yeah... dunno if you tried this, but if you're using 5ms frames, switch to 2.5ms frames to reduce your latency... of course this has the opposite effect of lowering overall throughput by about 10%.

PS. don't mind the red... those are scheduled downtime blips.

I have 15.1 on the AP and 15.1 on most CPE except for 3. Working on getting them updated as well, but so far not having much luck finding a channel that will work. I do notice clients connection quicker when I drop down to 2.5ms. From my testing it looks like there is interference around 907mhz even though the SA doesn't show it. Its too bad I was really hoping to get results like you posted with this gear. I'll continue to tweak the wireless and see what else I can do to improve performance.

One thing that might be happening is very narrow band interference that the spectrum analyzer in the 450i cannot resolve (i.e. 12.5 KHz energy), or even frequency hopping equipment.

To uncover this, you might try a longer spectrum scan (like 30m or an hour), although it might still not show anything.

I have no evidence to support this, but it's a guess on my part.  I agree with Eric that RSL, SNR, and all other statistics indicate that you should have a better performing link that you're getting.

I just did a little research on Trango's 900mhz product and it's a Direct Sequence Spread Spectrum (DSSS) radio. So that might be why you never saw or at least saw very little latency issues. If like what Matt says, that someone is using a very fast hopping radio over some or all of the band... a DSSS radio would have a better chance at randomly missing the burst of noise, which would cause a re-transmit aka latency or a dropped packet.

Do you see this latency regardless of where you put the channel? I assume you've carefully moved the channel and measured the latency across the entire band?


@CambiumMatt wrote:

One thing that might be happening is very narrow band interference that the spectrum analyzer in the 450i cannot resolve (i.e. 12.5 KHz energy), or even frequency hopping equipment.

To uncover this, you might try a longer spectrum scan (like 30m or an hour), although it might still not show anything.

I have no evidence to support this, but it's a guess on my part.  I agree with Eric that RSL, SNR, and all other statistics indicate that you should have a better performing link that you're getting.


I'll have to run that scan during our night maintnance to see if it can pick up something. When I first ran the SA and seeing the noise floor I was hoping to be able to get a decent upgrade in speed to customers in this area. If this is the case, is there basically nothing we can do to get around it or push through it better?

What kind of antennas are you using on both the AP and SM side?

We switched from Cambium's OEM 900Mhz sector antenna to a KP performance flat panel, and that helped boost our signals and cut down on noise quite a bit. KP also makes a 17dBi yagi.

That being said... like I said before, I think your signals/SnR/modulation looks great...


@Eric Ozrelic wrote:

I just did a little research on Trango's 900mhz product and it's a Direct Sequence Spread Spectrum (DSSS) radio. So that might be why you never saw or at least saw very little latency issues. If like what Matt says, that someone is using a very fast hopping radio over some or all of the band... a DSSS radio would have a better chance at randomly missing the burst of noise, which would cause a re-transmit aka latency or a dropped packet.

Do you see this latency regardless of where you put the channel? I assume you've carefully moved the channel and measured the latency across the entire band?


Last night I went through quite a bit of the band to see which frequency would work the best. I ended up having the most stable results on 920mhz. Anything below gives me high latency. I left it at a 10mhz channel with 2.5ms frame. Unfortunatley my upload connection is still low and will probably prevent me from adding many more customers to the AP.


@Eric Ozrelic wrote:

What kind of antennas are you using on both the AP and SM side?

We switched from Cambium's OEM 900Mhz sector antenna to a KP performance flat panel, and that helped boost our signals and cut down on noise quite a bit. KP also makes a 17dBi yagi.

That being said... like I said before, I think your signals/SnR/modulation looks great...


I'm using the KP 90deg x-pol sector at the tower and clients have a mix between the OEM yagi and KP's 17dBi yagi as well. Yeah the links look great they just don't perform as expected. Hopefully I can figure out a combination that works and allows me to really utilize the potential of this equipment.

We have seen issues with one or two clients who have poor upload eat up the available resources on the AP and cause high latency.  You can see this in cn Mastro if you are using it. You can also look on the AP under Tools -> Link Status for a good list of client stats.  But the time utilization chart on the AP on mastro shows it best.

Given that you mentioned the upload I would suspect that is a good starting place.  We had to revisit about 10% of our installs from the winter to correct these issues on the client side.  KP panels work great if you have enough room between tree lines and the client, otherwise the 17 yagi is the only option that helps.

I would be interested if you find this to be an contributing factor and if anyone else has had this issue.  It is a killer for us as we basically don't get to have a lot of upload to offer.  On one sector we did change the configuration to 60X40 DL/UL I think it was and that helped as well.

PaulTN

Fyi - We had similar issues  and remember the  450i 900 is dual slant thus the Spectrum Analysis will differ from othe r 900 products that are not offset. The biggest thing that helped is  to GPS sync all the APs you and to MATCH the Frame Period whether all 2.5 or all 5.0ms accross APs. If using with Cambium FSK 900 then set both the 900FSK and 450i 900 to a frame period of 5.0ms.

1 Like