You are here

AirGrid AG-HP-2G20??

8 posts / 0 new
Last post
N1TEW
N1TEW's picture
AirGrid AG-HP-2G20??


Am I reading the Supported Platform Matrix correctly in that there is no 2.4 AirGrid XW listed?  

What does that mean for the AG-HP-2G20?  I'm guessing it's not supported, but I'm not sure.  Please clarify.  

Thanks much!

K5DLQ
K5DLQ's picture
We aren't aware of any 2.4Ghz
We aren't aware of any 2.4Ghz "XW" models.  Do you have confirmation that your AirGrid is an XW model?  It will show it in AirOS as XM or XW.
N1TEW
N1TEW's picture
No, perhaps I've made the
No, perhaps I've made the wrong assumption.  I'm looking to buy the AG-HP-2G20 if it is supported.  However, I can't tell from the matrix.  All I see listed for 2g is the M2 while 5g has both.  Does that make sense?  
AE6XE
AE6XE's picture
Yes, this is an inconsistency
Yes, this is an inconsistency in the table.  We can look to correct that.    The airGrid has 4 models 2G16, 2G20, 5G23, 5G27.    Not sure on the 5G models if both of them are XM and XW (or just one of them)?    I think the feedhorn shows the same board ID, just the size of the grid antenna is different.  Fairly confident all these models work with AREDN, but don't recall 100%. 

Note, we shy away from the single antenna devices (not MIMO in the marketing terminology).   The dual antenna-polarity devices out perform the single antenna devices -- handle microwave fading issues in challenging environments much better and generally double the link rates give different data can be sent on both polarities simultaneously.  

I prefer the NanoBridges, e.g. 5G25 over the AirGrids, a MIMO device.  I just picked up a 6-pack of new equipment for $64.28 per unit cost (minus the power brick) for several meshers in the area.

Joe AE6XE
N1TEW
N1TEW's picture
Thanks, Joe.  

Thanks, Joe.  

We have several M2 Nanobridges that we picked up for around $35 each like new in unopened boxes.  That source has dried up, however.  I've used them as far as about 20 miles.  

I understand the benefit of MIMO, but didn't know how to quantify it.  For example, the 2g AirGrid has 2dB more gain and 5 dBm more transmit power.  I was hoping it might offer some advantages over the NanoBridges, but sounds like you don't think so.  Do you know of anyone that has made side-by-side comparisons?

I love the RocketDishes, but that's not in the budget for many.  

Thanks for your help!

-neal 

AE6XE
AE6XE's picture
The answer will be different

The answer will be different depending on the link distance.  We have a couple 5Ghz airgrid 27dBi in the area, my usage is based on.  

Short range -- out to ~15 miles.   Locally, we have ~4 NB 5G25 installed and showing ~39 TxMbps in the 10 to 15 mile range (TxMbps is slightly lower than raw link rate -- takes out packet loss to calculate.)   This is using a 10Mhz channel width.  In comparison, the max rate the AirGrid can achieve with 1 antenna or data stream is 36.1Mbps 802.11n MCS7  SGI link rate in 10Mhz.   We know the Nanobridges are performing better based on exceeding the max link rate of the airgrids.  The NanoBridges max link rate is 72.2Mbps MCS15. 

Mid range-- ~15 to ~30 miles.   In theory, as the distance increases then at some point the SNR at the receiver goes down enough and the NB will drop the transmit to MCS7- and a single data stream out both polarities (this is then combined  with a MIMO device on the other end before decoding the signal).   To compare with the single antenna device with higher xmit power with a signal on one polarity,  it would depend on the environment and conditions, but both performing comparably.  We're comparing which is better:

A) stronger signal on 1 polarity
B)  2 polarity signal that handles fading issues in the environment better

Long range -- ~30 miles +.   At some point the NB won't be usable with insufficient SNR at the receiver, but the higher power of the Airgrid may still link up.

Joe AE6XE

 

N1TEW
N1TEW's picture
Joe, 

Joe, 

Thanks much for the thoroughness of your response.  

We got into the M2's cheap and our RF environment is such that we haven't had any problem with 2.4g interference (that I know off).  Do you know any good sources for the NB M2?  I'm not finding much of a selection anymore.  That's why I was interested in the AG.  

Not to hijack my own thread, but since you bring up 10 Mhz..... I've experimented with the different bandwidths on the weaker hops, but I'm just kind of winging it.  At what point have you found it advantageous to drop back to 10 Mhz vs 20?

AE6XE
AE6XE's picture
Sorry, the occasional ebay

Sorry, the occasional ebay used NBM2 is all I've seen.   You're hitting on some key issues I see happening in a lot of places.    Getting long winded here and hope it is helpful for a lot of groups reading this.

When setting up a new cell or tower site station, I used to compare throughput (iperf measured)  on 5, 10, and 20Mhz to find the sweet spot.   10Mhz was always wining, except on short links <2 mile when 20Mhz rates (max 144.4) can exceed the max 72.2Mbps link rates of 10Mhz.   Now if we're talking a RocketDish this may be out to 5 miles or so where 20Mhz is the sweet spot and you could get 100Mbps+ links.    Now days, I sort of default to 10Mhz on a typical link and if I'm below 36Mbps (max for 5Mhz channel) and there's a lot of noise-interference (generally only on 2Ghz) I drop and test with 5Mhz (to get farther away from the noise and find the sweet spot).  

What is happening is there are always 64 carrier waves being modulated to carry bits (some are guard and don't carry data bits on the outside edges).  When we go from 20Mhz down to 10Mhz channel, there are still 64 carrier waves, only compressed (and getting taller) in half the bandwidth.   But the timing to send a  symbol is doubled (has to do with keeping the carrier waves 'orthogonal' so that adjacent carrier waves do not interfere with one another).   This means we are cutting the max bit rate in half because it takes twice as long to transmit each symbol in half the bandwidth still using 64 carrier waves.  

Another factor to consider with 2Ghz, there's really only 1 channel clear of the wifi allocation and noise.   The future options to optimize throughput have no where to grow.   With 5Ghz, there are many channels of clear sailing to change and isolate sections of the mesh network and get away from other contending links.  We often talk of the benefit of a 'mesh' and the many routing paths with redundancy.  We like that and still want to do that, but not everything on the same RF channel.  RF network implementations of a mesh is 'throughput' challenged. The RF links are impacted when a link or a routing path is directly or indirectly influenced by nodes on other links.

If there are 4 links that traffic needs to hop and traverse through to get from A to B, and each of the links contends with signals with the other 3, then only one link can transmit data at a time.   It takes 4 times as long (even less given handshaking) compared to each of these same links  on different channels.  When every link can be sending data at the same time, or on different channels, we get 4x the thoughput to get data across.  It scales.  

Here's where it becomes problematic for groups that have everything on 2Ghz with 20+ mesh nodes hearing each other.    Someone puts meshchat on an out-there-low-quality node-link.  The link is continually doing retries and lost packets with handshaking back and forth.  Now every other link in the area is hearing this signal and can't transmit at the same time.  RTS-CTS requests start happening as every node is trying to get the channel to transmit -- even more traffic handshaking.     If there are several low quality links, everyone hears everyone else, then the whole network throughput can come to a crawl.   If you find video or voip doesn't work beyond 1 or 2 hops, this is likely the situation :) .

Joe AE6XE

Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer