[HamWAN PSDR] The 3rd time is not charmed [was: 8dB drop/rise!]
Bart Kus
me at bartk.us
Mon Jul 21 13:14:20 PDT 2014
Are you sure that's graphing link speed? How did it dip below 6Mbit?
(MHz y-axis label?) The modems cannot operate slower than that in
only-n mode.
--Bart
On 07/21/2014 01:08 PM, Dean Gibson AE7Q wrote:
> The displayed link speed was also dropping. See
> http://www.ae7q.com/misc/media/5.9GHz/2014-07-xx_Data.png
>
> On 2014-07-21 11:24, Bart Kus wrote:
>> Have you looked at the possibility that your modem is negotiating a
>> higher link speed over time and that's the cause for the gradual
>> signal level loss? When you reboot these things and they aren't
>> pressed to move data, they opt for the slowest (and highest power)
>> modulation. As data comes in for them to move, if they build up
>> queues, they try higher order modulations, which make use of
>> amplitude modulation, and reduce the average power level.
>>
>> It is possible to instruct the modem to only support a subset of
>> speeds. For example, the slowest speed only (6Mbit):
>>
>> [eo at AE7SJ/Monroe-Paine] /interface wireless> set 0 supported-rates-a/g=
>> 12Mbps 18Mbps 24Mbps 36Mbps 48Mbps 54Mbps 6Mbps 9Mbps
>>
>> I don't think upgrading or downgrading firmware has anything to do
>> with RSSI readings as they're read directly from the Atheros chip and
>> not computed in firmware.
>>
>> The real test of the health of your connection would be if it
>> consistently supports the same ballpark speed range (bits/sec
>> actually moved) or if it is indeed dropping due to this 8dB RSSI
>> delta you're seeing. If the modem finds itself having too many
>> communication errors because it chose a modulation that is too fast,
>> it'll lower its negotiated speed.
>>
>> --Bart
>>
>>
>> On 7/21/2014 10:28 AM, Dean Gibson AE7Q wrote:
>>> OK, over the last three weeks both the TX and RX signal levels have
>>> dropped 8dB; see:
>>>
>>> http://www.ae7q.com/misc/media/5.9GHz/2014-07-xx_RX.png
>>>
>>> While this could be due to vegetation, the drop (somewhat gradual
>>> this time) looks suspiciously like the previous two drops of 8dB
>>> (see prior messages below). On Friday I noticed that MicroTik had a
>>> new firmware release, v6.17, and I upgraded to that. That fixed an
>>> NTP bug introduced in v6.14 but made no difference in dBm levels, so
>>> Sunday evening I downgraded to v6.13. That downgrade/reboot was
>>> successful, but also made no difference. So, I did a second reboot
>>> in preparation for upgrading back to v6.17 (using the same scenario
>>> as below).
>>>
>>> This time, however, "presto!" (the results below) does not quite
>>> apply. The 5SHPn shows a power-on LED, and when connected to a
>>> switch, shows an electrical network LED. However, *the 5SHPn is
>>> inaccessible via Ethernet* (using Winbox to access via both by last
>>> IP address, and by MAC address). PINGs fail; there is no
>>> corresponding entry in the ARP table. I can hold in the reset
>>> button while power-cycling the 5SHPn, and I get the flashing power
>>> LED indicating a "reset configuration" is happening, but to no
>>> avail: Winbox is unable to connect to the 5SHPn.
>>>
>>> *This is the third time I've had this 8dB drop*, and the second time
>>> I've had to go to the roof to reset the radio. So, I've removed the
>>> antenna from the roof. So, unless someone has a bright idea, the
>>> 5SHPn and the 31dBi dish antenna are available for $120 total. I'd
>>> like to sell them as a unit, because I've got a really nice (but
>>> easily removable) taped seal of the antenna/radio connection. If
>>> you can't get the radio to work, you're only out $20.
>>>
>>> -- Dean
>>> 425-338-4276 (home)
>>> 425-359-4276 (cell)
>>>
>>> On 2014-06-13 18:29, Dean Gibson AE7Q wrote:
>>>> Going back to v6.13 didn't improve anything, but it did seem to get
>>>> rid of the occasional variations in voltage (from 13.5v down to
>>>> 12.8v) that SNMP was reporting to "The Dude" software. So, after
>>>> letting v6.13 run for couple hours with no improvement, I
>>>> reinstalled v6.15.
>>>>
>>>> Presto! Instant dBm improvement back to my normally observed
>>>> values. Here are links to SNMP graphs of the drop @9:30am, and the
>>>> rise at 17:48pm. Of course, you can also see this in Nigel's Cacti
>>>> reporting, albeit with less horizontal (time) resolution.
>>>>
>>>> http://www.ae7q.com/misc/media/5.9GHz/2014-06-13_0940.png
>>>> http://www.ae7q.com/misc/media/5.9GHz/2014-06-13_1740.png
>>>>
>>>> Note that the reported sudden dBm "rise" to -60dBm (the upper limit
>>>> I've imposed to keep the graph scale reasonable) on two occasions
>>>> in the second graph, is when I upgraded and then reconfigured the
>>>> radio (from a cut-&-paste script), and should be ignored.
>>>>
>>>> Note that a very similar scenario occurred several months ago, when
>>>> I upgraded from v6.10 to v6.12.
>>>>
>>>> Of course, today is Friday the 13th. Maybe Jason is responsible.
>>>>
>>>> -- Dean
>>>>
>>>> On 2014-06-13 12:28, Dean Gibson AE7Q wrote:
>>>>>
>>>>> On 2014-06-13 12:13, Cory (NQ1E) wrote:
>>>>>> Did it rain on those trees you're pointing at around then? :)
>>>>>>
>>>>>> On Fri, Jun 13, 2014 at 12:12 PM, Dean Gibson AE7Q
>>>>>> <hamwan at ae7q.com <mailto:hamwan at ae7q.com>> wrote:
>>>>>>
>>>>>> This morning (well after the update to v6.15 last night) just
>>>>>> after 9:30am, I saw a 8dB drop in both TX and RX signal
>>>>>> levels, which persist at this time. I did a "reset
>>>>>> configuration", which did not solve the problem.
>>>>>>
>>>>>> Did anyone do a change at Paine sector 2 around 9:30am?
>>>>>>
>>>>>
>>>>> It's been raining gently all morning. Last night it rained a bit
>>>>> harder, and I saw no significant or unusual variation. I'm going
>>>>> back to v6.13 to see if that changes anything.
>>>>>
>>>>> The drop was very sudden. I have that antenna connection very
>>>>> well sealed, I thought, and the Ethernet cable runs uphill to get
>>>>> into the eaves. I did a cursory visual check of that this morning.
>>>>
>>>
>>> On 2014-05-04 13:00, Dean Gibson AE7Q wrote:
>>>> OK, everything is back working, and I have my dBm back!!
>>>>
>>>> Not a comedy of errors, but two factors intertwined to create this
>>>> problem:
>>>>
>>>> 1. Since configuring the radio after a configuration reset
>>>> requires the use of WinBox (until you get IP addresses set up),
>>>> I moved the network connection from the radio to its normal
>>>> router on the DMZ, and connected it to a router on my LAN via
>>>> an extension Ethernet cable and a "dual-RJ-45-jack adapter".
>>>> The port light on the router associated with the radio
>>>> connection was off. This morning when I went to disconnect the
>>>> cables, I saw the port light on the router flash momentarily.
>>>> Junked the dual-RJ-45 adapter and used another one (I had
>>>> previously tried a different cable), and I had WinBox access to
>>>> the radio! Everything worked except for the LEDs ...
>>>> 2. *After much more screwing around*, I found out that for version
>>>> 6.12,*the LEDs are not configured by default* !!! Remember, I
>>>> had done a "reset-configuration" ...
>>>>
>>>>
>>>> So, for others that update to v6.12, here are the relevant lines in
>>>> the setup:
>>>>
>>>> */system leds add type=interface-status interface=ether1-local
>>>> leds=user-led
>>>> /system leds add type=wireless-signal-strength
>>>> interface=wlan1-gateway leds=led1,led2,led3,led4,led5*
>>>>
>>>> Now that I know I can configure the LEDs with other options, I'll
>>>> try some.
>>>>
>>>> I have no clue as to why my dBm dropped, and why I now have my
>>>> prior values back ...
>>>>
>>>> -- Dean
>>>>
>>>>
>>>> -----Original Message-----
>>>> From: PSDR [mailto:psdr-bounces at hamwan.org
>>>> <mailto:psdr-bounces at hamwan.org>] On Behalf Of Dean Gibson AE7Q
>>>> Sent: Sunday, May 4, 2014 1:44 AM
>>>> To: Puget Sound Data Ring
>>>> Subject: Re: [HamWAN PSDR] Metal 5SHPn firmware 6.12 is current
>>>> (addendum)
>>>>
>>>> Well, this did not end well. ... I decided to reset the
>>>> configuration (using the command line) ... it appears to have
>>>> "bricked" the radio.
>>>>
>>>> It draws about 160ma (about the value from a week ago, when I
>>>> first measured it). There is no light on the side of the unit,
>>>> and the Ethernet port is dead.
>>>>
>>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> PSDR mailing list
>>> PSDR at hamwan.org
>>> http://mail.hamwan.org/mailman/listinfo/psdr_hamwan.org
>>
>>
>>
>> _______________________________________________
>> PSDR mailing list
>> PSDR at hamwan.org
>> http://mail.hamwan.org/mailman/listinfo/psdr_hamwan.org
>
>
>
> _______________________________________________
> PSDR mailing list
> PSDR at hamwan.org
> http://mail.hamwan.org/mailman/listinfo/psdr_hamwan.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.hamwan.net/pipermail/psdr/attachments/20140721/cd3525d4/attachment.html>
More information about the PSDR
mailing list