Router instability riddle is getting the better of me [SOLVED]

[ajax]

For the last few weeks I've been battling to find the cause of the instabilities I'm experiencing with my E1820 and MR3420.

When the dongle is plugged into the router, it disconnects often, sometimes it can't even get connected in over 15 minutes of waiting. During times of instability I would quickly plug the dongle into my pc directly and it would stay connected for hours. After those hours, I would plug it back into the router and it would be unstable again. This test I have repeated many many times.

In the above, no USB extensions were used. The router and dongle were under a fan for most of the tests. The signal on the router varied, depending on the exact location, from 45% to 77% (around -85dBm to -69 dBm).

Now here is what is mind boggling to me:
I took the router and dongle to work, approximately 3 km away from my home as the crow flies. I had the dongle plugged in on the router. In over 50 hours - no disconnects!! The router seems to connect faster too. Signal around 50 to 60%.

After these tests, I just plugged the power out and virtually walked with the gear back to my place and plugged it back in. Yes, you guessed it, unstable/disconnecting/lengthy detection times, etc. Again, plug the dongle into the pc, it connects first time and stays connected for hours.

So what the hell is happening at my place???

I'd appreciate any ideas, even if you think it may be stupid. Cause I've run out of ideas

 

[Axis]

For the last few weeks I've been battling to find the cause of the instabilities I'm experiencing with my E1820 and MR3420.

When the dongle is plugged into the router, it disconnects often, sometimes it can't even get connected in over 15 minutes of waiting. During times of instability I would quickly plug the dongle into my pc directly and it would stay connected for hours. After those hours, I would plug it back into the router and it would be unstable again. This test I have repeated many many times.

In the above, no USB extensions were used. The router and dongle were under a fan for most of the tests. The signal on the router varied, depending on the exact location, from 45% to 77% (around -85dBm to -69 dBm).

Now here is what is mind boggling to me:
I took the router and dongle to work, approximately 3 km away from my home as the crow flies. I had the dongle plugged in on the router. In over 50 hours - no disconnects!! The router seems to connect faster too. Signal around 50 to 60%.

After these tests, I just plugged the power out and virtually walked with the gear back to my place and plugged it back in. Yes, you guessed it, unstable/disconnecting/lengthy detection times, etc. Again, plug the dongle into the pc, it connects first time and stays connected for hours.

So what the hell is happening at my place???

I'd appreciate any ideas, even if you think it may be stupid. Cause I've run out of ideas

The problem is not with the router. It is with Cell C. Your problem is coverage.

 

[ajax]

Just some further info on all my testing:
Using an MTN or Vodacom simcard in the router is very very stable, even with a signal of around 40%. But VC/MTN runs on 2100 MHz UMTS unfortunately.

I have a friend who lives exactly halfway between me and my workplace. He has the same issues. His signal is around 70%, dongle plugged in directly on the router.

I thought it could be that when I plug in on the pc, it connects to a different and stable tower. So last night I did a test by the window where I used a short extension cable and fixed the dongle near the window. Same instability/stabililty issues with router/pc. Of course, I can't rule out the possibilty that the router can't use the extension cable but the laptop can.

Running a constant ping as I've indicated here keeps the connection on the router much more stable. On average maybe 1 or 2 disconnects per day. The same goes for the pc when I tested this a few weeks ago when I experienced frequent disconnects on both router and pc.

The router instability is more pronounced during evenings from say 18h00 to 23h00. I thought maybe I'm picking up interference from other users so I improved my signal by 12 dB with the corner reflector. In the corner reflector my signal is around -70 to -75 dBm. The tests at my work location included evenings by the way.

I had the router on a UPS thinking it was Eskom. That didn't help.

The router is set to 3G only mode, and nothing is connected to the WAN port. But I usually connect to it through ethernet and a switch. I did test without ethernet last night using its wireless only. No joy.

The tower at work that I connect to is a different one according to MDMA.

Possible further steps I can think of:

• 
  Borrow someone's Netgear and see if that is unstable too.

• 
  Drive to the tower that I usually connect to with a fully charged UPS and test the router there.

• 
  Load Openwrt firmware on the router. I see that USB is now supported. My first try was unsuccessful because after loading the GUI (Luci) there was not enough flash memory left for adding USB support. So I have to go without a GUI, and me and Linux aren't buddies

• 
  I have soldered a 10 micro Farad cap onto the power rails of the USB port. Inconclusive whether that helped.

 

[ajax]

The problem is not with the router. It is with Cell C. Your problem is coverage.

I don't agree - the router is unstable with a stronger signal than the pc.

 

[Viva]

I've had real trouble with disconnects every few minutes and sometimes I was even unable to connect at all. I don't use a router - only a short USB extension cable to the corner reflector. And suddenly around Thursday all the trouble stopped and my connection became really stable. Nothing changed on my side which indicates that the change is on the tower's side.

Do you connect to the same tower when using the router and not using the router at your house?

 

[Madhawk]

I'm using my E1820 with the TP-LINK MR3420. I have the dongle cable tied to my window where the signal is strongest.
The E1820 is connected to the TP-LINK USB extension cradle. The cradle is connected to the MR3420 router.
Once in awhile I do get disconnects, but not too often.

I also noticed that when I connect a 3m Belkin USB extension cable, the E1820 can't connect with the router.
So some USB cables are not compatible with some routers.

I also noticed that it takes sometimes over 5 minutes for the E1820 to connect to the MR3420.
Whereas my iBurst modem connects in less than 2 minutes.

I still think the problem lies with Cell C.

 

[Tobi-5520]

Have a similar problem - connection is far more stable through a pc than through my router (Dovado UMR). I have noted also that the modem (E1820) gets a lot hotter when plugged into my PC. Could there be a voltage (or something?) difference possibly causing poor performance on the router?

 

[ajax]

Do you connect to the same tower when using the router and not using the router at your house?

As I mentioned, one particular test was done with the modem held completely in a static position when switching between pc and router. But I can't be 100% sure of the towers. Would be great if the router could give tower info...

Could there be a voltage (or something?) difference possibly causing poor performance on the router?

It's difficult to rule out, but IMHO if VC/MTN simcards are stable with the same or weaker signal strength then it is unlikely to be a USB voltage issue.

Maybe you should also test in a different location? I left my router connected at work for a few days. I left a screencapturing recorder (Instant Demo) running to record the status page to capture the disconnects (which didn't occur...). Luckily I still have a WDS connection to my friend's 384k ADSL line at home as backup. It's not nice being without internet!

 

[The_Unbeliever]

I also noticed that when I connect a 3m Belkin USB extension cable, the E1820 can't connect with the router.
So some USB cables are not compatible with some routers.

See my thread which I've started.

Basically it's a combination of the correct USB port, the correct make and length of USB cable and the router's orientation which will make a big difference.

 

[ajax]

I also noticed that it takes sometimes over 5 minutes for the E1820 to connect to the MR3420.
Whereas my iBurst modem connects in less than 2 minutes.

Yes, same here.
I am wondering if it could be because the Cell C version of the E1820 was modified to be able to do UMTS 900, whereas the stock standard E1820 can't do UMTS 900, only UMTS 2100.

 

[ajax]

The problem is not with the router. It is with Cell C. Your problem is coverage.

It seems Axis was on the right track all along.

On Tuesday afternoon I decided to power up my router again after it’s been disconnected for a few days. I didn’t change anything, I just put it back in the same place I’ve always used it.
Now get this, I’ve had only 1 disconnect since Tuesday afternoon! Every morning so far I wake-up the pc from sleep and continue browsing. What’s even more weird, it seems I’ve lost a few dB’s of signal in the last week.

My friend with the same hardware (E1820/MR3420) in die Boord, Stellenbosch also reports a very stable connection. I can only conclude that somehow Cell C was responsible. It is now pointless to do testing with a different router. It remains a mystery why the pc was more stable than the router. I don’t think we can mark this one as “solved”, rather “deferred”. I’m happy, for now…

(I will update/edit my posts in the “routers to use” thread during the day to refer to this thread)

 

[Shred]

I am having the same issue. TP-Link TL-MR3420 with E1820 black Cell C Modem and the performance is terrible. When I connect straight to the PC it's perfect. I have updated firmware on both router and modem. Anyone got some more advice?

 

[Madhawk]

I am having the same issue. TP-Link TL-MR3420 with E1820 black Cell C Modem and the performance is terrible. When I connect straight to the PC it's perfect. I have updated firmware on both router and modem. Anyone got some more advice?

Have you got the router set to "3G only"? Also set "Max idle time to 0 (Zero)" in Connection Mode.

Try these settings and see if it makes a difference.

 

[Shred]

Have you got the router set to "3G only"? Also set "Max idle time to 0 (Zero)" in Connection Mode.

Try these settings and see if it makes a difference.

Thanks I will try this.

 

[Shred]

Have you got the router set to "3G only"? Also set "Max idle time to 0 (Zero)" in Connection Mode.

Try these settings and see if it makes a difference.

Did not work at all. The router also battles to identify the modem sometimes. I also get ping time-outs when connected to the router but none when connected to my laptop.

 

[ajax]

Did not work at all. The router also battles to identify the modem sometimes. I also get ping time-outs when connected to the router but none when connected to my laptop.

Are you connecting to the router via wifi? Just check that it's not an issue with wireless n. My laptop doesn't do wireless n and it was very slow until I changed the router to b&g only.

 

[Shred]

Are you connecting to the router via wifi? Just check that it's not an issue with wireless n. My laptop doesn't do wireless n and it was very slow until I changed the router to b&g only.

Connecting using Ethernet cable. I wonder if it is because I upgraded the firmware of the Cell C modem itself; maybe the MR3420 does not like the new firmware.

 

[ajax]

I wonder if it is because I upgraded the firmware of the Cell C modem itself; maybe the MR3420 does not like the new firmware.

Nope. I upgraded my E1820 too.

 

[Madhawk]

I suspect this is maybe a network issue.

A suggestion would be to take your modem and router to a friend's house with good Cell C coverage and test there.

I know it's a hassle, but that would rule out one problem.

 

[Shred]

I suspect this is maybe a network issue.

A suggestion would be to take your modem and router to a friend's house with good Cell C coverage and test there.

I know it's a hassle, but that would rule out one problem.

Good idea; I will try that.