Intermittent ATOM screen

[trident]

Hi all,

I don't know if others are having similar problems, but my ATOM screen for OGN-R LongSut has become very hit-and-miss and seems to work when it feels like it.

I can get the Metar, Login and other screens fine and the Home screen refreshes itself, but when clicking on the "Radar" button, it sometimes works but often doesn't and just leaves a blank screen. Logging in and rebooting the Pi sometimes clears the problem and I can then access the ATOM screen, but shortly after it often freezes. My network is fine and I can always connect to the Pi so is this a server or firmware issue? All T's and P's for the Pi seem to be fine apart from an undervoltage warning, but I've always had that and it's never had these problems before.

Over to you learned gentlemen.

Thanks,

Frank.

[steveu]

Hi all,

I don't know if others are having similar problems, but my ATOM screen for OGN-R LongSut has become very hit-and-miss and seems to work when it feels like it.

Where is this screen? On your phone, a Pi running in kiosk mode, or some other device? I assume this is not on the ATOM Pi itself?

I can get the Metar, Login and other screens fine and the Home screen refreshes itself, but when clicking on the "Radar" button, it sometimes works but often doesn't and just leaves a blank screen. Logging in and rebooting the Pi sometimes clears the problem and I can then access the ATOM screen, but shortly after it often freezes. My network is fine and I can always connect to the Pi so is this a server or firmware issue? All T's and P's for the Pi seem to be fine apart from an undervoltage warning, but I've always had that and it's never had these problems before.

Over to you learned gentlemen.

When you talk of rebooting the Pi, I assume you mean the ATOM Pi?

How do we know that the undervoltage hasn't simply got worse and tipped over the edge?

What sort of PSU is it? PAW support may correct me, but it needs to be capable of 2A?

If the Pi is running at a lower clock rate because of the PSU, then it could just struggle with the data processing? Poor hypothesis?

If it were mine I would resolve the PSU problem - thtat's never going to help, it might not be the fault, but may as well run it properly?

[exfirepro]

Hi Frank,

It’s been a while since we’ve heard from you on here, so I have just reread all your previous posts to refresh my memory.

Unless things have changed (which I’m not aware of), when you login ‘direct’ to your ATOM (at home for example) via the same router it is connected to, and then select the ‘Radar’ display option, your device loads VRS from the Pi each time before running it. This ensures that any settings selected are specific to that display device and don’t affect settings on any other device connected via the same router.

I presume a similar thing happens when logging in remotely via the station maintainer’s network/database login. This will therefore be susceptible to a degree of delay while the VRS software downloads and opens up, the actual delay depending on the speed of the connections you are using and the amount of data which needs to be transferred.

With regard to the undervoltage issues you have mentioned, I would certainly advise investigating and trying to resolve these as they could well be causing the Pi to throttle or reset at a critical point in the process. I remember lots of discussions when you were first setting up the station about using a WiFi connection (via a Pi3B) to avoid a very long Ethernet feed cable to and up your 12 metre mast, but can’t find any details of whether you actually went down that route, or how you decided to power the unit, except that you did at one point (after finally managing to configure the unit) say…

Right, what next? Get the hardware sorted and mounted, network and power cables up the mast and commission it properly.

…which makes me pretty certain you didn’t opt for the ‘now common’ route of using a +48Volt POE injector to squirt 48 volts up the ethernet cable with a matching splitter to reduce the voltage back to 5 volts at the Pi end. This raises the question as to whether the undervoltages are due to voltage drop over the length of the cable - or did you come up with an alternative solution to overcome that?

Best Regards

Peter

[trident]

Hello Steve and Peter,

Firstly, Thanks both for your replies and comments. I'm sorry for the delay in replying, but family matters have got in the way.


In reply to you first Steve, the screens are on my phone, laptop and one of several PCs all connected on the same local network as the ATOM Pi. I'm sorry but I don't know what you mean by "Kiosk" mode, but I access the ATOM Pi through a Chrome browser and direct connection via local address 192.168.1.175 etc.

Yes, I reboot the ATOM Pi through the login screen and selecting option 10. The Pi reboots and after a few minutes I see the "Reconnect" window and on clicking that I'm taken to the home screen with the message" Attempting to connect to the ATOM Grid". When that clears, usually I can select the "Radar" tab and the ATOM grid screen loads fine. However, occasionally I just get an empty browser screen and no amount of refreshing will load the ATOM data. Navigating back to the Home screen, everything else works fine except the Radar tab which just leaves me with the empty browser.

In respect of the PSU and undervoltage indication, I've always seen that but only fairly recently got the reported problems. I will check the actual voltage at the Pi but can you advise at what voltage does the Pi consider too low and trigger the undervoltage flag?

The Pi is powered by a dedicated 12V 2A d.c. adapter in my radio shack and fed up the mast by decent (1.5mm^2 if I remember correctly) cable to the ATOM Pi mounted inside a waterproof diecast enclosure 12 metres directly above. There it is fed to a 5V 2A linear regulator (LM317 if I remember correctly) and on to the Pi.

It is possible there is not enough headroom from the current 12V supply and I did change the feed from a 12V 8A ex-Tait PMR repeater PSU I had previously been using. I will do some investigation and probably swap back to the 8A PSU.



Hi Peter, yes, it's been a while since I was on here last. I'm not up-to-speed on the intricate workings of the software, but thanks for trying to enlighten me.

Yes, the same thing happens whether connecting locally or remotely via the web and I have noted the delay in response due to the internet connection speed.

It seems the undervoltage may be the cause of the issue, so as above, I'll investigate and try to fix that first. No, not using POE - just 12V fed up from the shack to a local 5V 2A regulator by stout lecky string!

The data feed to the ATOM Pi is a direct Cat 5 cable connection from an ethernet switch/WiFi bridge router in the shack. The switch/router has a WiFi link to another switch/router in the house 50 metres away which then connects to the main internet router through a pair of power-line adapters. It's a bit convoluted, but short of digging a 50 metre trench from the house to the shack it was the easiest method.

I'll update you both on my findings once I've had a chance to investigate, but a more pressing problem to solve is the intermittent tripping of the RCD feeding the barn where the ATOM Pi is powered from and where our solar panels feed into the grid 


Thanks and regards,

Frank.

[steveu]

In reply to you first Steve, the screens are on my phone, laptop and one of several PCs all connected on the same local network as the ATOM Pi. I'm sorry but I don't know what you mean by "Kiosk" mode, but I access the ATOM Pi through a Chrome browser and direct connection via local address 192.168.1.175 etc.

Understood. By kiosk mode, I'm referring to a second RPi which runs Chrome from auto boot, no mouse or keyboard, looking only at the VRS via the relevant IP address.

Yes, I reboot the ATOM Pi through the login screen and selecting option 10. The Pi reboots and after a few minutes I see the "Reconnect" window and on clicking that I'm taken to the home screen with the message" Attempting to connect to the ATOM Grid". When that clears, usually I can select the "Radar" tab and the ATOM grid screen loads fine. However, occasionally I just get an empty browser screen and no amount of refreshing will load the ATOM data. Navigating back to the Home screen, everything else works fine except the Radar tab which just leaves me with the empty browser.

I've seen something similar when there is a problem with bandwidth in the LAN. Low hanging fruit, easy stuff, a ten minute attempt at a fix, can you reboot all the IP chain? I'd say turn off the switch with Wifi at the ATOM/shack end, then the Wifi receiver for the same in the house, then last of all the two Powerline adaptors. Leave off for 2 minutes, then fire up starting from the Powerline adaptors, then the Wifi bridge at that end, then lastly the switch/bridge at the shack end. I suspect the Powerline adaptors have gone low bandwidth, I've seen this before.

Then you can post that I'm a grasper of straws...

I've done a fair amount of work with PoE and think especially for cheap IP cameras it can save a lot of time, and cope with long runs... Maybe this might be a long term solution.

I also think that the LM317 is up against it... says 1.5A in the data sheet and doesn't mention anything above that. One solution might be to set the output slightly higher.

RPi 3B specs say 5.1V & 2.5A. I'm going to say the LM317 is out of its depth...

[trident]

Hi Steve,

Thanks for your reply and comments.

Yes, I have seen Kiosk mode at Sywell if I remember correctly - Really useful in a clubhouse or similar setup.

Interesting about the LAN bandwidth. I do have the occasional problem of the PLA's losing link and needing a power recycle to regain connection, so it may be those throttling the bandwidth. I'm not keen on them as they produce RFI around the HF 80 metre band and as a I'm a Radio Amateur it rather goes against the grain. I plan to get rid of them at some point and just use directional aerials for the WiFi link between the house and the barn. I'll try a LAN reboot if and when the problem appears again and let you know if it's resolved. I've just had a thought that RFI getting into the PLA's would affect their performance and there is another amateur fairly local who uses a couple of KW on 80 metres!

To be honest, I can't remember which regulator I put in the box with the ATOM Pi up on the mast - It may be a 78S05, but I'll  have a look when I get a chance. I think I'll put a current clamp on the 12V feed up to the ATOM Pi and see what peak current it's drawing on the 'scope. That should show whether it's the regulator limiting. The 12V 2A adapter I'm currently using is a stop-gap whilst I sort out other things that need 12V up at the masthead such as RF preamps and coaxial relays, so I will revert to the 8A PSU. Maybe I'll upgrade to a 3A 5V regulator for the Pi tweaked to give 5.1 V anyway.
 
Just looking at the system monitor ports 8080 and 8081 I thought the Pi supply voltage was listed, but I can't see it. Is it readable remotely? I imagine it is monitored to trigger the undervoltage flag, so can the actual voltage be read easily?

I've not used PoE yet, but picked up a security DVR and some cameras at a radio rally recently and it's another job to put those up ASAP. They've got the option of PoE and the DVR can feed them directly without an injector, so I may use that.

Best regards,

Frank.

[exfirepro]

Hi Frank (and Steve),

All noted, but I’m out in India at the moment, so a bit limited in terms of what I can do.

I will monitor the thread and come back on once I get back.

Regards

Peter

[steveu]

I'll try a LAN reboot if and when the problem appears again and let you know if it's resolved. I've just had a thought that RFI getting into the PLA's would affect their performance and there is another amateur fairly local who uses a couple of KW on 80 metres!

An alternative method would be to plug a laptop into the switch with laptop WiFi turned off then SpeedTest on the LAN port?

[Keithvinning]

This sounds very much like a power supply problem.
It is showing as voltage low before it went down 

Regards

Keith

[Keithvinning]

In respect of the PSU and undervoltage indication, I've always seen that but only fairly recently got the reported problems. I will check the actual voltage at the Pi but can you advise at what voltage does the Pi consider too low and trigger the undervoltage flag?

The Pi is powered by a dedicated 12V 2A d.c. adapter in my radio shack and fed up the mast by decent (1.5mm^2 if I remember correctly) cable to the ATOM Pi mounted inside a waterproof diecast enclosure 12 metres directly above. There it is fed to a 5V 2A linear regulator (LM317 if I remember correctly) and on to the Pi.

It is possible there is not enough headroom from the current 12V supply and I did change the feed from a 12V 8A ex-Tait PMR repeater PSU I had previously been using. I will do some investigation and probably swap back to the 8A PSU.


Thanks and regards,

Frank.

Have you tried Power over ethernet? I use these regularly and they are very effective for long distance. Relatively inexpensive from Amazon Anvision are good.
Dont use cheap Chinese ones from ali express I had a bad batch.
 

Regards

Keith

[trident]

I'll try a LAN reboot if and when the problem appears again and let you know if it's resolved. I've just had a thought that RFI getting into the PLA's would affect their performance and there is another amateur fairly local who uses a couple of KW on 80 metres!

An alternative method would be to plug a laptop into the switch with laptop WiFi turned off then SpeedTest on the LAN port?

Hi Steve,

I don't think it's bandwidth issues as I run a separate PC in the shack sharing the bridge/switch/router that the ATOM Pi uses and most of the time it's fine running iplayer, youtube or spotify apart from fairly frequent internet freezes/dropouts 

[trident]

Hello Keith,

I've not used or tried PoE before, but as I previously mentioned I've already got a good 12V supply available at the masthead.

Dropping the PoE 48V-ish to 5V @ 2A would dissipate a lot of power as heat and I want to use a linear regulator because switch-mode regulators are RF noisy. The Pi (already noisy) is mounted in a finned diecast box on the mast and contains a small fan to help keep it cool, so I don't want to create more problems.

I've ordered a 78T05 (5V @ 3A) linear regulator to replace whatever is currently in the box feeding the Pi. Unfortunately the Motorola MC78T05 I ordered is not what I received from the seller on fleabay, so I'm in dispute with them as I think they're passing off goodness-knows-what brands as Motorola parts. I know they are essentially the same device, but I paid for a (hopefully) genuine part!

I'm also about to remove another potential weak-link in the chain by getting rid of the Power line adapters and putting a high-gain directional WiFi aerial for the remote bridge/switch and relocating the other end of the link in the house.

I'll update here once I make the changes.

Thanks and regards,

Frank.

[trident]

Hi Frank (and Steve),

All noted, but I’m out in India at the moment, so a bit limited in terms of what I can do.

I will monitor the thread and come back on once I get back.

Regards

Peter

Hello Peter,

Hope you're having a good presumably holiday in India?

I think I'm getting to the bottom of the problems, but we'll see once I've made some changes.

Best regards,

Frank.

[exfirepro]

Hi Frank,

I’ve been ‘incommunicado’ for the last 7 days up in the foothills of the Himalayas. Have had a great (though cold) time! Just arrived back in the land of 4G and WiFi last night, so trying to catch up before heading back to the UK overnight Monday.

All understood on your proposed modifications. POE is definitely worth considering if things don’t work out. As Keith says we now have most ATOM stations powered that way and running reliably and it is relatively inexpensive.

Need to go now for our last day trip out here (5:30am local time) but happy to discuss with you later on here or offline. Keep in touch.

Best Regards

Peter
(PilotAware’s temporary Indian Support Team)

[trident]

Hi Peter,

I wonder if you're home yet? Hope you had a great holiday, or was it a jolly on expenses as a remote Pilotaware support engineer?

Well I've now got what I think is a genuine Motorola MC78T05 so I just need a bit of good-ish weather to rehash the regulator in the box up the mast. I've also sorted out a high-gain panel aerial for the wifi link to the shack so I can get rid of the power-line adapters. Hopefully these changes will solve the problems.

I'll keep you all updated.

Best regards,

Frank.