Summer 2018 Maintenance Activities

[TerryJC]

It had been reported to me that the clock in the Tower was gradually getting further and further ahead of the real time; so much so that two weeks ago, the bells were ringing ahead of the real Minster across the road.

Weather and other commitments prevented me from looking into this until today. When I hooked up the monitor and keyboard 'date' told me that actually it was quite close to the real time again. This confirmed my suspicion that the problem was the same one as had occurred with the River System; an upgrade to the Raspbian software had overwritten the config files (see https://afterthoughtsoftware.com/products/rasclock). The Pi was using the processor clock ticks to keep time and these are prone to temperature change.

Interestingly, only one of the modifications in /lib/udev/hwclock-set had been lost, the commenting out of --stystz.

Anyway, I corrected that, set the clock using:

Code: Select all

sudo date -s "4 MAY 2018 10:22:30"  		# Set the system clock
sudo hwclock -w 				# Write the system clock time to the hwclock.
sudo hwclock -r 				# Confirm that the hwclock and the system clock are synced.

I then hung around for 5-10 mins to check that it all worked at 10:32.

[TerryJC]

After the above activity, I received an email from Greg, saying amongst other things:

I reckon you have fixed the crazy organist that plays 24-7. He was quiet first thing this morning and by 6pm the speaker was just buzzing but no music.

Since that wasn't the reason that I went in, I was intrigued at this, but thought bothing more of it.

Today, I went in to do some River System testing and Greg said that the audio in the Minster was stuttering. When I investigated, I found that the MP3 audio was restarting, just like it had been last year before I fitted the fan. After this was fitted we had two issues; the original PSU brick failed and the fan became sticky. When I fitted new devices the hum from the system became unacceptable, so I built some audio filers and slowed the fan down using a series resistor (see https://wmtprojectsforum.altervista.org ... t=10#p1014). Now it seems that i may have slowed the fan too much!

As a temporary fix today, I shut down the system for a few minutes to allow it to cool. I then opened the case and left the access door open (it is tied with string). I'm hoping that this additional airflow will allow the Pi to run much cooler.

If this appears to work, then I will remove the system from it's case and drill some holes in the bottom and back of the case to allow a reasonable airflow. I will then remove the fan and re-install the system with the case standing on runners to allow the air into the case to come in and get out after having gone over the heat-sink.

We will then monitor the situation over the summer.

One thing. We chose the ammo box solution for the case, because at the time it was cheaper than the IP65 cases then available and the Minster tended to get damp. The heat inside the case should be able to disspate the moisture while the system is running, but we will need to ensure that the box is removed after Winter Shutdown.

[Penri]

Hello Terry

Next time you are in can you take a look at the Quarterback, I though I saw only his right arm moving yesterday. It may have been me catching the movement too late as I wasn't really observing him closely.

Hwyl

Penri

[TerryJC]

Yes, I noticed it too. The problem is that to fix it I need to remove the Quarter Jack from the window opening. I thought I'd fixed this before, but obviously the solenoids are far to powerful for my puny repair.

[Penri]

Terry

I'm happy to take it out as and when you and / or Clive want to have a bash at a repair / alternative solution. We can talk to Greg about when would be best, I guess that anytime outside school holidays would be acceptable (but that's a guess).

Alternatively if to make the whole thing more robust it needs to be re-built:

• We could have a go at modelling the Quarterjack and get him 3D printed.

• We could get a new one carved from wood.

By we I really meant I with some help, I don't think either approach would be too difficult once we'd worked out a "better" mechanism for pivoting and actuating his arms.

Having a complete drop in replacement "window" could be attractive, a 3D printed Quaterjack may even be an attractive gift shop item.

Have a think and let me know, I'll have a little play at modelling now.

Penri

[TerryJC]

Penri,

We were never quite happy about the final arrangement for the arms on the Quarter Jack. What didn't help was that initially the solenoids were installed upside down and the design had to be pressed into working in the new orientation. Having said that, we were never happy with the 'shoulder-pivot' arrangement either, because the links have a tendency to work their way out of the locating holes. (I'll check for that the next time I'm in.)

What happened last season was that the block of wood that holds the solenoids in place pivoted and so the LH solenoid lost it's range of movement. This block is held in place with a single bolt:

QJ_Solenoids.JPG (92.69 KiB) Viewed 1345 times

You can see that in this picture the retaining block has already pivoted slightly out of alignment.

Over the Winter I put in an extra screw to stop this pivoting, but I guess it may have fallen out.

A new version would solve all of the above, but would be cheaper if we could struggle through this season, so that we could re-use the connector, the solenoids and the speaker.

[TerryJC]

Today, I went in to do some River System testing and Greg said that the audio in the Minster was stuttering. When I investigated, I found that the MP3 audio was restarting, just like it had been last year before I fitted the fan. After this was fitted we had two issues; the original PSU brick failed and the fan became sticky. When I fitted new devices the hum from the system became unacceptable, so I built some audio filers and slowed the fan down using a series resistor (see https://wmtprojectsforum.altervista.org ... t=10#p1014). Now it seems that i may have slowed the fan too much!

As a temporary fix today, I shut down the system for a few minutes to allow it to cool. I then opened the case and left the access door open (it is tied with string). I'm hoping that this additional airflow will allow the Pi to run much cooler.

If this appears to work, then I will remove the system from it's case and drill some holes in the bottom and back of the case to allow a reasonable airflow. I will then remove the fan and re-install the system with the case standing on runners to allow the air into the case to come in and get out after having gone over the heat-sink.

We will then monitor the situation over the summer.

One thing. We chose the ammo box solution for the case, because at the time it was cheaper than the IP65 cases then available and the Minster tended to get damp. The heat inside the case should be able to disspate the moisture while the system is running, but we will need to ensure that the box is removed after Winter Shutdown.

Sorry about the hefty requote, but it's worth re-iterating the history of this in view of what I found this morning.

When I arrived at around 0930, the music was playing and Hamish and Greg both mentioned that it had been playing from when they came in. Hamish also mentioned that when he went inside the Tower to repair the mains socket cover, the music seemed to reset itself after playing through much of the track. I noticed this too, so I opened the door in the case a bit wider and rebooted the Pi, whereupon it appeared to be playing through tracks OK when I left at around 1100. From all this I deduce that:

• Overheating is probably still the most likely cause of this.
• If that is accepted, then the work I did on the fan to quieten the hum has gone too far and the Pi is getting too hot again.
• If that is accepted, then putting holes in the case is the wrong solution; what is needed is to direct the cooling air over (and more particularly) under the Pi to ensure that the chips get enough airflow.

When I fitted the original fan last year, I used a miniature (40 mm) type. It ran at full power and generated some hum, but wasn't too bad. Unfortunately, it wore out after only one season and so I replaced it with a 60 mm type, which generated much more hum....

I have a suspicion that the problem isn't just because I had to slow the fan down, but also because it is blowing most of its air over the top of the Pi's surface. The centre axis of the smaller fan was below the bottom of the Pi, so it may also have some influence on the cooling effect. I'm going to look into using some kind of a funnel to direct the air below the Pi before I resort to fitting another 40 mm fan, which will probably become a consumable item.

[TerryJC]

What happened last season was that the block of wood that holds the solenoids in place pivoted and so the LH solenoid lost it's range of movement. This block is held in place with a single bolt...

I had a look at the Quarter Jack this morning as it sounded the quarters on the full hour. The solenoid is clearly operating OK and I could see the metal linkage moving normally.

In fact, the left arm is moving, but by not as much as the right arm (which is exactly how the real one was at the time we were doing this work ) Whether the block has moved again or the linkage has become bent is hard to see, but a proper fix needs the Quarter Jack Assembly removed and reworked as mentioned earlier.

[TerryJC]

I have a suspicion that the problem isn't just because I had to slow the fan down, but also because it is blowing most of its air over the top of the Pi's surface. The centre axis of the smaller fan was below the bottom of the Pi, so it may also have some influence on the cooling effect. I'm going to look into using some kind of a funnel to direct the air below the Pi before I resort to fitting another 40 mm fan, which will probably become a consumable item.

I started to look at how I could direct the air from the 60 mm fan to blow underneath the Pi rather than over the top and realised that the space available will make this difficult.

This morning I found a relatively cheap 5 V 50 mm fan on eBay which might suffice to:

• Lower the centre axis of the fan again, but not as much as 40 mm fan would.
• Run from the 5 V Pi supply instead of the 12 V which is also used to power the Audio Amplifiers.

If this fan runs off the Pi supply without causing excessive hum on the audio channels, then I can allow it to run at full power, which will hopefully solve the cooling problem once and for all.

[Penri]

Terry

Sounds promising but I'm left wondering is you are struggling to solve an issue being exacerbated by the enclosure being used. If there's no good reason to keep the box and we can identify a suitable and if it helps more spacious box, whether it would be better to take the plunge and buy one.

If I can help, let me know.


Hwyl

Penri