Thursday, August 16, 2018

I do ride occasionally

Due to a late meeting I missed my bus connection tonight and ended up taking Caltrain home. Train #264 arrived in Mountain View 3 minutes late, i.e. basically on time. I had a choice to wait in Mountain View for #268 which would take me down to Blossom Hill, or change trains in San Jose. San Jose has more variety of traffic, so I took #264 down to Diridon station and hung out on the platform until it was time to catch #268 for the ride home.

I have not transferred between trains in San Jose for a while and did not expect that south-bound trains are now leaving from the newly built platform 6 that opened several years ago.
I ended up asking a conductor who looked at me funny when I said that the Gilroy train normally leaves from track 5, no?
"Uh, no sir, it's been leaving from track 6 for quite some time now." 
Right. That was 15 years ago, when I was commuting to South San Francisco, before Caltrain introduced the Baby Bullet service ...

Tuesday, August 14, 2018

Dave Park's Cumberland West - Williams Street Yard

For tonight's B&O ops session, I opted for Williams Street Yard, which can get very busy and if the yard master is not careful he can completely gum up the railroad.  Williams Street "Yard" is basically a siding, a stub end track, and the River Interchange to the Western Maryland's Ridgley Yard in the foreground in the photo above.

Since this was the first time I ran Williams Street, I did manage to completely gum up the railroad and made several trains wait for 30+ minutes at the towers around Cumberland station while trying to service train BO11. It didn't help that a local was in town, and one of the towers had trouble with their JMRI CTC machine and couldn't throw turnouts for a while.

At Williams Street, the yard master needs to think ahead and position cars and switcher engines for the servicing tasks at hand before the respective train arrives in the station. This is very similar to how I envision operations at Backnang station on the future Murrbahn layout, so this was a very interesting exercise to try and see what such job would look like. I pretty much failed tonight, so I'm going to try this job again the next time it is available, and see if I can to it better.

Sunday, August 12, 2018

Digitizing Roco 43012 - More Space Planning

Slowly, but surely everything is finding its place. The PowerPack moved into the rear of the cab on top of the decoder. It turned out that the niche inside the loco nose didn't have enough space for the capacitor. I cut down the speaker baffle to fit the speaker between chassis and roof. I just need to decide how to keep it solidly in place, yet removable.
I will use the red cross bar behind the cab to consolidate wiring for head and tail lights as well as mounting the cab light.

Hand-made Coffee

Great coffee.
Hand-made. Hand-delivered.
Enjoyed in California.

Upgrade Orgy

I'm running several servers and network infrastructure at home. We rarely use desktop computers anymore, since the family has basically moved towards Laptops, Chromebooks, tablets, and phones for their technology needs (with the exception of the PC Gaming station). However, those devices along with entertainment equipment still want to be supported and connected to the Internet by ... something. At the very least for in-house backups and redundant storage. Cloud services for photos are all nice and good, but with volume they become expensive. Backup providers are shifting on an almost yearly basis in search of a viable business model.

I haven't touched much the infrastructure for quite a while. It worked well enough so there wasn't really a need. "Don't touch a running system" is for lazy system administrators that rather play with their trains. However, things started to get a bit creaky over time. Newer software versions complained about the OS being too old, that java needs this libc which ...

My backup server doubles as the railroad computer, so it has much more disk space than needed for the trains. A recent backup filled up the remainder of the archives disk ...
Alright, the combination of some older archives, plus the live backup from the server have exceeded the 2 Terabyte mark on the main server. Time to double the disk space and install 4 TB drives. But that requires a GPT-style partition, which requires parted and that's not installed on this machine, since it still runs Debian Squeeze.

All the signs point at the lazy system administrator to get off his behind and get moving. I procrastinated for too long already. Time to go all the way to Debian Stretch.

24 hours later, I'm still working through apt-get dist-upgrade cycles and resolving dependency conflicts. It would probably have been easier to install Debian Stretch off an USB stick.
Once that upgrade orgy is over I can copy the backup volumes from the 2TB array to the 4TB array, and start looking into getting JMRI working again.

Meanwhile the sole Ubuntu Desktop in the house is moving forward to Ubuntu 18.04 now.

After those two are done, I'll take another round of backups of the main server and upgrade it to Debian Stretch as well, which then allows me to modernize some of the monitoring infrastructure on the network.

Solar PV System Cleaning

It's the middle of summer. Time for the annual panel cleaning. I usually clean the solar panels in late July, early August. The winter rains keep them reasonably clean for the other half of the year, but by the middle of the summer enough dust, dirt, and gunk has baked on the panels that they can use some love. Cleaned panel on the left, dirty panel on the right.

I wash the panels with a hose, wipe them off with a rag, and finally wash off the remaining dirt.

It's getting pretty hot up here in late morning, so I should get back to work. The result is below. If you look closely, you can see that the two panels in the top left corner have some gunk left near the top edge, since I couldn't safely reach that corner with the rag.

Saturday, August 11, 2018

Digitizing Roco 43012 - Loksound 4.0 with PowerPack

I hate soldering an ESU PowerPack to the Loksound 4 decoder. Given the circumstances, this is pretty well done with reasonably sized solder pads, but I really don't like soldering on a $100+  decoder board.

Once that was done and visually inspected, I connected speaker and a motor to the decoder and fired up the Mobile Station to double-check proper operation. Let capacitor charge for a couple minutes, fire up the speaker, disconnect the lead, aaaaaand ... urgh. Nothing.

Yes, the motor is the one I replaced from Roco 43012
I verified the programming of CV 113, which controls how long the decoder pulls power from the capacitor. It was zero. Well, yeah, that's not going to work. Re-program to 150. Let charge. Try again, aaaaaand ... nothing. Urgh!

No matter what I programmed into CV 113, it was zero when I read the value back out. Alright. Something is preventing the decoder from accepting the PowerPack. I double-checked my solder connections. All good. Scratched my head. Didn't help. ...
Eventually, I turned to the Internet and found a hint that usage of AUX6 conflicts with usage of the PowerPack! That's ... unexpected.
To make the Powerpack work, AUX6 needs to be disabled by setting CV 315 to 0.

That worked great. On to the next try, the decoder sound continued to rumble along when I disconnected the power lead.

On to the next step: Space planning. Even though the unit has quite a bit of volume inside the aerodynamic shell, options for speaker and decoder placement are very limited due to the motor, flywheel, transmission, and gears, combined with seating space in the rear half of the locomotive and my desire to have a partially furnished cab with engineer. I am planning to add cab lighting, too.

Also, I want to keep the shell easily removable for service, so I can't have wires just go everywhere, otherwise it will be hard to get the shell back on.
I decided to mount the decoder on styrene sheet under the cab. There is barely not enough space to put the decoder next to the motor and I need the space above the motor for the speaker. The PowerPack will go into a niche in the engine nose. To organize cable runs a little bit, I'm trying to route lights and power pickup to a circuit board that will be located near the motor, roughly where the decoder is on the photo below.

I temporarily connected the power pickups, programmed the motor parameters, and did a short test run on the bench.

Sunday, August 05, 2018

Maerklin BR86 167 conversion to LokPilot

My very first digital Maerklin locomotive was a BR86 from a starter set. This model is known to have problems with power pickup, and sounding like it's grinding up its gears. Part of the issue here is the very low-end MM2 decoder built into these locomotives. Nevertheless, I lived with it for the last 10 years, but the locomotive got less and less use as other, cooler locomotives arrived on the Welztalbahn or were converted from analog.

When switching a local freight this afternoon, I got fed up about the noise and crappy driving habits of this locomotive and brought it to the work bench. The axles got some more oil, the gears some grease, and it ran a little bit quieter, but not a lot.
Alright, I do have a couple spare LokPilot 3.0 sitting on a shelf. Today's the day!

The old MM2 decoder made way for a Lokpilot, I connected power pickup and motor leads to see if this truly makes a difference... and what a difference it makes! The drivetrain no longer aches and croaks, but purrs like a kitten. I connected lights and Telex couplers. It all works. Off to the programming track we go.

Next, I spent a considerable amount of time debugging the bad power pickup. It's not the (known) bad power pick up from the outside rails, which I fixed soon after I got the locomotive by soldering cables between the spring that keeps the pilot trucks on the rails and the ground connector inside the locomotive.
It appears the pick-up shoe has trouble maintaining healthy contact with the center rail studs, and/or passing the current to the decoder. As a test I soldered a wire to the pick-up shoe base and fed it directly to the decoder.

That worked a bit better than using the contact at the bottom of the locomotive. I went even as far as routing the cable through the innards of the locomotive to the decoder bay and connecting it to the red cable feeding into the decoder.

When test driving the locomotive across the layout, however, it reliably stalled on the inside curve of curved turnouts and caused shorts because the pick-up shoe made contact between the center stud contacts and the inside rail of the outer curve. In the end what did make a difference was installing a replacement pick-up shoe. Supposedly, a Roco "silent" pick-up shoe should work even better.

Unfortunately, the grinding sound problem came back as well, not as bad as it used to be, but it looks like I will need to do some more tuning with the Lokpilot motor control settings to try and get this under control.

Sigh. Quite a bit of work for only moderate improvement...
For now, 86 167 is back in service on the afternoon local from Emsingen pulling cars for Prechtal into staging.