Feedback modules don't like Japanese trains

[Dani]

Hi,

 

I'm in trouble with my feedback modules... I'm using Uhlenbrock 63320 modules (English manual here), and they have an unexpected behaviour that on one hand seems "logical" but on the other hand is a trouble if all consumption modules work like that. I explain:

 

Basically, if a car whose bogies act to pick up current and both bogies are linked or bridged stops between the isolated section (connected to feedback module) and the previous track section (normal section, directly fed with dcc current) the detection doesn't occur. That was unexpected for me, but now I see it's normal because that car in between sections is joining or bridging the isolated section with a normal DCC one.

 

What exactly happens is that detection occurs when train is in movement and enters the detection section, but once it's full stopped (if a car has a bogie in the detection section and the other bogie in the normal track) the detection disappears, and I see all my hidden yard tracks free. Taking into account the distance between bogies of the same car (more than 6 cm) and the distance between bogies from distinct cars (less than 2cm) you have all the chances to stop always a car between sections.

 

This is something not happening with "normal" trains (a loco as first car digitalized, and normal cars behind with no pick up current wheels), but with Japanese trains it becomes a headache.

 

Anybody faced the same problem? Any suggestions?

 

I'm really angry because the builder and seller just agree this is normal and they don't make any advice about it and sell it like "with a 20cm detection section is infallible", without any kind of advice.

 

Thanks,

Dani.

[Martijn Meerts]

I had the same problem with the old layout using Selectrix and Minitrix/Fleischmann/Roco trains, but it's because of the way I did the blocks on that layout that caused the problem.

 

All the regular blocks had 3 sections. 2 of those I had hooked up to an occupancy detector, and 1 directly to digital power. There were 2 stop sections of around 15-20cm, and in between a regular digital power section of whatever length required. Whenever a train stopped part on a stop section and part on the digital power section, I had unreliable occupancy detection.

 

My workaround was to set up the computer control software so that trains would run a bit further, so they would be completely on the stop section. A better solution would've been to add detection to the non-detected section of the block though. Definitely something I'll be doing with the new layout.

 

Especially the Tomix all-car-power-pickup trains will cause issues in blocks unless the entire block has detection.

[Dani]

Yes Martijn! That's my case: one detection section entering the block (where I reduce speed of train), a normal DCC section, and other detection section at the end to stop it in the right place.

 

Do you mean that connecting also that central section of the block (the one with normal DCC) to the feedback module problem is solved? In this way doesn't matter if a train is between two detection sections?

 

This would be a simple solution as all my sections have a feeder that I only have to change from the normal DCC conector to the feedback module!

[Martijn Meerts]

That should fix the problem yes, because then the entire train will be in a detection zone. If you set up the control software correctly, it'll reliably set the block to occupied.

 

The only downside is that you need more detection modules, because you'll need 3 occupancy detectors per block, rather than 2. Depending on the number of blocks you have it could become expensive fast. (Cost is the reason I want with 2 occupancy detectors per block rather than 3, but having experienced the problems, I'd rather spend a bit more and use 3 detectors per block =))

[The_Ghan]

Hmmm,

 

I'll be curious to know how Digitrax occupancy detection works with my Tomix Shinkansen with the active couplers, particularly over the slow down / stop sections, which are wired like yours Martijn.

 

ATM I'm only doing my subway and none of those trains have active couplers.

 

Cheers

 

The_Ghan

[Martijn Meerts]

Glancing over the description of for example "BDL168 LocoNet Occupancy Detector", it looks like they use the same general idea as the Selectrix occupancy detectors (and most DCC detectors), which means you'll likely have the same problem as described here.

[The_Ghan]

Glancing over the description of for example "BDL168 LocoNet Occupancy Detector", it looks like they use the same general idea as the Selectrix occupancy detectors (and most DCC detectors), which means you'll likely have the same problem as described here.

 

I might have to experiment with isolating the cab cars. 

 

Cheers

 

The_Ghan

[Dani]

Glancing over the description of for example "BDL168 LocoNet Occupancy Detector", it looks like they use the same general idea as the Selectrix occupancy detectors (and most DCC detectors), which means you'll likely have the same problem as described here.

 

I might have to experiment with isolating the cab cars. 

 

Cheers

 

The_Ghan

 

Ghan, just and advice before you do a lot of work:

 

It doesn't matter if you isolate the cab cars, the same problem is with a normal EMU car as "out of the box". With a normal EMU car (or Shinkansen I guess) both bogies are linked together and used to pick up current if you want someday to install interior lighting on it. If the car is "stepping" at the same time on a isolated section (with detection) and a normal section, it's like if you remove the isolator. It's the same if they have decoder installed or not.

 

 

 

Cheers,

Dani.

[The_Ghan]

Ahh,

 

Digitrax claims that occupancy will be reported for both blocks.  If transponding is used it will report both blocks as well.

 

Cheers

 

The_Ghan

[Martijn Meerts]

If Digitrax says both "blocks" will be reported, then they're likely talking about a different situation.

 

The problem here is using multiple sections within 1 block, not multiple blocks. (A train too long to fit in 1 block and taking up 2 blocks instead will of course be correctly detected as long as 1 of the cars in the blocks is a power consumer)

[KenS]

I'm not following what you mean by "block" and "section", but I think I understand the problem: if you mix track fed through an occupancy detector with track fed directly (bypassing the detector) a car that can bridge the two (which could be a passenger car with both trucks connected, or a loco wired similarly) will prevent the occupancy detector from tripping.

 

That makes sense, since it's effectively providing any car in the detected section with an alternate path to the power supply (though it, which isn't very good in terms of capacity).  Electricity will always flow through the lowest resistance path (actually both are used, but it's proportional to resistance and one gets small fast as the difference in resistance increases), and it makes sense that the detector is adding a slight bit of resistance. If the direct feed is less overall even with the added resistance of going through the cars internal brass strips than going through the detector, then more current will come from the direct-wired feeder and less through the detector, for every car in the "detected" stretch of track.

 

Note that a corrolary here is that mixing detected and non-detected feeders this way could be danerous to your trains.  If current flows through one car between sections, those rails inside it will heat.  On a parked train the current is going to be trivial.  And on a moving train the higher current will produce a higher effective resistance, which may prevent the problem.  But it might not.  On a moving train the car will move on and have a chance to cool.  But if you left a car parked across a detector/non-detector joint while other trains were moving in the detected section, drawing power through the parked car, you might do some damage to it.

 

Detectors work by sensing current, but there's a minimum required to trip the detector.  With very low current (what you get when loads have an alternate path with lower resistance) there isn't enough to trip the detector.  Digitrax was probably saying that when a car bridged two detector sections next to each other, each would each have enough current that both would detect.

 

You might be able to fix this by adding resistance to the non-detected feeder.  But given the power involved even if the power draw was under 1A, you'd need something in the range of a 12-24W resistor (which is huge, and not cheap). And it would be impractical with a higher load through the feeder.

 

It may be cheaper just to add a detector to the adjacent feeder (you could potentially use the same one for a couple of feeders in different places if you don't care about the actual detection, as a way to cut costs). Then the two bridged sections should be close in resistance.  You'd get partial power draw through each, and as long as the smaller of the two was enough to trip the detector, both would work.

 

I've seen this in practice. With my test track wired to two detectors, there's a brief period as the DE10 I was using bridged both where both detectors reported "occupied" which could have just meant a delay in noticing the change, but likely meant it was drawing power from both.  I never tried parking it across both to see what happened (and I've taken the test rig apart to install the detector in the layout so I can't test it now).

[Martijn Meerts]

A block can/should be divided in multiple section. A standard single direction block will have 2 sections, both will have a detector. The first section is the brake section, the second section the stop section.

 

A block with movement in both directions will need 3 sections. A stop section on each end, and an additional section in the middle. The stop sections will function as brake sections as well depending on the direction of travel. The middle section has no real function, so it's often hooked up directly to the power loop rather than to a detector.

 

The problem here is indeed that once the train has come to a halt, and it bridges the (detected) stop section and undetected section, the block won't be occupied. As long as the train is still in motion the detection works just fine. I noticed some weird behavior with some locomotives though. When certain locomotives bridged the detected and not detected section, the occupancy detector would constantly go from detect to undetected and back to detected again, not sure what caused that ;)

 

Anyway, moral of the story is that you don't want to cheap out on occupancy detection, because it can cause plenty headaches. You might also want to test things before implementing it in a layout  

[The_Ghan]

Hi Martijn,

 

Just a couple of points:

 

1. I've lot track of the number of times you (and I) have regurgatated this information about wiring sections for stopping, etc,  in thread after thread.  I learned from your three-part special you posted a year or two ago and agree 100% with your philosophy.  I think you should dig those posts out, tidy them up, and make it a sticky at the top of the DCC page called "How to wire DCC for effetive occupancy detection".

 

2. I have whole-of-layout occupancy detection.  Yes, it's more expensive, but I have more hair because of it.

 

Cheers

 

Th_Ghan

[Martijn Meerts]

Yeah, it's been re-posted quite a few times by now :)

 

I've now had both whole-layout detection, and the cost-saving version as mentioned in this thread. Whole-layout detection is definitely the way to go for scalp preservation reasons :)

[inobu]

The feedback module is basically a script in a PIC (Programmable Integrated Chip) that scans the input voltage and current. The script looks for a voltage input that sets the reference value. As it samples the track section any current draw will trigger the register.    

 

The problem is the bogie allows the DCC voltage to jump the isolated segment and change the reference voltage. Most likely it causes the script to reset therefore changing the status back and forth. Any time the voltage/current threshold is met the occupancy condition will change with it based on what the bogie is injecting.  

 

The key is to establish a standard detection window and match to the them track segments. Isolate the bogies to prevent the signal hop and palace the load car in between two "dead bogie cars". This should make up your standard window size.

 

Inobu