GCA173 Hall-sensor and/or reed-switch converter.

ContentHardwareGCA


By Peter Giling


Introduction

Quite a few systems are available now to detect the Loc, passing a certain area.
Most of them require extra construction inside a loc.
Sometimes, it is already available in the decoder, like Lissy or Railcom.
But to receive these messages, it requires quite some expensive hardware.
One other system , Maerklin's MFX is also nice, announcing the train to be on the track, but it does not give
any information about the exact position.
The wheel-counter GCA141 together with Rocrail, is a system that will tell you at least
that a train has entered completely into the block.
But then, placing the IR transmitter and receiver is a very precise job, maybe not easy to fix for everyone.
All the other systems do not give you any info about if the complete train has entered the block, without losing
some wagons or couches on the track.
This of course unless you will build-in Lissy transmitter or an extra Railcom decoder and/or have the wagon/coach
equipped with some way of drawing current from the rails, where the current detection on the track will pass that information to the computer.
Reed-switches were very popular long before the digital era was emerged.
But due to the very limited possibilities of magnets, and the fact that faster running trains sometimes 'missed' the
reed-switch made the use of it less popular.
Even now, while we have many other ways of sensing the train on the track, a reed-switch might be sensible to use.
But we have a better, and much smaller option now : Hall-sensors.
These sensors, like the TLE4905L as described here,
are very small items, even usable for N-track, because this sensor only measures 4,2 x 3,3 x 1,5 mm.
It even can be inserted inside the gravel, making it totally invisible.
The other advantage we have now, is the availability of very strong and yet very small Neodynium magnets.

An example of a very small 1x4x5 mm magnet under this Fleischmann N-track 781283 Loc A small N-Track kipper with a (not glued !) 5mm cube magnet. A TLE4905 hall-sensor fixed in an N-track testing circuit
It can be fixed between the leggers !
magnets_122.jpg hall_magnet_182.jpg
A small example of all different kinds of magnets available A Roco NS24xx HO Loc with a 1 x 4 x 5mm magnet
Top middle is CAN-GC2 and
right top is GCA173 (prototype) connected to CAN-GC2
One other HO solution example


This sensor is 3-wire connected and could be connected directly to a CAN-GC2. or GCA50.
But why should we need this extra GCA173 ??
First of all the connection itself, This pc-board makes that much easier.
Than the pull-up ohm either GCA50 , CAN-GC2 or GCA-PI02 is 10K ohm, which should be much lower to have a solid distortion resistance.
Also a capacitor of minimum 47 nF is needed, conform the data-sheet.
CAN-GC2 needs pulses from feed-back with a minimum lengt of 25 mSec. For GCA50 even the minimum is 80 mSec.
The GCA-PI02 needs minimum of 10 mSec.
That makes them practically not suitable for this hall-sensor, or worse, a reed-contact.
To explain that let us make some approximate calculations. :
If we have a distance of 5mm between magnet and hall-sensor, the magnet will activate the sensor for a length of approx. 7mm.
A HO train, running in scale 120 km/h runs in reality 380 mm per second.
If the sensor only 'sees' the magnet for a length of 7 mm, the time the sensor is activated is 7/380 = 0,018 SEC = 18 milliSec.
A pulse like that is bound to fail, because the pulse will be only 1/4 of what we need.
The CAN-GC2 and GCA-PI02 are better (=shorter) but still way too critical.
GCA173 is here to take care for that
It has connectors for 8 sensors, either hall-sensor or reed-switch.
A small and cheap microprocessor on GCA173 checks continuously all 8 inputs, and will react when one or more inputs are going low, to forward this signal to either GCA50 or CAN-GC2.
Even if the input is no more than 2 millisecs, the output will remain high for 150 milliseconds, assuring that the message is passed correctly to your LocoNet or Can-network.
The hall-sensor itself is much faster than that (approx 1 uSec), so the chance that a train will be 'missed', is 0%.
One other extra advantage of using the GCA173 as a buffer, is that we will have enough connections for all eights sensors.
In case of direct connection between CAN-GC2(GCA50) and 8 hall-sensors, you will have to combine all +5V and 0V connections of the sensors together.


Features

  • Easy connection of up to eight individual sensors and or reed switches.
  • Each input will be passed through with a minimum pulse length of 100 mSec.
  • Low impedance input takes care for high distorsion immunity.
  • Even for reed-switches , the low input current is exactly right, to keep the reed-switch in good shape.
  • 1 cable connection to GCA50 / CAN-GC2, which also feeds this board with 5 Volt.
  • Very low power consumption when all inputs are inactive.
  • Led signal on each individual input to simplify testing.

Counting Rolling stock

The big advantage of this system is the fact that you will be able to detect if the train has entered completely into the block.
As you are used from Rocrail, this fabulous program also does have a standard setting for this sensor, called 'Wheel counter'.
Except here we are not counting wheels but Magnets. But the result will be the same.
At least two magnets should be mounted under the train, the last one under the last wagon.
When these magnets have announced themselves on entering the target block, the conclusion is safe to assume that
nothing is left behind in a tunnel or shadow station, so the computer will proceed with clearing the used route and use it again.
The sensor itself will be programmed as 'wheel-counter' in Rocrail.
Rocrail will automatically use this counter after the first initiation, and take over this info,
If you like, more than one extra magnet is an option, even different trains with different number of magnets is an option.
With that installed you could even distinguish which train did come in, but that is an option, not yet available in Rocrail. (Is not planned.)
Commuting trains should be equipped with at least two magnets, so sensors will work in both directions,
and make the train stop at the first detected magnet.

One other interesting option with these sensors:

Using Loc-combinations, where one is a s.c. 'dummy-loc', which is not taking any current from the track, this combination might stop too late, where current detection is used and the dummy is in front.
The use of two magnets -one at each front- will solve this problem simple and easy.

Handling in Rocrail

Rocrail compares the counted wagons from the From with the To block:

  • No comparison is done when one or both blocks are reporting zero.
  • Normal processing in case the To blocks reports equal or more than the From block.
  • In case the To block reports less then the From block the loco is cancelled from auto mode and the From block is closed.
  • After the lost material is connected to the train again, the From block can be set to action, and the train can be re-activated to auto mode.

Boards available

I nice double sided board , with through hole metallisation and silkscreen print is available now.

Magnets

A large variety of magnets is available from www.phgiling.net:

Type Size
1 Rod 4mm ∅ length 12,5 mm
2 Rod 4mm ∅ length 10 mm
3 Rod 3mm ∅ length 8 mm
4 Rod 3mm ∅ length 6 mm
5 Disc 5mm ∅ high 3 mm
6 Cube 5mm
7 Plate 5 x 4 x 1.5 mm
8 Plate 5 x 4 x 1 mm
9 Plate 5 x 2,5 x 1,5 mm
10 Plate 5 x 1,5 x 1 mm
11 Plate 10 x 4 x 1 mm


Magnets are only sold in sets of 8 pcs per size.

Correct position of magnets

Magnets have two poles, of which only one will activate the sensor.
Before putting magnets on your rolling stock, please check the correct position.
This can be done by checking with a sensor itself.
Sensors should be with marked (rounded) side facing towards the magnets.

Safety precautions

These very strong NeoDynium magnets are no toys !!!!!
In principle they are safe by itself, but please be careful with these magnets , do not allow children play with it.
There are quite a few safety remarks, downloadable here : Neodyne safety precautions

Connection Hall sensor

The connection wires between Hallsensor and GCA173 can be up to 6' or 2 meter.
Please be careful not to feed those wires parallel with rails and rail wires.
A bit of twisting these wires will inprove protection to any distorsion.
If that is inevitable, keep a distance of 4'' (10cm).
Straight angle crossing is no problem.
Below you can download a drawing of hallsensor connection.

How to connect TLE4905L hallsensor

Connection to reed-switch

Allthough GCA173 can work fine with reed-switch as wel, the option for wheel counter is NOT applicable.
This due to the fact the the number of pulses, generated on the passing of a magnet is never the same, and unpredictable.
connection to reed-switch

Cable to MGV50 / CAN-GC2 / CAN-GC4

Connection table of J1

J1 pin# Function Remark
1 +5V Supply GCA173
2 0V/GND Supply GCA173
3 Hall1 output sensor 1
4 Hall2 output sensor 2
5 Hall3 output sensor 3
6 Hall4 output sensor 4
7 Hall5 output sensor 5
8 Hall6 output sensor 6
9 Hall7 output sensor 7
10 Hall8 output sensor 8

Hardware

The schematics
The pcb and parts positions
The partslist
The Gerber files
N.B. Self made pc-boards are not supported!


Firmware

gca173_v1_4_firmware.zip
This latest 1_4 version is published, because it has a more accurate timing before resetting the output.
Timing is close to 80 mSec.
gca173_v1_5_firmware.zip
Version 1_5 has a better startup sequence. It will send all actual sensor positions

Pictures from satisfied customers

3 wires connected to sensor
middle wire isolated
Drill hole 4,5 mm between sleepers
feed sensor through hole
bend wires down ( round head sensor at top side ) and fix sensor
a first functional test with 5 volt
connected from any adapter

Personal Tools