DCC occupancy detections false positives

Old.Professor · 2008-05-17T20:27:39+00:00

I just installed DCC block occupancy detectors and am getting falsepositives only when the layout is in an operating session with lots ofpeople and trains. When I'm by myself there are no false indications.I'm at a loss what to do, or how to test if the changes have beeneffective.I'm using a variant on the VT-5 inductive (transformer-type) occupancydetector (see the variant by Richard Napper is described on page 18 in .During the operating session I saw the occupancy indication changefrom unoccupied to occupied and then back to unoccupied in about 30seconds. Most of the time it showed occupied when there was nothingon the track. Now that I'm by myself, it is steadily showingunoccupied.I had a thought that there might be stray fields. I was going to tryusing aluminum foil to shield the detector, but then I got towondering if that would be effective if the fields were magnetic.Suggestions would be very much appreciated.I also posed this request to

B

Bernhard Agthe 18 years ago

Hi,

In a way you're right. Opto's are nice. But I'm not sure they'd work in my case. Just assume, I generate some tickle-voltage oscillating with say 1kHz at a few Volt. How can I insert it into a non-powered track with an Opto? With a transformer this has been done often (light power on DC?), I'm not sure whether a capacitor will work (it should by theory ;-)

The detection is the other side, the tickle voltage is sent through say a non-lighted car with a 1K-Ohm resistor accross the axle. The inserted voltage will result in a current of a few milli-amps - hardly enough to "enlighten" an opto... I'm not sure if there's a way to uncouple that current from track power reliably (again with capacitor or transformer).

So in principle I'm sure my system should work, but I'm also sure, practical conditions will leave very little tolerance, if this works at all...

Actually I did have another idea last night. There are capacitance based sensors to detect whether a room is filled (by a person) or not. These systems work with the air of that room as dielectricum and sense a change in capacity. Although I'm sure this would work in a model railroad scenario (provided the train occupies a significant part of that section of track) there is lots of doubt if this will work with train power supply on the same wire...

Hmmm... I'm almost through track-laying, next step is wiring the layout and then some scenery, this leaves enough time to think about occupancy

- until then the "optical" solution must suffice: just look if the track is free ;-)

Ciao...

Vote

I

Ian Jackson 18 years ago

Optocouplers are a lot more sensitive than you're giving them credit for. My system is able to detect the leakage current when I put my dry finger across the DCC-energised track, which I haven't measured but will be perhaps a few tens of microamps. The real question is what the impedance is of the circuit at the optocoupler output. My circuit is built around microcontrollers whose input pins are at least

1MR.

Look at the datasheet for the optocoupler you're using. The figure you're looking for is `current transfer ratio'. I use TLP521-A whose datasheet quotes 50% minimum. The impedance presented by the circuit on the output in my system is dominated by a 100kR pullup resistor; that output (ie, the microcontroller input pin) will go from 5V to 0V if the optocoupler's phototransistor draws 50uA.

The actual trip point of the microcontroller input pin is somewhere around 1V but we can see that given the quoted current transfer ratio of 50% to 600%, my circuit will be trip at detection currents of approximately 8uA to 100uA.

Of course you can't run a track occupancy indicator on a layout diagram - even an LED - directly off this circuit. The current transfer ratio of 50% is hopelessly low. But this is easily fixed by (say) using a FET as an amplifier. FETs have gate impedances of 1MR or more.

Vote

K

Klaus D. Mikkelsen 18 years ago

Bernhard Agthe skriver:

Why use the oscillator and all that - it is way to advanced and difficult.

The optocoupler works, if you feed the track voltage to the tracks via a

10Kohms resistor and nothing else.

Klaus

Vote

D

Dale Gloer 18 years ago

Exactly, an oscillator is overkill. Just use an isolated low voltage AC supply, say 12 to 15 volts and a current limiting resistor to each detection section. I used Bruce Chubb Optimized Detectors, a 15 volt AC supply and 1K0 current limiting resistors. My rolling stock has 10k0 resistors, 1 axle per car. This worked just fine and did not cause any problems even after I converted to DCC but hadn't got around to updating all the wiring.

Dale Gloer

Klaus D. Mikkelsen wrote:

Vote

B

Bernhard Agthe 18 years ago

Hi,

[...]

Thanks for your input, I will add this to my list of things to try ;-) I will have to use a shunt to take the bulk of the train current and generate a voltage drop, if I understand this right?

Ciao...

Vote