DCC occupancy detections false positives

You need to check the 'sensitivity'

I.E. a DRY finger across the rails shouldn't trip things, a WET finger should.

There isn't any 'magnetic field' to shield.

Chuck D.

To expand on this a bit. The only 'magnetic fields' involved, are the magnetic coupling of the 'track feeder' passing through the 'sensor coil'. And those, you need, and they seems to be working. (Too well at times.)

The DCC signal is 'loaded' (current drawn) to indicate occupancy.

Something is loading things, the problem is WHAT?

The available DCC signal, can be loaded (some current drawn) by either capacitance or resistance. Capacitance can be as simple as the two wires paralleling each other over a long distance. Resistance can be as subtle as DAMP roadbed.

We need to determine what the cause of the problem is.

Chuck

I'm wondering if maybe he's got some sort of slightly conducting glue or paint on the tracks, or the ballast? I saw a layout where the ballast was made by some guy and then sold at a train show. You could get a solid ohmmeter reading just by touching it to the ballast and either rail. It got warm in some places when you put 12V to it for a while. This was back in the late 70's.

The guy tore it all up and redid the track and ballast. He was just about to start putting the buildings up.

On 5/17/2008 2:16 PM Charles Davis spake thus:

This is so unlikely (capacitive loading from just the 2 rails) as to be outside the realm of possibility.

If you had a pair of insulated wires tightly twisted together for a long run, you *might* start to see some effects due to capacitive loading, but even this is unlikely.

Not that UNLIKELY at the frequencies (& harmonics) generated by a DCC signal. Remember, they TRY for a 'Square Wave' with DCC. You can't look at things as if you were working with Analog Sine-Wave RF, and draw conclusions based on that universe.

When I typed the above ---- ">> Capacitance can be as simple as the two wires paralleling each other >> over a long distance." I wasn't thinking of a 10' run to a close block. more on the order of exceeding 75'. Chuck D.

On 5/17/2008 4:46 PM Charles Davis spake thus:

Sorry, TYPING IN ALL CAPS still doesn't make it any more correct. Even taking harmonics into account, a DCC signal is still well down in the kilohertz region, nowhere near radio frequencies. Capacitance effects, if any, are negligible.

Right, but the poster was talking about a 'bleeding edge' type of sensitivity. And no, the fact of the 'Square Wave' signal wave form generates all sorts of higher frequency harmonics. In face, those harmonics are a not insignificant portion of the power being pumped out by the DCC Power Station/ Booster.

Chuck D.

David Nebenzahl skriver:

Hmmm, can you tell me the harmonic content of a 10kHz square wave?

In case you need help, then read:

Especially the lines: "An ideal square wave requires that the signal changes from the high to the low state cleanly and instantaneously. This is impossible to achieve in real-world systems, as it would require infinite bandwidth."

Klaus

"Old.Professor" skriver:

Alone that you are trying to send sqare waves through a transformer can be enough to draw current and induct voltage to the secondary side of the transformer.

My recommendations is to use 4 diodes and an optocoupler for track occupancy detection.

Klaus

On 5/17/2008 8:52 PM Charles Davis spake thus:

So show me where I disagreed with that; I said above, if you bothered to read, "even taking harmonics into account".

So let's look at those harmonics, shall we? First of all we need to know the fundamental frequency of a DCC signal. Referring to the DCC standard itself (I'm reading from one of the PDFs on the NMRA site), the shortest interval defined is for a "1" bit, which they define as being about 100 microseconds wide. (They actually allow more time than that, but let's use the lower figure for the sake of discussion.)

Since freqquency is the reciprocal of bandwidth, that means the the highest frequency in a DCC signal *should* be about 10 kHz. For the sake of discussion, let's double that, to 20 kHz.

As we all know, square wave harmonics are the odd multiples of the fundamental frequency. These harmonics diminish the higher they go; in other words, the strongest harmonics are the lowest-order ones.

So we now have harmonics at the following frequencies:

- 60 kHz (3f) - 100 kHz (5f) - 140 kHz (7f) - 180 kHz (9f) - 220 kHz (11f)

and so on. So we can see that we have a diminishing set of harmonics that is *well* below a megahertz until we get well up into the series. Which says that any higher-frequency harmonics are going to be so weak as to have negliglbie effect. Which means that the capacitive effect of two conductors in the form of HO track can be disregarded.

On 5/18/2008 12:41 PM David Nebenzahl spake thus:

which should have read "Since frequency is the reciprocal of wavelength".

On 5/17/2008 11:29 PM Klaus D. Mikkelsen spake thus:

Yes, that's easy: harmonics at 30 kHz, 50 kHz, 70 kHz, 90 kHz, etc. See my other post in this thread.

Sorry, I don't use Wikipedia ("the encyclopedia any junior-high-school idiot can edit") for any serious information.

Misleading; you're forgetting that the amplitide of the harmonic series is constantly decreasing, so that the higher-order harmonics can be practically disregarded above some limit.

David Nebenzahl skriver:

Okay

I know this very well.

But that's looking at at square wave. What is the frequency content of a signal changing from 0 to 1 ?

Only the slew rate of the driver sets the limit.

Klaus

On 5/18/2008 1:49 PM Klaus D. Mikkelsen spake thus:

Well, that's basically half a square wave, right?

Correct, which is why we have to approximate the frequency. See my other post, where I derived a plausible upper frequency of 20 kHz from taking the stated parameters for the shortest-interval signal (a "1" bit) and doubling them. Maybe you could triple it for safety's sake.

Of course, this doesn't take into account the various "glitches" that DCC controllers inevitably output ...

David Nebenzahl skriver:

No, because the harmonics of your square is mainly to get the top of the square "flat". It really doen't tell you anything about the rising edge.

No, the leading edge of the square don't care if the pulse is 1uS or 1mS long, it still has to be very steep.

Klaus

On 5/18/2008 2:06 PM Klaus D. Mikkelsen spake thus:

Well, yes. We're really splitting hairs now, so this might be my last word on the subject. Basically, the steeper the rise time, the closer an approximation to a square wave it is, with all the applicable stuff about harmonic generation (odd-numbered harmonics). If it's less steep, then it's closer to a sine wave. So everything I've said applies, to some extent or other.

Note that in the real world, everything is only an approximation of these idealized concepts, like square waves, which don't really exist anywhere in reality. There's always stuff like overshoot, ringing, sloped edges, etc. Look at a "square wave" on a 'scope sometime for amusement.

On 5/17/2008 8:52 PM Charles Davis spake thus:

I'll let my last words on this subject be not my own, but those of the DCC experts at Tried & True Trains

At 10 KHz, the DCC signal is essentially immune from the problems with reflections and standing-waves which higher frequency tone systems can experience. The DCC specification requires that the decoders be able to reject input above 100 KHz. All useful DCC signal information is below 100 KHz, and the behavior of wiring at frequencies above 100 KHz is irrelevant to DCC operation.

David Nebenzahl skriver:

Thanks.

And in the ideal word you do not have problems with occupancy detectors

- real world is something else.

Klaus

Hi David; Sorry you are so hung up on "theory'. The 'Old Professor', was asking about a 'Real World' problem. In the 'Real World', observed problems, 'Trump' theory every time.

Capacitive loading cannot be ignored. It may be only a minute part of the detected load, but that 'minute' part may be the final 'smidgen' that trips the detector.

Go argue with Greg for a while.

Chuck D.