Low dcs signal strength

I don't think there's any one answer. Dcs can be tricky when it comes to getting a perfect track signal.

The first thing I would do is make sure the tiu's signal generators are actually working. 

 Hook up a small test and do a signal test on each channel with a know good engine. Should be a perfect 10.

(I don't think there's any other test  except the proto-cast feature...playing music )

Voltage... what goes in the fixed channesl comes out. they're a straight through connection.

Testing......  Disconnect the reds (leave the blacks connected) and add them back one at a time (test each time you add a feeder)   You may find longer blocks work better than a number of short ones.

Supper's calling.. good luck.

Does anyone know what type of signaling is used between the TIU and the track

The following is from The DCS O Gauge Companion 2nd Edition, page 55:

DCS Signal Strength

DCS signal strength is a number, from 1-10, that's calculated based on the number of data packets that the TIU sends and receives back from a PS2 engine that's performing a DCS signal strength test.

During the test, the TIU sends a continuous series of data packets to the PS2 engine. The PS2 engine is expected to reply to every packet that it receives, however, some packets are either not received by the engine or the response is not received by the TIU. Regardless, the TIU calculates DCS signal strength by counting how many response packets the TIU receives from the PS2 engine out of each hundred data packets the TIU sent, and then looks at a sliding scale to determine the signal strength that should be reported. The scale is not linear, rather, it is such that 87-100 packets equates to a DCS signal strength of “10”, 80-86 equates to a “9”, and so on.

and what would be a good voltage reading of the output signal. 

That question is irrelevant as regards DCS signal strength.

This and a whole lot more is all in MTH’s “The DCS O Gauge Companion 2nd Edition", available for purchase as an eBook or a printed book at MTH's web store!

Rad400,

   Read Barry's book, actual voltage has nothing to due with DCS signal strength.

Further track joins have more to due with signal loss than track length.  Follow the block building discipline and use the 12 track join rule for determining blocks, add the magic lights where needed and your layout will run with all 10's.  You are not looking at this correctly, right now.  DCS is an education in itself.

PCRR/Dave

Thanks all for the responses.  I will try the the "short track" test to make sure that all the TIU outputs are delivering 10's.  Once I know the TIUs are good, then I will start working on the layout to locate where I am loosing signal and make sure to follow the rules on page 55 & 56 in Barry's Book.

One other item I would like to verify:  The communication path between the DCS remote and the TIU is at 916.5Mhz band and the communication between TIU and the PS2/3 engines is at 3.72Mhz band.

Thanks,

Rad400 

Sure voltage plays apart, So does wire size. Set up a var channel with the dcs signal on. How low can you go before the dcs signal goes south.  I know dcs doesn't like loose connections.

Gregg,

Sure voltage plays a part

Actually, voltage does not really play a part as regards the DCS signal. Rather, it's the PS2/PS3 electronics that need a minimum voltage (about 10 volts or so) in order to operate properly and decode DCS commands.

Hi, Folks--

Barry is correct in a practical sense.  Actually, as with any signal, there is loss of strength with distance.  And, as you might guess, there is a formula to calculate it.

As you might further guess, in any simple form, this formula idealizes the signal conductor "pair."  Like it's the same along the path, and the path extends from the signal source to infinity.  Of course that is not usually the situation.

But if you use the formula, you'll find that at the rough distance of a football field, you have something like 90% of the signal strength.  I forget if that is power or voltage basis.  I forget because it hardly matters-- something else has got to be the problem.  Actually, Tony Lash' famous layout had as its longest block one 178 feet long-- it was Gargraves track and nominally the center rail and only one outside rail formed the DCS signal path.  It did not have signal from the feed end to the far end, only part way (it was not known exactly how far as the farther portion was covered).

The next longest block was about 135 feet, one example being center-fed and another end-fed.  These and shorter blocks had no signal problems, with the exception of one signal.  This was the signal to load an engine into the system-- this is a rather long exchange of several specific signals, and the entire set must be received end to end without error (inferred from the behavior in the usual block).

So one block, short and carefully idealized, was used for loading engines.  This was at the time a PS/2 system, so that the five-TUI limit for pass-off of engines applied, and thus the total blocks were limited to 20-- and 20 blocks were present.  An ability to run TMCC was overlaid, as was a power sourcing which could vary the ordinarily fixed 18 or 20 volts on track.  This enabled the running both of Lionel-style TMCC and conventional engines.  The power sourcing was split into two sources. thus with the three long loops (in large part it was a 3-track railroad) it was possible to run one of the loops-- perhaps any one-- in conventional.

By the cleverness of the arrangement, none of the DCS requirements were compromised, so we need not be concerned with the overlay circuits.  All power was routed through the TUIs; these were not all centrally located, but rather each one was central to a group of four blocks; each loop could run as many as three trains.  Of necessity, blocks were either end-fed or center-fed, so as to minimize wiring in the signal path.  Each TUI was between 4 and 8 feet downstream of its transformer; no wiring crossed any aisles.  Splitter boards were about 4 feet beyond TUIs.  On this layout individual wires went to every 3-foot section of Gargraves, becoming progressively fewer within each block--IIRC (unique to this layout, compared to wiring discussed here).  Again IIRC, each wire was #14.  That is, each wire was by itself able to carry the total normal output of the transformer feeding it, in case it should become the wire which shorted.

Well, it was wired by an electrician and was located in a commercial space.  Food for thought.  Sorry I can no longer recall the wiring size clearly-- been about 10 years.

The DCS signal is a set of discrete frequencies, where each frequency is orthogonal to all the other frequencies of the set.  Orthogonal means the magnetic and electrical vectors (which are themselves orthogonal to each other) are at right angles to all the other such vectors of all the other frequencies.  Because of this, no attenuation of any one frequency by interference (jamming perhaps) can affect the other frequencies.  [Any one who can explain this more clearly feel free.  After all this time, I'm thinking it may only be a mathematical analog of orthogonal vectors.]

These sets contain specific numbers of frequencies.  The allowed numbers are geometrically related to the powers of 2 (2 to the zero being 1) as follows:  1, 2, 4, 8, 16, 32, 64, and so on.  In WWII in the south Pacific, 2 to the tenth, or 1024 was used.  The equipment filled two farm wagons.  It took 3 or 4 days for the enemy to decrypt the plain language, so the talkers were colonists who had known each other before the war, and could reference locations etc by idioms known only among themselves in passing short-term targeting data.  The system was brought to the US by the Austrian Hedy LaMarr, when she emigrated to the US, with her band leader.  The two developed it for guidance uses, patented it, and donated the patent to the Navy.  (Musicians are often quite skilled in mathematics.)

In practice the groups of frequencies are one less than the numbers above, that is:  1, 3, 7, 15, 31, 63... etc.  [I'm guessing that the missing frequency in each group was zero (DC offset?) which cannot be broadcast, or had to be zeroed.]  MTH had in mind that the tracks in wired transmission were a very noisy (much static) environment, and that an orthogonal system would overcome this noise.  They choose to transmit 31 frequencies (and sampled at first at that rate in the receivers, which needed ideal conditions, did not work, and cause a delay of some time.  The double rate, here 62, always works, and did so for MTH.  The 31 frequencies begin slightly above the RS 232 highest frequency, at 121 kHz (IIRC above and 117).  The maximum frequency is slightly below the FM intermediate frequency of 10.25 MHz; the point of highest fundamental in the signal is around the 3.72 MHZ mentioned above (ie, ~121 x 31).  I believe the third harmonics, which square the waves, are normally included to improve signal recovery.  My reference was big on outline and small on detail.  There is also a formula to identify the point of maximum power.  I would characterize the band width as extending from 0.1 MHz to 10 MHz, more or less.

Now we come to the problem with all this.  As you will notice, the frequencies transmitted vary over a ratio of about 100 to one, or two orders of magnitude.  The waveforms of each frequency move down the conductive "pair" at about 3/4ths the speed of light, give or take, determined by the four electric and magnetic properties of the pair.  These are the capacitance, inductance, resistance, and leakage conductance (IIRC).  Capacitance and inductance are the main factors in the speed of transmission.  Unfortunately, capacitance varies with frequency in real life situations, while the equations we can handle without too much trouble consider capacitance to be constant.  This means that each of the 31 waves does not move at quite the same speed.  Communication pairs are selected to have low-capacitance insulation for this reason.

The decoding depends on detecting the zero crossings of the several waves retaining their order and not drifting too far out of expected position ("jitter"), if I may simplify.  MTH choose a test for accuracy such that 28 of the 31 waves had to contain the same information.  There is also the issue of keys, which vary in efficiency according to certain rules, meaning that there are not as many practical keys as the theoretical total.  So one might have gone from PS/2 to PS/3 by a bit of tweaking here, sacrificing errors per 1000 in command ID or concealment of the key.  There are other possibilities for signal improvement, and I am fairly certain MTH's CTO was aware of the jitter problem.

Any one who has dealt with wired master antenna systems will realize they are optimized (impedance matched) for one direction of transmission.  DCS requires an acknowledgement of receipt to complete each transmission, traveling in the reverse direction.  The ACK needs therefore to be short and well formed, and other requests for info avoided out on the distant road or in the forest of yard tracks.

Actually, it is possible to offset the signal path capacitance by adding some inductance to common across the line at intervals.  The famous light bulb of the light bulb trick had its coiled filament come under my microscope.  It has almost vanishing inductance (hence its effect is seen as magic)... only picohenrys to be exact, I've forgotten how many.  But if you calculate the capacitance of about 40 feet of center and outer rail of Gargraves on wood ties, you'll find it's only picofarads, about the same number.  This was about 10 years back, but I never posted the result.  You don't actually have to offset all the capacitance.  One bulb would probably set the 178-foot block right.

One of the great accomplishments of the early Bell System was to establish a 900-mile voice link from NYC to St Louis using 600-ohm open pair, #10s at 10" spacing, without any powered amplification in that distance.  The line was loaded by using passive inductors ("loading coils") at intervals (several miles?).  By accident Bell discovered that it was sufficient to offset only about 10% of the capacitance.  That was because the human ear had an amazing ability to correct the phase errors of the various frequencies.  [DCS has a very limited tolerance for phase distortion.]

--Frank  (Edit corrects spelling; revises and extends and paragraph regarding frequencies and band width; and makes conclusion regarding phase distortion, in conclusion.)

Frank, as I understand you, you are saying that the DCS signal is composed of electrical currents at varying frequencies.  Am I correct in this understanding?

Does anyone know how well Tony's layout operated or for that matter the NJ highrailers DCS part.?

rad400;

Looking at a club layout i would start with this list:

1) How are you measuring signal?

    Stop the engine, then check signal. A moving engine always shows lower signal.

    As a crosscheck, use some different engines to repeat the test. some work better than others.

2) How many sections of track per block?

    More than 12 joints in a block can lower the signal for the entire block.

3) Is the block truly isolated from the next one on each end.

    A short between blocks will create havoc.

4) How many blocks per TIU channel? Does this correlate with the blocks having low signal?

    Too many blocks can split the signal too much.

Russel;

I was taking the signal measurements with a slow moving engine.  Some of the loops I was getting 8-10 readings and other sections getting "out of range" & 1 & 2 readings.  I am going back to the club tomorrow and many of the items you suggested are on my list of items to check.

Thanks for the suggestions and I will report back on what I find.

Rad400

RJR--

Hello... It's been a long while since I posted here on DCS.  I recall posting some of the details before, probably more than ten years ago.  A lot of the actual numbers in the signal are taken from the patent taken out on this use for model trains.  A much more detailed explanation of the underlying principles can be found in a McGraw-Hill book, "Handbook of Telemetry & Remote Control," Gruenberg-- Editor; at least 2000 pages, maybe three.  I can't get close enough to my bookcase right now (rearranging basement) to read which edition it is, except that it is the _____th Ed.  I've had it at least 25 years and probably at least 30.  Thing is McG-H has the habit of stripping material out to make way for new, in its handbooks.  The section was already thin in some places, and may be largely gone by any current edition.

I was trapped in my car one day for four hours, with this book.  This is how I came to know how the DCS signal is produced, or at least, how it may be analyzed.  Before that, I had asked exactly the same question as Rad400.  I'd even posted that someone surely could put an oscilloscope on the signal and determine the voltage and frequency used.  Then we'd just see if the frequency could be transmitted over the various track types and branching configurations, just like master TV antenna wiring in an apartment building.  At that earlier point I was as much in the dark as anyone here ever was.  I just did not know I was in the dark.

Your question was:

"Frank, as I understand you, you are saying that the DCS signal is composed of electrical currents at varying frequencies.  Am I correct in this understanding?"

The answer is yes.  There are 31 frequencies; the lowest is about 121 kilohertz.

Well, it would be more clearly "yes" if your question had been "composed of...31 various fixed frequencies, transmitted *simultaneously." [*I think "simultaneously" as I have used it here implies the more familiar analog concept, as in musical tones.]

[But in a digital transmission I'd think that each pulse maps itself onto a particular position in the composite transmission that no other pulse may occupy.  That is, they are like discrete particles in one sense, yet seen like waves in another sense.  If that makes any sense...  I have trouble seeing this clearly, and I may take the lowest or 3-channel signal and try to map out the positions of each pulse.  In passing note that a 1-channel signal is actually the lowest but it is called the degenerate case, being the more familiar one fixed frequency on the track (at 121 kilohertz), similar to TMCC (at 455 kilohertz).]

Given the "3.72 MHz band" that Rad400 asked to have confirmed, band is not the term I would have used.  The number 3.72 MHz is actually the difference between the lowest and highest of the 31 fundimental frequencies transmitted.  These fundamentals are 121, 242, 363 kHz, etc.  Actually, they will be heard as noise across many of the conventional regulated bands--so it better be very faint noise.

It is conventional in a digital transmission as this is, to square the pulse rather than just sending a sine wave.  This is usually done by including the third harmonic.  Curiously these harmonics will lie on some of the fundamentals.  If we assume clipping at 10 MHz (to clear the FM intermediate frequency at 10.25 MHz) the last four fundamentals cannot be squared.  And some fifth (and greater odd) will be generated, unless some additional suppression is devised.  Food for thought.

All these harmonics increase the dv/dt at the zero crossings of the fundamentals, which is what the receiving decoders detect.  The more steeply the "wired radio" waves cross the zero voltage line, the more precisely they can be timed.  This makes successful decoding more likely.  More food for thought. (That "dv/dt" means rate of change, volts per second, d meaning difference).

That being realized, I may have to reverse myself-- increasing voltage will increase dv/dt also and so may assist reducing the jitter problem.  I had just assumed that the signal voltage was 5 volts-- transistor-transistor logic... I guess they just added a voltage doubler to the output of the TUI, to get a 10-volt signal.  Food for more thought.

Of course, I should mention that available power can be reduced by dividing the path-- the 12 sidings rule, is it?  Rarely seen, though Tony had 10 or11 in a yard.  Maybe that's why the MTH terminal board only has room for 12 taps (IIRC?).  Could we call it the "Use only one board per channel rule"?

In closing for now, I just want to say that I have discovered a simple proof that the explanation of the multiple frequencies is correct.  (Modulation by flipping the half waves, got to keep away from too much DC offset doing it, etc, etc, best ignored in any explanation at 2 am).  I'll try to post that as soon as we're snowed in again.

And BTW, there is a display device (computer screen, likely) called a spectrum analyzer.   Almost bought one back when (early 2000's).  $2500 here in my house $5 or $6000 Australian FOB Australia.  Just lucky AAA saved me by taking 4 hours to get to my car.  Just set that expensive frill on your point of signal-strength-1 and see the voltage of every one of the 31 signals.  Fits in a briefcase I'm sure....

Not sure it'll show the deadly jitters, though...  

Regards, Frank     (Edited in an attempt to clarify what my "yes" answer meant.) 

I wonder what's down the road for     DCS?.  any guesses ? How about each engine has it's own mini tiu built in  and the remote would talk directly to the engine eliminating the need for paired wiring and  such. You wouldn't need a tiu period.

Thanks for the description, Frank.  I have to digest it slowly.  With as much as 10" being forecast in our area, I should have time in the next day or so.

You can hear every command though a radio (am) including the watch dog signal  plugged into the same wall outlet as the trains  It doesn't mean anything to me except the tiu is working.  A series of short blasts sort of like a machine gun. Weird!!

Gregg,

You wouldn't need a tiu period.

Correct.

However, then you couldn't run conventional trains from the remote, either.

Barry,

    Most people do not understand that particular option even exists, and for me it really doesn't anyway, it removes way to much of my other train running options to even consider using it for any amount of time.

PCRR/Dave

Huh? what other option Dave.  If you mean conventional I agree . 

Gregg,

    Conventional in 2 different ways, the TR Z4K mode and the TR TIU mode both.

This would eliminate half of the trains moving on my layout.

PCRR/Dave

I think you still could  have The Z4K option.  It seems to me I spend too much time looking at the remote rather than running the train. There's just much to go wrong. dirty track . low dcs signal . Tiu generator failures.   operator error. Anyway that's just my opinion/ get rid of the tiu and talk directly to the engine.

On the generation of the DCS signal, theory and practice...

I had mentioned that I discovered a proof for the technical explanation of the nature of the DCS signal.  Actually, the proof depends on a more general theorem, and it's more that I remembered this theorem.  Despite a lot of effort, I could not find the mathematician's name.  Basically, it says that any two functions or processes that produce the same results are themselves the same, despite any outward difference in form or expression.  It is an equivalency theorem of some sort (I found a lot of discussion under numerical analysis theory--which was big in my day, although then there was a lack of computer capacity to make it generally useful).

In the patent, MTH describes enough about the frequency structure output from their electronic signal generator (TUI) that it can be seen it matches the output of the process described in the the Handbook of Telemetry, etc. for orthogonal signals.

MTH, in the patent, did not actually disclose the circuitry that produces the DCS signal they described.  But because of the equivalency theorem that applies, we do not actually need to know that circuitry.  Admittedly, it would be good to know the voltage and power of the signal output stage, but now we apparently have the voltage level for PS/3.  Given this and the 12-yard-tracks rule, and the FCC limit, we may be able to deduce the power of the source.

The significant thing is that we are in a position to analyze the DCS transmission path in much the same manner as the branching layout of the master TV antenna system.  The Barry Rules express much of this to a certain extent.  Still, it can help at times to borrow a more structured engineering approach.

If for no other purpose to generate more "Rules" (or less "magic"). 

For this we would need formulas for transmission equations and impedance matching.  The are a bit hard to find, so I'll post them as I have time for it.

--Frank