What the MRC says.....
NCE has noted that the problem
described below and has duplicated it in their workshop.
Tests show that this problem is limited to operation with sound decoder
only and the EB3 operates to specification with all other non sound decoders.
The solution proposed by
Marcus Ammann
with a globe that bypasses the EB3 with a globe in
parallel restricting the current to a couple of hundred mill watts is
currently the most practical solution and should not cause any foreseeable
problems.
Possible solutions:
Click on this link for a
better way to use the EB3 [MRC comment]
Click on this
line for Marcus Ammann
bypass circuit
In conclusion a mix of the two
solutions would appear to offer the most reliable operation of the EB3 in
a layout incorporating all of a high quantity of sound equipted
locomotives
Technical Discussion
EB3 and
limitations when operating with sound decoders
prepared by
Marcus Ammann
Since installing a EB3 I have noted
slow response times with sound decoder and have done some experiments.
With CV 49 set at 10 (100 m secs), system works ok only if one sound
equipped loco is in power zone, but with two locos in same zone, double
headed or second in a siding but still powered, the EB3 does not reset
within a reasonable time. Changing CV 49 to 11 to 14 still no good. With
CV 49 set to 15 (150 m secs) when there is a short in the zone then all
three status lights start flashing and after a few seconds, command
station trips. This is not a desirable result.
If I increase current trip limit to 3 Amps then sometimes ok. Put 3 sound
equipped locos in zone - no good, if I increase to trip limit to 4 Amps
still won't reset in reasonable time of what I would expect say within 5 -
10 seconds. See results below.
These are my test results. My sound locos in all tests, all have
Soundtraxx DSD100LCs. I don't have any BLI locos. I have wired up EB3 off
the layout and on my bench with no other track connected and away from any
other track wiring and using only one channel at a time.
Trip current 2 amps
1 sound loco = OK
2 sound locos = reset sometimes after at least 1 minute
Trip current 3 amps
2 sound locos = reset 5 to 40 secs and 50% greater than 1 minute
Trip current 4 amps
2 sound locos = resets after 5 - 10 seconds most of the time
3 sound locos = never resets in less than a minute
Above tests done on circuit breakers 1 & 2 and CV 49 and 50 set at 10
(100m
secs)
These results have been duplicated by
the MRC and should be taken into account when installing them in to a
layout running a large quanity of sound decoders.
Tests show that this problem is limited to operation with sound decoder
only and the EB3 operates to specification with all other decoders.
Why: Mark Gurries
Sound equipped locomotives have
presented challenges to DCC that were not anticipated when the NMRA
specification were written. The problem is the amount of capacitance that
needs to be charged up to allow the sound electronics to function reliably
with various types of DC power that is NOT pure DC. The capacitor charge
current is a huge spike involving amp levels that far exceed the current
capability of the both boosters and Circuit breakers. But it does not
take long to charge. But every time you add another sound equipped
engine, the problem grows in size to a point it will become a major
problem.
Prior to BLI, sound equipped locomotives were few and quite a show item
initially since it involved a lot a work to install a sound system. (But
what head turners the sound units were!) Anyway, the problem existed but
showed up at more of an annoyance level issue.
When BLI came on the scene, things changed quickly. Engines with sound
became "Ready To Run" along with good quality construction allowed people
to acquire more sound equipped engine faster than ever. Today most BLI
customer come back for more and having many sound equipped engines on the
layout has now become common place. Correspondingly, the problem is has
now become a BIG issue.
Here is how the problem happens.
Electronic based Circuit breakers use Current Level and Time Duration to
determined the difference between a normal momentary short circuit (normal
stuff rolling down the track) versus a real short (caused by a derailment)
where the short current can be sustained indefinitely.
Sound decoders have BIG capacitors in them that are used to store power to
keep the sound going un-interupted as the engine roles down the track make
less than perfect electrical power pickup at all times. These capacitors
must be charged up BEFORE the sound system will work. When they are first
charged up, they look like a short to booster or circuit breaker. The
short circuit current level fades quickly with time for it only
momentary. The current goes to zero when the cap is fully charged up.
If the capacitor current fades fast enough below the short circuit trip
level before the circuit break decides it is time to kill power, then
everything works like you expect. No problem. If the current trip level
is lowered or reduced, then the exact same capacitor current will not fade
fast enough to clear and the circuit breaker will trip.
Adding more sound equipped locomotives is the same as adding more
capacitors in parallel. Depending on your electronic circuit breakers
setting and the peak current capability of your booster, people will get
various degrees of success and failure with combinations of locomotives.
The layout wiring also plays a part here too. So there are lot of
variables involved on the layout side. What fails to function on layout A
may work just fine on layout B.
Your light bulb solution works because it adds resistance in series with
the engines limiting the peak current. The down side with the light
bulb is that if you have a lot of engines on the same section, the track
voltage will drop a lot as the light bulb starts to glow and the engines
will not run well.
Technical Discussion and possible
solution at the Manufacture's end...
The amount of capacitance to put into a sound decoder will have a minimum
and maximum requirement.
The minimum capacitance dictated for the circuit are typically concerns
that are covered by the datasheet of the parts involved or an engineers
experience with the circuits involved. But all of these specifications
assume the power is clean and un-interrupted (Pure DC). In this case, we
put the sound system on wheels with intermittent contact which changes the
capacitance requirement to work beyond the minimum. The maximum, however
is determined by the decoder sound system designer through actual real
world testing to handle the intermittent contact situation. Large
capacitance capacitors will store more energy and keep the sound going
through longer
durations of power dropouts or dirtier track so to speak. Unfortunately
there is no standard for "intermittent contact" in terms of time and
strength. Another factor determining capacitor size is the requirement to
have the sound unit work with old DC power. Many DC power packs have
pulse power or less than pure DC power that was targeted specifically for
motor control and not to run electronics. A BIG capacitor is again needed
to filter out all those pulses to allow generation of enough pure DC to
run the electronics. So the solution is all over the map depending on who
did the
design and what the design goals are. At the same time cost and space
issue
may become a factor in deciding what to do.
Soundtraxx discovered with the DSD that there was not enough capacitance
to make all the customers happy. Yet the size of the decoder was a big
concern since not every would have the space to fit a large capacitor if
it was factory installed. Although they never updated the DSD design (now
that Tsunami DSD is to replace it), the did address it with the DSX by
allowing one to optionally add extra capacitance. There is a technical
note about how to do just this on the Soundtraxx website. Since decoder
size is a Soundtraxx feature, having enough capacitance is a tough issue
to address.
I think the DSX approach is good idea that they should keep in the Tsunami
product when it comes out.
QSI, which does the sound for BLI and Lionel and perhaps others, has the
luxury of making decoders that are specifically design to fit in space
provided by the locomotive from the day the locomotive design was started.
Since the sound unit is guaranteed to fit, size is less of an issue and
unlike Soundtraxx, can use less expensive and bulkier components. Their
boards reflect just that design and thinking. They are huge compared to Soundtraxx boards.
From an electrical standpoint, there are two parameters that determine the
effective short circuit current level and durations.
1) The circuit impedance.
Using Ohms law, V = I x R and
re-writing it to I = V/R we can see that there is a direct relationship
between the current, track voltage and circuit resistance. If the
resistance goes down, the current goes up. The resistance is all the
resistance in the complete loop of current flow from the booster to the
track to the sound board through the cap back out all the way back to the
booster. Typically this resistance is less than 2 ohms and typical track
voltage is 14.5V. So the maximum current is really limited by what the
booster will provide. So clearly every time the cap charges up, we WILL
hit the booster current limit.
Part of that resistance is the resistance inside the cap which is called
ESR or Equivalent Series Resistance. Its resistance value can be high
relative to the layout wiring resistance. High performance caps will have
low ESR and cheap caps will have high ESR. Low ESR will result in High
peak Current Flow into the cap. High ESR will reduce or limit the peak
current to a lower value. So the choice of cap can also effect the peak
current value.
2) The capacitance value of the capacitor(s). Simply put, the more
capacitance you have, the more energy you can store. Its a bigger
rechargeable battery so to speak! That also means if the current available
to charge up the cap is limited, the longer in time it will take to charge
up to full.
So the worse thing to have is a low ESR cap with high capacitance. It
will draw high current and sustain that high current for a long time.
Just what the Circuit Breaker is looking for to shutdown. For a given
size, cheaper caps will have higher ESR and Store less energy. So there
are cost versus Size versus performance tradeoffs that must be made. The
total capacitor solution will then vary with the application requirements.
The thing that bothers me is that there is a simple solution the high
current cap charge inrush current. First install the minimum capacitance
needed by the parts on the board. Then add the extra big capacitance for
intermittent power holdup in parallel with the small cap but with a
circuit change. Put a combination diode + resistor in series with the big
cap with the diode in parallel with the resistor. The Resistor will limit
the peak current or power to a safe level when charging up that is well
below any trip limit. The diode allows the cap to bypass the current
limiting resistor and provide full power to keep the sound unit running
when power is momentarily lost. It cheap, small and simple to do. I hope
the Soundtraxx Tsunami has that or fixes the problem some other way.
If the NMRA DCC body was to do something, it would be to define a inrush
current profile that all booster and circuit breakers must pass and to
recommend the circuit in question be designed to minimize the inrush
current below this inrush current profile as best as possible.
