Hornby DCC Power Booster R8239

[St1ngr4y]

Having had my wrist slapped for going off-topic in Graskie's "Controlled Runaways" thread in the Railmaster forum, I have raised this topic separately to re-ask a couple of questions which no-one answered in the other thread.

I have a R8239 Power Booster connected between my Elite and my track bus. The connection is between the Elite's boost terminals and the PB's "boost in".

Q1 Are there occasions when a loco decoder needs to send a request TO the controller and if so, can the PB handle two-way traffic?

Q2 Is there any advantage or disadvantage in using an RJ12 cable to link the Elite and PB using Xpressnet ? Q1 also applies in this scenario too.

 

Ray

[PJR_slo]

Hi Ray

 

Apologies.

 

Q1: NMRA standards only specify one-way communications (except for programming track).  There is a separate document concerning "Railcom" communication, but for this both the decoder and controller (and booster if used) have to be Railcom compliant.

 

Q2: Xpressnet would possibly be very slightly better as it works at microcontroller level, rather than track level.  In practice, I guess there would be very little noticeable difference (if any).

 

Peter

[poliss]

Unless your layout is seperated into electrically isolated 'Power Districts' you do not need to use a 'booster'. It won't increase the number of locos you can use on your layout.

If your DCC accessories stop working properly, through lack of power, then you could power them directly from the booster.

Railcom enabled decoders can handle two way traffic. The controller also has to have a Railcom function.

The XPressNet port can be used for the 'booster' but it's mainly for connecting Walkabouts. Hornby say, "The BOOST output connection offers greater flexibility than the alternative RJ12 connection described above in that no special cables are required."

 

The NMRA standards are very limited. They only really cover protocols to decoders. Different makes of controllers use different systems to talk to their boosters etc. For instance, Hornby use XPressNet while ZTC use X-Bus.

Other makers use LocoNet, Cab Bus, CVP cab bus, Roconet, CAN-bus, DiMAX, ECosLink etc.

[Fishmanoz]

Just slightly off-topic but related Ray, poliss's suggestion above peaked my curiosity.  Is it the case that you aren't running 2 power districts, one with your Elite and the other with your booster?

[St1ngr4y]

Hi Ray

 

Apologies.

 

Q1: NMRA standards only specify one-way communications (except for programming track).  There is a separate document concerning "Railcom" communication, but for this both the decoder and controller (and booster if used) have to be Railcom compliant.

 

Q2: Xpressnet would possibly be very slightly better as it works at microcontroller level, rather than track level.  In practice, I guess there would be very little noticeable difference (if any).

 

Peter

Hello Peter,

No need for apologies - it should be me who should apologise for going off-topic. I installed the PB when I was new to DCC, and I thought (incorrectly at the time) that the Elite could only handle 1 amp. So I am using the PB for the track bus, but I have also an accessory decoder bus which is connected to the track output of the Elite. This setup has one advantage in that if I get a short on the track due to a loco fouling an incorrectly set electrofrog point, this short shows up on the PB, and I can still operate the points via the Elite. The downside to this is that the PB doesn't have an on/off switch, so to reset it, I have to pull the power connector and plug it back in.

The reason that I am asking questions about the PB at the moment is that I recently purchased a MyLocoSound decoder which I have installed into the tender of a Hornby Thompson Class B1. However the sound functions (including F1 Sound on/off) are often delayed anything between 5 seconds and a minute, when operated either directly from the Elite or from Railmaster via the Elite. It was suggested by the manufacturing company that I connected the Elite directly to the track, which I did, and lo and behold, the delays experienced with these functions disappeared. I can't find a satisfactory explanation as to why this should be, and it was why I asked whether the decoder ever requests from the controller the status of the sound functions, and whether these could be being blocked by the PB.

Ray  

[96RAF]

I think you may have answered your own question there Ray.

As I understand it your set up is Elite powers the points bus from its TRACK terminals and the PB powers the track via the Elite BOOST terminals.

I would suggest the Elite is still sending all loco commands to the points bus and is obviously getting no acknowledgement back so per DCC rules it sends it again and this is introducing your delay.

I haven't rerad the NMRA spec again to determine how many times this is allowed to go on but I guess not many and only for seconds not a minute depending upon max locos alive on track of course.

[Fishmanoz]

Are you sure that's right?  As a general rule, there is no acknowledgement of commands in a DCC system, Railcom excepted, and Railcom isn't involved in standard loco controls. 

[PJR_slo]

No, NMRA spec does not require acknoledgments, even for accessory decoders.  However, NMRA specs do state that loco commands should be repeated as often as possible.

 

A booster should simply repeat the instructions within a fraction of a second.  Any instruction buffer it may have would be limited to a very few NMRA packets.  Therefore there may be a possibility that the booster is not precisely echoing the controller output. 

 

There is no way that a booster could selectively block certain DCC packets as it probably would not have the circuitry to "understand" the data.

 

With the booster being connected to the track output, it should echo that exactly.  However, it may be worthwhile trying different connections.  It might also be that the booster is faulty (not high on the list of possibilities though...).

 

Peter

[96RAF]

Yes I'm in error there Fishy

Quoting from my DCC 'book':

"...The command station transmits to the decoders and only in rare situations when specifically asked does the decoder need to communicate back...". The book doesn't say what these situations may be - probably Railcom.

Where I was wrong is in the acknowledgement - it is in fact the decoder self checking that it has all the broadcast information and either acting upon it or rejecting it as incomplete and waiting for it to come around again.

"...Before a decoder can act upon a command it must first check the entire contents of a packet (specific DCC command intended for itself)...". In essence it does this with a checksum of the address and instruction information against an error checking byte - if correct proceed, if not the decoder "...watches the broadcast bitstream as before until it receives a properly addressed packet with a new command..." then repeats the check.

The bandwidth capability of the command station and the number of locos and/or accessories being addressed determines how long it is before a specific command is seen again.

Depending upon the command station important information like sounds can be prioritised and older less important activity like idle locos (e.g. stopped or constant running) can be skipped until an active command is generated.

 Remember that each loco is waiting for any of the following commands - short address, long address, simple or advanced consist, and/or broadcast address (i.e. all stop) and that each (loco) command has to specify speed, direction and lights/sound functions.

So I still think Ray's problem is the Elite pushing loco commands into his accessy bus district and the booster having to pick them up and send them to the track district. The method of connect of the booster may affect the response time and as PJ and poliss say Expressnet may be a better connection than the noisier track links.

[2e0dtoeric]

All I can add to this is - What a complicated way of doing things!

Unless you have a lot of things happening at once, all the time, and have a HUGE layout, what do you want the booster for?

I assume there is NO connection between the track BUS and the accessory BUS? (other than at the controller)

If you really feel that you must include it, I suggest you swap the connections over, and run the railway with the Elite, and the accessories with the booster.

[Fishmanoz]

I seem to be not seeing something here.  If we start with just the DCC data stream, it's all contained in the PWM signal, even if a lot is going on.  In fact, the maximum rate of active commands is reached when those data packets are being broadcast right behind each other, but still in that PWM signal.  Irrespective, they are being sent in series and, given the circuitry can handle the modulated wave form of a packet as well as the constant mark-space ratio without modulation, it won't matter or put a higher load on the system if a packet is sent only on the odd occasion versus packets being sent constantly.

 

All the booster is doing is picking up that data stream and re-broadcasting it at 4 Amps.  I can't imagine that when things get busy it starts to delay any of those commands/data packets.  It would have to store them and forward them later if that were the case, or miss them altogether.  I can't see this being the case though if all it is doing is repeating the signal it receives and it will be no harder to do so with lots of packets on it as it is when there are only a few.

[PJR_slo]

Fishy, a minor correction - it's not a PWM signal; there are 2 frequencies used, high frequency for the '1' and lower for the '0', on a alternating current which may be square-ish.  And I believe that you are totally correct in that the booster is basically a simple frequency follower, so what goes in should come our almost immediately.

 

Peter

[96RAF]

Given the command station will generally transmit data packets at 150-200Hz it doesn't matter if they are shortest 38 byte idle packets (i.e. a preamble to say a packet is coming, but with no instruction for any decoder, then packet end) or a much longer speed, direction, lights and sound function instructions to a decoder, they all queue until they go out (if necessary again and again) and its this queue that is picked up and mimiced by the booster.

So in theory if the command station is performing OK then so should the booster. As Fishy says the load on the system to send packets out doesn't change perse, but the work it has to do assembling these packets does increase according to activity demanded by the operator and its that performance which can vary among the various command stations.

I would be interested to see a paired HF oscilloscope comparison of the Elite at max chat versus the Booster keeping pace, and if the latter could possibly get out of synch, Elite queue to Booster queue, at times of high activity change by the Elite, thus introducing the noted delay. i.e. Do the Elite and the Booster have equal processing performances.

[PJR_slo]

RAF

 

You don't need a high frequency oscilloscope to see the packets - I used one connected to the PC set to record 10ms chunks, to capture about 60 bits of information.  And yes, it's possible to see the 21 header bits sent by the Elite.

 

As I said earlier, it's the ±4V range which is important to the decoding and as long as that is pretty much "square" there will be no problems.

 

Unfortunately, I don't have a Hornby booster, so I can't check the output from that, but it certainly would not require complex electronics to produce a good output.

 

However, I don't think that any of this speculation is helping Ray to find a solution, which I think has probably already been offered in this thread and that is to swap the usages of booster and Elite, or to remove the booster from the equation.

 

Peter

[Fishmanoz]

Fishy, a minor correction - it's not a PWM signal; there are 2 frequencies used, high frequency for the '1' and lower for the '0', on a alternating current which may be square-ish.  And I believe that you are totally correct in that the booster is basically a simple frequency follower, so what goes in should come our almost immediately.

 

Peter

 

 

Certainly if you look at the signal in the frequency domain, you are going to see I think 3 fundamentals being idle, 0 and 1, along with the harmonics needed to make them nearer square than sinusoidal. But surely the fact that you are producing only 2 pulse widths corresponding to 0 and 1 is no less PWM than if you can change it continuously from a mark-space ratio of 0 to 1?

 

But you are right, this won't solve Ray's problem, I was just trying to put some rigour into explaining it, or not.  And the solution isn't a real solution, rather just a workaround.

 

And just another thought to further complicate the issue - given the more general use of a booster is to have 2 quite separate power districts, rather then just a split between track and accessory buses as Ray has done, and given you can run locos between districts, surely there is a need to have phase lock between the controller and booster outputs?  Or put another way, the signals must be fully synchronised to avoid the deocder getting confused as it crosses?

 

[PJR_slo]

Fishy

 

The Mark/Space ratio is always (with one exception) 1:1.  As the loco can be put onto the tracks either way, the signal the decoder sees can be either phase.  The only exception is an idle zero which can have a long positive pulse and short negative pulse (or neg/pos depending on loco positioning).

 

A booster's output is totally isolated from the controllers output and therefore worst case would be that the decoder sees effectively DC for a fraction of a second (or longer).  This should be OK; the decoder would probably ignore that and continue to carry out its last instruction.

 

Peter

[96RAF]

To explain a little more for those interested the extended 0 pulse PJR is describing is for control of DC locos on a DCC track whereby either the positive or negative side of the pulse is longer or shorter than the other thus a DC motor will see an effective positive or negative voltage and travel in the appropriate direction - quite clever but simple solution really. When stationary the DC motor will see an average 0 volts but will feel the full cycle of +/-15VAC which is why it gets warm and hums.

[2e0dtoeric]

PJR is describing pulse width modulation, not digital control.

PWM is a way of applying dc to a motor at full voltage, but in  tiny 'squirts' so it doesn't run flat out, but does give full torque - the length of the 'squirt' (in milliseconds) decides the rpm's. ie 90 off/10 on gives dead slow, 50/50 is half speed, and 100/00 is flat out.

Digital is a whole chain of 'on's and off's' in a 'packet' that starts and stops with a long 'off' to act as a spacer/reset marker. The combination of on's and offs inside each packet tells which decoder chip to do what. A 'packet' lasts a fraction of a second, but the number of packets in a set time never varies, as they are all the same length, just with different contents.

A power booster is just an amplifier - no different in function to your hifi amp, or any other, it merely takes a signal from the input, and makes it bigger at the output. The signal itself will be unchanged.

This is getting rather too deep for most people to follow, and is totally irrelevent to the original question anyway!

[PJR_slo]

2e0

 

Sorry, I am precisely describing the digital control as specified by NMRA (other digital control protocols are available).

 

With the excption that I mentioned, the digital signal is positive for 50% of the time and negative for 50% of the time.  This in no way fits with PWM control which in most instances is a single frequency which varies the positive/zero (mark/space) ratio.

 

The number of packets in any given time does vary due to the different frequencies of signal bits.

 

The circuitry within a booster does have the capability to alter the input signal somewhat (as demonstrated by the various inputs, it is not a "simple amplifier") as I have described elsewhere.

 

Again, may I suggest that we limit discussion to the OP's question and should others wish to continue discussion regarding the intracies of the NMRA standards that they should open another thread on that topic.

 

Peter

[Howbi]

Well gentlemen, all respet to your electronics knowledge but after reading these posts my brain went numb talking about square sinusoidal harmonics and packets going here & there to a point to put me off DCC and go back to clockwork unless you can assure me that a decent size layout with 15 locos can be enjoyably operated with just one Elite controller, no boosters, no buses or computers? HB.

[96RAF]

Follow Fishy's basic measure howbiman - if its larger than a double garage you MAY need a booster.

The over-riding principle is "if it ain't broke don't fix it".

I reckon you are AOK using one Elite, no boosters, buses (full or partial) maybe installed if you are seeing any track unreliability and pc if thats what you want to control your trains, else stick to playing trains manually on a minmum complexity layout until you see trouble then think again.