How Many Bus Terminators is OK

[Greynut]

On reading different posts relating to suppressors/terminators, it certainly seems that they are the way to go in helping to reduce or stop spikes etc affecting decoders. I'm adding them myself and at present, I have 2 'T' shaped busses, separating from the same feed out of the elink. I have added a terminator to the end of each 'T' - amounting to 4 terminators. I intend to add a couple (might be 3) more such busses as I develop the layout, which goes round the walls of a room in a spiral with different conected levels. So ... there could be a total of 10 terminators.

 

As each bus terminator has a current draw, how many terminators would be considered OK using a Hornby 4 amp transformer?

 

Also, can a terminator be wired so that it can terminate more than one bus run to cut down on numbers used? (Sorry if that's a daft question but I thought I'd ask.)

 

It's very likely that I'll need a booster due to the eventual overall length of the current supply some stage - how might that affect it?

[Chrissaf]

If you have a look at 'PJR_slo' response in "Do Sidings need Electronic Trermination" he gives some maths that I concur with. He calculates each suppressor to draw about 50mA. From memory I think the value I measured on my installed suppressors was 47mA.

 

So 10 suppressors would equate to about 0.5A taken out of the DCC supply. But the issue is more than just a basic current consumption calculation. One has to factor in the effect of having so many 'high pass' filters in parallel on the DCC digital signals. That many filters could well have an adverse effect. My filter theory isn't good enough to make a judgement on this (maybe Fishmanoz has a view). A rhetorical question ( a question that does not need an answer) - do you really need 5 x T Buses?

 

I think there is a misconception that arises from people calling a 'suppressor' a 'terminator'. Sure, they probably work more effectively when used at the end of an open ended DCC Bus. But a true 'Terminator" is used to match the impedance of a signal transmission line.

 

This following is one bit of theory that I am confident with as it relates directly to my working career in the communications industry.

 

An infinitely long open circuit line will have (due to the effect of inductance and capacitance) a characteristic impedance. On shorter lines, the line degrades the signals carried over it less when the line is terminated in its own naturally occuring characteristic impedance. The Terminator limits the amount of signal 'reflections' being sent back into the line to interfere with the signals being transmitted. For example, some Coax lines are 50 ohm, some are 75 ohm etc. For Balanced twisted pair lines some are 100 ohm, some are 600 ohm, other values can also be found in different scenarios.

 

So the function of a 'Terminator' is to provide the correct impedance termination on the end of an open circuit line. A DCC suppressor is NOT a Terminator, it is a 'High Pass Filter'. Therefore in my view it is not mandatory that it needs to go at the END of an open Bus, although it may perform more efficiently at this location.

 

The following statement could cause some forum controversy so please readers treat it purely as my opinion.

 

Therefore if you follow this concept, it is not necessary to install suppressors on the ends of every T Bus. A couple of suspressors strategically placed within each DCC Power Domain should provide adequate voltage spike protection across all Buses located within that domain. In essence the suppressor should be located as close as is feasible to the problem area where shorts are likely to occur.

[RDS]

Why not install the Bus in one piece or just join the ends of the T Buses to make a ring Bus that starts at the elink and returns to the elink.  In fact thinking of it like that, would a terminator be needed at all.

I would welcome an input from Hornby Technical regarding the use of Terminators.

[Greynut]

Cheers Chris. It was reading the 'PJR_slo' response (and the conversations) in the sidings thread that prompted me to ask - I thought I'd better start a new thread on it though.

 

Ignoring the current draw aspect, I have read in different places on the web that the number of filters did not appear to adversly affect the overall performance. In fact, some advice seems to be to whack them in pretty much anywhere. Well ... nobody mentioned that it might not be wise to have too many. I have to admit, I mulled that one a bit myself but I haven't got the technical understanding to know if this is - or isn't - the case. I'm all for experimenting but it would be handy to know so that I could judge better if things don't seem right at some stage.

 

In answering your rhetorical question (lol) - I just think that using a separate bus for each level will be better than chasing a bus up the spiral - as it makes it more modular and I think that will help to keep tabs on things if something happens or needs doing. And I am a bit wary of having very long bus lengths.

[Chrissaf]

Greynut.

Whilst you were posting your last reply. I was performing a major edit of my earlier one. Please review my revised post again as it may be significantly different to what you last read.

Chris

[PJR_slo]

OK, if you increase the number of terminators, you are increasing the capacitance.  If you inserted 10 terminators, the capacitance would be 1 microFarad or about 20 Ohms impedance plus a 10 Ohm resistance = 30 Ohms total.

 

Theory says this would not degrade the relatively low frequency of DCC, but in practice a difference may be there. Whether you would notice it is another matter (personally I doubt it), but I would much rather use that half an amp for other things.

On the 2 layouts I now have, I used 1000 Ohm resistors to terminate the T on one (the bus is about 6 metres in each direction) and have used a ring (about 8 metres with spurs) on the other, without terminators.

So far, I have had no issues with either.

 

I personally think it more likely that accessory decoders running off the main track bus are more likely to put spikes onto the bus.  I use the Tawcraft decoders on a separate 18VDC bus with instructions only from the DCC bus.

 

Peter

[Greynut]

Thanks RDS. I'm doing it as separate busses as if I did a single bus, I would end up with around 130 foot of bus length - as I said in my previous bit - I think that making it modular is, perhaps, a better way.

 

I've been put off making it (or them) a ring bus because there appears to be quite a bit of thought saying that ring busses can be a bit iffy - especially if they are quite long. A ring bus would be appealing though, initially, that was the way I was going to go.

[Chrissaf]

Why not install the Bus in one piece or just join the ends of the T Buses to make a ring Bus that starts at the elink and returns to the elink.  In fact thinking of it like that, would a terminator be needed at all.

I would welcome an input from Hornby Technical regarding the use of Terminators.

 

General consensus by all the theorists seems to be that a DCC Bus in a Ring is not a good idea. Have a re-read of my (latest) re-written response to "How many Bus Terminations is OK" as it covers the misconception about a suppressor being a terminator. So in answer to your question "In fact thinking of it like that, would a terminator be needed at all", yes a suppressor is still potentially beneficial even if a Ring Bus was deployed.

[RDS]

Hi Greynut

I was just thinking about it in terms of a Household Ring Main for sockets, rather than say a Spur from the Consumer Unit.

It is made as a Bus as far as I am aware, to spread the current load both ways and therefore when installed as a Circular Bus on a layout, it would reduce volt drop along the length.

[Chrissaf]

OK, if you increase the number of terminators, you are increasing the capacitance.  If you inserted 10 terminators, the capacitance would be 1 microFarad or about 20 Ohms impedance plus a 10 Ohm resistance = 30 Ohms total.

 Theory says this would not degrade the relatively low frequency of DCC, but in practice a difference may be there. Whether you would notice it is another matter (personally I doubt it), but I would much rather use that half an amp for other things.

 

Peter, I put your values of 1uF and 10 ohms into my RC Filter calculator and it gave a cutoff frequency of 15.9Khz, so your theory is probably not far off.

 

[Greynut]

I seem to be a bit behind the postings every time I add something - sorry lads!!

 

But ...

 

Chris - thank you for that edit. Would that answer my question in that you can feed more than one bus through a terminator. I know that you say to put a filter near to a problem area (if known) - but could the left hand 'T' ends be filtered together through one filter and the right hand ones through another, for example? I know that I'm still advocating the ends - but with a 'T' shape, it seems logical. As for each power domain, are you saying that a filter put somewhere in one of the 'T' busses would handle the other 'T' busses as they have a common power source - i.e. they all connect to the same elink output? Then another filter to handle the other (connected) busses if a booster was used?

 

Thanks for your reply Peter. I agree, losing half an amp is quite a big price to pay for modulated DCC and that was one of the reasons I posted the question - to see whether there was a largish power cost in doing it. Plus, if there was - would there be a way to cut down on filters required. In a previous thread that I posted, I did ask the question whether things like point motors etc could give rise to spikes - so I reckon that we might all be thinking along similar lines.

 

RDS, a domestic ring main is a good way of looking at the way it could work - to counter voltage drop - but some of the advice that I have looked at points to potential interference of the DCC signal as being likely in this setup, especially over a long run. So I'm not sure on that one.

[Chrissaf]

Greynut.

 

 

Reading Peter's well theorised replies then it would appear that if you can stomach the current drain then there shouldn't be any major issues with having 10 suppressors. Personally, I am still of the opinion that "less is more" just to be on the safe side.

 

 

Peter made a good point about protecting the Accessory Decoders that may be wired permanently to the Bus. I myself had an R8247 that was damaged by a short. Hence why I added a couple of suppressors after the fact. In my case I split my R8247's into two groups and attached them to the ends of my T with an associated suppressor to protect them. So in my case, my fixed decoders and an associated suppressor were co-located at the T ends.

 

 

In your case, having say 5 x T's all in one Power Domain connected to a single controller, your Bus is technically a 'Star' rather than a T. So if you want to keep the number of suppressors to a minimum then, although not potentially ideal, it makes sense to centralise a suppressor to protect all legs of the star equally. This would put it near your DCC controller. Another suppressor could then go near where your Accessory Decoders (if you use them) are located. This I believe would give some good basic voltage spike protection without going overboard.

 

 

If you felt you needed to include terminators (as well as a central suppressor) on the ends of your open ended 'Star' Bus legs, then Peter's solution of using 1,000 ohm resistors is a good compromise. I have seen the use of 1,000 ohm DCC Bus termination resistors documented elsewhere on the net (but can't remember where).

 

 

Note: If you link together all the left hand T's together and all the right hand T's together and connect each bundle of T's to a single suppressor as you suggest, then the Bus stops being a 'Star' and takes on the appearance more of a 'flower petal' i.e a series of rings sharing two common points on their circumferences. Mark Gurries pages, to the best of my recollection, advocates a 'Star' configuration over a 'Ring' or 'Rings' as being preferred. As I believe you do too based on some of your other reply comments.

 

 

Canvas other people's suggestions and take a balanced view.

 

 

Chris

 

 

PS - Can we start calling them Suppressors not Terminators !!

 

EDIT:

In answer to your other question:

.

"As for each power domain, are you saying that a filter put somewhere in one of the 'T' busses would handle the other 'T' busses as they have a common power source - i.e. they all connect to the same elink output? Then another filter to handle the other (connected) busses if a booster was used?"

.

Yes & Yes

 

[PJR_slo]

When I used to install 10-baseT networks, it was essential it include terminators at each end (mainly to prevent echo), but occasionally, they formed a ring and terminators were then not used.

Now I am not an expert, but I would think that terminators were required due to the higher data transmission speeds and lower signal levels.

It is my opinion that terminators on a DCC bus are something of a luxury - 8kHz, 32V peak to peak, 1 to 4 Amps being carried aroung a decent size piece of nickel-silver (plus the occasional power bus).  The speed at which electricity travels is hardly going to cause echos at 8kHz even on a very large layout (I did once do the math to confirm).

The suppression of unwanted spikes is rather more essential, but surely the best method of removing spikes is to get rid of the source.

A short should not cause a spike; however, electric motors and switches can.  Most electric motors (in the locos) are sufficiently isolated (by the decoder) to prevent spikes on the track.

Chris, my issue with lower quality accessory decoders is that it may be possible to pass spikes back to the track from point motors - spikes that are generated as the magnetic field within switch the coils collapses - in a similar manner to which automobile coils generated sparks to fire the engine.

Hence the accessory decoders I use are opto-isolated and separately powered.  And no requirement for terminators or suppressors.

 

Peter

[Greynut]

Chris, I see your point about connecting several bus ends to one power outlet, in reality forming a star rather than several 'T' busses. My thinking here was that at the elink power outlet, the DCC signal is moderated by the elink itself. After I put the bit about connecting several bus ends to a suppressor, I realised that I would have formed a collection of interconnected rings which might not be a good idea.

 

Peter, I'm using DCCConcepts ADS8 decoders powered direct from the bus - and I do have a 4 way Tawcroft that I'll be using at some stage. So suppressors seem just as important for safegaurding these too.

Suppressors could certainly be overkill but shorts do happen and something that costs less than a quid to make and which could potentially minimise the effects seems to be a good thing. I'm rather interested in your bit about the collapsing field of a solenoid causing a spike - I asked a similar question a little while ago - it's got 'Can Solenoid Point Motors Affect Dcc Bus' as its subject title if you're interested.

I still haven't managed to get things up and running to see if I could detect anything like that happening though.

[PJR_slo]

I'm rather interested in your bit about the collapsing field of a solenoid causing a spike - I asked a similar question a little while ago - it's got 'Can Solenoid Point Motors Affect Dcc Bus' as its subject title if you're interested.

Without looking at the responses on your other thread (will check that out later to see if my thoughts are the same as others...), I think the answer to the question is yes, in 2 different ways, but depenadant on the design of the accessory decoder.

The Tawcraft decoders are powered via a separate (18V-22V) DC supply and that power supply, the output and the point motors are optically isolated from the DCC rails, which are connected to the decoder to simply supply the instructions.  My thoughts below do not apply to this type of decoder, only those which are both controlled and powered from the DCC Rail.

 

1.  Both during and after switching a point the accessory decoder needs power for the point motor or to recharge the CDU.  This power has to come from the track which is fed either 3Amps or 1Amp depending upon the Hornby PSU.  If there are several locos running at the time and maybe other items (coach lighting etc), the current required by the accessory decoder may cause a sudden voltage drop on the tracks  This could have knock-on consequences.

 

2.  As mentioned previously, any back EMF from the coil in the points motor - from both coils, not just the "live" one can only go back to the decoder.  If the decoder does not have sufficient protective circuitry, then it can also be fed back to the DCC bus.

 

The way an automotive coil is made is extremely similar to points motors with a common central connector and a coil on each side of this.

 

On my larger payout, I have run 6 locos with sound and light (+ 2 with smoke) at the same time, with 3 others on the tracks with sound turned on.  I have not experienced any problems with spikes or decoders losing settings.  Everything just works.  However, I have only run this with RM + Elite and 4A PSU.  I have been unable to get the eLink working on the Win 8.1 PC.

 

Peter

[Chrissaf]

Peter,

 

I take on board your comments about 10BaseT networks and the need for terminators in that scenario. The only reason I included all the detail about infinite line termination theory was to try and describe the difference between the function of a Terminator and a Suppressor. It is true what you say that the need for impedance termination is a function of frequency and distance but even speech frequency (300 - 3,400Hz) lines need to be terminated correctly, albeit the kind of speech lines I'm thinking about are several Km long and not just a few feet. With regard to shorts not causing spikes. There is a lot of documented theory on the net as to why shorts do cause voltage spikes on the DCC Bus. But I do concur wholeheartedly that electric motors and switches that are not spark quenched are a key source of electrical interference. I'm not so convinced though that back emfs caused by magnetic fields collapsing in solenoids is a major issue.

 

If it's of interest, where you used to install 10BaseT networks, my job in later years was to design them including 100/1000BaseT UTP cabling schemes associated with fibre backbones for corporate businesses. I also designed wide area TCP/IP Router networks using predominantly Cisco kit to connect it all together to make it work 'end to end'. Was glad to take an early retirement package in 2005 and get out of that business, the pace of technology change was very stressful.

 

Chris

 

PS - with regard to the other Solenoid post Ray mentions, my input is on there too. Bear in mind when reading it, the original question related to interference occuring through crossover inductance and not by feedback through a decoder.

[Chrissaf]

ADDENDUM as Edit opportunity has expired.

 

Peter, extract from the net:

 

NMRA did a study: "A Case for Using High Frequency Filters on DCC Layouts"Page 17 of the Feb. 2008 "Scale Rails", a publication of the NMRA, written by Didrik A. Voss, MMR.

 

Don't know what the conclusion of the study was, just that one took place. The title infers that there is a case (presumably positive) for using Suppressors (Filters).

 

 

The full web article can be read here:

https://sites.google.com/site/markgurries/home/dcc-general-best-practices/wiring-planing/snubbers-rc-filter

 

 

Chris

[PJR_slo]

Peter,

 

I take on board your comments about 10BaseT networks and the need for terminators in that scenario. The only reason I included all the detail about infinite line termination theory was to try and describe the difference between the function of a Terminator and a Suppressor. It is true what you say that the need for impedance termination is a function of frequency and distance but even speech frequency (300 - 3,400Hz) lines need to be terminated correctly, albeit the kind of speech lines I'm thinking about are several Km long and not just a few feet. With regard to shorts not causing spikes. There is a lot of documented theory on the net as to why shorts do cause voltage spikes on the DCC Bus. But I do concur wholeheartedly that electric motors and switches that are not spark quenched are a key source of electrical interference. I'm not so convinced though that back emfs caused by magnetic fields collapsing in solenoids is a major issue.

 

If it's of interest, where you used to install 10BaseT networks, my job in later years was to design them including 100/1000BaseT UTP cabling schemes associated with fibre backbones for corporate businesses. I also designed wide area TCP/IP Router networks using predominantly Cisco kit to connect it all together to make it work 'end to end'. Was glad to take an early retirement package in 2005 and get out of that business, the pace of technology change was very stressful.

 

Chris

 

PS - with regard to the other Solenoid post Ray mentions, my input is on there too. Bear in mind when reading it, the original question related to interference occuring through crossover inductance and not by feedback through a decoder.

Hi Chris

I do not have any issues with anything you have said, merely adding my thoughts to your excellent text.

 

As for me; I have been working with computers for almost 40 years now, but mainly on the software/development side of the business.  Having said that I have been involved in the design, build and install of things as well.  Currently using Cisco routers at the moment, and they are a real pain in the proverbial!  Cisco firmware faults do make life difficult.

 

Peter

[Chrissaf]

Peter,

 

 

I was fortunate I only had to work with designing the networks and NOT implementing them (no hands on so to speak - but not being hands on had its own issues with regard to learning about what can and can't be done technically). So if you have to configure the little devils (Cisco) as well, I take my hat off to you.

 

 

While I was away from this page I was reviewing what I wrote in the Solenoid thread mentioned above. I realise now in hindsight that I made an error. I stated in one of the replies that by reducing the Suppressor resistor below 100 ohms the pass through frequency lowers. I should have said it is raised. I've just run a couple of value combinations through my RC Filter calculator to check. Of course, what a lower value does do is increase the current drain on the DCC Bus. I mention it here in case you should spot my error, if and, when you read my input on that particular thread. The error doesn't actually impact on the gist of the thread from Ray's point of view as it was a supplementary comment.

 

Best regards

 

Chris

 

PS - One of my other job related tasks, was to write the technical input into the responses to 'Invitations to Tender'. Once you do that for a few years it's not surprising that one gets a knack for creating good prose.

 

PPS - I see in the bottom header of your replies that your layout is 6 x 4 METRES. That must be impressive. I wish I had that amount of space.

 

Any chance of posting a couple of photos here. Would be greatly interested to see it.

[Chrissaf]

Peter,

 

Just another comment regarding shorts and spikes. You are quite right, in a DC environment a short shouldn't cause a spike. If I understand what I have read elsewhere correctly the theory of shorts creating voltage spikes on DCC goes something like this:

 

The DCC signal is nominally 8Khz as you say. When a transient short occurs the current builds up (until it is sufficiently high enough to trip the controller protection). In the meantime prior to the protective trip the 8Khz signal starts to ring (a high voltage peak starts to form on the leading edge of the digital signal bit). Just as an aside, I know when I have created an accidental short on my layout as I can hear a distinct high frequency AC buzzing that lasts for about (a subjective estimate) 0.25 of a second, this was of course prior to me fitting my suppressors. After fitting my suppressors any short I had was virtually silent. In some circumstances this leading edge ring can achieve (according to what I have read) voltages that are anything up to three or four times the nominal DCC voltage. These voltages can then exceed the component voltage tolerances of some cheaper decoders with components that are engineered down to a price, hence causing failures that may be total or intermittent.

 

From the link posted in the ADDENDUM post above, there are further links that take you to some photos of Oscilloscope displays showing the effect of ringing on DCCsignals that may interest you.

 

Chris

[Greynut]

Peter, as you've probably guessed my layout is work in progress, interrupted by work in reverse, before work reprogressing in a different way. But I have had 4 locos running at the same time and points have been thrown via DCC control while they chugged around. I didn't notice any change or stutter from the locos when this happened but that's not to say that there wasn't a brief drop in power. The thing with most decoders is that their current draw and instruction come from the bus or track - it would be good if they all worked with option to split these functions - instructions through the bus ... current draw from an independant source. Maybe a relay arrangement - but even then the accessory decoder will be drawing current at some stage.

The back EMF is a problem which surely should be taken ito account by the decoder manufacturers - or am I being a tad naive on that.

 

Luckily, I'm still in the dark ages with Windows 7 - mind you, the elink and RailMaster can have odd moments of bafflement with that version as well.

 

Chris, I think I've heard a buzz like that on a few occasions, which I put down to a spark or suchlike generated and coming from the site of the short. So what you are saying is that it's the audible product of the 'ring' going through the bus. If the voltage can surge like that through a quarter of a second, it's not surprising decoder chips get deep fried. It can't do the controller much good either.

[Chrissaf]

If there is an audible buzz associated with, and at the location of, a short then there must be a spark making it. In this case a high frequency AC spark (the DCC AC signal). Think of an electric arc welder (an exaggerated example I know). The ringing probably intensifies what is heard. Hence with my suppressors fitted, my subsequent shorts have been less intense and virtually silent by comparison.

[PJR_slo]

Hi Greynut

My business workstations are all Win 7 (except those used for development testing); it is the better O/S for business.

I treated myself to an i7 Win8.1 (SSD) with 24" touchscreen for RM only. 

 

Chris - Many thanks for the info.  We take the layouts to exhibitions this is the larger one: 

and this is the latest smaller on (which is currently in the lounge): 

 

The larger one was designed to fit in one of the rooms in the converted loft area.  We are very fortunate in that our house has 4 floors, each with about 120 square metres.  When we converted the loft, the room we now use for the railway was used as a cinema with a 3 metre screen.  It then became the kid's games room and now the railway room.

 

Peter

[Chrissaf]

Peter,

 

Wow !!!

 

Loved the background automation. Can not even begin to think what was involved in automating the road traffic. Assume some form of motorised track under the baseboard using magnetism to grab the vehicles, under microprocessor control perhaps.

 

I think I even spotted a Seagull circling over the allotments.

 

One question. The themes seems to have a Scandinavian feel. Is that where you a based?

[Greynut]

Hi Peter,

 I go with Chris in saying "Wow"!! Both layouts are brilliant. It's good to see that you are using your rooms properly - i.e. - putting in a huge layout - I'll point that out to my dear wife!!

 

The automation is astounding, including the bird (sea gull) ... a great touch. And the stopping and starting traffic as well, it must have taken ages to plan, let alone to build it all. I am totally impressed and lost in admiration for what you have created.

 

For all the automation that you have running, I'm not sure that a Windows based machine (7 or 8.1) is the thing ... more like a super computer, I reckon.

 

Again - very well done and very many thanks for letting us have a look at it all.