ABC Braking

[WIMorrison]

Is there also a chance that these home made BM1s that you have (I think that they are all home made?) have been built to an incorrect circuit diagram and that is the root of your issues?

 

I would look at that before saying ESU or any other chip maker has made a mistake

[Ray H]

Is there also a chance that these home made BM1s that you have (I think that they are all home made?) have been built to an incorrect circuit diagram and that is the root of your issues?

 

I would look at that before saying ESU or any other chip maker has made a mistake

 

I'm about to go and dissect one.

 

I wasn't suggesting that ESU had made a mistake, I was wondering if unlike Zimo decoders which as I understand it will pass through the Brake Zone if they've not be configured to use it, ESU decoders default is to stop when there is a peak voltage variation between left and right running rails.

 

I haven't had a chance to check the ESU manual yet.

[Ray H]

I'm about to go and dissect one.

I didn't need to dissect one, all I had to do was to convert the stripboard layout to a circuit design and I discovered that I had been using yesterday's one back to front. That'll teach me to do a more thorough check prior to installation!

 

I tried an v3.5 ESU decoder equipped (OO) loco with the BM1 lookalike at home. There was no stopping as it entered the Brake Zone with the BM1 in either orientation. I need to check CV settings on the (O gauge) test loco this evening (and with the BM1 lookalike correctly orientated) but it does appear at first glance that my installation was the wrong way round rather than the ESU equipped loco failing to respond correctly.

 

I used an (OO) Zimo fitted loco for testing here last evening. It took a few attempts to get the loco to stop within the distance available - just under 2½ feet. I repeated the same test three times with throttle speed settings between 2 and 20 in increments of 2.

 

The extremes in stopping positions across the various speed settings on the first series of runs was over 10cms. The variations in stopping positions for each of the three same speed settings was generally less. I never managed to get the same stopping position (at the same speed setting) for all three runs and there were only a couple of times (at the same speed setting) when the stopping position was the same.

 

Save for tweaks to keep the loco within the stopping space available - the tests at home are on a dead end siding - the above series of similar tests were the first and without any attempt to tweak CV values further to improve stopping accuracy. That's today's task.

[Ray H]

I've spent a while this afternoon juggling around with CV settings and noting results.

 

No matter what CV I tweaked nor what value range was used, I consistently failed to get two consecutive "runs" generate the same result.

 

Each test run was over the same track section with the start location of each (and the distance between start and commencement of the Brake Zone not varying by more than 1cm). All test runs had the throttle set to 12 of 28 steps and used the same loco. Only one CV value was altered between each test run and the previous CV's value was not restored to its initial value.

 

CV 134 values of 6, 106 & 206 have been tried. There was no significant difference with any of the settings, the extreme values differing by just 3cms.

 

CV 142 values of 5, 18 & 24 were used. Again the extremes differed by a maximum of 4cms but the stopping distance increased by approximately 7cms with the value of 5 and decreased by a similar amount when the CV value was 24.

 

A value of 11 for CV 140  (with CV 142 = 24) saw an extreme variation of 6.5cms with the shortest stopping distance of 39.5cms differing from the 45cms stopping distance with the initial CV 142 setting (at 24) whilst CV 140 had been just 1.

 

Adjusting CV 141 to 11 and then 12  saw the extremes (at 11) of 5cms and 6cms (with CV 141 = 12). However that result ignores "a rogue had to be physically stopped" at 76.5cms on the second of four test runs. The abnormal run's result was 19.5cms greater than the next highest. 

 

Although we aim to let the trains stop (and start) themselves I did experiment with setting the throttle speed to 0 as soon as all wheels with pick-ups were on the Brake Zone. Again there was a stopping distance variation of 3cms over three runs.

 

I think that I have exhausted the tests that I can conduct at home.

 

We hope to get some testing time at the club either this evening or tomorrow afternoon on the target O gauge layout. The aim of those tests will be to see whether the extremes increase in value over and above what I found here on my OO layout and then look to use the CV values that consistently cause the trains to (a) stop over the sensor and (b) not pass an adverse signal - the end of the Brake Zone is a few centimetres beyond each signal.

[Ray H]

We've finally got to the stage of setting up a couple of O gauge locos at the club. Initially we went for CV 27 =1, CV 134 =106 , CV 142 = 12, CV 140 = 1 & started off with CV 141 at 10 - we didn't alter (or check) CV 157.

 

CV 141 at 10 didn't appear to cause the loco to stop soon enough on the test (dead end) siding so we dropped the value to 6. We experimented with speed values, starting with 4 and ending with 8 although we have decided that this wants tailoring per loco/train combination.

 

We've currently got 2 locos working (although their respective trains may change in due course). A third loco refused to play ball when we factory reset the decoder after it refused to even recognise the number we set it at a couple of weeks ago. Closer examination revealed a few mechanical problems so that loco has returned to works.

 

I had the task of setting the trains moving as we were also testing the behaviour of our sensors so I wasn't able to accurately check each train stop but I have to admit that from where I was standing it did look as though the same train was stopping in roughly the same places each time round. No SPADs were noticed!

 

We still have one tweak to do to the sensors and at least one further loco to set up for ABC braking but it does look as though if we get our (layout/censor) wiring correct the system will do what we want.

 

I'll be back when we've had a bit more of an opportunity to test the system more fully.

[Ray H]

We now have four locos to choose from including an aged Heljan Class 37 that's fitted with the large non-sound Zimo decoder that has a current peak of 2.5 amps.

 

We do get some strange occurrences.

 

We have set the brake zone on our test siding to use the same rail as all the others. However, we are finding that from time to time setting CV 27 to 1 has no effect so we change the value to 2 and the loco slows to a stop as required. This is the cue to move the loco on to the main part of the layout - once we're happy with (all) the CV settings.

 

This has led to a couple of instances where the loco fails to slow in the brake zone and unless we're quick it takes a liking to the rear of the train in front!

 

We change CV 27 back to 1 and all is in order.

 

Other than that all seems to be well.

 

Thanks to everyone for their assistance in allowing us to achieve a usable system. I'm now going to try to incorporate a similar feature on my home layout where I have a two road fiddle yard that is used by just two items of motive power.

 

Trains are always driven towards the "controlling" operator. Said operator can move trains out of the fiddle yard as normal. Adding brake zones will allow the same operator to return the trains from whence they came without having to monitor the rest of their movement. This will avoid the need to have another operator to drive trains towards the fiddle yard.

[Ray H]

The home layout now sports three home made "BM1s". The original "BM1" installation was at the end of a single siding at one "open" end of a U shaped layout. The more recently installed "BM1s" are at the ends of two adjacent sidings at the opposite "open" end of the layout.

 

The layout has three power districts with power to each district's track power bus taken from an NCE SB5 that is located towards the centre of the layout. All three "BM1s" are therefore at the ends of their respective power districts furthest from the SB5.

 

The "BM1s" are all fitted to the same rail, the rail nearest to the inside of the "U" shape.

 

It took a while to realise that despite the "BM1s" being fitted to the same rail of the layout, their orientation as far as the origins of the power district's track power supply is as if they were fitted to opposing rails. Consequently, usage of the original "BM1" requires CV 27 set to 2, whereas this CV's value is set to 1 for the two newer "BM1s".

 

In truth, since I've fitted the two new (and adjacent) "BM1s" at the opposite end of the layout the original "BM1" seems to have ceased to work but as it was only originally there for testing the theory this isn't currently a problem. The end of this siding is only used by a single (sound decoder fitted) Coal Tank.

 

One of the sidings with the newly installed "BM1" is used solely by a Park Royal railbus. The other siding is used solely by a Class 03 diesel. Both have Zimo sound decoder fitted. There are instances when the adjacent siding is occupied and instances when the other siding is empty when a "train" arrives.

 

The railbus's speed has to be kept at or below the 11^th step of the 28 step setting otherwise it ignores the ABC braking CV settings and only its progress is only halted (but with its wheels still turning) when it reaches the physical barrier at the end of the siding track. The Class 03 which is much slower at maximum speed will stop as required regardless of the throttle's speed step setting. 

 

On more than one occasion when the railbus has continued to the physical barrier (and still has its wheels turning), the Class 03 that had previously been happily sitting quietly at its stopping point having recently arrived in its siding, suddenly decides to move towards the siding's physical barrier.

 

It would be nice to understand what might be causing these oddities.

[Nigelcliffe]

Ray,

I'm not clear what is connected where, so this may not reflect what you have...

 

1 - orientation.  The Braking module (BM1) is directional, as you've found out.  By reversing the wires through it, the direction of bias applied to the DCC signal is reversed.  As the ABC brake is usually set to respond to a bias in one direction of the DCC signal, that explains why you have to reverse the CV setting.   Much better to fit the brake module (BM1) devices consistently !

 

2 - trains restarting.  Is there one Brake module (BM1) used in common for these locos, or are they on independent Brake modules (BM1) ?   If common, the reason is the stationary loco briefly sees a "no braking" signal as the wheels of the other loco crosses into the brake zone.    If independent, and possibly related to the behaviour of the railbus, may be the sensitivity setting in the decoder CV's, or could be that your diode choices for the BM1 are not ideal (below).

 

A DIY BM1 module needs fast-recovery diodes, and ideally a Schottky diode for the single return diode.  Number of diodes is "enough to make it work" but can vary with different diodes. The fast recovery is essential to get the DCC signal through, and the Schottky means there is negligible reduction of the DCC signal on the "not biased" half of the DCC waveform. 

[Ray H]

 

Does this (rough & ready) drawing help? I've only shown the relevant bits. The different colours represent the three power districts. The SB5's position is representative of its position relative to the three power districts.

 

Each of the BM1 inputs are connected (directly) to the relevant power district's track bus although for simplicity in the diagram they are shown connected to the rail (which is connected to the bus). The BM1 outputs are connected direct to the appropriate siding rail.

 

There are eight other sidings - not fitted with BM1s - on the left hand (red) side of the layout although I've only shown an additional two.

 

We've used UF5404CY for the single diode and 1N4001 for the multiples. As mentioned previously the boards were either built by a colleague or in one instance is a copy of the ones that he built using the surplus components he had left from his order. I believe that he got the details from one of the numerous Internet threads about the homemade BM1s.

 

The "red" section BM1 is the one the Coal tank wasn't too happy with yesterday but wasn't subjected to exhaustive testing to establish its grievance. The railbus uses the right hand of the two "black" section BM1s, the Class 03 the other.

Edited August 14, 2018 by Ray H

[Nigelcliffe]

1N4001 is a standard recovery diode, so might be part of the problem.    

UF5404 is a fast recovery (good), but also a standard voltage drop.

 

If there are parts around, I suggest making one entirely from UF5404's and see if that works.  Ideally the single diode would be a Schottky, but a standard fast recovery will be OK, just means a bit more voltage drop and potentially one more diode in the other half of the circuit (to get sufficient asymmetry, whereas with negligible drop on the single diode, the same asymmetry can be had with one diode fewer). 

[newbryford]

I made a few up using UF5404 for all the diodes - with 4 in series and then one across them in inverse parallel.

 

Seems to work well.

 

Cheers,

Mick

[Ray H]

Thanks Mick (and Nigel too).

 

In for a penny, in for a pound as they say and hang the expense! I've ordered some 1N5817s and some UF4002s.

 

The current homemade jobs that I'm using could do with being smaller and the use of different components will provide an excuse to re-design them. It will also allow me to evaluate both designs for whilst we've not had similar problems on the club's 7mm layout to date, it might be worth changing them as well to avoid in future problems - a 7mm 2-6-0 and 15 wagons rear-ending the previous train is not a pretty sight.

 

Interestingly today I ran the railbus into its siding at speed step 8 (of 28) and it failed to stop so I manually pushed it back and off the brake zone by a couple of inches and it moved forward again and stopped where it should have the first time.

[Ray H]

The two types of diode were ordered from different (auction site) retailers and both have just arrived despite neither being expected before the weekend.

 

Guess who is off to build a more compact "BM1" and try it out?

[Ray H]

An interesting afternoon and thank goodness for the physical barrier at the end of the siding used by the railbus, which is the siding that I've been testing on as the railbus has a top speed far in excess of the Class 03 which. so far - fingers crossed, has yet to fail to stop.

 

Both sidings are about 250mm long. There hasn't been any repetition of the Class 03 moving along the siding of its own accord and said Class 03 has generally been in its siding throughout today's tests.

 

Just to recap, the new "BM1" has been constructed on stripboard using 1 x 1N5817 plus 4 x UF4002 diodes (plus a screw [Molex?] type connector).

 

There have been numerous test runs during the afternoon with numerous slight tweaking of the CV values which began at CV 27 = 2, CV 134 =106 , CV 140 = 1, CV 141 = 6 & CV 142 = 12.

 

Once again and from the very start, any speed above 8 or 9 (of 28) caused a consistent over-run. Changing CV 140 to 11 didn't noticeably change anything. I even tried several low values for CV 141 to no apparent (reliable) avail. The odd adjusted value did allow me to go as high as speed step 9 but nothing higher. I did try 128 step runs but anything above about 50 (out of 128) produced an over-run.

 

On some of the faster test runs I found that I could manually push the railbus back (until it was a few inches) onto the main part of the layout and let it re-enter the siding and it did then stop before getting to the end of the siding. However, this wasn't guaranteed to happen - and its not something you want to have to do on a routine basis.

 

So I started playing around with CV 134 gradually dropping the value by 1 each time. This started to show results and currently, with testing finished for the day, I've got CV 134 set to 102 and the railbus stops before the end of the siding in the majority of runs. However, its speed doesn't seem to alter until it gets part way along the siding and then the (emergency) brake seems to apply.

 

I have no intention of running the railbus at warp speed myself but I'd like to be sure that anything between the railbus and the end of the siding (but beyond the desired stopping area) wouldn't end up in a heap if a guest operator wound things up too high.

[Parax]

On 01/07/2018 at 22:11, Junctionmad said:

I did some similar tests a while ago and Im trying to recall exactly what we found 

 

Firstly decoders can take some time to reliably detect the Asymmetric DCC signal , and if the pickup is mediocre or the track is bad , Then some decoders can take 1-2 seconds, Zimo has a CV That controls the recognition time, but as you make the recognition faster , you introduce some uncertainty into exactly when the decoder will recognise the Asymmetric Signal. This uncertainty translates into " jitter " in the stopping distance and is worse as the loco is going faster , hence sometimes programming in a long recognition time is better.

 

Secondly the issue of exactly when the loco is " inside " the ABC section can be variable , depending on how many wheels are picking up and exactly how, you could find that the exact point the decoder decides to activate ABC is a variable event , This is not helped by the fact that  any full DCC that then gets transmitted ( because say that wheel suddenly conducts ) , then resets the ABC system . Again this results in a variable stopping distance 

 

In the second case ( i.e. pickup and shorting  issues ) The effect is more pronounced the slower the loco is going through the transition section 

 

The Zimo decoder is capable of millimetre stops , but all this assumes that everything else is perfect , which it rarely is 

 

This is why I said ABC is a kluge , The system gets even more kludgy when you add a train of metal wheels, DCC coach lighting , etc as all these can confuse the exact point at which the decoder implements ( or keeps implementing ) ABC 

 

 

The reality is the only way to reliably trigger ABC is to detect the train in the section absolutely reliably ( light beams , magnets , current sensing  , and then switch in the ABC signal ( in reality a few diodes ) , you may need to have very long ABC sections to do this reliably as in reality , the whole train needs to be inside the braking section , before  the ABC signal is triggered.  One way to do this is to have TOTI sensing ( train on track ) that then switches the whole section under the complete train to ABC.

 

....

 

My own view is that if you want the simplicity of ABC, then you will have to except that you will get a degreee of unreliability in the exact stopping position due to a number of factors , not necessarily the decoder either . Anyway the prototype didnt stop in exactly the same place either !

 

The sad fact , is that " local Automation " , i.e. where you dont have a whole layout computer controlled , is very tricky to do with DCC , The system has huge flaws that were never addressed by NMRA ,particularly the issue of bidirectional communications.

 

....

 

dave

 

This is all true.. but ABC is a workable option when all the issues are addressed correctly..

 

I have fixed the many issues (metal wheels the long stopping distances and crucially push pull trains!) with my own home built modules.. I really wish that at least one manufacturer would make something that works properly like this..

 

I use on track detection in two sections, an approach and a stop block, but crucially if you want to stop at a red you've got to slow down at the amber first!  (see the full speed push into red light test for unrealistic emergency stopping!) 

The stop section I have in the demo is 45cm and the train is set to stop from the slow approach speed to stop in 35cm, (works with about 50mm precision.) The Approach section does need to be long enough to hold the whole train. (So it is simply from the previous signal..)

my modules are switched from the same DCC address that changes the lights..

(in the video below the left rail is continuous, the right rail is split into 2 sections per signal and there is a right rail bus wire along side)

 

Edited April 2, 2020 by Parax

[RAF96]

The Lenz page on ADCC tells that when using their BM-(any) modules only the right rail should be gapped - right being in the direction of travel.

[Parax]

34 minutes ago, RAF96 said:

The Lenz page on ADCC tells that when using their BM-(any) modules only the right rail should be gapped - right being in the direction of travel.

Yes this is correct, Right Rail with forward Diodes.. in order to use the forward only Braking mode. 

All the retail modules use forward diodes and must be installed in the correct direction on the right rail.

 

But if your module has Reverse diodes it will need to be on the Left rail to get the same forward only braking mode to work.

So Right Rail with forward diodes = Left Rail with reverse diodes. 

 

If you are not using Directional Braking modes and are just braking on any asymmetry then you can have any of the four combinations.

Right Rail Forward Diodes

Left Rail Reverse Diodes

Right Rail Reverse Diodes

Left Rail Forward Diodes

 

 

 

 

[Ray H]

I've seen a section of a circuit diagram that incorporates the ABC braking module but that uses a 3.3v Zener diode in lieu of the 4 series diodes.

 

I want to add the braking module to a Arduino based "device" and a Zener would require less space than four diodes.

 

Is a zener based build a viable option?

[Crosland]

8 minutes ago, Ray H said:

I've seen a section of a circuit diagram that incorporates the ABC braking module but that uses a 3.3v Zener diode in lieu of the 4 series diodes.

 

I want to add the braking module to a Arduino based "device" and a Zener would require less space than four diodes.

 

Is a zener based build a viable option?

 

No, not really, is the short answer.

 

Normal diodes are used for their forward voltage drop, hence you need a few in series for the decoder to detect the voltage difference.

 

Zeners drop voltage due to the reverse breakdown voltage and you would connect them the other way around.

 

The problem is that Zeners (or any diode in reverse breakdown) can handle A LOT less current than even the cheapest 1N4001 diode (which also have a much higher reverse breakdown voltage). By the time you have sourced a power Zener it will probably be as large as, and way more expensive than, a few switching diodes.

 

I wonder if they tested the response of their zener when there was a short circuit in the braking zone.

[Ray H]

Thanks for the speedy response.

 

I had in the back of my mind that it was probably a non-starter otherwise it would have been a more popular option.