Point Control

[StoneMonkey78]

Hi.  I've been lurking for nearly a year, but this is my first post.  Apologies if this is not the correct sub-forum for my questions, but I wasn't sure.

 

Background. I'm returning to the hobby after an enforced absence of nearly 20 years - my, how things have changed!  I sold all my OO scale stuff in the late 90s to switch to N gauge (scale) and bought a CVP EasyDCC system to control the proposed layout.  Having dug it out, it's working fine, and I'm content with controlling locos with the couple of panel-mounted rotary controls, ie I don't want to go down the route of controlling locos from a computer.

 

Layout.  All still in planning with nothing built, but nearly all the track bought (Peco Code 55 on the visible bits/Code 80 for the hidden sidings).  I need the layout to be portable so it will be a series of 10 x 1200mm x 600mm boards.  The proposed layout has 91 points (97 point motors allowing 2 for slips and 3-ways) and 15 semaphores, and whilst I probably won't need to automate them all, I'll plan on worst case.

 

Original Intention.  Back in the day, point motors were invariably solenoids controlled with push buttons on a mimic panel.  When I started to look around at present day solutions, I was impressed with Dave Fenton's MegaPoints Controllers and a shift to servo-based tie-bar movement.  However, having costed a MegaPoints solution at something over £1.3k to put in place, a solution that worked out at something over £12.50 per turnout seemed excessive.  Accordingly I have spent hours looking at other micro-controller solutions, reading hundreds of posts on multiple threads and viewing dozens of YouTube help videos.  However in all this research, whilst I got much useful input, I didn't find a single solution that met both my needs and my ability to understand the solutions being discussed (I am tech-savvy, but so much has changed such that a new language has emerged that I have yet to learn).

 

Computer Control.  The one main issue my research called me to question concerned having a physical mimic panel.  The hours needed to design, construct and wire such a beast, and the lack of flexibility if changes are needed, drew me to the screen-based alternatives.  It is clear there are many programs out there that might meet my requirement to specifically control my points and semaphores, but I stress that I am not looking for a program that controls locos, does DCC programming, does block control with interlocked signals, etc - all I want to do is control the points and semaphores.

 

Proposed Solution.  My research has led to a possible solution that works out significantly cheaper - if it works!.  Nevertheless, as I am dealing with hardware I have never used, and being a Yorkshireman who won't spend a penny if not sure that what is being bought is going to work, I wanted to run my ideas past this community.  I welcome any input, alternative suggestions, highlight any errors, and solutions for the bits I have not yet worked out.

 

Servo-Control. SG90 servos mounted below the baseboard, controlled from an Arduino Mega clone via a 2-wire I2C bus and a PCA9685 16-channel servo controller.

 

Frog Switching. The DCC Bus signals are polarity-switched using a 5V relay switch, triggered via an 8- or 16- pin I/O expansion board, sitting on the I2C bus.  I have identified 3 suitable expansion boards: the PCF8574 and PCF8574A 8-pin boards and the MPC23017 16-pin board.  I am conscious that there is only a finite number of I2C addresses that such boards can utilise, so will use the 16-pin board(s) where required for any individual baseboard, rather than the 8-pin board shown in the diagram.  I am also conscious that there are 16-way relay switches available - I think that the 8-way relays will be easier to mount and wire (inside the baseboard frame rather than under the baseboard surface).

 

Wiring. By mounting the required servos, servo controllers, relays and I/O expansion boards on each baseboard, then (excluding any wiring for lighting circuits) there will only be 6 wires passing from one board to the next: the DCC bus, a 5vDC supply, and the I2C bus.

 

Costs.

 

Arduino Mega2560 clone £7.89 (~0.08p per point)

PCA9685 16-ch servo controller £1.53 (0.10p per point)

SG90 servo (bulk buying gets cost to under £1.00 ea)

MPC23017 16-pin I/O board £1.70 (0.11p per point)

8-ch Relay Module £3.60 (0.45p per point)

Total - £1.74 per point

 

Control.  Now the bit I haven't yet worked out.  I have looked at JMRI Panel Pro; the graphics aren't great, but I can get a touch screen laptop to switch the point graphics on screen.  Apparently the Arduino can be controlled by JMRI, though I am unsure of how the linkage is made; is it the same USB connection that is used to load the Arduino sketch, is there another bit of hardware that needs to go between the two?  I note that some Megas have wi-fi; is a wireless option viable?  Are there better control programs than Panel Pro that will work with Arduinos? Is there anyone out there implementing this solution, in whole or in part, who has suitable sketches to share?

 

Thanks in anticipation for any contribution.

[Robin2]

I'm building a turnout control system for a club layout but it does not have nearly so many turnouts. I have built several standard interchangeable circuit boards using stripboard with Atmega328 MCUs (like in an Uno) and they will be distributed under the different baseboards to control servos for turnouts and uncouplers and LEDs for signal. The control panel has been built in conventional style (not my own choice) and the switches will feed into a Mega which will send messages by Serial to the other circuit boards. In fact there will be two Megas in the control panel and the other one will receive wireless signals from hand controllers and control the DC power to the tracks (not DCC). It will also be responsible for operating LED lights on the control panel. This system will mean that there are only 3 serial wires (Tx Rx and GND) connecting between the boards to control all the turnouts, uncouplers and LEDs

 

I2C is a short-range technology - maybe 30 cm. I don't think it will be suitable for connections from one board to the next.

 

One thing to keep in mind is that the polarity of the frogs must change when the turnout blades are in the middle of their move, other wise there will be a short circuit.

 

I suspect people will recommend JMRI for the control system, but I have no experience of it myself. I like programming so I have created my own mimic panel in a web page using SVG graphics.

 

...R

[Nigelcliffe]

To make JMRI work, you need a mechanism in the software to talk to the turnout addressing bus structure.  For the supported systems (commercial and DIY documented ones) within JMRI, that might be via USB, or it could be via any number of other communication methods.    As the proposed design is DIY, then that's another thing you've got to write - people will try to help on the JMRI support lists, but its down to you to do it.  (And like Robin, I'm not convinced about IC2)

 

JMRI's graphics are as good as the ones you choose to draw yourself - what appears on a Panel can be completely replaced with graphics you draw yourself.   That the standard off-the-shelf set doesn't seem attractive needn't limit its use.

 

 

I wonder why MERG's CBUS wasn't mentioned ?   It would appear to do what's required, its a build-it yourself setup, at a reasonable price (servo control alone is about £1.60 per turnout, but that hasn't got the frog relays).   There are Servo modules (probably CANMIO as starting place), it works with JMRI for computer control, or there are other options for that side.    CBUS was originally designed as a layout control bus - to control signals, turnouts and receive layout feedback.   Its not primarily a DCC design, and the components which enable it to issue DCC instructions to trains were added later.

 

 

Also, look up the "Internet of Toy Trains" on YouTube, for an enthusiastic Swiss-American who's pushing at the boundaries of communication around a layout. 

 

 

Robin's comment on frog polarity timing is important for N-gauge turnouts , at least until the new "unifrog" design becomes more widespread in Peco's lineup.    MERGs forums have discussed the issue of timing relay movement with the mid-point of servo movement with respect to the CBUS system. 

 

 

 

[StoneMonkey78]

Thanks Robin and Nigel for your input.

 

I hadn't appreciated the distance limitations associated with I2C buses.  Having now researched this issue, and for the benefit of others reading this thread, the limitation is solvable.  The factors that affect this bus centre mainly on line capacitance that distorts the leading edge of both clock and data pulses (rise time), such that as line distance increases there is a point where the slave boards can no longer reliably determine each pulse.  The effect can be mitigated to some extent by using lower data rates on the bus. A single I2C master/slave arrangement should work at distances up to 7m, but for every 5 additional slave boards on the bus, this distance reduces by 1m.  As my layout involves about 20 slave boards, then I will have to look at using I2C extenders to distribute the signal.

 

I found 3 x I2C extenders: Sparkfund's Differential I2C Breakout (£8.98 ea from a UK supplier), the Bausatz I2C-Extender (available is a kit at E11.00 ea from Germany), and Sandbox Electronics Active I2C Long Cable Extender (£14.66 ea from China).  All can use a star distribution, so I could probably get away with 3 or 4 boards.  Assuming the Sparkfund board was used with appropriate cables, this would add about 40p to the cost of each point.

 

One other thing I discovered regarding the I2C bus, although it is described as a 2-wire bus, all the boards need to share a common ground.  All the advice is to use 4-core wiring to include VCC, Ground, Clock and Data lines.

 

I was aware of the need to ensure the relay switching of frog polarity is appropriately timed.  As an aside, I had forgotten that I had bought Insulfrog Code 80 points for the hidden/fiddle lines, so the number of relays needed is much less than at first thought.

 

Nigel, I hadn't heard of CBUS, so I will look into that.  Also, although I had tinkered with Panel Pro's Panel Editor, I hadn't tried the Layout Editor - the Layout Editor seems to give much better control of the graphics layout.

[Robin2]

10 hours ago, StoneMonkey78 said:

Thanks Robin and Nigel for your input.

 

I hadn't appreciated the distance limitations associated with I2C buses.  Having now researched this issue, and for the benefit of others reading this thread, the limitation is solvable.  The factors that affect this bus centre mainly on line capacitance that distorts the leading edge of both clock and data pulses (rise time), such that as line distance increases there is a point where the slave boards can no longer reliably determine each pulse.  The effect can be mitigated to some extent by using lower data rates on the bus. A single I2C master/slave arrangement should work at distances up to 7m, but for every 5 additional slave boards on the bus, this distance reduces by 1m.  As my layout involves about 20 slave boards, then I will have to look at using I2C extenders to distribute the signal.

IMHO, Serial is so much simpler.

 

...R

[Nigelcliffe]

I agree with Robin, Serial (perhaps RS485 ?) is easier.  

 

But, I'd do it with CBUS unless you want a DIY electronics and code project, including writing the JMRI interface code.  

CBUS is a working setup, buy the kits for a few modules and the USB interface and you've got all you need.  Supported in JMRI for layout control.   You need a Windows computer for module setup (the "FCU" program), there is a non-windows option under development.   CBUS is effectively a four wire system: a twisted pair for the data, plus ground and supply volts. 

(  I don't currently use CBUS, having picked an alternative for my own projects many years ago.  ). 

 

 

On the JMRI side, the Panel Editor produces a logical description of the layout. Which is useful for things like setting up signalling rules (signal associated with turnout settings and track occupancy), and automated running.   It also gives a quick mimic diagram, which is why I think a lot of people use it.  But, because its origins is in setting up the logical rules of how one turnout connects to another, its graphical capabilities are limited.    The Layout Editor is a free-form graphical panel, with any graphical appearance you choose.   You can use both, use the Panel Editor to create the logical rules, then "hide" that panel when running the railway, doing all controls on the carefully drawn graphical Layout.  (Or can have multiple Layouts for different elements of control). 

 

[Junctionmad]

Given you seem to have a reasonable technical understanding , I really recommend MERGs CBUS solution , CBUS is directly supported by JMRI , the kits are very cheap and canbus is more more suitable then  I2C for board to board . There are servo control boards , general IO and all flavors in between .  We use it to control 50 points and 30 signals 

 

I suggest you give it a very close look 

 

Dave 

 

( a MERG member ) 

Edited August 2, 2019 by Junctionmad

[Ian Morgan]

Plus, there is a CBUS Arduino library so you can create your own Arduino projects that will connect to CBUS and be supported by JMRI and FCU.

[StoneMonkey78]

Thanks for all your inputs, and apologies for the delay in responding.  I have spent the last few days researching/costing a MERG CBUS solution and wanted to get my head around what was involved/options/numbers etc.

 

CBUS undoubtedly offers a solution that doesn't involve a lot of programming, centred around the use of the CANMIO boards to directly control either servos or relays, married to a CANUSB4 module to link with my computer and thereby JMRI.  The PMP18 single servo kit was also of interest in that a number of my baseboards have only a single insulfrog turnout or semaphore, so such a servo controller becomes very cost effective when using a 2-wire trigger from a relay on an adjacent baseboard.  I felt that I would also need a CANGIZMO to test the CBUS connectivity.  Whilst the cost of such a CBUS-based solution is less than the MegaPoints solution, it is still coming out at a cost of over £550 (excluding the cost of wiring and servo mounts which were not included in my estimated costs in my original post).  Also there is the significant disadvantage of having to build nearly 30 circuits.  My soldering skills are quite moderate, and therefore the time, effort and angst of such an undertaking is not something that appeals.

 

However, Nigel/Robin, thank you for the heads-up concerning RS485 - not a board that I had heard of.  This merits further investigation as a possible bus solution.  I understand the boards are very cheap (in comparison to the I2C-Extender solutions I identified above).  Having watched a number of YouTube videos in this area, there may be an easier solution involving an RS485 connected network where a JMRI/CMRI-connected Arduino Mega, acts as a master to a series of slave Arduino Nano controllers.  The Nano controllers control the PCA9685 16-ch servo controllers and 8-ch relay modules with fairly simple sketches, addressed directly from JMRI via a USB connection to the Mega.

 

The disadvantage of this solution is the potential need to upload revised sketches to multiple Arduinos under different baseboards (though hopefully once working, repeated crawling might not be needed!).  I also accept that time will be needed to perfect the sketches.  However, I feel more confident tackling this area than having to build ~30 circuit boards and getting them to work correctly.

 

Apart from the PMP18 single servo controllers (which I will be using), I was also impressed with MERG's servo mounts.  I have seen other laser-cut and 3D-printed mount designs, and feel that the MERG design/cost is something to seriously consider.

 

I leave on holiday for 4 weeks shortly, so whilst I will continue to monitor this thread, I will not be able to follow up any research suggestions for some weeks (my wife wants me to have a digital detox!).

 

Thanks all. David

[Buhar]

Hi David, 

You're doing some very useful thinking and research on this issue. I have no knowledge of boards and interfaces so I rely on others to resolve such issues and would then copy the results. 

I'm hoping for a large layout with a 16 road staging yard so controlling costs is essential.

 

Thank you. 

Alan 

[Robin2]

8 hours ago, StoneMonkey78 said:

Arduino Mega, acts as a master to a series of slave Arduino Nano controllers. 

If you have a large system it may be more economic, and simpler to use Mega clones rather than nanos.

 

if you want to control servos with Arduinos you are going to have to make up some sort of circuit boards to facilitate the connections. I am not familiar with the Merg products, but if the only obstacle is the need to construct circuit boards you need to balance that against the construction you will need for an alternative system. The Megapoints people do all that construction.

 

Megapoints and MERG (AFAIK) also provide the system to facilitate setting up servos. If you go for an Arduino system you will have to write the software (and maybe create the hardware) for that set up system. Writing the software to make the set up easy will probably require far more lines of code compared to the core code needed to operate the servos. In the system I am building the servos settings will be "hard wired" into the Arduino program to avoid the need for special set up software.

 

What is the greatest distance that the serial (or RS485) signal will have to travel?

 

...R

Edited August 6, 2019 by Robin2

[StoneMonkey78]

 

13 hours ago, Robin2 said:

If you go for an Arduino system you will have to write the software (and maybe create the hardware) for that set up system.

 

...R

Hi Robin.

 

The overall dimensions of the proposed layout is 4.8m x 1.2m, increasing to 2.2m at the other end (designed to fit in a garage); assuming a roughly centre-feed from the controls (plan to have option of connecting controls from either side of long face), so maximum distance from control to furthest board will be in the order of ~4m.

 

I am encouraged by Nigel's steer to look at IoTT - didn't get the Swiss guy's name - his Arduino sketches seemed logical (notwithstanding his occasional typing errors!) and easily understood.  This solution has the potential to be scaled for what I want, though I'm not sure about powering everything from the DCC bus - there must be a better way of doing this.

 

Similarly, I very much like the look of Nico Teering's Mardec software (it is a really clever Arduino sketch where the user doesn't have to write the sketch, merely enter keyboard inputs to on-screen prompts) and his Arduino-based DCCNext boards.  If the web site is to be believed, a 12 servo controller (or 8 servo + 8 relay switch controls) for E7.50 - excluding the cost of the offered enclosure - could offer a solution at about £2.15 per point, including the cost of the servo.  His videos are here.  His DCCNext boards come in kit form the same as MERG kits, but the greater number of channels would result in fewer boards and therefore less soldering!  I have asked some clarifying questions from Nico, but have yet to hear back.

 

Plenty to think about.

 

[Edited - forgot to add the cost of the frog relay]

Edited August 7, 2019 by StoneMonkey78

[Robin2]

Based on those dimensions I would not expect any difficulty with a simple Arduino Serial communication system. If you ran into problems it would not require much extra work to add RS485.

 

...R

Edited August 7, 2019 by Robin2

[Junctionmad]

Note for that of volume , you could just get the canmio pcbs and source the  parts directly, each MIO board can drive up to 16 servos or 8 servos and 8 relays , the setup software is windows based and free to MERG members 

 

the other big advantage is you get a fully worked up layout bus , so you can fit train on track detectors , point position feedback etc and all manner of other io ( turntables etc ) 

 

your route will involve  100s if not 1000s of software development hours , much testing and debugging , simply to invent another ( possible inferior ) wheel. I do not  see the point of trying to save every last penny creating a custom solution when several exist

 

( and I say this as a professional embedded design engineer ) 

[StoneMonkey78]

33 minutes ago, Junctionmad said:

Note for that of volume , you could just get the canmio pcbs and source the  parts directly, each MIO board can drive up to 16 servos or 8 servos and 8 relays , the setup software is windows based and free to MERG members .

The MERG kits currently being supplied are designated CANMIO-SVG and they only drive 8 servos, not the 16 you mention (there are only 8x3 pinouts on the board), so with a kit cost of £12.76, that works out at £3.19 per point (assuming one channel for servo control and one for frog switching) - this figure is just the board costs and excludes the cost of servo, relay, bus and power wiring.

 

I did look at the option of sourcing components for all the MERG boards.  Where I needed only 1 board (CANUSB4, CANGIZMO) I couldn't source the parts cheaper that the kit price.  I would need ~26 CANMIO boards (just for the turnout servos/frog switching and semaphore servos); if I sourced the PCBs and PICs from MERG, and the rest from Farnell, then the per board price came to £12.01, not really much of a saving over the MERG kit price of £12.76.  I have no doubt that I could drive that price down maybe another 10-20% if I sourced some components from multiple sources (Ebay/China etc), but the real savings only occurred if I was buying enough components to make 100+ boards.

 

If the DCCNext/Mardec option mentioned above does what the website suggests, then my cost just for the board and free software at E7.50 works out at E0.94 per point (87p), less than 1/3rd of the price of the CANMIO board.  The Mardec sketch seems to be well thought out such that user programming is very simple (no sketches to be written).

 

It is not about saving money, for me it is about not spending money unnecessarily - the pedant might suggest there is no difference, but to me there is.

[Nigelcliffe]

41 minutes ago, StoneMonkey78 said:

The MERG kits currently being supplied are designated CANMIO-SVG and they only drive 8 servos, not the 16 you mention (there are only 8x3 pinouts on the board), so with a kit cost of £12.76, that works out at £3.19 per point (assuming one channel for servo control and one for frog switching) - this figure is just the board costs and excludes the cost of servo, relay, bus and power wiring.

 

 

Re-read what JunctionMad said, carefully.....   Don't assume that the MERG kit-locker option is the only configuration of the device - its just one of many options.    With the appropriate build and firmware, the CANMIO supports 16 channels, so that's 8 servos with 8 relays for frog switching.  So, the cost per turnout is approximately half your estimate. 

 

There are a dozen ways to solve this, few of them are really wrong-headed.  Some have more support locally than others. 

 

Any layout involving 91 turnouts is complex, large, and expensive (£12 per turnout including the Servo = £1,100).  It also needs a decent number of trains: say 25 trains at £200 each = £5,000.   Its going to take quite a bit of time to construct.   The price range estimate for control systems seems to vary between £0.87 per turnout through £3.19 (or £1.60?)  to £12.50 (Megapoints off the shelf),  so the price range is £80 (Mardec, not including all parts)  through £145 (MERG at 16 outputs, not including all parts) , through £290 (MERG at 8 outputs, not including all parts) , through to £1200 (Megapoints).    I'd be counting the time in developing any control arrangements as being a major drag on the time to complete the project.  

 

- Nigel

[Robin2]

1 hour ago, StoneMonkey78 said:

It is not about saving money, for me it is about not spending money unnecessarily - the pedant might suggest there is no difference, but to me there is.

 

21 minutes ago, Nigelcliffe said:

I'd be counting the time in developing any control arrangements as being a major drag on the time to complete the project.  

 

A lot depends on whether you like developing the electronics and programming (as I do) or whether your real interest is in getting trains running asap. For me, once I have proved that the trains work I have very little further interest.

 

...R

[StoneMonkey78]

2 hours ago, Nigelcliffe said:

 

Re-read what JunctionMad said, carefully.....   Don't assume that the MERG kit-locker option is the only configuration of the device - its just one of many options.    With the appropriate build and firmware, the CANMIO supports 16 channels, so that's 8 servos with 8 relays for frog switching.  So, the cost per turnout is approximately half your estimate. 

- Nigel

As I don't have any of the hardware I can only go by the information provided on the MERG website.  The website says that until the alternative versions of the CANMIO board have been produced, all orders will only be supplied with the CANMIO-SVO (Kit 98-S).  The build instructions (dated 1a March 2018) clearly state 'This kit is for the CANMIO-SVO which is configured to drive up to 8 servos.'  The circuit board includes an array of 8x3 pin headers to which the 8 3-wire servos can be attached, and, as an alternative, the option of fitting either a 2.54mm or 3.5mm 1 x 10 pin array - 8 of these alternative pin header holes are electrically connected to the centre pins of the servo pin headers (the other 2 are marked 0V and + .  So this particular kit only supports 8 outputs.  If there is a MERG kit that supports 16 servo channels I would be interested, but I could not find it on the MERG website.  Nigel could you provide a link? Many thanks

[Nigelcliffe]

11 hours ago, StoneMonkey78 said:

As I don't have any of the hardware I can only go by the information provided on the MERG website.  The website says that until the alternative versions of the CANMIO board have been produced, all orders will only be supplied with the CANMIO-SVO (Kit 98-S).  The build instructions (dated 1a March 2018) clearly state 'This kit is for the CANMIO-SVO which is configured to drive up to 8 servos.'  The circuit board includes an array of 8x3 pin headers to which the 8 3-wire servos can be attached, and, as an alternative, the option of fitting either a 2.54mm or 3.5mm 1 x 10 pin array - 8 of these alternative pin header holes are electrically connected to the centre pins of the servo pin headers (the other 2 are marked 0V and + .  So this particular kit only supports 8 outputs.  If there is a MERG kit that supports 16 servo channels I would be interested, but I could not find it on the MERG website.  Nigel could you provide a link? Many thanks

 

You need to dig around the MERG forums and MERG Software Wiki,  for which you may need to be a MERG member to gain access.   There is a software version for the CANMIO called the "Universal CANMIO Firmware V2".       As JunctionMad said, you need the CANMIO PCB, the components, and the software.

 

MERG is a club of people interested in electronics.  Amongst the thousands of MERG members are a variety of interests and different things get developed.   The Kitlocker only offers kits which have been through the long process of documenting, adding instructions, packaging, testing, etc.. All of which takes time of a small band of volunteers prepared to put kits together and keep assembling kits.   Individuals developing things can be a long way from what is in a kit.

 

 

- Nigel

[Ian Morgan]

The Merg CANMIO Universal Firmware can be easily loaded onto the CANMIO-SVO kit using the Flim Config Utility (FCU) or a dedicated PIC Programmer. Alternatively, there are many Merg members who could offer to do this for you.

 

The Universal Firmware makes use of the 'normal' 8 outputs which have 3 pin servo connectors on the PCB, plus the extra 8 I/O lines that are available on the 'box' connector on the PCB. All 16 I/O lines can be configured to operate servos.

 

It is also possible to configure various combinations of I/O pins for input, output, analogue inputs, servo control (points, signals with bounce and multi position), delays, sequences, RFID readers, and Hall effect magnetic sensors, depending on what hardware is connected.

 

 

[Junctionmad]

Quote

Quote

 

Edited August 12, 2019 by Junctionmad

[Junctionmad]

Quote

As I don't have any of the hardware I can only go by the information provided on the MERG website.  The website says that until the alternative versions of the CANMIO board have been produced, all orders will only be supplied with the CANMIO-SVO (Kit 98-S).  The build instructions (dated 1a March 2018) clearly state 'This kit is for the CANMIO-SVO which is configured to drive up to 8 servos.'  The circuit board includes an array of 8x3 pin headers to which the 8 3-wire servos can be attached, and, as an alternative, the option of fitting either a 2.54mm or 3.5mm 1 x 10 pin array - 8 of these alternative pin header holes are electrically connected to the centre pins of the servo pin headers (the other 2 are marked 0V and + .  So this particular kit only supports 8 outputs.  If there is a MERG kit that supports 16 servo channels I would be interested, but I could not find it on the MERG website.  Nigel could you provide a link? Many thanks

The situation around the CANMIO can be very confusing to " outsiders" as there are multiple variants.  The SVO variant uses the same circuit board as all the main CANMIO variants ,  Hence the additional 8 way connector can be fitted , however its not conveniently arranged for servo pins as the main connector is , but it can be used to drive 16 servos using the Universal CANMIO software ( in reality thats the only software to use ) .  Furthermore the second 8 way connector can be used to drive a relay Board ( either MERGs or the ubiquitous Chinese relays boards )  allowing the alternative 8+ 8relay configuration . For example I have a 8+8 servo variant as a custom PCB , which is available , as we use these on  a big layout , you could build these up yourself, uses standard Universal CANMIO software from MERG , relays are all onboard  ( I also have a 4+4 board, which actually I think is a better way as its more compact , but you need twice as many ) and many MERG members make all sorts of PCBs available , usually at a  Cost basis. The actual range is huge . Designs are effectively open source , so no long term issues 

 

The advantage is all the software is readily available and fully developed to do what you want , no rolling your own.  You can connect JMRI into this via a CANUSB4 implementing virtual panels and switches and equally you could add in parallel physical push buttons etc , should you ever want a " hard" panel or switches for route setting 

 

Unlike DCC , CBUS is a two way layout control bus, so for example, you can rig  an event to be sent when the servo reaches its end travel and that event can update a panel , providing feedback that a servo actually moved , you can go even further and fit point movement detection and feed it back over the same bus if so desired.  Also track occupancy , which is very useful for computer control can easily be added, whereas with DCC , you need another bus ( like Loconet) because DCC is essentially one way 

 

Given the complexity and cost of your layout I urge you to not fall into the "penny wise , pound foolish " type of thinking and implement cul-de-sac solutions or ones that pare costs to the point where future mods or expansion are problematic ( or expose you to large amounts of personal time ) 

Edited August 12, 2019 by Junctionmad

[StoneMonkey78]

Thanks everyone for your follow up posts, and apologies for not responding before this, but holiday preparations have taken priority.

 

Thanks for the steer regarding  the Universal CANMIO Firmware, Nigel; armed with a search term I was able to find things on the MERG website that I would never have found otherwise.  Although the assembly instructions for the CANMIO-SVO kit include installation of the 10-way box header (J3), there is no mention of its function or reference to the potential attachment of daughter boards, and thereby the board's full potential.

 

Accordingly, as described by your responses, I can achieve the necessary servo control and frog switching with 13 CANMIO boards;  this makes the cost much more affordable.  I'm still a little perturbed at taking on the board construction task - the MERG forum seems to be littered with calls for help when constructed boards don't work once assembled, though I acknowledge that no one will post when everything works as intended!

 

On a separate note, having been unimpressed with the graphics available in JMRI's Panel Editor software, the steer to looking at the Layout Editor software, reveals a much better graphical interface.  I've not yet finished all the block designations, and getting my head around how to get JMRI to construct multi-headed semaphores, but here is a screenshot of my point/semaphore graphical control panel:

[Johnbeynon]

I’m a first time MERG builder this week and have just built CANUSB and CANPAN and the test board...all worked first time. Still figuring out what next....

[Junctionmad]

You must remember many MERG builds are by people with very little electronics build experience and the typical faults are wrong component identification, wrong component orientation and bad soldering 

 

the circuit itself works extremely well and I’ve built 30 units myself , I would suggest you wait for the crystal oscillator version 

 

in the meantime of you want an 8 way servo version without relays,  , I have  a canmio  pcb   which supports the crystal