[lordjw]

Hi!

If you have comments/questions of the thread DCC design/spec thread, started by Nuro, please post them here.

Thanks

Julius

 

[Xlot]

Minor comment - I beg to differ on maintaining current controllers :
- they are crude and comparatively expensive power devices
- you need to add the LC button anyway
- we've seen that SSD ones are cheap, it's likely that even cheaper ones are available off the shelf from videogames
BTW, how many speed steps are you contemplating ? Don't tell me I can see that from the message structure

More relevantly, I'm afraid I don't understand how you are using Hall sensors to locate the car (or why you should be doing it) :
- how does the sensor know a magnet means LC1 or LC7 - or the start line ?
- why should the sensor care if it's LC1 or LC7 ? Once you push the button, the sensor should just send a message to the base station the moment it "sees" any magnet

edited I now understand the need to identify the LC, since you are centralizing everything in the base station - the question is do you have to? Couldn't you decentralize like SSD, where each LC is autonomous ?

- that leaves open lap counting/timing. Maybe you could have a second, dedicated sensor on the left side of the car - and the two (one per lane) start line magnets would be similarly offset

Hope the above isn't too dumb

Beppe

 

[Midas the King]

Hi,

Just couple of comments.

1.

I would also like idea of free hardware/software for "do it yourself/allmost yourself" digital slotracing. There is many benefits compared to SSD or other systems. Main benefit is that free system can be designed more or less freely. No such economical compromises needed like in SSD. It is quite same if components costs 2 euros or will they cost 30 euros. SSD has had huge economical pressure and solutions has been limited by the fact that PCB can't cost more than couple of euros if it's going to be sold to people. Free system doesn't have same pressure. If i'd design a system for free it would be quite same to me that how many will use it, so prices of the components are not so relevant as long as we are talking under 10 euro/component.

2.

Technical comment:

We are living year 2004. DCC for slotracing is dead. No matter which generation of DCC we are talking. If designing a system using technology from 80's it will be just one poor bugging piece of nice little PIC-board and perhaps website among all other like Davics, DCC's and so on. Biggest drawback is limited speed of communication and two way problem which is forcing to use timeslots and it will generate lots of limits and problems.
Nowadays it is possible to send and receive at same time and use differend kind of modulators (chips from 2euros) to make wide spectrum communication links without having to worry of DC. Signal is going to ride on DC like in modern GPS antenna arrays. These systems can be also suitable for electrical hobbiers.

But anyway. Offcourse it is worth to try and build one for fun. From other applications i have learned that words PIC, unmodulated serial communication, fast and reliability doesn't belong in same sentence

I think that in DCC case it is just easier (and cheaper) to buy SSD, drill a hole into cars and enjoy instead of developing allready old and outdated system which is going to sink under commercial systems.

(just an opinion)

Best regards:

Sami

 

[lordjw]

Well, I will first reply to Sami

COST ! I have about 300 cars. If I want to equip 40 of my favourite cars with digital chips I don't want to spend single extra buck for a decoder. Makes no sense to add cost if you can do with less components.

I agree that DCC is an old technology BUT ! nowadays you can beef it up with the latest transistor technology. Power FETs have come long way and are greatly improved since the introduction of DCC.

I know that there are better ways for communication such as RF. At present I have NordIC, Siemens and Cypress single-chip RF-tranceivers for evaluation. RF could be superior - BUT ! It is out of reach of regular electronics hobbyist. Also these special modulator chips require some know-how and are also out of reach for regular electronics hobbyist. Sure I have 1GHz oscilloscope and 10GHz spectrum analyzer at work and I'm able to inspect what happens if something goes wrong but regular hobbyist cannot do that. I want to be able to design specs that will enable most of electronics hobbyists to build CarDCC themselves. Lets not be elitistic electronic engineers, lets be realistic and keep things simple for hobbyists.

All systems so far (SCX, Davic, Scalex?, Carrera?) are most propably using DCC type communication so it is very unwise and arrogant for you to state that it doesn't work. In my opinion thousands of RR-DCC hobbyists cannot be wrong either. They really have working systems and RR-DCC works also with slotcars as RC45 has demonstrated. And do you know what chip is inside nearly every decoder ever built? Microchip PIC. You have quite wild claims.

Comments are always appreciated but please try be constructive.

Julius

 

[lordjw]

Xlot,

I like the feel of my NINCO variable resistor throttles and would like to keep them. Many people have their favourite Parma controllers and I'm sure they would like to use the controller they are used to. Just add button for LC or it is possible to make automatic such as with blst system. 6-bits for speed means 64speed steps.

Lets consider how the system should work.
1. Driver wants to change lane he/she pushes LC button
2. Command station detects button press and the car has to change lane at next available LC
3. Command station starts asking the car where it is going
4. Car detects next available LC spot with HALL sensor (1-7)
5. Car tells now CS that it has detected that it is approaching for example LC3
6. Command station activates LC3 for the car
7. Track IR gate can be used for more accurate LC, anti-collision possible to implement with software
8. Car changes lane

Car needs to determine its position becouse there will be several LC pieces. When LC button is pressed it has to simply tell the command station which LC to activate. Car has to also detect starting line.

It is (hopefully) possible to form either binary codes or pulse sequences for HALL sensor by placing magnets under the track.

Lap counting/timing is used with your any excisting sensors. These sensors will produce excact moment when the car has passed starting line. CarDCC command station will provide information which car it was that just passed startline.

Here's a picture of a HALL along with magnets and mystery car.

Remember that there are NO stupid questions, maybe only stupid answers

Julius

 

[Xlot]

QUOTE there are NO stupid questions

Since you say so

64 speed steps seem exuberant (remember my bread-and-butter comment on number of diodes in a PM controller, or of coils/wiper positions in a 35 ohm resistor ?)

If I understand correctly, you would cluster magnets to form a "bar code" - I wonder if this is done for railroads
But it does look complex - if I remember correctly you said that the momentaneous short circuit for 3rd gen DCC is local, so the information presumably could be kept limited to the LC in question, without the need of Davic-style dead strips
Anti-collision protection on lane change is again a local, simply hardwired thing - and you do need info on traffic on the other lane > more stuff to feed the command station, your way
And, how does the command station communicate with the LC actuator ? If you again use the rails, don't you add to the number of addresses (and decoders)? If you hardwire, you add cabling

Beppe

 

[Midas the King]

QUOTE (lordjw @ 22 Aug 2004, 20:02)Well, I will first reply to Sami

I agree that DCC is an old technology BUT ! nowadays you can beef it up with the latest transistor technology. Power FETs have come long way and are greatly improved since the introduction of DCC.

I know that there are better ways for communication such as RF. At present I have NordIC, Siemens and Cypress single-chip RF-tranceivers for evaluation. RF could be superior - BUT ! It is out of reach of regular electronics hobbyist. Also these special modulator chips require some know-how and are also out of reach for regular electronics hobbyist. Sure I have 1GHz oscilloscope and 10GHz spectrum analyzer at work and I'm able to inspect what happens if something goes wrong but regular hobbyist cannot do that. I want to be able to design specs that will enable most of electronics hobbyists to build CarDCC themselves. Lets not be elitistic electronic engineers, lets be realistic and keep things simple for hobbyists.

All systems so far (SCX, Davic, Scalex?, Carrera?) are most propably using DCC type communication so it is very unwise and arrogant for you to state that it doesn't work. In my opinion thousands of RR-DCC hobbyists cannot be wrong either. They really have working systems and RR-DCC works also with slotcars as RC45 has demonstrated. And do you know what chip is inside nearly every decoder ever built? Microchip PIC. You have quite wild claims.

Comments are always appreciated but please try be constructive.

Julius

Oho,

I think that you understand my opinion differently than i. Comments were asked and IMO i just said my comment and without dissing your of anyone others system. This isn't matter of being right and wrong and throwing claims. Simply matter of opinion. PIC's will work, but PIC is too slow and it has too much limits compared to other brands in same price. I have learned it hard way by doing things with it

It is perfect MC for controlling/counting/monitoring things that are not so demanding but when you have to measure, process and transmit datastream it is outdated and creates many limits no matter who uses it in trains or cars. It is simple fact. In my opinion it was constructive fact.

Anyway it is not so important. Second thing is that i don't dream RF technology to be used in hobbiers building projects. IMO such system should work using rails and still there is plenty of easy ways to do it. Enduser just have to know what chip to buy and he will get things going with same knowledge as he builds PIC board if he'd have good instructions. F.E Cirrus has many plug and play s-pdif tranceivers and other stuff. 3Mb/s isn't elitistic goal. I know that many hobbier has built systems (audio, gaming, RC-model airplane etc...) using todays technology without being engineer.

I'm just littlebit suprised of your reaction

I just think that it isn't arrogant or unwise to mention different kind of options for DCC. Sorry

Let's also define DCC. When i'm talking of DCC i mean that it is serial protocol which is used by modeltrain hobbiers. Serial protocol, unmodulated with shared timeslots and without errorhandling.

Only motivation what drives me to think own digital system is that SSD or any other commercial system isn't quite what i expected. SSD is cheap and it will be nice system in home. I think that it is made littlebit further than DCC but it has own limits too which disturbs me. SportWorld will be even more advanced but.. We have plans to build a clubtrack with couple of friends and the digital system must be able to be at least as good as analog system and flexible at sametime. I have also made my homework and done some checking but none of available system supports all needs what we want. So that's why i'm thinking of possibility to desing one myself. That's why i have closed DCC case as a solution.

Time will show what comes up. I hope that you'll make DCC work!

Best Regards:

Sami

 

[Xlot]

Hey guys, I thought only Texans had a short fuse !!!

MTK :
QUOTE SSD or any other commercial system isn't quite what i expected

since rather little has been released - and nothing at all about the electronics - could you elaborate ?

Beppe

 

[lordjw]

Hi guys!

I knew that there will be some interesting debate when I started designing. Luckily debate is now here and not at DCC design thread.

Xlot, you got it right, I want to test if I am able to form barcodes out of few magnets and that way detect car position at the track. I will try to compile description of the whole system. Actually that is the first thing that I should have done.

Sami, I appreciate any comments. Please keep them coming. I have been designing electronics for more than 10 years now. At work I design laser displacement sensors and industrial line-scan CCD cameras that have clock frequencies up to 80MHz. Communication speeds are up to 10MHz from camera to control electronics, which is my design also. I use Microchip PIC extensively for receiving the data from my laser sensors. I know it works. I realize that there are better and faster processors available but why use them if PIC is able to do the job.

During all this time of designing I have seen various chips come and go. If you select some special chip to do the communication you may encounter the following problems:
1. You will have to buy 1000 pieces to get it
2. There is a danger that chip goes out of production
3. If you replace the chip with compatible from other manufacturer, you cannot be 100% sure that it works
4. Hobbyists may be able to get these specialized chips, but price per chip will be very very high (they dont buy 1000pcs)
5. Sometimes it is hard, nearly impossible to even get these chips

With DCC I mean indeed RR-type DCC. Its proven, requires simple components. Almost anyone can build a decoder. You may be able to use Railroad decoders if you update the decoder software.

Julius

 

[Midas the King]

QUOTE (Xlot @ 22 Aug 2004, 23:17)MTK :
QUOTE SSD or any other commercial system isn't quite what i expected

since rather little has been released - and nothing at all about the electronics - could you elaborate ?

Beppe


From released information it is quite easy do solve out those tech specs, methods what they use and get a picture of what SSD is capable. I think that quite many have guessed those allready. They use light for communicating back etc...

I think that SSD will be best of (IMO) available systems and i'm going to use it at home with kids. 100% sure that it will be fun. Much cheaper and easier than build DCC by oneself.
But everyone knows that if one is going to build 4 lane clubtrack one could easily wish more flexibility and options.

Not short fuse here. I think that i just let this one go

 

[Xlot]

Exciting to see you making physical progress, Julius

Couldn't your perceived lack of power be due to the 12 V feed voltage - which as you previously pointed out is decreased by the drop in the final rectifiers ?

Beppe

 

[Maltese]

Julius,

I am so interested in the amount of code which was required for the command station as well as for the decoder. I am working on a decoder too and would love to know where some sample decoder code might be.

I' m very curious about the sequences/logic for implementing the on-chip PWM I plan to use as well as how the decoder "reads" the bits over the regular DC power. This is ridiculously fasciniting to my little mind.

I am looking at using the ATMEL 20Mhz chip using BASCOM AVR as the dev tool. I hear their chips are faster and the code compiles really tightly.

(I NEVER thought that I would be interested in programming chips!!)

Oh, and that board you show the picture of...what was the source/cost? It's pre-built, right? It looks similar to an IC programmer/developer board, eh?

Thanks

-Maltese

 

[lordjw]

Hi!

Xlot,
I will try to increase voltage as very crude first cure for the lack of power. I need to install some protective components to the decoder datalines before I try this.

Maltese,
Everything is built from scratch. PCB has been manufactured by using regular methods, UV, natrium hydroxide, ferrocloride. Cost is quite reasonable for command station, pimicro 10euro, FETS 3euro/1piece, additional components maybe 20-25euros. Maybe command station could be something like 50euros. Decoder is naturally much cheaper, maybe it is 10euros, but if you make bigger production run you may be able to lower the price.

There's not much code to show actually. It's better to have on-chip PWM generator than to implement it with software as I did. I can post you all the code if you wish, but I'm afraid it's not very useful - you will find better code examples from the net. I have heard good things about AVR. I use Microchip becouse of wide selection of processors and becouse I have programming tools for PICs.

Julius

 

[uksqueezea]

OH Please continue

, if its quite alright with you chaps, I'll chime in when you catch up a bit more.

 

[astro]

Wow, lordjw! Your work on CarDCC is impressive and it is fantastic to see the BLOG of your product development!

I shall most likely be going down the SSD route - due to laziness and my belief that they will be writing interesting fun telemetry and racing mode type stuff in due course, as well as compatibility with present and future mate's systems.

I mention that because I do have thoughts on your developements, and should contextualise them as those of an interested observer rather than someone who is trying to get actively involved. Also, the points I make are supposed to be constructive, but are obviously issues that I question! So although its a list of negatives, I am not in any way trying to knock your amazing work and research!

I can understand your choice of using standard controllers, though I think it is unfair to criticise the commercial companies for making dedicated controllers. The dedicated controllers can be cheaper, because the resistors or encoders they use do not need to have the power handling of analogue controllers; this gives them a greater freedom in design regarding shape, size, weight and cost of the controllers, in other words, a better long term design solution.

I think your decision not to include active reverse polarity braking is that hackers might be able to reprogram their cars in order to cheat? Does this mean that cars will have no braking at all? Adjustable reverse polarity brakes has added a lot of fun and play value to my analogue experience, so personally I think it would be great with a digital system. People with a desire to cheat will find a way whatever you do, so I don't think eliminating brakes to stop this is worth while.

I like the way the car can tell the base station where it is, but maybe this ought to be sent all the time, and depending on sector, rather than on 'about to reach LC'. Sector information will allow great telemetry software.

Are you thinking of doing the LC changes from the control centre rather than from the car? I don't think this will be quck enough to be reliable and to prevent cars from following the LC, also it seems to limit the track to 2 lane... not entirely sure what your plans are on this one

Thats all for now, once again thanks for the Blog, its fascinating!

 

[Xlot]

Yes Julius,

exciting reading indeed - and easier than Stephen Hawking !!

As you know, I share Astro's comment on resistor controllers : how much did your Ninco cost ? Does anybody know the price of the simplest videogame controller one finds in supermarkets ?

Number of speed steps : if you look at the Italia Slot feed (SF front page bottom left) you'll see pictures of MB's latest electronic controller - I counted the coils in the resistor, and came up with about 14 in one picture and maybe 20 on another

I don't understand the initial speed response you got - if your curve is linear, there should be an initial portion where you press the trigger and nothing happens until you reach the a certain voltage at which the car crawls - that's the classic Vo Configuration Variable, which is set precisely to avoid this dead excursion, so it should be the opposite of what you experienced
I'm not crazy about setting Vmax, but no harm - this would be set into the decoder and not add to message length, right ?
On the other hand, I do believe that Vo setting and choice of response curves are essential - you do want drivers to have something to fiddle with, and these would change from car to car

Auxiliary button :
- even with automatic LCs (a' la BLST) you would still need it to go back into the pit lane
- I'm (OK, personally) deeply skeptical about using it in a complex mode under actual racing conditions - in my running tests, I find it already distracting to use it for lane changing - and I'm alone ! So, I would suggest forgetting about flashing the lights (hey, I thought only we Italians did that, anyway !

) and turn on the main lights via a microswitch during a mandatory pit stop

Number of cars : egoistically, I'm delighted with 16 - although I believe that on a conventional track this will really be limited to endurance racing - but then I suppose the booster station can handle at least 10 A , would more be possible ?

Again, molti complimenti

Beppe

 

[lordjw]

Hi!

Thanks for your kind words, Astro. Please do comment the DCC thread. It might seem that I was short-tempered if you read my earlier conversation with Sami. Well, there is a history behind that earlier conversation and believe you me I always take critisism very well. I hope that Sami will help me to find out if RF can be used with the system.

Controllers, well I just love the standard NINCO controller and would like to keep it. You are exactly right about the fact that "digital" controller can be built by using more simple and accurate technologies. Hopefully new Scalex digital controllers have good feel.

My first decoder has standard braking feature. Transistor Q4 shorts the motor poles as the trigger is released. Braking can be made adjustable, I am considering this. I am not against active braking but it increases the component count and size of the decoder. Maybe I should design deluxe decoder first that has it all. I am surprised that people actually would like to have active braking.

I plan to produce intelligent LC track pieces. A LC piece is simply told which car wants to change lane. LC listens via RF?? which car/cars are approaching. LC piece could also determine whether not to allow LC if there are other cars at adjacent lane. Additional sensors can be used to make LC as accurate as possible.

There is no limit for number of lanes but you will have to manufacture quite interesting LC pices for 4-track.

Thanks for your comments, keep them coming

Julius

 

[Xlot]

Addenda :
(please add IMHO, I believe, perhaps... to each sentence - we all know there are no hard "facts" yet ! )

- from an operative standpoint, the change button has to be kept pressed until the car actually passes over the LC (what if you change your mind ?)

- the LC doesn't need to know the car ID , but just that there's a car wanting to change. There's no advantage, and you pay for it with wiring or additional addresses

 

[lordjw]

Hi!

Beppe, my speed response has to be something to do with PWM. Controller voltage is converted and fed to the PWM generator. I just have a feel that the car accelerates too quickly. I will try to program logarithmic look-up table inside the command station to correct this.

Lights, it takes only few small components and small piece of software so I will include flashing of the lights. Driver can always deside not to use it.

Number of cars, I just wanted to find out whether my decoder equipped car would go crazy and read other cars data packet. Running of 16 cars needs more examination of the FET drivers. I will have to load communication very forcefully and see if the decoder still is able to synchronize with the communication. Now I just found out that it can be done and that the decoder software is robust.

Many thanks for your comments

Julius

 

[Maltese]

As far as the Ninco controller is concerned, you changed the cable, right? Are you running 14-16V through that RJ11 cable?

Once you have connected the Ninco Controller to the CS board, how many wires total are being used in the RJ11 cable? Do you feel the same number of wires can be used when you add the lane change functionality?

Great work thus far, sir.

Maltese