Board index ‹ Talking T Gauge Forum ‹ T Gauge Custom Rolling Stock

[dvdouden]

Hi there, newbie here. I've just taken my first steps back into the scale trains hobby by purchasing some T gauge items. Last time I touched a scale train was two decades ago when I was in my teens (my dad's H0 Marklin).

Anyway, I'd Ike to build my own DCC-like system from scratch. Ideally, I'd like the decoder to fit inside the chassis, but since my order hasn't arrived yet I have no way of measuring the available space in the chassis.
Does anyone know the width of the motorized chassis (so inside the body shell)?
The 19 meter and longer chassis' have some empty space behind the motor. What is the height of this space and the length for the various chassis'? (19, 20, 21 and 23 meter)
What is the vertical distance between the motor and the bus-bars?

And the last question: is the construction of standard locomotives similar to the motorized chassis'? I.e.: is the body shell of standard locomotives removable and is there similar space available underneath?

My order hasn't been shipped yet, otherwise I would've measured it myself (ordered a 23m chassis), but I'd like to be able do some pcb designing in the mean time and having some target dimensions would help a lot, especially the width of the chassis is crucial.

Thanks in advance for any answers!

[dkightley]

Welcome to the world of T!

I think your questions can be sort of answered by referring you to a couple of existing threads...

Firstly, this thread: viewtopic.php?f=11&t=493&p=7820&hilit=dcx65#p7820

You'll see what space is available....but not the dimensions, as all will be revealed. There's only one DCC decoder chip available at present that stands any chance of being fited inside any existing T gauge power unit..the DCX65 decoder from CT Elektroniks. I have had some on order for three months now....and neither myself nor the UK distributer can get any form of explanation from CT as to why they are not supplying! The DCX65 by the way is 6mm x 5mm x 1.8mm and will fit in the 23mm chassis...but at present, I'm not sure how the smaller units fair!

The second thread is viewtopic.php?f=10&t=525

This thread documents what I'm currently doing with DCC....

To answer one of your questions....all of the power units are constructed in exactly the same way....so the "gap" you refer to , if space is available, will be there on both the unbodied and bodied power units.

[dvdouden]

Welcome to the world of T!

I think your questions can be sort of answered by referring you to a couple of existing threads...

Thanks! I did read those threads before, and made some guesses on the dimensions based on the picture I found here and on tgauge.com, but couldn't find any info on the width of the chassis, nor deduce it from photo's.
I saw the DCX65, that's ridiculously tiny! I won't be able to make my board that small, since I can't solder lead-less IC's by hand. The current design is 14 by 7 mm's. I guess it's too wide to fit, I can bring the width down to 5 mm I think, but that will increase the length of the board by a couple of mm.

To answer one of your questions....all of the power units are constructed in exactly the same way....so the "gap" you refer to , if space is available, will be there on both the unbodied and bodied power units.

Cool, I was hoping for that

Thanks for the reply!

[dvdouden]

Well, it's here!

First thoughts: TINY! And very cool to see something of that size running.

I didn't order a PWM controller since I wanted to build my own. Wired an Arduino and an H-bridge to the track, wrote a few lines of code and here's the result!

It seems to have some difficulty with running at low speeds especially in curved sections. It tends to stop running, requiring a small push to get going again. I guess I should clean the track and wheels properly first since it appears to lose its electrical connection to the track (motor stops buzzing).
It also seems to wobble a bit in corners at low speeds, but I guess that's the gears not meshing properly due to the small radius of the bend (120mm)?

Measured some lap times and tweaked the controller a bit to get a scale speed of about 130km/h (75mph). Then added some code to slowly ramp up the speed, trying to get an acceleration of 2.2m/s/s and a deceleration of twice that, which works out to roughly 16 seconds from standstill to top speed and 8 seconds back to standstill.
Shot a video, uploaded it, and then realized the Wikipedia article mentioned acceleration in km/h/s instead of m/s/s... DOH
The 16 seconds looked about right. When you're in a car It should've been 57 seconds and a bit.

Anyway, to answer some of my own questions:
Width of the chassis without body shell: 5.1mm
Dimensions of the space behind the motor on the 23m chassis: L=12.2mm, W=5.1mm, H=3.4mm. The bottom of this space has walls running back to front which are 1.0mm high (so in the center the space is 4.4mm high). I'm unable to measure the thickness of these walls without taking stuff apart, but if I had to take a guess then I'd say 0.5mm (so inner width would be 5.0mm)
The space between the motor and the bus-bars measures ~1.0mm, so it should be possible to slide a thin PCB between the bus-bar and the motor contacts; this would be ideal for a plug and play controller.

All in all I'm pretty pleased. Got it working pretty fast, seems to run good and the quality is great. Still need to sort some things out:
-fiddle a bit more with the controller to make it run a bit smoother at low speeds; haven't touched the PWM frequency yet)
-fix the acceleration
-rework the decoder design; it doesn't look like it's going to fit right now. The width is okay, the length is somewhat problematic (it will fit the 23m chassis, but I'd like it to fit the 20 and 21m chassis as well), but the height is the real problem here. Back to the drawing board...

[dkightley]

This is good stuff!!!

I do like the smooth acceleration you've been able to achieve so quickly.

What peak line voltage are you using for the PWM output? And what frequency?

[dvdouden]

Voltage should be somewhere around 4.5V, frequency is 100Hz, duty cycle is 32% at top speed. I'm just increasing the duty cycle from 0% to 32% linearly in 16 seconds. I tried going for the correct acceleration (57 seconds), but that makes it a lot less smooth. But there's plenty of variables to play with when you're building your own gear

[dvdouden]

Now with the correct acceleration

[msimister]

That's brilliant. I'd like to get my rolling stock accelerating and decelerating like that but for this very non-electrically minded bloke, where did you add code and to what? Are you using a standard controller from Tgauge.com?
Thanks.
Malcolm

[dvdouden]

That's brilliant. I'd like to get my rolling stock accelerating and decelerating like that but for this very non-electrically minded bloke, where did you add code and to what? Are you using a standard controller from Tgauge.com?
Thanks.
Malcolm

Thanks for the compliments!
I'm not using the standard Tgauge controller. Instead I opted for building my own from some parts I already had.
I'm using an Arduino ( https://www.arduino.cc/ ) and a dc motor driver board ( https://www.banggood.com/Dual-Channel-L ... rehouse=CN ). The Arduino controls the motor driver, and the motor driver is connected to the track. The code is uploaded to and executed by the Arduino microcontroller. Dumbed down, it looks something like this:

Code: Select all

start applying power to motor driver
wait for 3200 microseconds
stop applying power to motor driver
wait for 6800 microseconds
repeat (100 times per second)


One run of code takes 10,000 microseconds, or 1/100th of a second. So you can say it operates at 100Hz.
Within this 10,000 microseconds, power is applied for 3,200 microseconds. So that's 32% duty cycle.
By varying the duty cycle you can control the speed of the train (increasing makes it run faster, decreasing makes it slow down and a duty cycle of 0% makes it stop). I found 32% to be the right amount of duty cycle to make the train run at 80 mm/sec (or 75mph/130kph). This may vary from train to train and also depends on the input voltage.

This practice of turning power on and off very quickly with varying ratios of on and off time is called Pulse Width Modulation or PWM, and it's exactly what the official TGauge controller does. By turning the speed knob you can vary the duty cycle and make the train run faster or slower.

To get a smooth acceleration, the code slowly ramps up the duty cycle. So initially it may start with 2 microseconds on and 9998 microseconds off, then 4/9996, 6/9994 etc until it reaches 3200/6800. It will take 3200/2 = 1600 increments to get to full speed, but since the code runs 100 times per second it means full speed is reached in 16 second flat. And that's exactly what happens in the first video. In the second video I reduced the increments to 0.5 microseconds, so it takes four times longer to reach full speed.

There's no user input at the moment; it's run completely by code. But it's not that difficult to connect some knobs and buttons to the Arduino and write some code to handle them, or to make sure that the train never accelerates faster then a set amount, even when you open the throttle instantly.

[msimister]

Er thanks. Think I need to study electronics...