Brushless Axle Hung Motor System

[IanT]

Well a quick progress report (although perhaps not too much of that at the moment)

I finally had my test rig all set up and ready to go - and had checked that the expected PWM signal was still appearing from the MMite. So double checked the wiring (just as well), set the speed to zero (1ms pulses) and then took a deep breath and connected the battery.





Nothing (but no smoke or bang either - lets breath out) - although I got some beeps from the ESC, which was a bit encouraging. Typed in '100' for full speed (2ms pulses) and up she fired, so reduced to '50' (1.5ms) and the speed dropped, down to '30' and the motor stopped spinning.






It seems that I have some control from '31' up - but nothing from '30' down. I had to stop at that point but hopefully will get some more time in this evening. Thinking about it, 'calibration' is usually done by setting the Tx first 'full on' (high), followed by 'full-off' (low) - so that's the next thing to try and see whether that helps even out the control range.

So a small step forward. The good news, there was no "magic smoke" and my Micro can generate PWM signals that the ESC recognises. Beyond that, clearly I've nothing anywhere usable at the moment, so it's very early days and whether I can get these inexpensive BLDC to work still remains to be seen.

My next step is to see if I can 'calibrate' the ESC (or modify my PWM programme) and see what that brings - and then start looking in more detail at the SimonK firmware and its setup options. Clearly, Gavin has had much better results with 'Gimbal' motors (80t vs 13t) and his L6234 driver. I've already been looking forward at Option '2' - and some of the latest BLDC intelligent 'chippery' but I think it's worthwhile sticking with this approach for a little while yet before jumping off into another one. Then I can give a better idea about whether Option 1 is viable (or not) and maybe what could be further explored to make it so.

Regards,

IanT

[IanT]

OK - after 'calibration' the motor now starts at '8' - with no motion between '1' & '7'. Much better but it still spins fairly quickly on starting - too fast I think for our use.

So I tried to fool the ESC by switching between '1' and '8' quickly (trying a few pause combinations). Too short pauses between switching and nothing happens, too long pauses and it just starts and stops in bursts - starting quickly and then stopping just as abruptly. So I suspect SimonK is introducing a short delay before recognising a signal, as well as probably having a 'minimum' level before it fires up the motor too.

So the next step is to get the USBasp & Atmel 'probe' out and start looking at the SK version number and settings using the various drone ESC utilities. It's certainly not going to be as simple as just making the ESC 'Reversible' I'm afraid.

Regards,

IanT

[Doddy]

Thanks Ian,

I recently found an article by Will Sweatman with links to articles and PCB designs for a L6234 based three phase motor driver. Everything worked well at high speeds but for slower speeds the motor was choppy.

The problem was solved by changing over to a sine wave to drive the motor, but instead of the Arduino calculating the wave he moved away from using 36 steps in the software and used a lookup table of 128 steps for generating a sine wave modulated PWM signal.

This video shows just how slow the brushless Gimbal motor can move.

https://www.youtube.com/watch?time_continue=2&v=667XSSxieXM

Somebody posted on Will's blog that as BDLC motors are structurally similar to stepper motors, just with less steps, it occurred to me that stepper motor technology might be worth investigating for G3 applications.

This is a completely new area to me, but I think I am beginning to catch up with some of your terminology and the processes that you are applying to your respective designs.

Thanks to you and Gavin posting on your work, this opens up a whole new area of the hobby that would have otherwise remained in the dark. 

[IanT]

Good morning Doddy,

Yes, I'm aware of the BerryJam site and he's got some very useful thoughts on it. He does seem to be using a 'Gimbal' motor though and these are a little different from "power" motors, in that they have more 'teeth' (Gavins' motor is an 80t - mine are 13t) and apparently use a finer/thinner wire on the rotors.

Without going all technical (because I frankly don't know enough to do so) my limited understanding is that a 'Gimbal' motor is somewhere mid-way between a 'Power' BLDC and a 'Stepper' motor. All work on the same general principles but the 'power' one is designed for continuous rotation, the 'stepper' is designed to - well - "Step" (defined amounts). A Gimbal is somewhere in-between the other two, designed to react quickly to feedback 'loops' but not defined steps per se. It generally drives light loads over short distances - keeping a camera level for instance.

So we have these different design characteristics, plus size, weight and cost to consider. We are not too concerned with weight - although generally with extra weight comes increased size and that may be an issue for us. Cost is important, because we also need to add on the extra potential cost of the electronics involved in BLDC - so the real cost of a BLDC 'solution' is the motor + its electronics.

I don't know where we will end up with this but I think the context is that we don't really need 'ultra' slow speeds, just "slow enough" (with sufficient torque) to enable an engine to move off with a good load. We will have multiple motors at our disposal and (at least in my case) they will be lightly geared (4:1). I've already run some speed estimates in fact.

With a 4ft diameter wheel, an engine (at a scale speed of 1 mph) would require a wheel rotation of 420rpm, which (at 4:1) gives a motor rotation of 1,680rpm. That's 28rps (seconds) which on a 6 rotor-pair BLDC engine (e.g. mine) needs commutation at 168rps. I've a table of higher scale speeds but you can easily work them out and you will find we are moving up into quite usable motor speeds at a scale walking speed.

So I think the issue is that initial 'shove' and low end torque required - till we get moving.

There are some other very interesting videos out there covering some aspects of this problem but perhaps it's best if I point to them as we reach that area and explore them in more detail. Other duties (and my Manager) now call I'm afraid...

Regards,

IanT

   

[IanT]

Well, it's been a week since I last posted - so maybe I should update this thread.

I've had a few other things needing my spare time recently, so efforts in this area have been both limited and somewhat frustrating. From the videos, it looked easy to examine and configure the ESCs firmware. However, it has proven to be not quite so simple. My first attempts to use BLHeli just didn't work. So I decided to take things 'step-by-step' and recheck everything methodically.

My first thought was that the new 'socket' cable might be wired incorrectly for my USBasp version, as I wasn't using the (arduino) cable provided with the USBasp and as it didn't come from Hobbyking (like the 'socket' cable) they were from different sources - and neither had any documentation that I could find (the downside of using low-cost parts). The USBasp didn't have any signal markings on its 10pin O/P at all and terminated (at the end of its cable) in a 6pin connector board (which was marked). So I had to trace the pins 'backwards' to identify the right pins on the USBasp and then back again out into the new 'socket' cable. As it happened (after admittedly getting myself somewhat confused) I decided that everything was actually OK.

So I looked again and then realised (Dummy!) that there was no USBasp 'driver' loaded. The new driver was easy to find and download but my laptop couldn't see or use the driver, because it was "unsigned" - and Win10 steadfastly refused to use it. It turned out that there are several solutions to using "unsigned" drivers but in this case the best one was to use another (free) utility called "Zadig". Once this was downloaded, I was able to "sign" the driver and the USBasp now appeared (in my device list) when plugged in. Great (I thought) that should finally sort it!. But still no joy with BLHeli and its 'verify' routines.

OK I thought, I'll try KKMulticopter instead. This was downloaded (and unzipped ) but it wanted the Java Runtime Environment (JRE) loaded to run. So I downloaded the latest Java version and installed it. KKm couldn't see it. More headscratching/googleing ensued and eventually I was pointed at the Win10 Environment Variables, which can be reached via the Control Panel. This is where I had to leave off but I think I'm in the right area and a little more effort may well get me there. It's the usual thing, it's always hard the first time around, until you know what to do & where to go and then (with hindsight) everything becomes much easier. Unfortunately, you still have to plod through learning everything (and going down dead-ends) that first time around...

Whether or not, this is a good use of my time, I'm not really sure but I will stick with it for a while longer. I have been looking at other BLDC 'driver' options (and have found some interesting devices to research) but that's probably enough news for tonight.

Regards,

IanT

[Doddy]

In the absence of Chinese brushless motor and controller deliveries, I had a look at an alternative mode of power.

https://www.youtube.com/watch?v=fDq82yJGW3U

[John Candy]

That brought a smile to my face!

It's a wonder some "do-gooder" hasn't reported them to the RSPCA!!!

Regards,
John.

P.S. Anyone tried linear induction? At school, in the mid-60's, we set up a short track in the science lab!

[Doddy]

Anyone tried linear induction? At school, in the mid-60's, we set up a short track in the science lab!

Not myself John, but the Tomy Toy Company of Japan has released its magnetically operated Maglev Train. 














https://www.youtube.com/watch?v=xirz7Zm8t6w

https://www.amazon.co.uk/Linear-superconducting-linear-system-Special/dp/B00Y1UOL7I

[cabbage]

We built a gauss gun in the 6form. It could embedd aluminium pencil sharpener in the plasterboard walls. The impulse capacitor was made from cooking foil and waxed bread paper. I did get to see the flying tea tray at BREL Derby.

Regards

Ralph

[John Candy]

I can't remember a lot about the linear induction train we built....it was more than 50 years ago.....but we used brass curtain rail and "runners" for reasons of economy!

John.

[IanT]

I suppose we could always try breeding smaller Hamsters?  (it might be quicker!)



IanT

[IanT]

Before I did anything else tonight, I set-up my test rig and checked the ESC was still working. I was concerned that I might have fried the Atmel along the way. Fortunately, everything still seemed OK.

So I went back to BLHeli and finally decided that my laptop still wasn't really seeing the USBasp device - although it was there according to Win10. After an evening of messing around trying different driver versions/sources - I finally found (and re-installed) one that has allowed me to 'Verify' the firmware as SimonK version "tgy.2012_09_30" - comparing it with the tgy.hex file on my laptop. BLHeli confirmed that the ESCs firmware was identical. Progress!

So in simple terms I now know that the USBasp driver and socket cable can 'read' the Simonk firmware on the ESCs Atmel chip. I also know that the SK firmware is not the latest (2015) version - so the next step will be to try and update (re-write) it - and then modify it.

But not tonight, that's enough for one evening I think.

Regards,

IanT

[IanT]

I was a bit tired last night, so (before I forget the details) and just in case I (or anyone else) wants to replicate this process...

The driver issue was solved when I found an Australian supplier (Protostack) that has a USBasp device that looks remarkably similar to my eBay 'LC Technology V2.0' one. Their website lists their own (AU$18) device but also has a 'Manual' for their version (complete with circuit diagram/cable pin-out) and also "signed" drivers. These drivers (in my case x64-v3.0.7) are 'embedded' into a .exe file that loads the selected driver (either 32 or 64bit version) onto your Win10 device. So I couldn't use Zadig - but this process did apparently overwrite the other driver I'd used with the new one.

https://protostack.com.au/shop/accessories/usbasp-avr-programmer/

I did have to run it twice to get it to work though, not sure why - maybe finger trouble on my part.

Regards,

IanT

[Doddy]

A Radio Controlled Test Rig for 'Brushless' Motor Technology

Well, I got tied of thinking rather than doing, and I decided that if I don't start doing, then nothing will get done. And after the recent arrival of my LGB Ballenburg Steam loco, and in the absence of my own G3 scale train to run, my mind turned to how I can use my G Scale locomotives on Gauge 3 track, and then I found this video for inspiration . . .      https://www.youtube.com/watch?v=VVFk-qHMutU


So! getting back to reality and the possible use of dual gauge track (G and G3), and after some re-evaluation of Deltang products by inadvertently watching RC Trains version of the same 'rebranded' Deltang system, and with some co-incidental input from Ralph's 'Cabbage' and Tim Gleed-Owen at the time regarding the building of, and operation of the Deltang radio control system. Additionally, the many thanks I owe to IanT and Gavin's technical posts on this forum, I both read and understood the terminology they were using, and managed to put together a block diagram of 'off-the-shelf' components.





I am now looking forward to testing a potential solution for the integration of 2.4G Radio Control, but also leaving the door open for the consideration of integrating new technology developments such as Bluetooth or Radio distributed DCC (Digital Command Control). A rather nice demo of the Deltang Radio Control system is to be found here . . . https://www.youtube.com/watch?v=dvPks5HAfpQ

The outcome is looking encouraging, as not only is the price of the Chinese based solution cost effective, the whole kit and caboodle (less the brushless motors) comes in cheaper than the cost of a single Revolution Trains 2.4G receiver. And it has the potential to provide far more integration possibilities than the Aristo-Craft Revolution based solution I was looking at previously.


All the components for the bogies,

   
• Bodyshell,
• Chassis design,
• Wheelsets,
• Digital Sound card,
• ESC's,
• Brushless motors and controllers,
• RPM counters and a KMH speedometer

are either on order and/or in the process of being designed, manufactured or delivered. With the Deltang transmitter and receiver to be ordered very soon, the project is well on its way.

The lovely thing about this design architecture, is that with some minor adjustments, the design features can be used for the technically more complicated designs I had for a Revolution controlled 'Peak' and GCR/LNER 8H steam locomotive that I wanted to build.


The ADtranz built 12X is the mothership in a long line of AEG, ADtranz and now Bombardier locomotive developments from which I have a large selection of liveries to consider painting it in.




The model will have the following features:

   
• Deltang 2.4G Radio Control
• Manually operated acceleration and braking profiles
• Distributed Brushless Motor Control
• Slave Motor Controllers
• Servo Operated Pantographs
• Directional Lighting to Swiss (SBB) operating standards
• 16bit (CD quality) Digital Sound Card(s)
• User operated Whistle* and Airhorn sound samples
• Stereo pair of miniature Hi Fi speaker technology
• Optional Distributed 'Bluetooth' DCC Control*
• Optional Centrifugal Clutch Drive System*

Many thanks for all the contributions to encourage me get my finger out and finally have a go.

Best regards,


Bob (Doddy)

[Peaky 556]

Thanks very much Doddy, language I can understand in this thread, for a change!
The RC Trains Tx demo video was very good I thought, to introduce the G3 fraternity to these smaller, dedicated and low cost R/C solutions for model trains.  So much better in my view than aircraft "stick"based systems, but I know that is a personal preference.
Is it true that the Tx mentioned in the video is also available from Micron Radio Control as the Tx22, as are others with differing features?
Regards, Tim