Nibbles
|
| posted on 25/2/13 at 02:09 PM |
|
|
Hi and Gauging interest.
First of all, I would like to say Hi and introduce myself.
I am a friend of Strontium Dog on here and fellow car enthusiast. I am a keen motorsport competitor in various forms, including rallying, karting,
autosolos and autotests. I currently own a couple of Celica GT4's, one is modified for my hobby, the other has an LPG conversion and is my main
road car for the regular long trips I do for work.
I am also an electronic engineer, and run my own business designing & manufacturing specialist electronic equipment, mostly for specific
customers.
I am currently developing a number of aftermarket automotive products which are 'own products' for me to market directly. By cutting out
any middle men, I can avoid the extra mark-ups and keep costs down to an affordable level.
These products will be partly 'open source', however I want to retain control of the product and avoid the marketing nightmares of
megasquirt, linux etc. where there are hundreds of versions and there is so much hassle involved in picking through the muddle that customers just go
elsewhere.
Anyway, at this stage I would like to gauge interest in the following products, and ask for suggestions as this will influence the design. I hope
it's OK to include links to the development threads on my main club (I believe these are viewable without signing up) so people can get a better
idea. If there is interest, I will sign up as a trader once the first product is ready.
ECU. http://www.gt4dc.co.uk/forum/viewtopic.php?f=13&t=5973
This will be a fully mappable ECU (dual map facility), 4 injector and 4 ignition drives for full sequential injection and coil on plug on a 4 cylinder
engine. Cam & Crank triggers can be opto or magnetic (one of megasquirts failings). It will have a knock sensor input and knock control software.
Expected cost about £300 ish (incl. VAT) Expected ready to market sometime late 2014, though guinea pigs will be welcome later this year.
Digital Dash. http://www.gt4dc.co.uk/forum/viewtopic.php?f=13&t=6240
This changed direction from my original post and we are now looking at 2 distinct items. If there is demand in the kit car world for a stand alone
full dash, then I will hapily look at this also.
1 - sensor interface box. This can be stand alone with it's own small alpha-numeric display or can just provide the interface to product 2 or to
a standard OBD2 device and tablet / smartphone app. Expected cost about £250 incl. VAT.
2 - Display box. This provides a colour graphic display and links to the above unit via canbus. It can also plug into an OBD2 / canbus system to
provide 'aftermarket gauges' and warning system on a newer type ECU. Expected cost £200-£200 incl. VAT.
The prime purpose of this system is to carry out the 'gauge watching' for the driver, and provide an alert when something needs his
attention. It also allows the driver to view sensors not on the normal dashboard display without cluttering up the dashboard with loads of gauges.
Expected ready to market later in 2013.
Canbus power distribution / switching system. http://www.gt4dc.co.uk/forum/viewtopic.php?f=13&t=6277
This should be of particular interest to kit car builders. The concept of canbus is to massively simplify the wiring loom requirements of a car by
placing canbus interfaces on (or locally to) sensors, actuators, bulbs etc. and then pass all the signals about on a single pair of wires as serial
data. The only wiring then required for the car is a single power wire plus a single signal pair.
Expected cost £200 - £250 per 8In+8out module, probably about 7 per average car, maybe less for kit cars.
Expected to be on the market in 2014, subject to interest.
|
|
|
|
|
shindha
|
| posted on 25/2/13 at 03:33 PM |
|
|
Sounds like a great idea and welcome aboard, I was looking at the Canbus options although I am not an auto electrician or that good with electrics but
the information I was coming across seemed to suggest that the Canbus would be very expensive.
Could you make a single product with the ECU, Canbus and Dashboard as modules that could be deployed as stand alone modules but with greater and
easier interconnectivity
Like I said I don't know much about electronics and therefore might be miles of the mark.
|
|
|
Nibbles
|
| posted on 25/2/13 at 04:03 PM |
|
|
The intention with all these products is that they can be used 'standalone' or can be interlinked to reduce wiring - e.g. the digital dash
can link to the ECU and pick up all ECU sensors without need to duplicate wiring or sensors. It would also enable easy diagnostic / error code
readouts and reset.
|
|
|
Not Anumber
|
| posted on 25/2/13 at 04:15 PM |
|
|
Would you also be supplying a loom to support the modules and canbus as builders could struggle to use and adapt the standard kit car looms supplied
by Premier and others ?
|
|
|
Confused but excited.
|
| posted on 25/2/13 at 04:22 PM |
|
|
Sounds like an excellent idea.
Don't forget to register as a trader first.
Tell them about the bent treacle edges!
|
|
|
Slimy38
|
| posted on 25/2/13 at 04:36 PM |
|
|
Would a CAN bus introduce any IVA complexities for new builds? Or will the examiner just look for the right bulb being lit regardless of how the power
gets there?
|
|
|
MikeRJ
|
| posted on 25/2/13 at 05:02 PM |
|
|
Interesting ECU project, and I wish you luck though I hope you know how much work you have let yourself in for! I absolutely agree with your comments
on MS, the amount of information you have to wade through is simply ridiculous.
I understand this is the first iteration of your project, but I have some (hopefully constructive) criticisms 
I totally understand why you have gone for the D-sub connector, just as the original Megasquirt team did. However it is a poor choice for an ECU,
primarily because the pins are too close together to be able to easily connect wires of the appropriate gauge for high current use (coils, injectors
etc.). Even the MS people recognised this issue when they developed the Microsquirt, and used a proper automotive grade connector. Obviously they
are a bit pricey and not as easy to source as D-sub, but they are designed for the job.
The distributed processing architecture using an 18F for the less critical tasks is interesting. However be careful when using I2C comms in this kind
of situation, it's quite possible for a glitch to lock the bus up entirely e.g. an unexpected reset of the master device whilst the slave is
still addressed. You can usually work around this, i.e. implement a timeout in the master, which bit bashes dummy clocks out to try and free the bus,
but it's a bit ugly. I work as an embedded firmware and design engineer at a large company, and we try to avoid I2C in any critical/high
reliability areas in favor of SPI for exactly this issue, as well as for increased throughput and ease of slew rate control.
The injector flyback diodes are probably not a good idea. Dumping the flyback current into the 12v rail will significantly slow the injector closing
time, the voltage should be allowed to rise to 70v or so to speed up closing, which the VNP5N does anyway. Easily fixed by omitting them when you
stuff the PCB. Are you planning on high speed PWM for peak/hold on low impedance injectors?
The VR interfaces for the trigger wheels are single ended, so will have relatively poor noise immunity. Again this is something the MS people have a
lot of experience with, which is why the Microsquirt uses the MAX992x differential VR amplifiers.
The vanilla 7805 is not rated for automotive use, though it can be reliable if adequately protected from the kind of voltage spikes you can see in an
automotive system. Ideally you want to look at using something designed for automotive use such as the LM2940 which will survive "load
dump" events that will fry a 7805.
Putting inductors in the ground path (i.e. L1) is risky. Any current transient (e.g. from MOSFET gate charge, or from your VR input) will be
converted into a voltage drop which could cause all sorts of problems. Likewise, ensure the LC circuit on the input to the +5v regulator has
sufficiently low Q, or you could see a large amount of ringing and potentially high voltages.
|
|
|
Nibbles
|
| posted on 25/2/13 at 07:42 PM |
|
|
Thanks for the feedback so far, all very useful.
I certainly won't forget to register as a trader & donate as & when. I'm not a freeloader.
I would be happy to look at looms for specific cars, but that would cost obviously. The canbus system was aimed more at people scratch building as it
will cut the amount of wiring massively. If the wiring is already done, then it doesn't seem logical to bin it and start again unless
significant modification is required anyway.
Interesting point about the IVA, and not one I can answer. Maybe other members on here would know.
Mike RJ:
My original plan was to use the original Denso type connector and make the unit plug & play for the celica. The Dsub was a fallback in order to
get things moving. I have subsequently managed to sort a source of these connectors, plus I have also stumled upon a suitable box / connector combo
made by cinch which may well be the answer for a generic version.
Re. I2C, I would always put timeout / recovery on any operations. I2C was chosen as it uses only 2 pins for all devices, whereas SPI is many more by
the time you've included enables.
Injector flyback diodes were put in the design in case they were needed, I was assuming they wouldn't be. Easier to leave off than to knife
& fork them on later.
I wasn't planning on PWM peak/hold for injectors. My intention is to produce a low / medium end ECU which could replace a standard ECU in a
'90's car, retain all the safeties and more besides, and provide extra functions such as remapping, dual maps, launch control, etc. There
is a huge hole in the market at this point as the current low end ECU's don't have the safeties and self-learning capabilities of the
standard ECU's, leaving many people with blown engines.
The trigger interface I have designed was aimed at the celica variable reluctance sensors (which megasquirt can't handle), plus the ability to
accept a 5V pulse or similar. If you are able to post up information on the VR sensors, I can ensure the next iteration of PCB can handle them.
Thanks for the information on the 7805. I have used them extensively and found them pretty bulletproof but will happily use the alternative if they
are better.
I have included small inductors between power & signal supplies as past experience has shown that spikes otherwise pass easily into the processor
causing corruption & resets. Depending on application, the Earth one is normally just a wire link with a ferrite bead round it. I have been using
this setup for many years on systems which are switching high currents and found it solves all previous problems encountered.
[Edited on 25/2/13 by Nibbles]
|
|
|
skov
|
| posted on 26/2/13 at 09:30 AM |
|
|
That's a massive amount of work to do there!
I think you'll probably be ok on the IVA side of things. The IVA manual only really covers the security of the wiring and doesn't say
anything about electronic modules if I remember rightly.
Though technically you'll need all your modules e-marked to sell them for road use, which doesn't come cheap (I think it cost me around
£3k last time I put one of my designs through  ).
I do have to wonder if your CAN distribution modules really would save any wiring.
I was going to do something similar on my own build (using some modules I already have), and decided if anything it increased the number and
complexity of the wires.
Best of luck with the project!
|
|
|
