[JrZook]

Well time for a small update:

Managed to finally track down some Maxim VR interfacing chips in quantities less than 1500!! Man these suckers are small!! 10 legs with each leg only 0.2mm wide. Gonna be fun to solder on to a test board



So while I'm waiting for all my EFI gear to be sorted I've started writing up the coding for ignition timing part of it. Calculation resolution is 0.1 degrees at 8000RPM so that should work ok. At the moment im just implementing a linear next-tooth predictor algorithm so we will see how accurate that will be in terms of ignition timing.

Got all my basic PC framework running for a simple user interface to the micro in real time. This will eventually be extended to alter run time variables such as the ignition and VE tables.

Its all coming together super slowly.......

[atari4x4]

i have no idea what all ^^^ that means, but my hat goes off to you for being able to figure it all out.

FYI, I'm pretty sure they just sold you a 11yo Millennium Bug instead of Maxim VR interfacing chips

[Damo]

Yeah that will be fun to solder, maggy lamp FTW

Your test bench is still not finished, Thursday i'm hoping.

[JrZook]

Woooooooooo cheers man! Pm'd ya! I got the bearings.

[JrZook]

Well turns out those LM1815 VR chips I had previously were in fact duds!!! Both of them!! Many hours of frustration and questions to Rhinoman trying to figure out if I wired them wrong but both of them were bloody DOA! How did I figure that out? Cause I got another pair from a different supplier, threw then into the same circuit and walla! Alive!!

So with these proper zero crossing detectors conditioning the VR signal, the output is much cleaner!! Even get a descent data log at cranking speed finally. Still not 100% tho, as I do have a few slight issues with noise and/or possibly the set-up on the lm1815 but a large improvement!



New cranking log! Much clearer!! You can actually distinguish which cylinder is on its compressing stroke and which is on its exhaust (TDC of cyl1 is the 2nd tooth after the long and TDC cyl2 is the 3rd). Compression stroke is distinguished by the longer tooth interval of the TDC tooth.

As you can see I've still got an issue around the #37 tooth count. Most likely due to electrical noise or a false trigger on the LM1815. If I can get these cranking intervals reliable and clean I may not even need a cam sync sensor to determine the engine cylinder phase.

[atari4x4]

atari4x4 wrote:

i'm pretty sure they just sold you a 11yo Millennium Bug instead of Maxim VR interfacing chips

[JrZook]

atari4x4 wrote:

atari4x4 wrote:

i'm pretty sure they just sold you a 11yo Millennium Bug instead of Maxim VR interfacing chips

Lol you mean they sold a pair of bugs instead of the LM1815's

[Rhinoman]

Is the polarity of your VR sensor correct? I didn't notice before but in the previous trace the signal goes negative then positive. In a 'conventional' circuit it is the other way up with the interface arming on the positive going sedge then triggering on the zero crossing of the negative going edge.

[JrZook]

Which trace you referring to James? This one?



If so the VR signal is traveling from left to right here and thus it goes from positive peak, zero crossing then negative peak. That other signal was the output of my very primitive VR conditioning circuit, not the LM.

I have actually setup the LM chip for the moment with pin 14 grounded, which gives me a falling edge upon a positive going negative zero crossing of the VR. I checked this with a sine wave. I still have to have a play with the LM circuit to see if I can get it perfectly clean and reliable. I was just happy to finally see the IC actually do something for once!! . I have some SSOP to DIP breakout boards on there way so I can test out the maxim's.

Looking over the whole project I reckon I probably picked the hardest engine and flywheel tooth setup to play with! Its an odd fire 60deg v-twin, which is a very jerky engine at RPM's less than 3K. Combine this with the ultra course 6-2 tooth setup on the flywheel and RPM and tooth prediction algorithms get ultra tricky!

Guess I wanted a challenge, I got it now

[Zookyliz]

Did I have a stroke or are you typing absolute gibberish?

[JrZook]

Thanks for the complements Lizzo! Should come help me drink rums at night and swear at it when it not working properly hehe . Maybe after a few you will understand what I wrote better.

[Zookyliz]

JrZook wrote:

Thanks for the complements Lizzo! Should come help me drink rums at night and swear at it when it not working properly hehe . Maybe after a few you will understand what I wrote better.

OK. see you in 6 hours?
hehe

looks like i'll need to be on acid to get what's going on here

[Rhinoman]

JrZook wrote:

If so the VR signal is traveling from left to right here and thus it goes from positive peak, zero crossing then negative peak. That other signal was the output of my very primitive VR conditioning circuit, not the LM.

The scope sweeps from left to right, that would mean that the edge on the furthest left is the first edge to arrive so that signal is going from negative to positive. Take a look at the 'oscillogram' in figure 2 in the datasheet.

[JrZook]

Rhinoman wrote:

JrZook wrote:

If so the VR signal is traveling from left to right here and thus it goes from positive peak, zero crossing then negative peak. That other signal was the output of my very primitive VR conditioning circuit, not the LM.

The scope sweeps from left to right, that would mean that the edge on the furthest left is the first edge to arrive so that signal is going from negative to positive. Take a look at the 'oscillogram' in figure 2 in the datasheet.

Ah yea you got it right! What was I thinking mmmmmmmm. Ill throw the scope on my new setup and post it up. Get rid of the confusion from my previous failed attempts.

[JrZook]

Well thanks to Mr Damo for turning up a shaft and drive flange for the spare flywheel, I now have a setup for a spin bench!

This is going to make life a lot easier to test my firmware and hardware as well as give the starter a rest on the bike. At the moment I will just run it off a drill which will give me speeds up to 2000RPM while I try to sort out a more permanent electric motor set up to test higher RPMs.

With the spin bench set-up ill be able to run and scope out the stock ignition unit together with mine to check timing accuracy.

[JrZook]

Add el-cheapo 240V drill with variable speed controller and rubber mount it to the test bench and done! Simple spin bench that will produce upto 2700RPM

[Damo]

Love it!

[JrZook]

Notice what my awesome rubber mount is?

[Damo]

JrZook wrote:

Notice what my awesome rubber mount is?

Yes I noticed, that's why I approve

[Rhinoman]

Nice, I'm looking forward to seeing some test runs. I found my LM1815s the other day and I now have the dev. board from Maxim too, I just need to find some time to do some tests.

[JrZook]

Well had a few mins to spare to managed to pull the stock igniter off the bike, wire it up, connect it to the test bench and scope the output to see how Yamaha set it up from factory. The output was pretty simple but interesting.

Firstly Yamaha has large exciters on the flywheel, much larger than the core of the VR sensor (which may be an issue as I'm triggering of the zero crossing). The stock igniter uses the leading positive peak output of the VR sensor (which corresponds to the leading edge of the exciter) to set the base 8degrees BTDC ignition timing.

At low cranking RPMs <750, the dwell is set by the previous tooth edge and fires at the leading tooth edge corresponding to TDC of the specific cylinder. So dwell at cranking RPMs could be upto 33ms at 300RPM! Pretty excessive by at least it will guarantee that the ignition coil is getting fully charged for reliable firing at cranking.

Once the rpm rises above 750, the dwell time is set to 5ms and the stock igniter starts to advance the ignition timing.

Stock ignition timing is pretty low for a 2 valve hemispherical combustion camber engine. Its stated at 8D BTDC up to 1300RPM and peaks at only 27 degrees at 4000RPM. I guess they do this to ensure reliability to suit the worst possible scenarios such as low octane fuels, high loads and high heat. Will be interesting to see how see performs with a bit more advance!

So for getting this beast started and the RPM stabilised I will have to fire the coils at TDC as I'm triggering off the zero crossing of the waveform, not the leading edge. This should get her ticking over at least, till I can start dialling in some advance above 750RPM.

This is what tonight efforts looked like.


And the stock igniter being molested by a team of alligator clips


Making progress

[Rhinoman]

I picked up an SJ dizzy from my storage place at the weekend so I'll try and get it hooked up to a drill and the MAX dev kit this week and see what I get.
Did you mange to measure the voltage coming out of the VR sensor on your set up?

[JrZook]

Would be great to hear your findings and thoughts on your MAX dev kit!

Haven't yet scoped out the VR voltage yet tho I did destroy a few long time lurker bugs in my code which wouldn't have been helping.

Had a bit more of a play with the LM and have connected pin 5 to 5V, mode 2. Arming threshold minimum I believe now is 250mV, which has got rid of quite a bit of noise and false triggering at higher RPM's tho it's caused a few issues at lower cranking speeds.

With the ECU's your involved in that have VR interfaces, what is the absolute minimum RPM you set for reliable triggering and stable reference signals? 200?

[Rhinoman]

Poor performance at high RPM is usually a grounding/decoupling issue. The input is clamped at under a volts and then you're pushing current into the power rails. You need to measure the input voltage to check how much current is being handled by the clamping circuitry. You may get some improvement by increasing the input resistance.
Poor performance at low RPM is usually a noise problem.
You shouldn't have any problem with low RPM as you have 5 teeth, the XBR/GB500 has only one. The real limitation is a function of the input voltage/frequency and the time constant of the peak/hold circuit. In the past I've run down to around 150mV and up to 100V or more.

[BAZZOOK]

I feel quite discombobulated

[JrZook]

Yea so do I mate, not an easy project


Measured the VR output with no load. I managed to get a peak to peak of ~150V @ 2500RPM so I assume that will climb even more at redline of 8k!
So it seems like that will overload the input current of 3mA with only the 18K external resistor.

Issue is I can't spin it upto 8k to find the max Vpeak of the VR. What about placing a resistive load across the VR to try and reduce its maximum output voltage?

[Rhinoman]

JrZook wrote:

Measured the VR output with no load. I managed to get a peak to peak of ~150V @ 2500RPM so I assume that will climb even more at redline of 8k!
So it seems like that will overload the input current of 3mA with only the 18K external resistor.

Issue is I can't spin it upto 8k to find the max Vpeak of the VR. What about placing a resistive load across the VR to try and reduce its maximum output voltage?

It'll come down once you've got a load on it. The output is linear so if you get 50V at 4000rpm then you'll get 100V at 8000rpm. You can put a resistor across the sensor if needed.

[JrZook]

Well after an absolute epic rum fuelled night I got inspired to smash together some more code and what do you know. Got the bloody thing to fire up and run off my ECU!! Digital ignition only atm, but it works!

Code is a bit dodgy, PCB is mutilated, VR IC chip is hanging off a breadboard and there's miles of wires hanging all over the place. Looks kinda like some crazy science project, amazed she still worked. Gave me a confidence boost to say the least!


Quality hacking into the stock wiring loom, gotta love them alligator clips!



Rest of my awesome hungover setup, surprisingly I had no issues with electrical noise at this stage. Yea I didn't believe it either

My PC diagnostic interface for now


And a lil clip to see where it's heading.

Yep them be some miss and back fires in there! It still runs tho! Time to clean up some code and start fixing this horrid mess of wires and other semi-implemented ideas.

Overall I think its a huge win!

[Damo]

Good stuff man, it's taken a fair bit of work to get this far. Hopefully it's downhill from here on!

[monley]

I don't under any of this, but any way