Page 1 of 17

Fred's "Hotel Hyundai" 1988 Stellar! (3rd)

Posted: Sat Mar 12, 2011 3:50 pm
by Fred
Ign only, dizzy, hall effect, 4 wires and a hose :-)

EDIT: A couple of photos of it on my first night homeless and living in it:

Image

Image

Successful code for this decoder/vehicle.

Posted: Sat Mar 26, 2011 5:15 am
by Fred
This post will be a log of the code that I've run the car with. I forget what I used before today, but I'll keep a diary of hashes and/or versions from now on.

5pm 26 March 2011 NZST 0215ea5d3cbf2ba5753a9736a566ebf24e803e73 built with HOTEL=1 and working well!

Fred.

Re: Fred's "Hotel Hyundai" 1988 Stellar! (3rd)

Posted: Sat Sep 29, 2012 3:06 pm
by Fred
I'm clearly doing a spiffing job of keeping this thread up to date. Therefore I add this image of how the ignition input capture circuit will look by the end of tomorrow NZ time:

Image

Note, the capacitor should be small and sized appropriately for appropriate rise times with the pull up used. Lower value pull up = larger cap for same rise time. 390 is what is currently in use in the old broken setup.

Tonight we figured out the primary cause of the noise issues on the hotel: junction voltages! The BJT in the hall output was dropping about the same as the LED in the opto, so was intermittently not turning it off upon pull down.

Code he's currently running is:

Firmware Version: 0.2.0-SNAPSHOT-174-g38cc5a9-HOTEL
Interface Version: IFreeEMS Vanilla 0.0.0
Location ID count: 37
Decoder Name: HallOrOptical-Distributor-4of64
Build Date: Wed Sep 26 13:42:53 CEST 2012

Which is a custom branch with the hack for starting based off of public hash a5ebb527bc8235bfd6.

Once the circuit is fixed I'd like to see a WOT run through 3rd from idle, marked with any knock, and lift off coast down to stop again (or to the next corner/intersection). Reason being, not sure that the angles used are currently correct. Also not sure that it's possible for them to be, though. Logs look dodgy.

Fred.

Re: Fred's "Hotel Hyundai" 1988 Stellar! (3rd)

Posted: Wed Oct 03, 2012 12:31 pm
by Fred
Sorry Peter! :-(

Turns out the orange cap is just 0.01uF / 103 marked. This is too low for the RPM and pull up involved.

Game plan:

1) Increase cap to 0.1uF/104 or even 0.22uF/224 - key change!
2) Try moving opto input pull up (only) from 12V to 5V (leave hall input pull up on 12V) - experimental change.
3) Upgrade 12V pull up to 150ohm 2W resistor, WILL get hot :-) / or multiple paralleled smaller ones / eg 2 * 390 @ 0.5W will be OK, just.

0.1uF + 390 = ~1* timing retard @ 6k
0.22uF + 390 = ~2* timing retard @ 6k
0.1uF + 2 * 390 (parallel) = ~0.5* timing retard @ 6k
0.22uF + 2 * 390 (parallel) = ~1* timing retard @ 6k

lower value resistor = more heat & stronger pull up = more noise resistant = faster charge for same cap or = larger cap with same retard

Fred.

Re: Fred's "Hotel Hyundai" 1988 Stellar! (3rd)

Posted: Tue Dec 10, 2013 8:58 pm
by Fred
Drove this last night, but without timing control. It sucked. Guess what today's priority is? Fix it!

Re: Fred's "Hotel Hyundai" 1988 Stellar! (3rd)

Posted: Thu Dec 12, 2013 6:49 am
by Fred
Didn't quite nail it yesterday or the day before, but have mostly nailed it now.
  1. Completely replaced the ignition output ignitor drive circuit with one that DOESN'T LEAVE THE COIL ON WITH THE KEY ON and also ALLOWS THE SPARK TO ACTUALLY FORM BY TURNING OFF COMPLETELY lol :-o
  2. Ripped an ignition lead and had to fix it.
  3. Got the dizzy timed up properlyish, needs revisiting after point 7 is fixed...
  4. Added a new hack to the old hack to be more safe, coil wise
  5. Reduced ign cut to 50 RPM, sounds much better now - less violent shaking, more modern high-performance frequency
  6. Secured the battery with a small ratchet strap - was floating around destroying itself before that, can't find mounts...
  7. Changed decoder tooth size angle to 83 from 69 which I believe is WORSE than it was before, BUT, couldn't possibly run it as it was with that circuitry. SO, I am pretty sure the circuitry SUCKS and will be looking at that in the near future so as to be able to tune the ignition properly and get decentish economy.
  8. Wore the starter out a fair bit more. Tone changed. Hmmmm.
Basic summary: Presto changed the circuit at some point and made it less or more accurate (likely less) and this affected the apparent tooth angle of the dizzy, which Preston didn't realise needed changing to match. I will scope the circuit to see how both traces compare and just how shit/awesome it is.

Drivable for now, more circuit work later.

Fred.

Re: Fred's "Hotel Hyundai" 1988 Stellar! (3rd)

Posted: Fri Dec 13, 2013 8:57 am
by Fred
So I threw a scope at the mighty hotel input and output of input and found that it's pretty fucked :-)

First screenshot, near idle

Image

Time of a cylinder cycle: 7.6 * 5 = 38ms
Time of a revolution: 2 * 38 = 76ms
Engine speed in RPM: 60000 / 76 = 789.473684211
Lag of leading edge: 1.6 * 5 = 8ms
Lag of trailing edge: WTF

Could this be lagging by an entire cycle or more? :-o Or am I missing something?


Second screenshot, high idle

Image

Time of a cylinder cycle: 7.9 * 2 = 15.8ms
Time of a revolution: 2 * 15.8 = 31.6ms
Engine speed in RPM: 60000 / 31.6 = 1898.73417722
Lag of leading edge: ?
Lag of trailing edge: ?

/me is completely baffled


Third screenshot, medium revs

Image

Time of a cylinder cycle: 8.8 * 1 = 8.8ms
Time of a revolution: 2 * 8.8 = 17.6ms
Engine speed in RPM: 60000 / 17.6 = 3409.09090909
Lag of leading edge: ?
Lag of trailing edge: ?

Yep, /me is puzzled.

In any case, an asymmetrical input circuit is 100% not OK. So the plan, then:
  • 5V power to hall setup (currently 12V, may be necessary, to be experimented with)
  • 5V pull up on input of maximum strength (currently 12V including possible spikes)
  • Small capacitor for rapid charge/discharge and some noise immunity (currently has 102/1nf on it)
  • Schmidt trigger IC with hysteresis, same as Dan has used in his skybucket ECU (currently has a dirty slow opto in it)
This should:

Provide fast reliable honest noise-immune input that allows the engine to run as it did way back in 2010? Or when ever I got it running. As bad as the Puma is/was, it did a better job than a shitty 4n25 opto a la Mega$quirt :-)

May drive it despite all of this, as below 3k it seems reasonable.

Fred.

Re: Fred's "Hotel Hyundai" 1988 Stellar! (3rd)

Posted: Fri Dec 13, 2013 9:58 am
by Fred
Whatever the lag on this circuit is, it's way excessive. Some may remember issue 499 which greatly concerned me. Issue 499 was only 0.1 of a millisecond, not 8 or more. This has to be fixed up. Unfortunately I need to use the car, so I'll just have to keep the revs down, I guess. Ugh.

Re: Fred's "Hotel Hyundai" 1988 Stellar! (3rd)

Posted: Sat Dec 14, 2013 10:19 am
by Fred
I think I may have screwed up the above tests... and you will LOL when you hear how, if I'm right... /me goes to check

Confirmed. I fucked up. The second trace is the dwell out, not the input at all. Will check it tomorrow and repost the same data, but correctly... sorry Presto/everyone.

Fred.

Re: Fred's "Hotel Hyundai" 1988 Stellar! (3rd)

Posted: Sun Dec 15, 2013 11:28 am
by Fred
Retested tonight, no visible difference in edges at all. Woops. Good screenies for setting up the angles, though.

5ms period, probably, like the first screeny last time, but corrupted:

Image

CPU high 20ms exactly
CPU low 19.3333ms roughly
762.711864407 RPM

Next images:

Image

1757.8125 RPM
CPU high 8.73333ms
CPU low 8.33333ms

Image

1757.8125 RPM
CPU high 8.6666666
CPU low 8.4

Image

1717.55790746 RPM
CPU high 8.33333ms
CPU low 8.73333ms



Final images:

Image

3351.95530726 RPM
CPU high 4.25ms
CPU low 4.7ms

Image

3488.37209302 RPM
CPU high 4.1ms
CPU low 4.5ms

Will finish this post tomorrow, post about and setup correct decoder angle, and try the car again.