An Engine Simulator is Born - Part 1

[S2central.net]

Good point - the early V8 certainly will be the same... never seen 1Z ecu though !

[S2central.net]

Xmas came early today in the shape of a nice 2 channel oscilloscope and a function generator. This scope is a long awaited replacement for the super expensive one that I blagged off Des for ages. Looking forward to getting some traces off my motor running to progress the ESIM project.

[mcandmar]

Tease, so what did you get? ..need pics.

[S2central.net]

Here it is - cheap and cheerful device from Hantek, but I found some great reviews that encouraged me to try it - basically 200 quid which is an order of magnitude cheaper than you would pay for the big names... Obviously its early days and this geared for the hobbyist not professional use day in day out but seems perfect for audio & automotive applications... The 'auto set' button is fabulous - my kids could use it - will be awesome bit of kit for projects and diagnostics !

This one has 100MHz bandwidth - there is a cheaper version at 20MHz which is fast enough for automotive use. What amazed me the most was this one is easier to use than the Tektronix one I borrowed before.

My only tiny niggle is that it is only two channels and not four but for the money it is great.

[mcandmar]

Looks great for the money, i've been thinking about getting one of these fancy new digital storage scopes so i'll expect a more in depth review in the future. My old Tek has no capture functionality at all which can be very limiting, and the main reason i'm thinking of buying something digital. I have another one that is so old it has vacuum tubes inside it!

[Arnout]

Looks great for the money, i've been thinking about getting one of these fancy new digital storage scopes so i'll expect a more in depth review in the future. My old Tek has no capture functionality at all which can be very limiting, and the main reason i'm thinking of buying something digital. I have another one that is so old it has vacuum tubes inside it!

Same here, it's so old that the power cord is desintegrating and no new connectors for this round connection can be had easily. It's only 10 MHz however does have an analogue storage function! Works still fine despite it's age. Especially the plug in modules for operational and differential amplifiers can be handy in some cases.

[S2central.net]

As my travels have curtailed for a while, I managed to get ESIM on the bench again - nothing new at the moment other than wiring up to a proper JPT-55 connector (as opposed to my breakout box lash up). Verified the following for sanity.

Permanent 12V into pin 18
Switched 12V into pin 27 - this activates internal relay for 12V output on pin 37
0V hooked up to pins 10,14,19,24
5V reference observed at pin 12 wrt 0V reference at pin 30
OBD wires hooked up to VCDS port and PC talking to it

Total current draw below 1A, but it only plays on the bench with test supply while I get past the next few hurdles... when I have it capable of generating sparks (to test BIMs & COPs) it needs to be connected to a car battery.

Few pics attached - I need to plan a kind of control box with switches, pots, indicators etc as well so its somewhat robust.

I have high level designs done for G4 and G40 simulation, and will be using the function generator for G28 simulation.

A bit of screened audio cable is needed for the G28 input into the ECU... haven't found any lying around yet !

Once I get G28 looking good (verified on ECU by fuel pump relay activation), then I can start on the circuits to emulate the G4 (crank reference) and G40 (camshaft) sensors. That needs to be all verified on the scope before next connection to an ECU.

F-U-N

[S2central.net]

Bit more lunchtime ESIM action today... I chopped up a stereo audio cable to make screened input for G28.. bit of care required to join common 0V with G4 input across pins 47-48-49. Hooked up G28 to signal generator set to create a sawtooth wave with zero offset and 4V pk-pk amplitude at around 50Hz. I also connected a 12V test LED between pins 37 and 03 so I can see when fuel pump relay (FPRy) is triggered - that will verify shape & rate of G28 is good enough for ECU.

Sure enough - hooked it all up and the LED illuminates... dropped the G28 rate slowly and its clear to see that FPRy activation is good at 30Hz. You can see the blue trace at lower speeds for pin 3 not being a solid 0V.

Anyway - doing some quick math to check the logic here... flywheel has 135 teeth which equates to 2.25 teeth per second per rpm

rpm = G28_Frequency/2.25 = 30/2.25 = ~13rpm which is the number I got before with different ECU on breakout box etc.

I also discovered that G28 input is perfectly useful (at least for FPRy activation) down to ~1V pk-pk which is pretty amazing given how large the real signal is from the car.

Next step is to put G28 into some counting logic so I can start the process of simulating G4 and G40 at the right times.

[S2central.net]

Also interesting that you don't get all these faults listed on VCDS until the ECU is happy with G28 input... note how it doesn't mention the absence of G4 signal, but a bunch of other problems are correctly identified (as they aren't wired up yet).

I see this is a stock immo ECU that I will need to tweak so it 'starts' - virtually anyway !

[edge_devil]

Screened Cable

That's looking like fun Paul well jealous... love the idea though!

..you set my brain thinking about screened cables as I remember repairing a broken wire in my ECU connector... do you know which ones are screened, and are they screened all the way to the plastic connector or just up to the plug cover then split to 0v?

[S2central.net]

The G4, G28 and knock sensors are definitely screened cables to avoid interference. Same goes for G39 (O2 probe) actually.

[S2central.net]

Made a bit of progress on this over the weekend... Highlights below -

- potential divider & schmitt trigger circuit used to convert G28 sawooth waveform into TTL square wave (G28_clock)
- built an 8-bit counter driven by G28_clock which along with logic to restart counting after 135 teeth and provides a G4_ttl_pulse
- the G4_ttl_pulse also drives a latch which flips high and low with alternate G4 to provide useful 'phase' signal for G40 processing

I will post some scope images later, but the result is that frequency of G4_ttl_pulse is 1/135 of the G28 input which is what we want.
I can ramp the engine rpm up / down quickly on the signal generator and see the G4 period widen and narrow accordingly.

I might try to see if the TTL version of the G4 pulse is acceptable by the ECU, but first of all I need to simulate the G40 signal - that one is a bit trickier but seeing as I know what 'phase' the engine is in, and where I need the pulse to start and stop, then I can create some logic around that to provide a G40 signal that envelopes alternate G4 signals. I had to order a few more chips for the required logic functions.

Must draw up all the diagrams properly for posterity as well !

[S2central.net]

Some more pics attached...

1 - G28 conversion to TTL (sawtooth input is yellow, TTL version is blue)

2 - Really tidy breadboard (NOT)

In the following pics, its actually an inverted version of G4 shown in blue as it comes straight from 8-input NAND gate - another inverter creates a +ve pulse that is always one flywheel tooth wide.

3 - Simulated G28 & G4 at idle (800 crank rpm), G4 period = 75ms
4 - Simulated G28 & G4 at approx 4000 rpm, G4 period = 15ms (one fifth of idle period)
5 - Simulated G28 & G4 at approx 8000 rpm, G4 period = 7.5ms (one tenth of idle period)

Remember that G28 frequency in Hz determines engine speed by the calculation RPM = G28/2.25

The 2.25 comes from 135 teeth per minute (60 per second).

I took a video of a sweep from 800 to 8000 rpm (1.8KHz to 18kHz in terms of G28) which will get onto youtube later - really exciting stuff for all the electronics nerds out there... makes about 10 of us I reckon - maybe in binary !

[S2central.net]

Here is the video of progress so far then :

http://www.youtube.com/watch?v=ztERYOHIX38

[irishs2]

Very neat.

Are you using discrete logic just because you have it to hand, or are you going to shift it all into fpga/cpld later on?