Fake - F7R Motorsport Daily

[Evoman]

brillant, thank you

[Chris_C]

Yup, each dot in one "knock", but the knock will occur almost continuously whilst the data is being sampled at x squillion samples a second, so it isn't really that much knock.

Zombeh, it's worse for some reason top end since I've had it all back together, or I just didn't notice something before. Narrow band is showing 0.8V ish when it knocks, but absolute max it ever gets to is also 0.8V so I'm thinking there is an earthing problem on the lambda stopping it getting to 1.0V. That would also explain why it would be different. I'm hoping to get it out with the Ostrich in later in the week... need to make up some working time now you have the TVR back!

As to island fuel... no idea! Prolly exploded by now.

I'm so pleased to have this working, I've been talking about it for ages and being able to actually see problems is so helpful

[zombeh]

I don't have it back at the moment though do I, I'm still waiting to see my shocks again. Then I need to properly sort the bonnet out and then at last I can get on with making it better rather than just making it work.

[Chris_C]

Ever have the feeling the more you understand something, the less about it you actually know

Full EPROM emulation working now, I can read data in and reflash an EPROM with the same software. Now just to get my fake EPROM into Fake's ECU!

[Chris_C]

I'm not sure many people really care... but in my mind this engine swap won't be complete until it runs like a car bought from a manufacturer AND making the power it should. There are so many valvers now in 300's that run, but so many we see are such a bodge too which I think is a shame. In my eyes this is truely an option which if the marketing men had allowed would/could have been made and been either epically successful or more likely at least just way above the competition in terms of what the package offered the driver and my interpretation of that won't be complete until all these niggles are sorted.

Ok... been trying to get my head round this a bit more and still failing. Well, ish. I think that some knowledge might have been gleaned last night, but we'll see. I need to hook up a sillyscope to the car again really.

Fuel map. Currently runs off a "really special wtf" map, as that was the thing it ran best on originally.


The thing that was noted, was that my reading for MAP was pretty much the inverse of this, i.e. thats giving loads of fuel at low pressure (so it must be what it thinks is WOT) and my data was showing WOT to be high numbers of MAP. Something random and backwards. However, checking my advance tables, my MAP sensor is correct, I get high advance values where I should.

So. That could be several things, either Monsieur Le Frenchie can't write software and has got an axis backwards or has written a bad map or lots of other things that came into my head. As the data log stream gives me Injector Duration (so it looks at least, the heading is something like Temps Injectuer ) as well as fuel correction (2 of these, happy google translations of these give "the variation of economies of the rich") I should be able to see whats going on. So, back to the plotting
Injector times plotted from revs and MAP


Which... is a thing with a slope completely different to the original map. Nice. So, either the car is learning it's map so completely it just uses the base map to turn over the first 3 turns, or Temps Injecteur is infact just showing 2 correction factors (1 factor times revs, 1 factor times MAP would give a flat surface) and this gets overlaid on the base map, or the third option is it can't see the base map at all to know what to add to the correction to overlay.

So, to quote again above, the more I learn... the less I know!

[Chris_C]

Ok... so about a week ago I came up with this graph from the datalogging I've been doing on Fake.


I've since done a fair few more runs and she always returns that same pattern when I graph it.

So. Why not draw some lines over it, to show where the bins are that the ECU is comparing it too (the bins differ for fuel and advance, so lets take just fuel)


Then... if I take the map I'm running... and subtract the OEM Williams map, I'll get the difference between them and can see if that has any effect on where the knock is


Then, put that all together along with some sparks (clicky for a clearer bigger picture) The bottom graph cells lay on top of the top graph cells if that makes sense, I couldn't suss a quick way of changing cell size. Or, to put it another way, cell A1 in the top graph is the same as cell A1 in the bottom graph.


Looking at that, the knock I see at 4800rpm is 6 bins down in the fuel map, so, looking 6 bins down in the delta fuel and can see that a lot of that row in blue (less fuel compared to OEM) Infact, one of the worse bins is 600mBar on the Manifold pressure at 4800rpm, which corresponds to one of the most dark blue squares in the delta table. So, more fuel required there then!

Same is seen at 4-6000rpm at 400mBar to my mind, this area again has more advance and less fuel than an OEM Williams map. A lot of this is pointing to me trying a Williams map again now I've got a lot of other teething issues with my install fixed.

It doesn't explain the WOT stuff though, which I'm still very confused about. Nor does it explain the backwardsness of that map that I plotted yesterday so I still need to hook up the oscilloscope.

[Speedy88]

Very interesting stuff but likewise, I'm very confused too Something to note though is that I've experienced WOT knock only recently as well, despite all the racing and other fun I've been doing. A side-effect of ethanol based additives?

[foggyjames]

Bump. Just clearing out old thread notification emails, and found this...

I can't help but think there's a simpler way of going about this. Start by mapping the fuel correctly (this is fairly easy once you have a wideband), that way you can take that out of the equation as the cause of the knock. Then you can focus on ignition advance, which is probably the cause, looking at that chuffing huge step in advance in the ignition map (at "8" RPMs). The hole in the middle of the fuel map is probably adding some heat to the party, but fuel maps can look strange, so a wideband is required to confirm whether or not that's valid.

Dyno time is probably the most efficient way to get your ignition map right. Maybe we could have a chat with Iain and arrange a dyno day sometime? Maybe that'll encourage Ryan to get "the big gun" finished (or started, for that matter!).

cheers

James

[Speedy88]

What's the news on Fake btw?

[Chris_C]

Start by mapping the fuel correctly (this is fairly easy once you have a wideband), that way you can take that out of the equation as the cause of the knock.

Yup, well, this is the plan, only to start with we have no idea how Fenix works to sort out the fuel. There has been some progress I havn't yet posted, I *think* I now have got that sussed.

2 new problems have cropped up (well, 3, as those in the know know... I will update the thread about the third very soon) firstly is probably the cause of my pinking. She pinks mainly when she gets hot, looking at it seems as though my air intake temp sensor is reading a bit lower than I'd expect. Need to fix my intake and that should sort that.

Second thing seems to be my crank sensor is playing up, I'm not sure if that is down to the sensor playing up or if it's noise in the system (well, I do have a few 1-5kHz signals near that point on the ECU and I've realised that I havn't used co-ax cable.

So, both of those could cause some fairly interesting "noise" that could be whats confusing the ECU.

[foggyjames]

I've realised that I havn't used co-ax cable.

So, both of those could cause some fairly interesting "noise" that could be whats confusing the ECU.

Haha...yes...very much so. We had MS showing some seriously dirty signals before we used co-ax.

cheers

James

[jtbo]

I remember using microphone wire for ignition signal of MS, without it there was some rather odd things going on, so it is certainly worth to do that. I think that shielding was meant to be connected at one end only in that, which was weird for me, however it did make occasional misses go away. Who knows what other benefits such may have.

[foggyjames]

That's basically what we used - unbalanced audio cable. It's a lot more flexible than the sort of co-ax you're likely to have lying around for wiring up TV aerials! I suspect Chris may have some nicer stuff lying around anyway.

cheers

James

[Chris_C]

I suspect Chris may have some nicer stuff lying around anyway.

We have the technologies (think there should be a 50m reel somewhere!)

Good call on mic wire though Jani, not thought of that and it's super flexible.

[jtbo]

If I'm not horribly mistaken, that microphone wire stuff were at Megasquirt's manual somewhere, however it is many many years since I did that stuff so I can remember horribly wrong, actually it does not matter if it would be yesterday, my memory is so shot these days