Apex AFC Neo: Confirmed Programming Bug - Karman Function Doesn't Work

[Jesters Deadd]

So this explains some wierd things happening the last week or so while I was trying to tune my car... Does this bug effect the settings that are at 0% correction?

[IttyBitty9Bees]

Ok, excellent...i was starting to wonder if it was just me...i'm seeing my hertz (and therefore pulse width modulation) go down to almost nothing just a hair after 4000rpm right as I make full boost...starts breaking up and the knock count goes through the roof...but only for about 1/10 of a second...but more than enough to ruin the rest of the pull....i was wondering if my ecu or my maf was going haywire. I'm hoping they resolve this by the end of this month or early July.

[erikgross]

Originally Posted by Jesters Deadd

So this explains some wierd things happening the last week or so while I was trying to tune my car... Does this bug effect the settings that are at 0% correction?

I'm running my car with the Neo correction turned off (probably the same as making the corrections 0% across the board) by pressing the down key on the screen where you set the correction numbers. See the manual for how to do that - when successful, the whole screen gets grayed out, which means the correction is turned off.

When doing the above and running wastegate boost, my car seems to behave normally. But ~250bhp really isn't all that fun

[erikgross]

Originally Posted by Dj4play

If you have actual raw data to send me i can expedite the issue with apexi. if you want to email or PM it to me

Thanks for the offer - the raw data is in the link in post#1 of this thread. I've been in regularly contact with the US Operations Manager in California, and indirectly in contact with the lead electronics engineer in Japan. They've confirmed the issue and are working on a solution - I get the impression they're working on getting a solution available as soon as humanly possible. The only other thing I've thought to do was to offer to help with the new firmware, as my background is in computer hardware design/verification. I haven't gone that far yet ;-)

[Kibwe Walker]

What would take so long if they have the old firmware from the safc2 somewhere right? Thats known to work with Karman setups isn't it?

[q2418130103p]

Originally Posted by Kibwe Walker

What would take so long if they have the old firmware from the safc2 somewhere right? Thats known to work with Karman setups isn't it?

The safc and the neo dont necessarily use the same hardware, the firmware is likely not compatible.

frankly, im amazined this issue can be resolved with a firmware fix in the first place, it looks like a hardware issue to me, but I guess they are confident they can resolve it with software.

[Kibwe Walker]

Originally Posted by q2418130103p

The safc and the neo dont necessarily use the same hardware, the firmware is likely not compatible.

frankly, im amazined this issue can be resolved with a firmware fix in the first place, it looks like a hardware issue to me, but I guess they are confident they can resolve it with software.

I know they don't use the same hardware but this is a software problem, so whats wrong with getting the software from a safc2 which is known to work well with these setups and dowloading that to the NEO.

The only real difference between the NEO and afc2 would be the hardware (color screen), vtec control in one unit, a few more options and more points of adjustment in pro mode. Don't really see why they had to mess with the settings and conversions for fuel correction.

[erikgross]

I suspect that they used a new microcontroller (processor) for the Neo and did a questionable job of porting the microcode from the older generation to the new generation. Depending on the identities of the old and new microcontrollers, they could have similar, but slightly different, instruction sets and capbabilities... or they could be completely different. In any case, it's not like firmware is often written in C++ or some high-level language - that means that if you change the hardware, you often have to significantly change the software that runs on it. You probalby don't just get to recompile the same source and expect it to work.

At one of the chip design companies I've worked for, we have a "law" in the verification community: "if it's not tested, it's broken". Apex obviously didn't test the Karman functionality enough (at all?) prior to product release, and they got burned. And as a result, so did the rest of us...

[q2418130103p]

Originally Posted by Kibwe Walker

I know they don't use the same hardware but this is a software problem, so whats wrong with getting the software from a safc2 which is known to work well with these setups and dowloading that to the NEO.

The only real difference between the NEO and afc2 would be the hardware (color screen), vtec control in one unit, a few more options and more points of adjustment in pro mode. Don't really see why they had to mess with the settings and conversions for fuel correction.

I see what you mean, you are saying they already have the correct forumlas, so whats the big deal? I imagine the problem is not really with their formulas though. I am still not convinced that this is a software problem.

It looks like they have a problem with the ADC or DAC. The data eric posted looks a lot more like there is a problem sampling or recreating the analog signal.

I have never seen a software bug cause larges spikes in output voltages (unless they chose the wrong sampling rate using software), unless you totall botch the formula you use to do the conversion...

I dont doubt they can resolve it with software, possibly using a software filter to try and calm things down a bit with their output.

Just my .02, not trying to provoke anyone.

[Kibwe Walker]

All good theories and I will leave it up to the computer savy (which I am not), to figure it out. I just want this done ASAP before I go to Colorodo or atleast have the Neo sent off by then so its done by the time I get back and ready to be reinstalled.