Any possibility of creating a small condition board fast enough to help ECUs handle the crappy sensor?
With a cost of less than half a sensor it sell like hotcakes

I'm dreaming of a shrinkwrappable board to be placed in the harness.

 

Mrpham linked to it in post #7:
http://jbperf.com/dual_VR/v2_1.html

I've not tested to verify that this does indeed fix problematic NB sensors, but it uses the same MAX9926 chip that I use in my own builds, and which DIY uses in their MS3Pro and PnPPro ECUs.

 

The problem reading forum in the app, read (but not understood) posts are well hidden. Any results on improving function on flaky sensors would be nice.

 

I'm running a DIYPNP 1.5B, so Microsquirt based. I am NOT using the LM1815 VR Conditioner.

My CAM sensor is wired into VR2, SG and 12V. So following the official instructions. There's a 470ohm pullup resistor to 5V for VR2 as well.

Car runs amazingly well until the sensor gets heat soaked. Only really happens during the hotter months in Australia.

All this stuff about the MAX9926 is above my knowledge level, but willing to try it with some help

 

This is where my knowledgebase becomes hazy, as DIY has never published the schematics for the DIYPnP main board.

From reading their documentation, it's unclear to me what the topology of the cam sensor input circuit is. They make reference to a "hybrid circuit," but don't actually say what it is, so I don't know what chip is behind it. Specifically:
I have an idea, and it hinges on my admittedly vague recollection of having scoped one of these sensors years ago. Try replacing R48 on the MicroSquirt Module with something in the general vicinity of 70k (the standard value is 100k). This will raise the "off" detection threshold for that sensor from 1.8v to around 2.25v, which if I'm correct about the problem being the sensor's "off" voltage going up when hot, should fix it.

EDIT: if you don't want to fiddle with removing a surface-mount resistor from the board, you can tack a 1/8 watt 270k resistor across R48 (in parallel with it), which will create an effective resistance of 73k, and move the threshold to 2.2v. You can get a 1/8w 270k resistor in this pack: http://www.radioshack.com/1-8-watt-c...k/2710003.html (red-violet-yellow-gold.) This method has the practical advantage of allowing you to easily remove the resistor and go back to stock if it makes things worse.




I've emailed DIY asking if they'll help shed some light on the subject. But suffice to say I'm fairly certain that you're experiencing the same issue first brought to my attention by Emilio a few years ago on the MS3X.

For reference, I am guessing that they are using this circuit on the MicroSquirt Module:

 

Thank you so much for that!

I'll grab some resistors today and give it a go

 

I'll be very interested to see the results.

As I noted, the above is mostly guesswork. If it were my ECU, I'd remove Q7, build an external conditioning circuit, and feed its output to the R43/R58 junction. That, admittedly, is more work.

 

Joe is correct - the DIYPNP has no conditioning on the main board and wires it straight into the MicroSquirt module. We have a slot for a pull up resistor, but that's it.

 

It's a known problem on stock VVT cars.

AFAICT the sensor position on a VVT head (oil) heats em up more than on a non-VVT head (timing belt).

 

Another possible fix is doubling the value of the pullup resistor that the sensor sees.

 

Also have this problem. And I can anecdotally confirm joe's heat related drop off. I used to swap sensors out when they started being shitty. But recently I started just hosing the sensor down with the tire sprayer between runs after i get a few misfires and it seems to correct the issue. Though sometimes no amount of hosing fixes the sensor and they just completely **** the bed once hotish. I still havent flipped the cam sensor input on the EMS4 in order to try the evo8 cam sensor I have to see if those are better.

Joe how did you do your temperature test? Did you just hook it up to a logger with power and ground and move a piece of metal through the sensing range while pointing a heat gun at the sensor? I'd rather test the evo8 sensor to see if it has the same issue without having to put it in my can and then wait for summer.

 

This.

Signal changes too much with heat. Even the stock ECU has trouble with it. NB2 sensor runs hotter because it's bathed in oil so the instances of sync loss there are an order of magnitude higher than in NB1's.

 

Looked at one on a scope while attached to a running engine, first cold, then later after it had been driven hard for a bit. Paid close attention to the on-state and off-state voltages, and noticed a change between the two tests.

 

Interesting! I've wanted to try it but never had a car that both had problems and who's time wasn't too valuable to dink around with. I've wanted to create a heat sink for the sensor but all the important stuff is on the other end.


Hmmm, what's the sensor like? Why do you like it?

I'd love to come up with a fix for people, I've just not yet decided which I'd like.

 

The evo sensor looks exactly identical to the mazda one, except they dont seem to have this issue even though their sensor is also mounted in the valve cover being exposed to oil (like ours) and is right next to the turbo.

 

Evo 8 sensor



Miata sensor

 

That rock auto picture doesnt seem right. I confirmed with a local evo racer and the miata and evo 8 sensors are externally identical and the used evo sensor I bought looks identical to the mazda one.

 

It makes sense that the evo8 CAS is the same sensor. NA's use the 1g DSM CAS as well.

Speaking of, I am still running an NA CAS after ironing out the bugs on a NB2 swap, but all this talk of failing sensors (especially on track) has got me concerned. Is there any gain to be had to move to an NB2 CAS? What is the failure percentage rate on these?

 

NB's do not use a CAS. Abe modified and retrofitted a magnetic NA8 CAS onto an NB1 engine for us for the 2012, Thunderhill 25hrs. OK for <7000 rpm, stock internal engine. For a higher revving or high power engine, you want the timing resolution of a crank trigger.

 

I see... CAS = crank angle sensor? My apologies if so, i was referring to cam angle sensor.