Joe Perez

iTrader: (8)

The terminology is confusing.

We (MT crew) typically use the term CAS to refer to the distributor-like device which plugs into the back of the camshaft on the NA cars. This device produces two seperate outputs, one of which is effectively crank position (two equal pulses per crank revolution) and the other is effectively cam position (two pulses per cam revolution, with unequal duration.)

The use of this device was discontinued with the introduction of the NB (model year 1999 in North America), and replaced with two separate sensors of the type being discussed here. While they are electrically compatible, the timing marks which they read on the 1999-2005 engines produce a completely different signal pattern.

For people using Megasquirt (or similar) ECUs, you can get away with using an NA-style CAS in an NB engine, however there is a disadvantage to doing this; because it is driven by the timing belt, there is considerable jitter in the signal. This decreases the accuracy of the ignition timing, and is the primary reason that Mazda started installing a discrete crank sensor in mid-1995, and did away with the CAS altogether in 1999- they needed a much more stable signal in order to do OBD-II misfire detection.

emilio700

iTrader: (3)

Mazda refers to the sensors as
CAS Cam Angle Sensor
CKP Crank Angle Sensor

NiklasFalk

iTrader: (1)

or in the 2000 Wiring diagram:
CKP Crankshaft Position Sensor
CMP Camshaft Position Sensor

Ian

iTrader: (2)

So, no more acronyms

Thanks for the detail Joe. Makes sense now. I'll consider doing a more accurate crank tooth wheel when conversion happens and hope that I get lucky with the NB cam angle sensor. I am getting intermittent mild sync loss on my brain DIYPNP, but needed to keep the NA sensor while troubleshooting other issues.

Leafy

iTrader: (1)

Are you certain you got an evo8 and not like an evox one? The OEM used evo8 sensor I pulled out is externally identical to the miata one. the connector points up and plugs in and it fits in the miata hole. I dont think the one you got will work, it might be too far away from the pickups.

This is exactly what the one I bought looks like and the one on the local evo 8.
03 06 Mitsubishi EVO VIII Cam Angle Sensor CAS | eBay


yes, looks like you got one for the evo x
Mitsubishi OEM Cam Sensor - EVO X

kaisersoze

iTrader: (8)

Leafy is right.
The mitsubishi part number is
MD327107 or J5T23071A
I think these sensors have also been used in some Volvos, Kias and Hyundais-even PT cruisers, eagle talon, dodge stratus
They appear very similar to the miata CAS

emilio700

iTrader: (3)

Crap. Yup, Evo 10 sensor. Ordered 2004 model Evo 8 sensor just now.

Joe, you wanna test it?

emilio700

iTrader: (3)

Do over. Beck Arnley 184-0401 CAS for Evo 8. $37.41 off Amazon.

Observations. It's magnetic, where the Miata sensor is not. For this reason I didn't test it. I'll defer to Joe or Abe for advice on whether it's safe to test with either a stock ECU or MS.

The locating ribs on the plug are about .008 thicker, but in the same location. To use it, one would have to trim the thickness of those rids down a bit. Otherwise, it lines up.

Sensor protrudes about .020 more.

Rib on sensor is a skosh bigger than Miata sensor, maybe .002. Fits into valve cover with a bit of a shove. Bolt center is the same

Evo sensor has three ports that one can clearly see the pins through. Presumably for live diagnostics without unplugging the sensor.

Joe, you interested in testing this sensor?

Reverant

iTrader: (10)

No need to try with a stock ECU or with an ECU that expects a hall sensor.

It may also not work with a hall sensor on the crank and a VR on the cam, due to opposite rising/falling edges.

Leafy

iTrader: (1)

It wont work with the stock ecu or any ecu that doesnt let you setup custom cam and crank trigger settings. Most stand alones let you do this. Should just have to change the cam trigger from rising edge to falling edge.

The magnetic part seems weird. VR sensors are magnetic, but hall sensors are not. Hall sensors use 3 wire, and VR only use two. Very interested to see what joe finds when testing it.

AbeFM

iTrader: (3)

Er, the OEM mazda ones - as best I understand - are VR sensors internally and they have a conditioning board to *emulate* a hall sensor on the output.

Emilio,
I'm happy to look at it - either it's open collector or it isn't. For something like a megasquirt, it's easy to change the polarity in software.

I'll test it on the bench and stick it in my car. Either send me one, or if it's just a FLAPS part I can go get it and try it - I'm supposed to send you a bill anyway. :-)

Guess I'd better resubscribe to this thread.
-Abe.

emilio700

iTrader: (3)

I'll send you the Evo 8 sensor Abe.

AbeFM

iTrader: (3)

Cool! This weekend's already shot but I'll take a look as soon as I'm back. Hopefully I'll drive it back up to you running my engine. :-)

Tlaloque

If the polarity has to be changed, could you possibly switch the output and ground wires for a stock-ECU?

AbeFM

iTrader: (3)

Ah, just to close the loop - the Evo 10 didn't fit. The OEM one works, obviously.

The interesting one is the Evo8 sensor. When measured, it pulled down my 1k pull up to well under a volt - I recall something like 0.2V.

The OEM one pulled down only to ~2V.

That means it should be a MUCH cleaner signal (I didn't look at switching times but I should!).

On my MS3Pro, it works just fine. Better? Hard to say, but certainly solid. However, on a stock NB I couldn't get the motor to sync up. So it might be worth a try for some people but I have one verified failure trying to run it on the stock Mazda ECU.

I've been running the Evo8 sensor (no mods, just bolted in) on my VVT motor for a month or two and it's been fine - but that's maybe 2 tanks of gas, a little early to say much.

I did have a couple hot days where there'd be a ~20% chance of a skip and I didn't notice any. Not terribly definitive, but if you're running MS3Pro (or any setup using the same Maxim input chip) it's probably the way to go - or at least try out. I'll update if something happens but if you don't hear back assume it's working fine since.

(Edit: I can't recall if I checked the timing/rising/falling edge. I imagine my cams could be phased off a bit and I'd not know, I'll have to measure at LEAST the response between the two sensors - I can unplug VVT connector and see if there's a shift relative to OEM)

Joe Perez

iTrader: (8)

No.

Sensors with open-collector outputs can only be wired one way.

Raw VR sensors are reversible, but sensors of the type used in the Miata engine are not.

aidandj

iTrader: (5)

<br />
<br />Why? (Want to hear the Joe explanation. My transistors class has all left my brain.)

Joe Perez

iTrader: (8)

Primarily because current can only flow one way through a transistor.

Sensors like the cam and crank sensor on an NB don't output a voltage, they provide a switched closure to ground. It is for this reason that external pullup resistors are necessary- they provide the voltage from which the ECU derives the signal.

I drew a quick schematic to illustrate the point. Consider the following simplified diagram of a sensor (at left) and ECU (at right):




That transistor in the middle is what drives the output pin of the sensor which connects to the ECU. Its collector terminal is wired to the CPU (hence the term open-collector) and its emitter terminal is connected to ground. Within the ECU, a pullup resistor connected to 12v (could also be 5v) puts voltage on that line. When the sensor is "on", the transistor conducts all that voltage to ground, and the voltage on the signal wire drops to 0. When the sensor turns "off", the transistor opens up and the pullup resistor causes the line to float up to 12v (or 5v). This process creates a squarewave signal which the CPU can read.

Now there are two problems with trying to invert the way this works. The first is that transistors only conduct in one direction, so if you reversed it nothing would happen. The second is that the ground which the transistor is connected to is the same one that the sensor's internal signal-conditioning electronics use, so if you disconnect that pin from ground externally and instead connect it to the signal line (which has voltage on it via the pullup resistor) the sensor won't work at all.




I think you're also trying to solve a problem that doesn't exist.

aidandj

iTrader: (5)

Beautiful. Its all coming back.

AbeFM

iTrader: (3)

Okay then... I guess this is what I get for "quick-and-dirty" testing: the Evo8 sensor only seems to catch one tooth out of the doublet - meaning the motor always thinks it's on the first half of the engine cycle.

This somehow doesn't trigger a 'lost sync' in Megasquirt firmware. I've reported it and there's a chance in a thousand they'll look into it.

Anyway, the point is, when you do it, VVT doesn't work since the cam phase is reported as giberish.

I'll try to get it wired up on my drill or lathe and see what's really going on, but if you're running sequential you can't just put the sensor on.

That said, the motor ran fine, it wasn't half fueling, it was just sluggish as one might expect with the cam at full retard all the time.