ChainHartMachin

Has anyone tried adapting a Nissan Q45 MAF sensor to the 165 ECM? I read about it here http://www.microsquirt.info/mafmap.htm#ms2.

Three points I realize from searching the boards.

1) The code uses an 8-bit buffer to store the current mass airflow reading, which effectively limits the useful range to 255 grams/second, so I'm not going to increase the airflow I can read.

2) The analog MAF input on the 165 ECM has a 1kohm pull up resistor, how do I tell if this is going to give me a problem? (besides just trying it) I suppose the output impedance of the sensor signal channel would have something to do with it?

3) I realize that I would need to change my intake plumbing a bit since one side of this sensor has a 4-bolt flange.

4) I would need to change the MAF tables.

The main things I hope to gain from this are:

A) The Nissan MAF has better granularity at low airflow readings than the stock MAF.

B) The sensor is supposed to be one of the highest flowing stock MAF sensors available and they are cheap on ebay. Even if I'm stuck w/ the 255 grams/second limit in the code, at least the sensor won't be a bottleneck to air flow the way the stock MAF is.

C) It's probably a better design of MAF (all though that's an assumption based on the idea that it's Nissan).

D) I have the door open for an easier conversion to megasquirt EFI.

Thanks!

dimented24x7

iTrader: (2)

Another thing you might consider is using an LT1/LS1 MAF with a translator. Some of those allow you to adjust the response to maximize teh resolution. There is one issue with the 165 ECM that is hard to get around, though. That is the actual code used to store the MAF flow. The 6 8-bit tables/scalars are more difficult to tune, and can hinder high flow resolution. For whatever reason, GM did not go with a single 16-bit table and lookup procedure, which is surpisingly easy to implement in an 8-bit ECM.

You can get by the 255 g/sec limit with the 165 by manipulating the injector constant to make the values represent flows beyond 255 grams/sec. This will allow you to use a larger MAF. But, you will want to have a translator to compensate for some of the resolution loss that you will encounter.

If you just want to run the Q45 one, it will probably work fine, and it should be much better than the bosch unit, which was not well known for its reliability and flowrate. It looks like one of the probe type MAFs, with the sensing elements suspended in the airstream. One thing that it does not appear to have, is a flow straightener in front of the sensor (it looks it might just be a screen). You will need to be careful with teh ducting so you dont generate turbulence, or uneaven airflow thru the MAF. That will really screw up its response.

As far as teh megasquirt, it only supports voltage based MAFs, but Im sure that they probably use 10-bit ADCs, so you should be able to run larger MAFs with >255 g/sec airflows. They also appear to support a derivative of GMs Dyna-Air fuel algorithm, which can use the MAF and MAP to calculate the fueling. I personally did not really like using the blended fueling. The only time I use it is during hard transitions, as the MAP responds much more quickly than the MAF, which tends to lag a bit.

ChainHartMachin

The big thing that attracts me to the idea of using the Q45 MAF is the cost.
Used ones are really cheap on ebay. I'm not sure what cost of the LS1 MAF + translator is by comparison.

Is the trick with the injector constant to basically under-report the flow rate to the ECM so that for a given MAF value a wider injector pulse-width is calculated?

I'm not certain my engine will even pull enough air to exceed 255 grams/sec, though with a less restrictive MAF maybe it would. I suppose the fall back plan is to use the PE tables and a wide-band O2 (which I have).

The setup I have for air induction ahead of my existing MAF is basically a K&N cone filter setup that is connected to a piece of aluminum pipe w/ 30 degree bend. There's about 3 inches of straight pipe before the actual MAF sensor.

dimented24x7

iTrader: (2)

Yes, this will alter the PW calculations. For example, if you put 50% less flowrate for the injectors, then the airflow in the MAF table will now represent 2x more airflow. Its an easy and cheap trick to getting more out of the 8-bit system.

ChainHartMachin

So I finally got around to this conversion. The main things I've learned so far are:

1) The stock input A/D channel doesn't work as it has a pull-up to 5 volts which the stock MAF simply pulls down. This thread has the information I used to get around that problem https://www.thirdgen.org/forums/diy-...alog-ford.html

2) The Maf doesn't really fit in the stock location, at least in my case since I have a Firebird. It might work fine in the stock location on a Camaro. I ended up attaching it to a K&N cone air filter and then using a piece of 3-inch pipe where the stock sensor was (I also welded a bracket to the pipe so I could use the bolts from the stock maf sensor to attach it)

3)I used the information on this page to generate a calibration curve (the "MAF analyzer" application http://www.microsquirt.info/mafmap.htm). I got some of the airflow points a little differently than the method they mention. Instead of trying to get every point by free revving in the driveway and adjusting maf entries, I used some data logs taken with my initial best guess table and chose the points where the BLM and INT were stable at 128 +/- 2 for several frames. I recorded the airflow value at these points, then worked backwards to the maf voltage by interpolating between maf table voltage entries.

dimented24x7

iTrader: (2)

In regards to #1, have you verified that? If it does have a pull-up, then youll see +5V on the MAF line with nothing connected. Id be surprised if that is the case as when there is no MAF signal or the MAF is unplugged, its output will max out, and cause a no-start due to all the excess fuel.

ChainHartMachin

Here's what I observed personally with regards to #1.

The MAF signal voltage was about 2.58 volts with the Q45 sensor plugged in and the sensor itself un-powered, but the ignition in the "on" position (the MAF relay is closed when the fuel pump relay is also closed).

When I tried to run it like this, a MAF hi code (33 I think) was set which caused the ECM to use assumed values and allow the engine to run (the motor ran really rich for a brief period of time). After I read the post in #1, I quit trying to use the normal MAF input on the ECM.

The 5V pull-up is something I remember from looking at an ECM schematic prior to writing the original thread. I think the thread referenced in item #1 confirms it.

I also had to disable code 36 in the ECM (no burnoff for the q45 maf)

RBob

iTrader: (1)

The voltage MAF input of the '7165 ECM has a 1K PU to +5 volts. Then a 10K series resistor and on to the ADC unit.

RBob.

dimented24x7

iTrader: (2)

Thats a rather strange way to have set it up, with the stock MAF varying its resistance. It may confer some noise-immunity as the signal is essentially a current rather than a voltage, but if you have a dirty contact, forget it.

Just a side thought, but it does bring up the possibility of using an LS1 MAF directly. Way back when, I was experimenting with a circuit that provided +5 volts continuously with a pull-up resistor and an RC circuit to average out the voltage. As the output from the LS1 MAF is a series of chopped fixed duration pulses, an increase in frequency will result in a reduction in voltage at the input as the duty cycle of the +5V signal decreases. It worked pretty well if I recalled, but never made it off the bench.

ChainHartMachin

I ended up lowering the MAF tables and injector constants from the "real" values (both constants at 30lbs/hr, which matches my injectors, MAF tables tuned for 124 - 133 blms). The main reason was the LV8 values at WOT were too high (sometimes 251). The spark table stops at 208 LV8 and the code takes actual LV8 values greater than 208 and sets them to exactly 208 to do the look up.

Is 208 LV8 supposed to represent 100 KPA? If so then maybe it makes sense to for my logs to go a little above 208 (but not to 251) since where I live the current altimeter pressure on NWS is 101.4 KPA.

alvanwie

If you look at the code there is an LV8 scalar factor used to convert MAF to load. the default value is 80 (represents 80/64 = 1.25 scalar). You can change this value to change the relationship between flow and load.