maf install success!!!!!!...so far
#47
I've been doing a little work on this problem too - I tested my pre-owned Ford MAF and the reason it was so cheap is that it's toast.
So, I went back to the drawing board - and looked for specs for a MAF that would fit the application. I found the specs for the Bosch HFM here;
http://www.guzzimental.com/documents...%20Sensors.pdf
Along with discussion of how it craps out and is expensive - and people have been using a unit made by Pierburg in it's place.
Looking at the transfer function it's obviously very different to the late AFM. The signal for the late AFM was trimmed to be linear with respect to airflow. (from FRWilk). The curve for the MAF is logarithmic and sits at about 1v at zero airflow which would not do good things for your idle :-)
So, I started looking for ways to convert the signal. Most people choose a digital solution - program a PIC, it reads the signal and chooses the nearest value from a lookup table. Relatively easy to build, and a universal solution - this is the like the split second programmable controller. This guy built his own...http://www.carlton24v.co.uk/afmmaf.htm
But, there is an alternative - an analog circuit. I started to think about this circuit as a variable gain amplifier - the gain varies with the input voltage.
I put the FRWilk data and the MAF data into an excel spreadsheet, built a model of each of the curves, and calculated the conversion factor. I then modelled this curve (lots of modelling) - and found that it had an exponential form. From memory it's something like y=0.06e^0.007x - but I'll post the whole spreadsheet if anyone is interested.
So - end of the story coming up - I've been working on a circuit design, and testing in 5SPICE (electronics simulator) - it's a bit tricky, but you can get an exponential curve by amplifying the forward switching behaviour of a diode or transistor. The issue is that temperature variation is a big issue - 10 degrees makes a big difference. The work around is to use a matched pair of transistors with one working as a reference, or you can use a thermistor bonded to the transistor.
So that's where I'm up to - I'm running final simulations to try and get it to be temperature stable - but this can be built with about $20 or so worth of components, so I'm hoping it will be worth it.
Incidentally, this could be used to scale the voltages for the much bigger MAF which flows up to 1000kg/hr - or slightly more than double what I calculated the NA needs at max rpm.
On that note - I've been looking at crudely modelling VE on the 944 engine based on the gradient of the horsepower curve. I assume the peak of the horsepower curve corresponds to where VE starts to suffer, so I got the idea that rate of increase of RPM must be a good indicator of the VE at that RPM.
Could that mean that we can assume VE is proportional to or a function of the differential of the horsepower curve? If it's true that would prove useful.
When I take the final circut mod to the dyno then I'll have a log of airflow with RPM/hp and I'll go further with that.
Any thoughts?
So, I went back to the drawing board - and looked for specs for a MAF that would fit the application. I found the specs for the Bosch HFM here;
http://www.guzzimental.com/documents...%20Sensors.pdf
Along with discussion of how it craps out and is expensive - and people have been using a unit made by Pierburg in it's place.
Looking at the transfer function it's obviously very different to the late AFM. The signal for the late AFM was trimmed to be linear with respect to airflow. (from FRWilk). The curve for the MAF is logarithmic and sits at about 1v at zero airflow which would not do good things for your idle :-)
So, I started looking for ways to convert the signal. Most people choose a digital solution - program a PIC, it reads the signal and chooses the nearest value from a lookup table. Relatively easy to build, and a universal solution - this is the like the split second programmable controller. This guy built his own...http://www.carlton24v.co.uk/afmmaf.htm
But, there is an alternative - an analog circuit. I started to think about this circuit as a variable gain amplifier - the gain varies with the input voltage.
I put the FRWilk data and the MAF data into an excel spreadsheet, built a model of each of the curves, and calculated the conversion factor. I then modelled this curve (lots of modelling) - and found that it had an exponential form. From memory it's something like y=0.06e^0.007x - but I'll post the whole spreadsheet if anyone is interested.
So - end of the story coming up - I've been working on a circuit design, and testing in 5SPICE (electronics simulator) - it's a bit tricky, but you can get an exponential curve by amplifying the forward switching behaviour of a diode or transistor. The issue is that temperature variation is a big issue - 10 degrees makes a big difference. The work around is to use a matched pair of transistors with one working as a reference, or you can use a thermistor bonded to the transistor.
So that's where I'm up to - I'm running final simulations to try and get it to be temperature stable - but this can be built with about $20 or so worth of components, so I'm hoping it will be worth it.
Incidentally, this could be used to scale the voltages for the much bigger MAF which flows up to 1000kg/hr - or slightly more than double what I calculated the NA needs at max rpm.
On that note - I've been looking at crudely modelling VE on the 944 engine based on the gradient of the horsepower curve. I assume the peak of the horsepower curve corresponds to where VE starts to suffer, so I got the idea that rate of increase of RPM must be a good indicator of the VE at that RPM.
Could that mean that we can assume VE is proportional to or a function of the differential of the horsepower curve? If it's true that would prove useful.
When I take the final circut mod to the dyno then I'll have a log of airflow with RPM/hp and I'll go further with that.
Any thoughts?
#48
Intermediate
Join Date: Apr 2007
Location: Boston, MA
Posts: 43
Likes: 0
Received 0 Likes
on
0 Posts
it would seem to make sense that the VE curve would relate to the horsepower curve, but in which way is going to be pretty hard to figure out. But i do have a question, if a curve of VE were graphed how would the efficiency be plotted? i mean i understand we could graph flow against hp curve pretty easily but the efficiency of the flow at every moment of horsepower being graphed is hard to correlate, if its even possible to correlate.
but back to the VE curve, it seems plausible that it would be a derivative of a horsepower graph but when the horsepower drops it might not always be the VE it could be lack of fuel, lack of exhaust efficiency, weight of the piston/crankshaft... lots more reason hp decreases than just ve, but it could be a marker.
just editing instead of posting again.
Do MAF's output air temperature signals? and in what form of communication is the signal? you could probably use the same variable gain amplifier idea mated to another variable gain (for the airflow map) because i would assume temperature is linear to the density which is linear to the amount of oxygen being put in a combustion chamber.
but back to the VE curve, it seems plausible that it would be a derivative of a horsepower graph but when the horsepower drops it might not always be the VE it could be lack of fuel, lack of exhaust efficiency, weight of the piston/crankshaft... lots more reason hp decreases than just ve, but it could be a marker.
just editing instead of posting again.
Do MAF's output air temperature signals? and in what form of communication is the signal? you could probably use the same variable gain amplifier idea mated to another variable gain (for the airflow map) because i would assume temperature is linear to the density which is linear to the amount of oxygen being put in a combustion chamber.
#49
Do MAF's output air temperature signals? and in what form of communication is the signal? you could probably use the same variable gain amplifier idea mated to another variable gain (for the airflow map) because i would assume temperature is linear to the density which is linear to the amount of oxygen being put in a combustion chamber.
Agree with the VE and the cylinder filling not being the only restriction. I'd say that graphing airflow against linear increase of RPM will give a better idea.
Anyhow - the VE thing was just a side project because it was an approximation in my calculations of peak flow. In order to "fit" in the curve of the late AFM the engine must be flowing around than 480kg/hr of air, based on cool dry sea level air weighing 1.2kg per cubic metre. This means that VE would be about 90% at max RPM. I reckon that VE will be higher (over 100%) further down the RPM curve where the resonance in the intake tract is more favourable.
Anyhow, the MAF is the first step in getting the intake really working.
Does anyone out there have the output curve for the Ford Taurus MAF - or any others? This would all be a lot easier with a linear output :-)
#50
Race Car
#51
Intermediate
Join Date: Apr 2007
Location: Boston, MA
Posts: 43
Likes: 0
Received 0 Likes
on
0 Posts
well keep working, if i had some know how or spare funds id be in your position for sure. if you do stumble across a well working set of equations and input/output tables you could surely make alot of 'listers days, so keep on keepin on.
#52
For the MAF adapter I have equations that work perfectly and a circuit design that's not quite there yet - and once it's built I guess there will be a lot of testing to be done. I'll post the excel models and the schematic if I run into any trouble - that way any real electronics guys on the list should be able to help.
The thing I'm sacrificing is flexibility - the circuit will work one way, and one way only - hand built for the MAF curve in question.
The thing I'm sacrificing is flexibility - the circuit will work one way, and one way only - hand built for the MAF curve in question.
#53
Addict
Rennlist Member
Rennlist
Small Business Partner
Rennlist Member
Rennlist
Small Business Partner
bad_monkey - look at the FRWilk page again, you have mis-read it.
The Transfer Function for the late DME is also a curve, not linear. Don't confuse 'flap-angle' with 'flow-rate'.
-Rogue
The Transfer Function for the late DME is also a curve, not linear. Don't confuse 'flap-angle' with 'flow-rate'.
-Rogue
#54
Also, the graph is flow in m^3 / hr rather than kg/hr. I'm using a density of 1.2 kg per m^3 as mentioned before - and doing the all translations leads us to a weird conclusion..
Assuming the flap angle / flow relationship is the same for the late AFM as the early then the upper part of late AFM curve is very similar (within 3% or 4% and parallel) to the HFM 5 curve - but deviates considerably in the lower 6th of the flow axis.
The reason appears to be that the HFM5 is designed to detect backflow - the reversal of direction of the intake air as the valve shuts off. This is due to air being elastic- i.e helmholtz resonance in the intake.
So, in order to indicate backflow there needs to be sensor resolution for that backflow - and the HFM 5 signal crosses the y axis at about 1v in order to provide that resolution. 0->1v indicates backflow.
So, I'm going to look at correcting this part of the MAF curve to match the AFM curve - and also trim the upper part to an exact fit.
But if Rogue Ant hadn't raised that point I'd have been quite confused when I came to test the circuit I had designed....
So, cheers -
#56
Thought I'd share my spreadsheet of mad science.
The graph pretty much says it all - the serious variance is all below 60kg/hr or so, and I'm designing a circuit to reduce the curve proportional accordingly below this point.
Also of note, if you increase the density of the air from 1.2 kg/m^3 to 1.3 kg/m^3 things line up much better.
How are folks going with the restrictor idea?
The graph pretty much says it all - the serious variance is all below 60kg/hr or so, and I'm designing a circuit to reduce the curve proportional accordingly below this point.
Also of note, if you increase the density of the air from 1.2 kg/m^3 to 1.3 kg/m^3 things line up much better.
How are folks going with the restrictor idea?
#57
Three Wheelin'
You may find this article interesting regarding vane type air flow meters and tuning.
http://autospeed.com/A_109878/cms/article.html
It appears that the AFM may not be such a bad thing after all.
http://autospeed.com/A_109878/cms/article.html
It appears that the AFM may not be such a bad thing after all.
#58
Addict
Rennlist Member
Rennlist
Small Business Partner
Rennlist Member
Rennlist
Small Business Partner
bad_monkey,
I would also design an 'averaging circuit' to condition the MAF input/output signal. A MAF signal is going to be much more 'jumpy' then an AFM (due to much faster response and sensitivity). And considering the low polling rate of the DME (88hz), the signal needs to be smoothed some.
-Rogue
I would also design an 'averaging circuit' to condition the MAF input/output signal. A MAF signal is going to be much more 'jumpy' then an AFM (due to much faster response and sensitivity). And considering the low polling rate of the DME (88hz), the signal needs to be smoothed some.
-Rogue