MAF sensor accuracy? 1

[fastline12]

I am finishing up on a pretty big CFM flow bench I am looking at possibly using MAF sensing to digitize the setup and drive a VFD through a PLC loop. I am kind of second guessing that whole idea. I killed the idea before for some reason I can now not remember.

If that will not do it, what would you guys reach for? Basically the system is running 2 8-71 helical lobe superchargers driven by a large motor and VFD estimated to pull approx 2000cfm at 28in.

 

[Warpspeed]

O/k so your MAF is putting out exactly 3.053 volts.

How many CFM is that EXACTLY, and how can you be absolutely sure ?

And what if the ambient temperature changes, how many CFM is it then ?

You do also realize that the MAF output voltage versus airflow is a hugely non linear curve.

 

[fastline12]

Yeah, I realize that there would need to be calibrations and I am not real sure that is even the direction I want to go. Most systems will use a calibrated flow plate and monitor pressures but I am kind of thinking out loud here. Obviously simpler is better but accurate is rather important too.

 

[Warpspeed]

Repeatability and sensitivity to change is what you want.
Do it properly, build an orifice flow bench.

 

[Lou Scannon]

Not an LFE?

 

[Warpspeed]

LFE is good, but very expensive for mere mortals.
It also clogs and gradually loses accuracy over time.

A flow measurement orifice costs nothing, and always remains the same.

 

[gruntguru]

MAFs are not sensitive to air temp. The bridge cicuit maintains the filament at a fixed temperature difference above ambient. They measure "mass flow" directly as the cooling effect on the filament is proportional to the number of air molecules that "strike" it (not always what you want when measuring Cd of cylinder heads etc).

They are effected by deposits (dust etc) so they need periodic cleaning (usually by "burn off").

The non-linear output is deliberate and beneficial (maintains % resolution) unlike orifice plates which have a pathetic turndown (they are non-linear in the wrong direction) - you need a few sizes and switching if you wan to cover a wide flow range.

My choice would be MAF, with an orifice or two to spot-check the MAF calibration from time to time.

Engineering is the art of creating things you need, from things you can get.

 

[Lou Scannon]

What I would do for an inexpensive setup is build up the airflow sensing portion of my bench with an MAF sensor installed with 10+ diameters upstream and 5+ diameters downstream of straight pipe. This will be the permanent flow measuring element of the bench, never to be altered. I would take it to a commercial flow bench and get a calibration curve for it. In use I would have a high grade filter ahead of the MAF to keep the element clean. Depending on the accuacy required for individual projects, I might get the calibration curve redone from time to time.

 

[Heatison]

Use a hot wire anemometer and be done with it.

http://www.instrumart.com/ProductLi...30&Landing=1&gclid=CIWXpufglacCFchl7AodcykTcQ

 

[1gibson]

An automotive MAF sensor is a hot wire anemometer... Although that is a good suggestion to receive a calibrated curve.

Without calibrating or checking with an orifice, closest you could get with a MAF sensor is to find the MAF table programmed in the vehicle's PCM. Use a common sensor (GM 85mm, LQ4) and average tables from multiple vehicles (they will have different intake ducting and slightly different values) if possible.

That sensor has a built in screen to straighten the inlet flow. Honeycomb a few mm thick so it does have a straightening effect, it's not just a round wire mesh screen.

 

[Lou Scannon]

My first hand observation is that the screen or honeycomb (depending on the model in question) does more to reduce actual real time measurment "noise" due to turbulence than immunize the sensor output transfer function against different inlet configurations.
A series of screens or honeycombs spaced strategically apart can help somewhat with the latter, arguably at the cost of length, added complexity, and flow restriction (which may or may not be an issue).

 

[Warpspeed]

Yup.
All hot wire anemometer can do is measure the air velocity in the exact physical vicinity of the wire.

If the flow profile is non laminar, or biased to one side, or surging, the hot wire will give a false reading.

 

[kpt4321]

Warp, they don't measure air velocity in the vicinity of the wire, they measure mass airflow in the vicinity of the wire.

The two, of course, are related, but with the potential for nontrivial air density changes under normal operating conditions, they are not the same.

 

[Warpspeed]

It may read accurately where the small heated wire is actually placed, but that may not be representative of conditions across the wider flow area.

You can certainly calibrate a particular hot wire sensor placed in a particular flow path, and this is routinely done in vehicles with some type of lookup table.
Provided absolutely nothing is changed, it may work reasonably well over a span of time during final engine tuning.

As one of the longest term members of the Flow Bench Forum, I can tell you that nobody there, and no commercial flow bench manufacturer uses a hot wire MAF for flow measurement.
The consensus there, is that a non corrected ratiometric measurement against a reference orifice plate is definitely the way to go for a flow bench.

 

[Warpspeed]

The second paragraph should read:

You can certainly calibrate a particular hot wire sensor placed in a particular flow path, and this is routinely done in vehicles with some type of lookup table adjusted during final engine tuning.
Provided absolutely nothing is changed, it may work reasonably well over a span of time.

 

[fastline12]

It certainly sounds like the variables for MAF sensing for this application are too great to get repeatable responses day after day. I realize OEMs either refer to a factory calibration or they create their own for use in engine management.

Warp, what is the easiest yet accurate way to digitally "sensorize" a flow bench? I am trying to stay away from analog stuff so I can capture and review test data.

 

[Warpspeed]

Fastline12,

I suggest you check out the Flow Bench Forum.
There you will find many like minded people, most of them non wealthy enthusiasts, and a sprinkling of well known top professional "names".

Forum members from around the world have solved all the problems and built their own professional quality computerised digital flow benches for basically the cost of the materials.

It is by far the most friendly Forum to be found anywhere on the internet.

http://www.flowbenchtech.com/forum/

 

[fastline12]

Thanks for that. I will venture in that direction.

 

[patprimmer]

Thanks warpspeed. Bookmarked and will inspire a build if budget permits.

Regards
Pat
See FAQ731-376 for tips on use of eng-tips by professional engineers &

http://eng-tips.com/market.cfm

for site rules