ChassisDyno (2005-02-27 08:35:54)

A dyno is used for mapping the torque output (on the tires or crankshaft) against loadsites (RPM and MAP and other factors).

Dynoing can be done in 2 ways:


If you want to build a water-brake dyno, here are some hints. You have to do more research, of course:

dyno.jpg


There are some code in MS-AVR GenBoard (on uC for logging and perl that runs on the PC for processing) to help mapping the ignition advance of a factory ignition and tuning the engine to max performance without a classical dyno:

Here are some logs from MembersPage/MichaelKristensen's car (this is converted from binary to text with a perl oneliner):

http://home.caffrey.dk/megasquirt/dump/log1.gz

I'm not sure if csv can contain formulas, so in xls form (oops, this is many megabytes, avoid this, use the log1.gz above...): http://www.x-dsl.hu/genboard/dyno/log1.xls


here is the comm.c part (see UartComm for more details) that generated the (original binary) log:\nÿ1ÿ


The ugliness comes from the traditional megatune-protocol compatibility:

the fine precision rpm is available: RPM = 100*( v13 + v21 / 256);

See comm.c (above) for the slight differences relative to the traditional 22byte megatune 'A' output. It contains precision RPM (a must for dyno-ing) and also the measured ignition advance (measured with GenBoard INT6 input relative to the missing tooth index signal of the 36-2 trigger on IC3 input).

If someone feels like playing with it, extract the

Publish programs or formulas used (perl or octave (==matlab) preferred, but others also good)

Partial results also welcome.

What can you expect? If you collect data per loadsites (RPM/MAP values), you will find 4 or 5 angular acceleration (dRPM/dt) layers. These are according to the gear ratios. At 4th gear you will find much lower acceleration at the same (MAP,RPM) loadsite than in 1st gear. If you have a VSS (vehicle speed sensor) data logged, this is simpler. Even than, there will be some noise (layers not very narrow) because of hill-climbing and downhilling. This can be corrected if you log GPS data at the same time (GPS modules are available <90 Euro). Even with flat-areas or GPS-elevation-correction you will have finite-width layers because of changing wind-vector. This can be decreased with averaging.

To know the output of the engine, besides the acceleration force F1 = a * m = dv/dt * m = dRPM/dt * const/gearslowing * m you need to add another force: F2=friction force

which can be measured by logging fuelcut deacceleration. You deaccelerate over the useful RPM-range in all gear-ratios. In this fuelcut mode, MAP does not count, only RPM and gear. You will use the same F2 for any MAP. From this F2 you can find out the CW air-drag coefficient of your car: you need to find the 2nd order coefficient in the F=a*v*v + b*v + c formula (hint: simulated annealing of above a,b,c coefficients and v-initial variable targetting the minimal square of error of diff between measured and calculated deceleration curves - I got some perl code - Marcell).

Unfortunately I do not know Mik's gear ratios, tiresize and weight.

See also GenBoard/TuneAndDrive and MembersPage/RedMist/SoftwareDyno

[opensource street-dyno software and some theory]

A non-opensource software to evaluate data: http://www.tweecer.com/StreetDyno/index.html

There is a chassisdyno yahoo group.


See also