Date: Tue Jan 18 10:43:44 1994
Subject: filing down spark plugs
I recently read in the latest Hot Rod magazine about a method to
increase the performance and fuel economy. The method was to file away
the top electrode till its edge was concident with the axis of the
central electrode. Any luck with this method.
[Probably not and in any event, none that Hot Rod could measure. When
you read increased horsepower claims such as the 5 hp claim
SplitFire likes to claim for their plugs, consider this.
I have a nice little side business repairing SuperFlow dynomometers, the
overwhelmingly dominant dyno in the US. Every magazine article I've
ever read used a SuperFlow. The standard SuperFlow is rated at 1000 HP,
10,000 rpm and 800 ft-lbs of torque. The RPM signal is converted to a
voltage by a tach chip before being submitted to an A/D converter. The
torque signal is derived from a strain gauge attached to the absorber.
This signal is also applied to the same A/D converter through an analog
mux. Horsepower before SAE correction is the simple calculation:
(torque (ft-lb) * RPM ) / 5252
This computation is done in an analog multiplier for the analog readout
and by the CPU for the digital readout. So good, so far. But here's
the kicker. The A/D converter is an 8 bit unit. That is, it digitizes
the incoming signal into one of 256 binary values. For torque, that is
800 ft-lbs / 256 = 3.13 ft-lbs per bit. For RPM, 10,000/256 = 39 rpm
per bit. At a constant 6000 RPM, the best HP resolution is 3.5 hp. At
a constant 500 ft-lbs of torque, the best HP resolution is 3.7 HP. This
lack of precision results in the best theoretical HP measurement at 6000
RPM being +- 3.5 hp. Worst case is 3.5 + 3.7 = 7.2 hp. The
root-sum-square (much more representative of the real world) is 5.0 hp.
The precision varies, of course, with RPM. The important point is any
horsepower variation less than about 5 hp is meaningless and is more
likely attributable to quantitizing error in the electronics. Understand
that this does NOT include other systematic error terms such as the
errors associated with the analog electronics or the torque sensor
calibration. I personally attribute no credibility to differences
less than 10 hp.
The other thing to keep in mind when viewing published figures is that
the most frequently published numbers are corrected to SAE Net. This
correction for ambient temperature, humidity and barometric pressure
is only approximate and is really suitable for generating numbers for
ad copy where they are legally required. We have conclusively proved
that the correction is only approximate using a client's dyno cell
that is equipped to control temperature, humidity and baro pressure.
To illustrate the problems involved, I've spent considerable time with a
client because his dyno isn't "producing the numbers he wants". His
engines, which he sells to racers who make buying decisions largely on
dyno sheets, are considerably down on power compared to what his
competition claims. His dyno is spot-on calibrated. He has carried an
engine around to two other shops, one of which is Bill Elliot's shop in
Dawsonville, GA. The span of readings on this engine among the three
dynos is over 80 HP on a 500 hp engine! I have personally checked two
of the dynos and know them to be properly calibrated. The difference is
in the buildup of error terms in this inherently inprecise measurement
system and in the SAE net compensation between Florida at sea level and
here in Atlanta at about 1000 ft elevation.
Bottom line - take any claims of small increases in HP due to "tricks"
with a LARGE grain of salt.
From: Dave Baker
Subject: Re: ALUMINUM FLYWHEEL? yay or nay....
Date: 08 Oct 1998
>From: Mike Kohlbrenner
>And in the same vein -- have you ever worked with a DynoJet?
>Those devices are effected by the rotating masses in the
>vehicle being tested. As a result, they do not give an
>absolutely accurate measure of a car's performance. They
>certainly don't measure real engine horsepower. They also
>do not give an accurate portrayal of acceleration
>performance. For them to be really accurate, the ratio of
>the inertia of the testing drum (in terms of horsepower
>requirements for acceleration) to the rotating inertias of
>the car being tested would have to be exactly equal to the
>ratio of the car's linear inertia to the rotating inertias.
We don't have the Dynojet machine over here as far as I know but there are 3
types of rolling road dyno which I am familiar with.
1) - Clayton type water brake dyno's. 2 rollers in the normal way and an
internal viscous type torque converter to measure the applied force to the
2) Electric eddy current dyno's - Hoffman and Sun are the usual makes.
3) Massive flywheel Bosch system which a good friend of mine has in his race
workshop. A big cast iron "flywheel" is on the end of the driven roller and the
machine's electronics knows the inertia and calculates power from this. You
can't unfortunately hold a car at steady rpm and load as there is no "brake" as
such in the system.
All have their good and bad points but IMO they are all capable of reasonably
accurate results if calibrated and used properly. I have a beef with systems
with small diameter rollers which seem to suffer from tyre slip more often. The
Claytons fall into this category and I have seen weird results with highly
tuned engines. A guy I met once had a race car with a supposed 150 bhp wheel
figure on a Clayton system i.e 175 or so at flywheel. He was not competitive
with this engine and the dyno operator sent him to me. I built a complete new
engine for him which was set up on a Superflow engine dyno at exactly 141
flywheel bhp. He was most disappointed as the power curves kept coming and used
to wander off behind the dyno hut for a ciggie and a sulk. He took the car out
for the first time a few weeks later and knocked 4 seconds off his previous
best lap time and set 4 new UK track records in the next 6 races.
On another eddy current rolling road the car showed 116 bhp at he wheels some
time later which would be spot on for the measured flywheel figure.
My best guess is that his original engine had no more than 130 bhp true
flywheel figure but the engine builders never had it set up on an accurate
engine dyno so he doesn't know. Claytons have always made me suspicious after
My BIG bugbear though is with so called transmission losses measured on the
overrun. Most dyno firms here now insist on giving "flywheel" bhp printouts and
often the measured wheel figure is never even shown. IMHO it is IMPOSSIBLE to
measure trans losses on a rolling road dyno.
What happens is this - you run the car under load until just after peak power
then the operator drops the clutch and lets the car freewheel down against the
roller which supposedly measures the "negative" power absorbed and treats this
as a trans loss.
Now one thing is for certain - trans loss is a percentage of power input to the
system. If the car is in neutral there is no power being fed in. Also the gears
are now backlashed against the wrong face of the gear teeth as the roller is
driving the car not the other way round. Under these circumstances I see no way
in which the trans loss shown can bear any relationship to actual losses when
the car is under power.
Tweaking the electronics to generate big overrun losses or just using a touch
of brake pedal while the car is winding down makes for big trans losses and a
nice fat flywheel power curve to keep the punters happy. I have a number of
rules of thumb about transmission losses which I have established over the
years from known engines.
On most Front Wheel drive cars trans losses are between 15% and 17% of the
flywheel figure. VW themselves quote 15% as being an average transmission loss
for their cars. (by trans loss I mean all losses between the flywheel and the
road so it includes gearbox, final drive and tyre losses). Low powered cars
tend to have a higher % trans loss because tyre losses are more of a constant
than a % of power input and so represent a bigger proportion of the engine
power than they do for powerful cars.
Rear Wheel drive cars can have 2% or so higher losses due to turning the drive
through 90 degrees before it gets to the wheels but they also often have direct
drive in 4th gear which cancels this out.
Trans losses will vary with the gear in which the car is tested. This is mainly
due to the higher wheel speed in a higher gear leading to greater tyre losses.
Always use the same gear and same tyre pressures to make comparisons meaningful
for power runs taken at different times.
ALWAYS use the measured WHEEL bhp figures and if you are desperate to know a
flywheel figure then add a notional trans loss yourself. The ONLY WAY to
accurately know flywheel bhp is to take out the engine and put it on an engine
My chart for trans losses for FWD is as follows:
wheel trans flywheel
bhp loss bhp
80 17 97
100 19 119
125 22 147
150 26 176
175 30 205
200 34 234
If the dyno you use shows much higher losses than this then you are being told
As to your original point about rotating inertia - I agree that the dyno system
cannot "know" this about a particular car but some of the systems I have seen
take an input for the vehicle weight and use some compensation on the basis
that a heavy vehicle will have larger rotating bits. If the run is taken in a
high gear then the effect of rotating components should not be large due to the
slowish acceleration rates generated. I think the error in measured engine
power should not be more than 2% and I see bigger variation in power figures on
different rolling roads than this anyway. I tend not to use rolling road power
figures as absolutes as I don't think the machines work to close enough
tolerances. They give a decent guide to power and are of course mainly of use
in setting up fuel and ignition curves.
For accurate work you need an engine dyno where you can control oil and water
temps to get repeatable results.
Dave Baker at Puma Race Engines (London - England) - specialist flow
development and engine work.