AFM measurement/backpressure (was Turbo speed sensor)

[Mike (Perth, Western Australia)]

At 12:01 AM 9/03/2000 +0800, you wrote:
>>I suppose in those conditions a simple pressure sensor off the exhaust
>>before the turbo would be OK.
>
>>ie Short (cooled) piece of pipe to carefully mounted sensor.
>
>Most pressure sensors will handle only about 180C.

Oh yes - have one of those Motorola silicon bridge units, max is about
75deg C

> Cooling the pipe
>reduces the pressure as far as I know; you need to know the amount
>of cooling (temperature sensors?) to determine the actual
>backpressure.

Doubt it - flow is effectively zero and the sensor need only be about
300mm from the exhaust and behind the heat shield. I'm not thinking of
using any forced cooling normal convective is fine. Once its in steady
state the pressure at the sensor will follow the point at which the pipe
enters the exh manfiold (I would exspect) quite quickly, actually 200mm
behind heat shield would make it OK (torch into garage at 1am, trip over
bin I left there - ouch),

>That's a bit more number crunching than you might want to do. Even
>then, the pressure you determine will be the pressure at some time
>in the past - that might be as long as 10 milliseconds ago. You
>almost certainly will not see the pulsations resulting from the
>opening and closing of valves.

Not interested in pulsations (at moment) but the average pressure over a
exhaust cycle (though any pulsations could be interesting, though I'm
not sure of the b/w of the sensor, but IIRC its better then 1mS), I'd be
plumbing the pipe into a point just before the turbo flange and the length
of pipe to the sensor may very well dampen any fluctuations well major
peaks anyway.

>>The speed wouldn't be much of an issue for this reason. Once its installed
>>and calibrated then the conditions which lead to a immediate ignition
>>retard can be 'pre-emptive' ie. a differentiator off the pressure sensor
>>could initiate an interrupt for immediate retard by a set number of
>>degrees - or even better, the differentiator sets the interrupt pulse
>>width depending on rate of 'upward' change of pressure in exh before turbo.
>
>The condition which is being avoided is that of knock. A knock
>sensor is apparently out of the question to aid in learning the
>engine's response under a range of conditions. It is theoretically
>possible that a single knock event can destroy a fragile engine.
>However, you don't need to run the engine to severe knock
>conditions, and once you've learnt the response, you can use that to
>avoid the knock condition by a safe, but minimal margin.

Yeah sure, but you can see that a 'pattern' of AFM, speed, advance,
exh pressure - could be recognised as one which leads to knock as learned
from previous cycles. Hence when this 'pattern' or a probabilisitc part
of it appears then the level of advance is reduced - the greatest weighting
factor would likely be combination of load and exhast backpressure, in that
case a 10mS response time would be adequate (quick mental calc) but I
expect it to be as good as 1mS.

>An alternative may be to build pressure sensors into the head of the
>engine; those devices must exist (presumably you could buy them if
>you had the money). Another option may be to monitor the speed of
>deflection of the crankshaft - the knock pressure will cause the
>angular velocity to exhibit a characteristic "jitter". You won't be
>able to measure that easily at the flywheel end - the inertia will
>"squelch" the jitter, but a measurement at the other end could be
>worthwhile if there are no inertial loads there.

>A PLL-type logic could be used to detect the jitter from a
>"crank-angle" sensor if the resolution were high enough.
>
>Of course, that's an empirical approach...

mmmm <groan> seems a bit complex when it should be fairly predicatable for
a given engine when the bulk of knock inducing parameters are known
in advance, unless its a test engine with frequent alterations,

>The exhaust backpressure determines the internal exhaust gas
>recirculation and hence the remaining displacement to be filled with
>fresh mixture. That's the air-fuel ratio - remembering that the two
>differ by about 400C (if not more) in temperature, making for a more
>volatile mixture (hence the need to retard with more EGR).
>
>At high speed, we can consider the two gas mixtures to be
>stratified; they won't change significantly in volume, nor will they
>mix to a significant degree.

mmm With you so far,

>We can measure "exactly" how much fresh mixture enters the cylinder
>by measuring the flow through the inlet tract, but only if the
>exhaust valve has already closed. What we need for that is a
>high-speed differential pressure sensor and two pressure taps ahead
>of the inlet valve. The inlet temperature is fairly constant and we
>should already know what this is for calculating the amount of fuel
>to inject...

mmm But surely the AFM - even a slow one will point to the overall
flow, cycle to cycle variations are interesting but I can't see
how its information value is of immediate use given the engines
unlikely to have instantaneous speed changes - hence we can use
the s/w to pre-empt the knock condition.

Don't get me wrong, its an interesting proposition but, I seem to
have got myself fixated on the pattern matching pre-emptive value
in AFM, speed, advance, exh pressure etc. Especially as I hope to
push my own setup a bit higher and not having a lot of engine
instrumentation has me at a loss - exhast backpressure sounds like
a useful qty to know - at least for experience - that and EGT.

>If there's an exhaust valve overlap with inlet, and we don't know
>the exhaust backpressure, then we must measure if the overlap causes
>exhaust to be drawn back into the cylinder, or if a significant
>portion of the fresh mixture is being "scavenged" into the exhaust.
>Enter your trusty O2 sensor. You need it to react fast enough to
>detect a rise in O2 after the exhaust valve to indicate scavenging,
>or to remain "level" indicating only exhaust gases downstream of the
>exhaust valve (obtaining such a sensor's a problem for higher engine
>speeds).

mmm OK got it, but why measure it when its known from the engine
layout that its going to happen at combination of AFM, speed, etc

>	Even knowing the mean exhaust backpressure wouldn't be
>	sufficient on it's own to determine the degree of EGR. Flow
>	dynamics of the exhaust valve closing will often result in a
>	disproprotionately-high exhaust flow for the closing valve
>	as the columns of exhaust gas "sucks" harder through the
>	diminishing opening.

mmm Interesting,

>Any exhaust gases actually drawn back will add to the base EGR of
>the engine and will be "cancelled out" by a reduced fresh mixture
>intake. We can still calculate the AFR based on our knowledge of
>volumetric efficiency of the engine.
>
>Prior testing with the engine will indicate appropriate ignition
>timing to avoid knock at a particular AFR for a specific load/boost.

Yes, does sound like a potential issue thats helpful, I need to think
a little more how far I go on my own setup.

I found out the other day, my block as 3 knock sensor locations along the
length of the block though only the center one is used.

Tah,



Rgds ~`:o)

Mike Massen         Trading as "Network Power Systems" and "Network Computers"
Perth, Western Australia  Ph +61 8 9444 8961  Fx +618 9264 8229 (fax -> email)
Products/Personal/Client web area at http://www.wantree.com.au/~erazmus
 (Current pics - trip to Malaysia to install equipment in jungle power site)

Some say there is no magic but, all things begin with thought then it becomes
academic, then some poor slob works out a practical way to implement all that
theory, this is called Engineering - for most people another form of magic.

----------------------------------------------------------------------------
To unsubscribe from diy_efi, send "unsubscribe diy_efi" (without the quotes)
in the body of a message (not the subject) to majordomo at lists.diy-efi.org