Turbo boost gauge

[sawood12]

Sorry Graham, though I can't claim to be 100% sure about my current understanding I'm not sure I understand your explanation. How can the AFM detect overboost when it is fully open and delivering max voltage at about 4kprm on high throttle positions? With airflow being invisible to the DME at 3/4 throttle position to WOT all there is to detect overboost (which you're only really going to get at high throttle settings and high RPM) is the KLR's MAP sensor. Hence it is the case that though it is the DME which actually commands overboost protection, it can only do that once the KLR detects it via the MAP sensor and lets the DME know. Also i'm unsure about how the CV "control's" boost - implying some form of closed loop control system. I've tried to understand this from various threads and descriptions, but they always fail to fully explain it for me. For example, if you were to fit a Boost Enhancer the pressure signal is hidden from the CV rendering it useless until such time the BE cracks open which then will open the wastegate. If the ECU was commanding the CV to control boost and it was not seeing any effect of its inputs, then this would surely trick the ECU into thinking there was something wrong and trigger overboost protection. Also if the CV was somehow controlling boost then why do you get leaky wastegates affecting boost performance? If the CV was monitoring this degradation in boost performance then it would surely end up behaving like a Boost Enhancer and hide the pressure signal from the Wastegate unit such time max boost was achieved?

 

[GPF]

ORIGINAL: sawood12 How can the AFM detect overboost when it is fully open and delivering max voltage at about 4kprm on high throttle positions? With airflow being invisible to the DME at 3/4 throttle position to WOT all there is to detect overboost (which you're only really going to get at high throttle settings and high RPM) is the KLR's MAP sensor. Hence it is the case that though it is the DME which actually commands overboost protection, it can only do that once the KLR detects it via the MAP sensor and lets the DME know.

OK, this is probably not the kind of thing to debate having just got back from the pub, so I'll probably have to edit this tomorrow![] The "AFM maxes out at 4K" is a popular misconception, although the rate of voltage change does tail off. When the dme senses WOT (which does not relate to throttle position, but to load via the AFM voltage) it switches to it's WOT maps for both fuel and timing. These tables only have one value for each rpm band, of which there are 16 spanning from 1000rpm upto 6240rpm. The arflow voltage is used to scale the fueling values in these tables - not sure about the timing. There is another table with allowable AFM voltages at 12 rpm bands from 1480rpm to 6240. If the dme sees a voltage from the AFM that exceeds the values/rpm in this table for longer than a set time, or if the voltage is rising faster than it considers normal it triggers the overboost protection. This is not the "limp mode" but an instant fuel cut. As I said earlier, the KLR and it's map sensor have nothing to do with this. Aftermarket chips raise or defeat the overboost protection by increasing the allowable AFM voltages in this overboost table - the 12 values start at address 16EE in the binary on the DME chip. If you have bypassed the CV, you do not need an aftermarket KLR chip - just the DME chip......

ORIGINAL: sawood12 Also i'm unsure about how the CV "control's" boost - implying some form of closed loop control system. I've tried to understand this from various threads and descriptions, but they always fail to fully explain it for me. For example, if you were to fit a Boost Enhancer the pressure signal is hidden from the CV rendering it useless until such time the BE cracks open which then will open the wastegate. If the ECU was commanding the CV to control boost and it was not seeing any effect of its inputs, then this would surely trick the ECU into thinking there was something wrong and trigger overboost protection. Also if the CV was somehow controlling boost then why do you get leaky wastegates affecting boost performance? If the CV was monitoring this degradation in boost performance then it would surely end up behaving like a Boost Enhancer and hide the pressure signal from the Wastegate unit such time max boost was achieved?

The KLR controls boost by directing more, or less, air pressure to the top port on the wastegate by switching the CV in and out. It is indeed a closed loop control system. That's what the CV is for - it's just like the solenoid valve on an aftermarket EBC. If the KLR & CV were not in the loop, the full (air) signal would go straight to the wastegate and you would only ever get the maximum boost that the wastegate spring can resist, which would be much less than standard boost levels. When you fit a boost enhancer, you are blocking the signal to the wastegate that has been triggered by the KLR in the first place to prevent the wastegate from opening gradually - which is what the KLR is trying to do. A weak wastegate opens early because it cannot resist the pressure that the exhaust gasses exert on the back of the wastegate valve. This opens the wastegate before the pressure from the inlet side pushes it open - hence it opens too early and you can't make high boost.

 

[barks944]

ORIGINAL: GPF The "AFM maxes out at 4K" is a popular misconception, although the rate of voltage change does tail off. When the dme senses WOT (which does not relate to throttle position, but to load via the AFM voltage) it switches to it's WOT maps for both fuel and timing. These tables only have one value for each rpm band, of which there are 16 spanning from 1000rpm upto 6240rpm. The arflow voltage is used to scale the fueling values in these tables - not sure about the timing. There is another table with allowable AFM voltages at 12 rpm bands from 1480rpm to 6240. If the dme sees a voltage from the AFM that exceeds the values/rpm in this table for longer than a set time, or if the voltage is rising faster than it considers normal it triggers the overboost protection. This is not the "limp mode" but an instant fuel cut.

Whats the maximum flow rate of the AFM on the 951? I wonder if the DME also uses this table of AFM voltages to calculate the load of the engine when its not operating from its wide open throttle maps? In order to calculate load using the AFM the DME must understand what flow rate indicates maximum load at the current engine speed. This overboost information achieves that rougly by recording what the maximum expected flow rate should be at any given engine speed. If say the current flow rate as read from the AFM is half the maximum at the current RPM then you could calculate that your at 50% engine load and read from the corresponding position on the AFR and Ignition maps.

 

[GPF]

ORIGINAL: barks944 Whats the maximum flow rate of the AFM on the 951? I wonder if the DME also uses this table of AFM voltages to calculate the load of the engine when its not operating from its wide open throttle maps? In order to calculate load using the AFM the DME must understand what flow rate indicates maximum load at the current engine speed. This overboost information achieves that rougly by recording what the maximum expected flow rate should be at any given engine speed. If say the current flow rate as read from the AFM is half the maximum at the current RPM then you could calculate that your at 50% engine load and read from the corresponding position on the AFR and Ignition maps.

Don't know what the maximum actual flow rate is - as you say the dme is deriving, or implying, that from the voltages it sees. There is a set of 3 tables that defines the AFM transfer function and it uses these, probably in conjunction with some coding get to "load" - but that is probably a notional value anyway. It doesn't use this overboost table in any of the Part Throttle calcs. You can put max values in the overboost table and part throttle performance is completely unaffected. For part throttle there are 2D tables for fuel and timing with rpm on one axis and load on the other. The load comes from the afm voltages as translated by the transfer function as above.

 

[barks944]

I believe the transfer function is what converts the measured voltage into air flow rate rate. The reason its not easy to do a MAF conversion is because this transfer function cannot map the voltage curve of a MAF sensor to the indicated mass air flow rate. This is due to the shape of a MAF curve so even if you change the transfer function parameters you cannot get the mapping correct. This info came from Rogue_Ant. There is another calculation involved to give load at any given RPM. Perhaps it use's volumetric efficiency, it could calculate that knowing air flow and engine speed and the capacity of the engine. Edit: To further clarify what I'm babbling on about. The key parameters for ignition mapping are engine speed and load. Load is essentially the cylinder fill, for a NA engine full load would mean that the cylinder pressure reaches atmospheric pressure * VE during its intake stoke. Cylinder fill (assuming good AFR) is the major factor in determining burn rate and thus is a major factor in determining when the cylinder makes its maximum pressure. Higher load (cylinder fill) means a faster burn rate. In order for this maximum pressure to occur at TDC for peak performance you need to consider engine speed as well so you can start combustion at a point in time that will mean the combustion completes as the piston hits TDC. The reason I think you need to consider more than the AFM voltage is that it doesn't tell you the cylinder fill without considering the engine speed. To get the same cylinder fill, or equally the same load, at 1000rpm as 2000rpm the air flow rate at 2000rpm would have to be double that at 1000rpm. This is also where vacuum advance comes from, an increase in manifold vacuum which is related to engine load or cylinder fill indicates a reduction in load and thus an increase in ignition advance ensures peak pressure is reached at TDC. However calculating cylinder fill is a more relevant parameter for ignition mapping as engines typically reduce in volumetric efficiency as engine speed increases. Therefore manifold vacuum does not directly relate to cylinder fill as restrictions come into plat at high revs that reduce cylinder fill despite having a consistant manifold vacuum. Although saying that manifold pressure does provide a consistent parameter to map against assuming the engine does not alter its volumetric efficiency at a given load site.

 

[sawood12]

ORIGINAL: GPF OK, this is probably not the kind of thing to debate having just got back from the pub, so I'll probably have to edit this tomorrow![] The "AFM maxes out at 4K" is a popular misconception,

Well in that case that blows my understanding out of the water. I must drink more beer!

 

[barks944]

The lux afm can measure unto 400 m3 per hour. Can anyone convert that into a rough hp estimate? Would be interesting to know if it can measure up to 160hp.

 

[Diver944]

ORIGINAL: robwright they must be based on models but how can I put are pretty generic maps.  So a custom map based on your own dyno figures is the best way to go then.  Question is how much does one of those cost?  And who does it? 

Most responsible chips contain a generic map for a given level of boost and assume everything else in the system is working properly ie wastegate, injectors, fuel pressure etc. Just as the factory chips are set for 0.75 bar of boost then the afftermarket ones are set for 1.0bar or maybe 1.1 bar. As long as the wastegate is set for that figure (and not something else) they will work fine. This solution will be circa 95% perfect for a mildly modified car To gain as much as is possible from your specific setup on your specific car then a full rolling road remap would be the way to go and the most popular, well thought of guy in 944T circles is Wayne Schofield of Chip Wizards in Rochdale. Last I heard he charges around £400 for a full day on a 944T. He has lots of experience with tuning racing Porsches and has good mechanical skills too so can spot many of the common problems that maybe causing your car to underperform before he even starts.

 

[GPF]

ORIGINAL: barks944 Seems like theres a lot of differenent understanding of these systems. I can tell you the following things I am very sure are true. I know this as I have been reverse engineering the circuits to do my project. The ignition signal from the DME is sent to the KLR which sends back a signal which directly fires the coil circuitry in the power board of the DME with no input to the microprocessor on the DME. The DME does not know if knock has occured. The DME does not know boost pressure. It derives load from the AFM signal and assumes full load when it sees a full load signal which is generated by the KLR. My Assumptions: I believe, although without certainty that the KLR manages overboost protection. This is why you must chip the KLR even if a manual boost controller is used. Otherwise it will retard ignition or w/e. I think knock protection is simply triggering limp mode, the KLR does not manage knock or adapt ignition to prevent it, it just protects the engine if it detects a certain frequency of knock. Edit: As part of my work on this engine management system I am compiling somthing of a disection of the 944/951's management system. Would appreciate a people reading what I write and giving some feedback in order to create a good shared understanding of the system. Any takers?

Sorry Tom, missed this. I agree with most of this but there are one or two things I'm not so sure about. Some of them have come up in later posts, but just for completeness....

ORIGINAL: barks944 Seems like theres a lot of differenent understanding of these systems. I can tell you the following things I am very sure are true. I know this as I have been reverse engineering the circuits to do my project. The ignition signal from the DME is sent to the KLR which sends back a signal which directly fires the coil circuitry in the power board of the DME with no input to the microprocessor on the DME.

I believe that this is how the KLR takes the initial signal from the dme to either pass it straight through or delay (retard) it by either 3 or 6 degrees if it has seen knock.

ORIGINAL: barks944 The DME does not know if knock has occured. The DME does not know boost pressure.

Agreed.

ORIGINAL: barks944 It derives load from the AFM signal and assumes full load when it sees a full load signal which is generated by the KLR.

Not sure about that. It derives load via the AFM signal, so why does it need a "full load signal" from the KLR? In fact, where would the KLR get the information from? It is not using pressure (KLR vac line can be disconnected and plugged with no effect) and it doesn't get AFM volts. I think that once load simply goes past the highest loading in the Part Throttle maps (which is way before 100%), the DME then just switches to it's WOT maps. In fact, consider an early NA, do they even have KLRs? They still need to transition from Part Throttle to WOT though...

ORIGINAL: barks944 My Assumptions: I believe, although without certainty that the KLR manages overboost protection. This is why you must chip the KLR even if a manual boost controller is used. Otherwise it will retard ignition or w/e. I think knock protection is simply triggering limp mode, the KLR does not manage knock or adapt ignition to prevent it, it just protects the engine if it detects a certain frequency of knock.

This is not correct - the DME manages overboost via it's overboost tables by cutting fuel and you do not have to chip the KLR if an MBC is used - because you have removed the CV there would be no point in any case. My first venture into modifying my car was to fit a dual port wastegate and MBC and get Wayne Schofield to chip it. Wayne really knows his stuff and did all sorts of things to it, but he did not replace the factory KLR chip. The KLR does try to manage knock though - by retarding the ignition in 2 stages ; by 3 degrees initially and then by another 3 degrees. If that doesn't work it then triggers limp mode where it reduces boost - but it can only do that if the CV is still in play. (Just think of all the pissed off KLR's out there!)[] Not 100% sure if any other, boost related, circumstances also trigger the KLR limp mode but I don't think so.

ORIGINAL: barks944 Edit: As part of my work on this engine management system I am compiling somthing of a disection of the 944/951's management system. Would appreciate a people reading what I write and giving some feedback in order to create a good shared understanding of the system. Any takers?

I think that's a great idea. I'd love to understand more about how the system actually works, so I'd happily read anything you come up with.[] PS - Sorry if the post comes across as a bit brusque, I'm just in a bit of a rush!

 

[barks944]

ORIGINAL: GPF

ORIGINAL: barks944 Seems like theres a lot of differenent understanding of these systems. I can tell you the following things I am very sure are true. I know this as I have been reverse engineering the circuits to do my project. The ignition signal from the DME is sent to the KLR which sends back a signal which directly fires the coil circuitry in the power board of the DME with no input to the microprocessor on the DME.

I believe that this is how the KLR takes the initial signal from the dme to either pass it straight through or delay (retard) it by either 3 or 6 degrees if it has seen knock.

I agree with the above conclusion although I am not sure how much the signal from the DME is retarded.

ORIGINAL: GPF

ORIGINAL: barks944 It derives load from the AFM signal and assumes full load when it sees a full load signal which is generated by the KLR.

Not sure about that. It derives load via the AFM signal, so why does it need a "full load signal" from the KLR? In fact, where would the KLR get the information from? It is not using pressure (KLR vac line can be disconnected and plugged with no effect) and it doesn't get AFM volts. I think that once load simply goes past the highest loading in the Part Throttle maps (which is way before 100%), the DME then just switches to it's WOT maps. In fact, consider an early NA, do they even have KLRs? They still need to transition from Part Throttle to WOT though...

The full load signal from the KLR replaces the WOT switch on the normally aspirated engines. On the NA engines the DME reads a WOT switch on the throttle sensor and switches to WOT maps whenever it sees the WOT signal. The wide open throttle switch does not exist on the turbo's throttle sensor and is replaced by a potentiometer type measurement which is instead fed to the KLR, for throttle based boost control I suspect (a feature lost when going manual). The KLR then generates a WOT/Full load signal which it sends to the DME when it detects full throttle or perhaps some manifold pressure, I'm not sure. But anyway this causes the DME to switch to its WOT maps. The cars may also switch to their wide open throttle maps above some engine load as well. I edited one of my earlier posts with the reasons I think that the DME uses the AFM to measure load, you might of missed the edit but I'm still convinced this is the case.

 

[barks944]

Basically the 951 management system is a hack job done on the NA system. For some reason they decided to *add* another computer system into the mix rather than designing a single integrated system, which is what I am planning to do. You would have thought that the effort it took to re-design the DME and design the new KLR they may as well have built a single integrated unit to do everything in one box...... I wonder if they had issues with the ammount of I/O on the microprocessors of the day....

 

[GPF]

ORIGINAL: barks944 The full load signal from the KLR replaces the WOT switch on the normally aspirated engines. On the NA engines the DME reads a WOT switch on the throttle sensor and switches to WOT maps whenever it sees the WOT signal. The wide open throttle switch does not exist on the turbo's throttle sensor and is replaced by a potentiometer type measurement which is instead fed to the KLR, for throttle based boost control I suspect (a feature lost when going manual). The KLR then generates a WOT/Full load signal which it sends to the DME when it detects full throttle or perhaps some manifold pressure, I'm not sure. But anyway this causes the DME to switch to its WOT maps. The cars may also switch to their wide open throttle maps above some engine load as well. I edited one of my earlier posts with the reasons I think that the DME uses the AFM to measure load, you might of missed the edit but I'm still convinced this is the case.

So the TPS goes to the KLR rather than the DME? Now you've said that, it is ringing a faint bell. When I installed the MAP kit I took one of the signals from the KLR and it could well have been the TPS. I did see your edit btw, and I agree the DME uses AFM within it's load calcs. Goodness, does this mean we are all actually on the same page now? []

 

[TTM]

ORIGINAL: barks944 Basically the 951 management system is a hack job done on the NA system. For some reason they decided to *add* another computer system into the mix rather than designing a single integrated system, which is what I am planning to do. You would have thought that the effort it took to re-design the DME and design the new KLR they may as well have built a single integrated unit to do everything in one box...... I wonder if they had issues with the ammount of I/O on the microprocessors of the day....

It was probably more cost effective to proceed this way.

 

[sawood12]

I think people are a little too quick to slag off the standard 951 ECU. It was an era where these systems were still in the process of development in a time where most of the competition were still messing around with carburettors. The standard ECU has, in the hands of skilled and talented modders, proved to be able to do 98% of the job of a fully top spec standalone system with all the bells and whistles. It may take alot of effort and some creative thinking to get it that good, but it is capable. Not bad for a 20yr old car and was an ECU that was much more advanced than pretty much anything out there in the late 80's and certainly more advanced that what was on the 911's of the era. It is not really fair to look back to a 20yr old piece of technology and judge it my today's standards.

 

[robwright]

Guys you have talked a lot about retarding the ignition timing but it is puzzling me as to how this is actually achieved. Maybe one of you could explain it to me in a nutshell. Just trying to get my head around it as our engines use distributors. Typically this would normally have been done by physically moving the backplate of the dizzy, usually by some kind of vacuum system. Obviously this is not the case on our cars.

 

[sawood12]

ORIGINAL: robwright Guys you have talked a lot about retarding the ignition timing but it is puzzling me as to how this is actually achieved.  Maybe one of you could explain it to me in a nutshell.  Just trying to get my head around it as our engines use distributors.  Typically this would normally have been done by physically moving the backplate of the dizzy, usually by some kind of vacuum system.  Obviously this is not the case on our cars.

Though the contact points in the distributor are fixed, there is a distance of arc where the rotor is in contact with the dizzy contact. The actual signal that delivers the spark is turned on and off at a specific point and is controlled by the ECU and therefore controls the actual point along the circumference of the arc of contact when it commands the spark. I think the control range is plus or minus 6 degrees from TDC so the circumference of the arc must be 12 degrees and the ECU controls exactly where along that arc the spark is delivered. With the old vacuum system the electrical power was always on at the king lead and the spark would be delivered at the exact point of contact between the rotor arm and the dizzy contact so you had to move the position of the dizzy contacts to achieve ignition retard and advance.

 

[robwright]

Thanks Scott makes sense now you put it that way. Now I can stop googling!!!! [&:] But not drinking however []

 

[barks944]

ORIGINAL: sawood12 I think people are a little too quick to slag off the standard 951 ECU. It was an era where these systems were still in the process of development in a time where most of the competition were still messing around with carburettors. The standard ECU has, in the hands of skilled and talented modders, proved to be able to do 98% of the job of a fully top spec standalone system with all the bells and whistles. It may take alot of effort and some creative thinking to get it that good, but it is capable. Not bad for a 20yr old car and was an ECU that was much more advanced than pretty much anything out there in the late 80's and certainly more advanced that what was on the 911's of the era. It is not really fair to look back to a 20yr old piece of technology and judge it my today's standards.

It does work well as we all know. But there is significant room for improvement and it is a hack job! The 911's of the same era used ML3.1 as in the late model Lux and the 930 turbo used the same system as the 951 I think.

 

[sawood12]

There will always be room for improvement Tom. The 930 system was far more basic. I don't think it had a knock computer and it doesn't have batched fuel injection - the injectors continually spray fuel and it is regulated electronically by the AFM position - just a small step on from mechanical fuel injection where there was a mechanical linkage between the AFM controlling a tap to regulate the flow rate of the fuel.

 

[barks944]

Ah right I did wonder about the 930 as its a much earlier car, was just going on info from FRWilk's site.