Transaxle

Right. The cold start behavior is terrible without NTC.

That was our first approach. Did not work.

It seems not to be a flow problem. It is caused by different pressures.

The point is the "overshooting" of the flap. The signal speed is not a big problem. We tried to do as much as possible in "cheap" software...

Pauerman

Tom,

So your CPU handles the mass air to volume conversion that is typically accomplished by piggyback controllers more commonly used on the 951?

What kind of hp/tq will 1000 kg/hr support?

With the reliefs you've machined into the piston dish area, are you retaining the stock rod length? If so, what compression ratio do you think you're running?

Vic

Transaxle

I think it works the same way like other products (Mafterburner, etc.). One difference is the price. Parts for our solution are about $49 - others costs 10 times more. Since we plan to use it on about 20 cars - it is a major difference.

The other point is that we want to be able to control the whole software - not only the mapping.

Depends on the engine... ... 1000kg/hr should allow 12PSI@7.000 rpm with a 3l motor. But you can easily scale it with the diameter of the sensor.

The engine was constructed by turbo-uli. He's on this list too. He may explain some mechanical details. The compression is below 8:1.

Duke

Sweet!
Detailes about what's done to the pistons, rods and head would be much appreciated

Transaxle

It is a stock M44.41 block and head. The pistons are modified to reduce the compression. The rods are modified M44.52 versions. The block is modified to prevent oil-pressure problems. The pistons are oil cooled. The turbo is a modified K26/27.

For detailed questions on the modifications you may contact Turbo-Uli directly. But he's currently "a little busy" since he has to support several cars in the racing season:



And there are a few other 3l turbos waiting for completion.

Jfrahm

Why not use an atmospheric BOV to avoid the backflow?

Transaxle

We use the open BOV in the racecar. There is no backflow. The "stock-like" version uses the closed BOV.

Tomas L

An open BOV should only be used in a MAP system. With AFM or MAF an open BOV will let out air thats been measured by the flowmeter. The DME will think that the airflow in to the engine is larger than it really is and will deliver to much fuel. The cool side of this is that you should get large flames out of the exhaust at each shift...:-)
If the BOV is confusing the MAF you should get a similar effect, is that the problem you have?

1000kg/h is enough air for 356 hp with a BSFC of 0.5 lb/hp*hr and an A/F of 12.5:1. The flow is rated at an output signal of 4.33v, you should be able to run it to at least 4.5v and, as the sensor is logaritmic, measure air flow that is perhaps 20-30% larger.

How does this problem show up?
If I remember correctly the flap is designed to open "to much", to give an accelerator pump effect, during full throttle, mostly in low to mid rpm.

Tomas

Transaxle

The bad effect is that the rich exhaust gas will destroy the catalysator. No problem with the racecar - but bad with the streetcar.
The BOV confuses the MAF in a similar way because the measured air also leaves the system. We could solve that problem with a sensor with backflow measurement capabilities.

That's exactly the effect. The flap still moves up or down while the MAF already measures the real flow. But that is not a big deal to simulate in software. Just sum up the last values. Again - not so important with a racecar - but you can feel the difference in normal traffic.

Tomas L

I've been thinking about using a MAP signal just to get an accelerator pump effect. The MAP sensor is faster than a MAF and it senses pressure in the intake which gives it even greater speed advantage. The MAP sensor though has other problems making it unsuitable for a turbo car. If one just uses the MAP sensor to sense changes in manifold pressure and uses these changes as a correction to the MAF signal, you should get the best of both worlds.

Tomas

Tomas L

Which microcontroller are you using?
One with both A/D and D/A built in from what I can see in the picture.

Tomas

Transaxle

On the picture is a Motorola 6805. That chip is totally oversized, but it is my daily tool for everything to measure or control in the car. It has 8 8-Bit A/D input channels and 2 8-Bit D/A PWM output channels.

The actual version is based an a AVR chip with 10-Bit input channels and a 150 kHz, 8-bit high-speed PWM output. That 8 pin chip fits into the MAF adapter cable....

And yes - MAP has some advantages for racecars or if you are using a modern engine management. With the old Bosch DME/KLR you'll need a lot of helper-stuff to get it work correctly.

If I would like to keep the old Motronic, I would go the MAF way. With a modern standalone system I would take the MAP option. Our approach is not a additional piggyback system but a AFM compatible MAF device.

azmi951

is anyone going to have a full how to write up on this conversion?

What other benifits are there besides more power when going to a MAF?

Are there any cars out there that use the HFMs and what might be the right size for a 2.5L 951?

Where do I get the moterola processer and do I need to do any programming? I sounds like it is a universal processor.

Sean

cgmonster

Your motor looks fantastic. Great job!
What really caught my eye was the MAF conversion. When I was looking for MAFs for my car I was not convinced that some of the vendors out therehad the best products so I went about building my own.

Originally I envisioned a Ford MAF sensor with my electronics that converted its signal to what the 951 Motronic expected. I built this circuit using a PIC18 (www.microchip.com) processor. However I did not want to use the built-in A/D as I wanted a much higher resolution (because I sometimes lean on the side of accuracy overkill), so I used an MCP 3201 and built an opamp filter circuit in front of it for filtering noise from the A/D converter. The A/D converter then went into a PIC18 that outputted the signal value to a serial port using a MAX232 driver. I tested it with
the regular AFM and it worked beautifully. At half open it was showing 2.5V and at full open 4.5V.

I was then planning to go to a friend's local tuner shop that had one of http://www.superflow.com/flowbench/sf1020/index.htm. I was going to measure for the same airflow the corresponding voltages on the AFM and MAF and then change the aforementioned PIC18's program to a simple lookup that sent the AFM's voltage for the MAF's based on my test above.

The only catch here was to choose a MAF with a diameter that did not exceed the AFM's airflow by a large value. I was thinking about a 70mm MAF but before I got to choose it I was swamped at work and never got back to my MAF project. If you want more details let me know.

The best part of all this was that this project would have cost me MAF + $100 max. Also I found a manufacturer at www.ussensor.com that could make a NTC resistor with the same characteristics as the
one in the AFM. They even sent me two samples. They said that if I bought a hundred of those it would cost me less than $100.

Transaxle

@cgmonster:
You took the same way for your approach. I don't concern to much about absolute accuracy - the AFM isn't very precise and the 25 year old DME computer isn't too. The standard A/D port works fine.

The HFM5 includes a compatible NTC.



The creation of a lookup-table is simple: