alexjc4

I can't help but be a bit pedantic here, so please take this in the lighthearted/cheeky manner it is intended, I can but dream of being in possession of your skills and experience:

Strangely the 959 didn't have a centre differential at all, just a multi-disc clutch (PSK). Cutting edge as it was, by the time the 964 was released the Japanese had learnt from the 959 and released an arguably superior system in the BNR32, where the focus of the 4WD design was definitely on performance rather than safety (the addition of 4WS was their sop to safety).

The big win for manufacturers in systems like ABD is that it combines the mechanical and control systems for ABS and stability control into a single system. That's production engineering, not performance engineering in my mind.

Active centre diffs were banned in WRC and F1 - they only tend to ban things that work.

Re the skew towards safety; from Porsche Allrad, Manfred Bantle (964 4WD development engineer) "A new type of differential controlled all wheel drive with regulated axle revolution differentiation and automatic adjustment of response threshold for the lock effect ..... It was conceived for road use, in consideration of safety and simplicity". Reinhard Seiffert (ibid) "Drive systems like those of the Porsche 959 and Carrera 4 show that electronic control, especially for cars with higher performance, permits optimum safety ..."

Go drive a 964 C4, with a factory fresh diff, on the edge (the point where the electronics come in) and discover why many feel they gave safety primacy over performance. Multi-million dollar lawsuits in the 1980s related to the 930 may be unrelated of course.

End of pedantry


Sadly my interest in the theory of these things (and pedantry) isn't matched by my electrical engineering skills. So unless we can tickle someones fancy that has the required skills we may not get anywhere. Still, it's fun to discuss.

dlpalumbo

Concerning the expense of the C4 accels. What makes them expensive is that they are DC accels. They can sense the more or less constant g-loads that occur during cornering, starting and stopping. Most commonly available accels are AC, that is they measure an oscillating g-load. One supplier I found, Dytran, has units from $500 to $800 depending on the sensitivity required.

With a spec on the C4's accels sensitvity (mv/g), one of these might be substituted.

alexjc4

This military grade digital sensor, I'd guess you could convert the digital output to an analogue one with the addition of a microcontroller, is just $8 a piece.
http://www.analog.com/en/mems-sensor...s/product.html

My initial search that many accelerometers were available in both analogue or digital forms much cheaper than $500, in fact I get the impression the native form output is analogue and the digital outputs requires on board processing. For example: https://www.sparkfun.com/products/9269 (not sure that's explicitly up to spec but it's the kind of component that raised my interest in this initially.)

Something, that's both analog and specificaly purposed for automotive use, might be this:
http://www.anglia.com/st/literature_...lacce-1004.pdf
Which is ~$16 a piece but maybe a little more tricky to wire up

We can easily derive the sensitivity of the stock sensor by turning on its side giving voltage @ 0.98g

alexjc4

There's another sensor option that may not have been available when the C4 was conceived, in the form of gyros like this https://www.sparkfun.com/products/9801? which would allow you to do yaw control.

Edit: Aha, Analog Devices do specific purpose automotive ones http://www.analog.com/en/mems-sensor...s/product.html again fairly cheap ~$10. And they offer a breakout board http://www.analog.com/en/evaluation/...DXL203/eb.html for $30 which looks like it allows you to tune the analogue output.

dlpalumbo

I like the MEMS devices, such as the LIS3L02 you show from anglia. I assume the Analog Devices is MEMS also. They work, I believe, on much the same principle as the unit in the C4, there's an arm cantilevered in space and its displacement due to g-load is measured electronically.

We'd have to spec out the 964 unit over a range of g forces to test linearity. We can't be sure if the ECU is compensating for some non-linearity in the device. The test could be done on a rotating table if we could rig the wiring. Or, we could tap into the C4's wiring and run the car in circles at varying speeds and radii.

That almost sounds fun.

FeralComprehension

I think coming up with a modern 964 accelerometer replacement is a good place to start; it should be relatively easy to discern the output of the stocker and then replicate it via a little signal processing on the output of whatever replacement sensing device is chosen.

At that point things get tougher. The stock system has PDAS and ABS integrated both mechanically and electrically, so you either need to come up with a freestanding controller replacement that will cover both systems (yow, talk about liability!) or simply be satisfied with gaming the sensor output in such a way that the stock controller gives you the response you're looking for. Kind of a kludge, really, but unless you're willing to take on the ABS issue....

alexjc4

The minute you touch the brakes the 4wd locks disengage and the system disarms. Really its only the fault detection that will spot that the factory pdas brain isnt connected to the pdas solenoid that u need to work around. .So the systems are fairly independent tho they both use inputs from the g sensors. So any g sensor simulation device would feed both systems; existing abs +pdas brains and any new pdas solenoid controller u put in.

FeralComprehension

I'd love to have something more user-configurable, but I don't believe it's as simple as you're making it out to be. The stock system uses the ABS sensors to determine the need for traction control; how would you go about integrating that function?

alexjc4

Lol everything is more complicated then it seems, no excuse for not trying though.

Anyway

You could always leave either one of the lock solenoids connected, and just intercept the signal for the other lock solenoid. As it goes, wrt traction control aparently the Skyline aftermarket systems seem to be using TPS and G sensor for traction control type functions quite effectively.

You are dead right about the liability element esp in the US. But these existing suppliers figure out a way to make it work comercially if we can put the right info together we might get one of them involved I guess.

alexjc4

You are also right that a simple like for like replacement of the gsensors is simplest and most likely to get delivered.

The next most simple would be simply to modulate the gsensor signals to fool the pdas system in which case you would need to pass unadulterated signals to the abs.

Finally you get the full on replacement pdas controller, in which case you may leave the pdas controlleer in place and just disconnect the solenoids from the factory pdas brain.

kgorman

Very interesting. I have done quite a bit of arduino programming. The idea of spoofing the accelerometers is intoxicating. Subscribed.

kos11-12

Hi Alex, very interesting....
Have You tested so far any of those parts mentioned on your car...?

kos11-12

I was reading this thread from 2007, quite interesting,

https://rennlist.com/forums/964-foru...e-on-c4-2.html

alexjc4

I'm just at the thinking stages at the moment. I don't have the skills to complete even the basic g-sensor replacement project under my own steam.

I understand the theory of how these modern components can form a system but I also know that the electronic engineering required is a bit more complex than my W=EI and V=IR level of education. Though in some ways that's the beauty of the packages like arduino which do a decent amount of the work for you.

Also ideally we'd need a car with a "known good" PDAS setup to compare with; I can't tell how tight my diffs lock up (i.e how worn they are) or how accurate my accelerometers are compared to factory spec. We'd also need to do thorough research and analysis of the operation of the current system to avoid any unintended consequences.

It's all do-able its just a matter of time.

Maybe the next step would be to reach out to some of the guys who've done this work for other cars and see if they would be interested in doing something similar for our cars.

There is some sense in also looking at using mechanical Torsen/ATB/TBD diffs rather than plate LSDs for the front (and maybe rear) diffs, leaving just the centre diff under active control. I guess these would come in at £1000 per end fitted, if the correct fitment is even available. Edit: They are for the rear: http://www.quaife.co.uk/shop/products/qdf2q-0 £790