Let's design a do-it-yourself DCCD!

[porcupine73]

Hello, anyone want to design and build a do-it-yourself DCCD (driver controlled center differential)? It would be for AT vehicles only obviously. This circuit would be a build on the manually controlled locking center diff.

 

We'd need to find out the waveform the frequency at which the TCU drives the duty c. It could be measured using an oscilloscope on the duty c drive wire.

 

We'd need a variable duty cycle (via potentiometer) circuit oscillating at the proper frequency with the proper waveform capable of outputting at least the required current to drive the duty c.

 

In 'auto' mode, we'd just leave the TCU output to duty c connected as is.

 

Anyway this may be a crazy idea but what do you guys think?

[ericem]

Can we pull it from a wrx STi?

[porcupine73]

Can we pull it from a wrx STi?

Possibly, but I would suspect in that case the dial knob is an input to the TCU rather than instrument directly driving the duty c.

 

I suppose another risk of this project would be energizing the duty c when it is not supposed to be energized (i.e. duty b energized)? We'd have to review the charts in the service manual.

[Manarius]

Possibly, but I would suspect in that case the dial knob is an input to the TCU rather than instrument directly driving the duty c.

 

I suppose another risk of this project would be energizing the duty c when it is not supposed to be energized (i.e. duty b energized)? We'd have to review the charts in the service manual.

Pull it from a STi? No. It's far too complex to even think about pulling. It's too wired into the system.

[NoahDL88]

Pull it from a STi? No. It's far too complex to even think about pulling. It's too wired into the system.

 

Yeah, sure whatever:rolleyes:

 

 

Its been done a bunch of times, at least in motor swaps using the JDM EJ20 turbo motors and the DCCD transmission. I think the USDM would be harder, but they do have aftermarket controlers for the JDM stuff, so i don't see why it wouldn't be possible.

 

Just don't know why you'd want to.

[ericem]

Pull it from a STi? No. It's far too complex to even think about pulling. It's too wired into the system.

 

possibly its a simple system, maybe all we need is the TCU, and controller which adjusts the split between front and rear diffs. But then again the STi can do 40/60 so more rear wheel drive power. SOmeone with more experience is hiding some where

[DerFahrer]

Umm, the USDM DCCD has a $h1t-ton of sensors on the transmission itself, a sensor at each wheel (can't remember whether it's an individual sensor independent of the ABS sensor, or whether it's integrated into the ABS sensor itself), a sensor in the rear differential, and more I'm surely forgetting.

 

Ditch any of this stuff, and you will have an improperly-working DCCD system, or maybe a NON-working DCCD system.

[Manarius]

Umm, the USDM DCCD has a $h1t-ton of sensors on the transmission itself, a sensor at each wheel (can't remember whether it's an individual sensor independent of the ABS sensor, or whether it's integrated into the ABS sensor itself), a sensor in the rear differential, and more I'm surely forgetting.

 

Ditch any of this stuff, and you will have an improperly-working DCCD system, or maybe a NON-working DCCD system.

Thanks for confirming what I said .

 

Nobody seems to believe me even though I'm absolutely right....

[ericem]

Thanks for confirming what I said .

 

Nobody seems to believe me even though I'm absolutely right....

 

OK, lets find a scrapped WRX STi shouldnt be to hard to find? Then we will have our own STi right guys? Well i already know everyones subaru is a STi

[BruceyWV]

I wonder how this one works?

 

http://rallispec.com/prod_elecsys.htm

 

It doesent show any sensors for each wheel, cant it just use the ABS?

 

I've also heard its possible to reverse the split on automatics, so its rwd bias, and 10/90 most of the time. (90 on the rear wheels)

[Manarius]

I've also heard its possible to reverse the split on automatics, so its rwd bias, and 10/90 most of the time. (90 on the rear wheels)

Not any 4EAT Subaru transmission. Either end of the duty cycle of Solenoid C is as such: 0% or 100% is 100/0 or 50/50. And, I don't know how you'd do it with a non-STi manual because all non-STi manuals are viscous coupling units - 100% mechanical.

[BruceyWV]

Why you gotta be a hater?

Now I'll have to dig up the few posts I can find on it.

[Manarius]

Why you gotta be a hater?

Now I'll have to dig up the few posts I can find on it.

Not being a hater, just stating how it is.

[BruceyWV]

Except you're forgetting that at least in the VTD 4EAT in the WRX and H6 Outbacks, its essentially an electronic differential, and is plenty possible, I'll try to find the vague info about it, some guy does it at his shop.

[Andyjo]

mmm yeah, the DCCD on the STi is a manual, and you want to put it on an automatic... that'll run into issues... sure you could steal the switch out of an STi and use that.... I've looked into making this type of circuit... you need two pulse width modifiers, and a 555 timer to do it up i believe... not that hard of a circuit... just remember... you're making the tranny do something that it might not want to do.. could lead to problems

[BruceyWV]

http://forums.nasioc.com/forums/showthread.php?t=251957&highlight=10%2F90+4EAT

 

First relevant link I found, I guess the next step is to have someone contact Rally Knight?

[nipper]

http://forums.nasioc.com/forums/showthread.php?t=251957&highlight=10%2F90+4EAT

 

First relevant link I found, I guess the next step is to have someone contact Rally Knight?

 

actually 2700 for a manual shift 4eat with full control isnt a bad deal.

 

nipper

[RallyKeith]

Not being a hater, just stating how it is.

 

 

For once I'm gonna go ahead and agree with Manarius here. I agree with what he's said so far.

 

Also, the USDM STi DCCD is tied into the ABS system. It's one of the reasons you can't easily disable ABS in an USDM STi. It is one computer that looks at each wheel speed sensor along with sensors in the transmission, and a sensor on the ring gear of the rear diff, and controls the center diff lockup along with the front/rear brake bias and the ABS.

 

Keith

[BruceyWV]

I dont plan to do it, just trying to get the info out for others that might want to. I'm MORE than happy with my transmission, espaically lately, there seems to be a lot less auto lugging with it with the rebuilt finally running. What about VTD 4EAT's that natively send 45/55 and can send as much as 10/90 to begin with?

[99obw]

Hello, anyone want to design and build a do-it-yourself DCCD (driver controlled center differential)? It would be for AT vehicles only obviously. This circuit would be a build on the manually controlled locking center diff.

 

We'd need to find out the waveform the frequency at which the TCU drives the duty c. It could be measured using an oscilloscope on the duty c drive wire.

 

We'd need a variable duty cycle (via potentiometer) circuit oscillating at the proper frequency with the proper waveform capable of outputting at least the required current to drive the duty c.

 

In 'auto' mode, we'd just leave the TCU output to duty c connected as is.

 

Anyway this may be a crazy idea but what do you guys think?

 

What you propose doing is relatively easy from an engineering standpoint. I might be willing to contribute my EE skills, but my free time is somewhat limited. I have access to 'scopes and EE tools.

 

Rather than a pot I used a quadrature output rotary encoder for the last PWM controller I built. That allowed me to implement menus with an LCD and have digital repeatability, worked like a charm.

 

One thing that comes to mind is a semi-auto mode, where the computer still controls the torque split but said device allows adjustment of bias.

[porcupine73]

Alright thanks for the input 99obw! Do you know of any chips that can, with a minimum of external circuits, handle the variable duty cycle from 0-100%?

 

Such a chip probably wouldn't handle the amount of current we need, so what would we use for current gain? Maybe on of those honking TO-3 case transistors? A FET?

 

Based on an other post about the 'buzzing' sound I'm going to guess the frequency may be as low as 200Hz. Initially I thought it might be like 20kHz or about to prevent humans from hearing any sound from it.

 

Probably not critical at this phase, but do we know if the power to rear is linear with the duty cycle? We know 0% duty cycle=50% power to rear, and 100% duty cycle = 0% power to rear. But does 50% duty cycle = 25% power to rear, or is this a non-linear relationship?

 

Any other thoughts on making sure we don't drive the duty c if it isn't supposed to be driven at the time?

 

Hm, I just found a caveat in the service manual. Operating this circuit under wheel slippage conditions could cause the TCU to learn odd torque parameters thus causing strange operation in the future when in 'auto'. It says

The control unit stores optimum transfer clutch torque data for a variety of driving conditions. When actual driving conditions are detected by the sensors, the control unit selects a duty ratio most suitable to the given condition from the memory.

For example, suppose the circuit is operating at a fixed duty cycle and rear wheel slippage occurs. Unless the shifter is in '1' or something, the TCU probably starts outputting a higher and higher duty cycle to reduce power to the rear to stop slippage. However the TCU's duty cycle action has no impact because the circuit is operating. Thus the TCU may learn an improper duty cycle value for the condition.

[99obw]

Do you know of any chips that can, with a minimum of external circuits, handle the variable duty cycle from 0-100%?

 

Sure, the microcontroller I used for the last one has a built in quadrature port and four built in PWM controllers. There are plenty of analog ways to generate PWM, but I prefer microcontrollers for a number of reasons.

Here's what I would use. I would be willing to fund acquisition of prototype quantities of the components needed.

 

I prefer to use an external crystal or XO for processor clock rather than the low-accuracy oscillator built-in to the micro. We would also need a voltage regulator and a few caps and resistors.

 

This chip running at 40MHz could probably also be used to close the loop with the wheel speed sensors, but that's a huge increase in software complexity. Let's just try open-loop at first.

 

Such a chip probably wouldn't handle the amount of current we need, so what would we use for current gain? Maybe on of those honking TO-3 case transistors? A FET?

 

I'd use an n-channel FET in a TO-220 or TO-247 package. In addition a clamp diode in a similar package will be necessary when driving an inductive load. I'm assuming that subaru supplies 12V to the solenoid and PWM's the ground, but I haven't looked at the wiring diagram.

 

Probably not critical at this phase, but do we know if the power to rear is linear with the duty cycle? We know 0% duty cycle=50% power to rear, and 100% duty cycle = 0% power to rear. But does 50% duty cycle = 25% power to rear, or is this a non-linear relationship?

I don't know. I'd say that assuming linearity is a good starting point.

 

Hm, I just found a caveat in the service manual. Operating this circuit under wheel slippage conditions could cause the TCU to learn odd torque parameters thus causing strange operation in the future when in 'auto'. It saysFor example, suppose the circuit is operating at a fixed duty cycle and rear wheel slippage occurs. Unless the shifter is in '1' or something, the TCU probably starts outputting a higher and higher duty cycle to reduce power to the rear to stop slippage. However the TCU's duty cycle action has no impact because the circuit is operating. Thus the TCU may learn an improper duty cycle value for the condition.

 

I think it will be difficult to prevent messing with the TCU's trims, but honestly I don't know much about it.

[nipper]

The first thing we really need to do is get somone to put a scope on the duty c solenoid and see what it is doing. They should be in snow country so they can play with things. Ideally it would be two (maybe three) inputs. One for the C solenoid, and the other two for the speed sensors.

Once we get that information, we would know more about how the duty c cycle varies. My gut feeling is that more pressure is applied at low speed then high speed, and that the period is shorter (to allow for torque bind and higher Delta between sensors) then at highway speed.

 

If we just stick with doing this at low speed, we may be able to go one better, and using a VSS have it unlock at 20mph and go back to the tcu.

 

nipper

[nipper]

Not any 4EAT Subaru transmission. Either end of the duty cycle of Solenoid C is as such: 0% or 100% is 100/0 or 50/50. And, I don't know how you'd do it with a non-STi manual because all non-STi manuals are viscous coupling units - 100% mechanical.

 

The duty cycle is never 100% (100% being fully on or off depending upon how you look at it) It always pulses, its just a matter of how much and for how long.

 

nipper

[nipper]

Alright thanks for the input 99obw! Do you know of any chips that can, with a minimum of external circuits, handle the variable duty cycle from 0-100%?

 

Such a chip probably wouldn't handle the amount of current we need, so what would we use for current gain? Maybe on of those honking TO-3 case transistors? A FET?

 

Based on an other post about the 'buzzing' sound I'm going to guess the frequency may be as low as 200Hz. Initially I thought it might be like 20kHz or about to prevent humans from hearing any sound from it.

 

Probably not critical at this phase, but do we know if the power to rear is linear with the duty cycle? We know 0% duty cycle=50% power to rear, and 100% duty cycle = 0% power to rear. But does 50% duty cycle = 25% power to rear, or is this a non-linear relationship?

 

Any other thoughts on making sure we don't drive the duty c if it isn't supposed to be driven at the time?

 

Hm, I just found a caveat in the service manual. Operating this circuit under wheel slippage conditions could cause the TCU to learn odd torque parameters thus causing strange operation in the future when in 'auto'. It saysFor example, suppose the circuit is operating at a fixed duty cycle and rear wheel slippage occurs. Unless the shifter is in '1' or something, the TCU probably starts outputting a higher and higher duty cycle to reduce power to the rear to stop slippage. However the TCU's duty cycle action has no impact because the circuit is operating. Thus the TCU may learn an improper duty cycle value for the condition.

 

For what year is that?

 

nipper