Andial Splitter, simpler alternative solution.

[scarceller]

I've had several inquiries about building an alternative solution to the Andial Splitter that's becoming very hard to source these days.

The good news is that after reviewing what's needed to fire the 964 twin coils from a 84-89 Carrera DME it turns out I found a better simpler approach that does not require you to splice into the DME harness near the DME (under the seat).

I've come up with a new tinny circuit module that can go into the engine bay, then all that's needed is to build a custom harness to mate up to the 964 coils, with coils mounted on the standard 964 coil bracket and the 2 ignitor modules mounted on the bracket. Basically you source these parts from a 964 exactly like in this picture:


Then I create the small circuit module and you can easily build this harness:


The entire solution then wires up in the engine bay with 4 simple wires! I added that 30amp relay to insure the 2 coils get powered from a solid 12v source, you could power the setup without the relay from the stock coil's '+' black wire but that wire is feed via the ignition switch. I felt that a relay was in order, so I added one. I actually advice getting main constant power for the relay from the starter, this way you have solid supply of 12v for both coils. The rear fuse block is also a decent source of power but may require a few small mods, I'll outline these mods soon when I post pictures of the setup fully installed in a test car.

[scarceller]

Post #2, is a place holder for in car pics

[mysocal911]

Originally Posted by scarceller

I've had several inquiries about building an alternative solution to the Andial Splitter that's becoming very hard to source these days.

The good news is that after reviewing what's needed to fire the 964 twin coils from a 84-89 Carrera DME it turns out I found a better simpler approach that does not require you to splice into the DME harness near the DME (under the seat).

I've come up with a new tinny circuit module that can go into the engine bay, then all that's needed is to build a custom harness to mate up to the 964 coils, with coils mounted on the standard 964 coil bracket and the 2 ignitor modules mounted on the bracket. Basically you source these parts from a 964 exactly like in this picture:


Then I create the small circuit module and you can easily build this harness:


The entire solution then wires up in the engine bay with 4 simple wires! I added that 30amp relay to insure the 2 coils get powered from a solid 12v source, you could power the setup without the relay from the stock coil's '+' black wire but that wire is feed via the ignition switch. I felt that a relay was in order, so I added one. I actually advice getting main constant power for the relay from the starter, this way you have solid supply of 12v for both coils. The rear fuse block is also a decent source of power but may require a few small mods, I'll outline these mods soon when I post pictures of the setup fully installed in a test car.

Nice effort to replace the NLA Andial Splitter, but not any simpler and requires the addition
of a relay.

[scarceller]

I did not set out to copy the Andial design, I set out to build what I saw based on specs for the 964 modules and the current draw for 2 coils instead of 1. I could have left the relay out and simply powered both coils from the line that powered the single stock coil but I did not like running this much current through the ignition switch so I felt the relay was needed. By the way the Andial setup grabs power for both coils via the existing black coil wire, I decided against this.

As far as simplicity this design is far simpler to install than the andial design. What I envision is a special harness that has built into it the Relay and the module. You get the harness fully built and then simply wire up 4 wires, all done in the engine bay. I also supply a nice terminal block for the harness and you install the terminal block onto the rear fuse panel tin. Pictures will be coming soon. From the end user's perspective this solution is much simpler than the Andial setup.

You basically get a pre built harness that has a 4 post terminal block on one end and the connectors for the ignitors and the coils on the other end. You only hook up 4 wires to the terminal block and all 4 sources for the wires are readily available in the engine bay. Not sure if it could be done any simpler than this. If I don't use the relay then it becomes a 3 wire hookup.

[darrin]

Sal: appreciate the transparency in your efforts to solve a problem while developing a product!

[mysocal911]

Originally Posted by scarceller

Not sure if it could be done any simpler than this.

Well, I've seen a design using just two 964 modules and a few resistors, as I mentioned
on the other thread, installed on a twin plugged 911 3.2 at a well known SoCal race
shop and running reliably. Unfortunately the resistors were installed on the module
connectors under the connector rubber covers (very simple approach), so I wasn't able
to determine their values during the time the connector covers were pulled back.
But I'm sure there're many competent EEs who post on this forum who could easily
calculate what the required resistor values would be.

[scarceller]

Dave,

I do see how this could work but I would not run things this way in my personal 911 simply because it's not exactly the best practices design. The second coil with have a 'slight' delay introduced by the extra drive transistor and the drive circuit for each coil is non-symmetrical. Meaning the drive circuits vary significantly between the 2 coils. If it's possible to build a symmetrical drive circuit for both coils why would we not do that? I respect your input and I'm sure you are correct that the approach you outlined would work but I would not use this approach in a hi-performace Porsche motor. I have the ability to calculate those resistor values, I've already done the calcs but I opted for a better symmetrical design.

Furthermore, here's the Bosch design notes for firing BIM modules, notice what they recommend for the drive stage for the BIMs. I basically followed the Bosch design practices recommended in the design docs for integrating these modules with Bosch Motorsports ECUs used in professional race cars.

The above picture is right out of the Bosch Motorsports specs, I basically followed this design but with a PNP transistor on the hi-side instead. Then a 7805 5vdc regulator to meet the spec they outlined. Please don't just build this circuit as is as it won't drive 2 BIMs at once, but it gives you the basic parameters for current needs within each BIM.

I'd rather take the design tips from the Bosch engineers that built the modules. It just seems reasonable to follow that design no matter how HARD or COMPLICATED it may be.

Originally Posted by mysocal911

Well, I've seen a design using just two 964 modules and a few resistors, as I mentioned
on the other thread, installed on a twin plugged 911 3.2 at a well known SoCal race
shop and running reliably. Unfortunately the resistors were installed on the module
connectors under the connector rubber covers (very simple approach), so I wasn't able
to determine their values during the time the connector covers were pulled back.
But I'm sure there're many competent EEs who post on this forum who could easily
calculate what the required resistor values would be.

[scarceller]

I'll even show you what I built at a high level, I don't mind documenting my basic circuit design, if someone else cares to do the homework on the R and I values for the components feel free. I also had help from member 911st another great source of electronics knowledge in these forums. The design was the result of several approaches considered by us. Here's the basic high level circuit:

We had a very specific reason we went for the PNP design because of the final stage within the DME we felt the PNP design would be slightly more stable since the input is not driven from a TTL or Buffer device.

Dave, you really are making this solution more credible.
Thanks.

[mysocal911]

Originally Posted by scarceller

The second coil with have a 'slight' delay introduced by the extra drive transistor and the drive circuit for each coil is non-symmetrical. Meaning the drive circuits vary significantly between the 2 coils.

1. The delay is negligible.
2. The waveforms are symmetrical based on a properly designed drive signal,
i.e. using the correct resistor values.

Originally Posted by scarceller

Furthermore, here's the Bosch design notes for firing BIM modules, notice what they recommend for the drive stage for the BIMs. I basically followed the Bosch design practices recommended in the design docs for integrating these modules with Bosch Motorsports ECUs used in professional race cars.

The required drive signal is basically a current source (not a voltage level) to achieve
the proper coil current as also noted from Bosch, i.e. the module is basically a power
transistor (darlington). That is what you've done by having a resistor (R3) being
pulled-up to 5 volts. In your circuit the drive current is 5 volts minus the input voltage
of the module (~ 1.5 volts) divided by R3. So the first module replaces your inverter PNP circuit
with its own R3 driving the second module which drives the second coil. Again, it's that simple!

Furthermore, your circuit using a PNP and R4 has a poor turn-off time for the module
versus using just a NPN and R3, which would reduce the turn-off time of the module by
not having the R4 to remove the darlington's base charge. This is shown in the recommended
circuit you posted from Bosch, i.e. a NPN with a pull-up and a very small series resistor
to remove the base charge. Also, there is really no need for a regulator when a proper
R3 is selected for a worst case analysis of the module's drive requirement.

[scarceller]

The NPN design is also not a good choice because the Darlington within the DME only pulls down to about 0.5-0.7vdc when it turns on. This we felt could result in the NPN not activating (saturating) fully. With the PNP design the issue goes away because while the coil is not charging both the base and the emitter of the PNP are at 5vdc. Then when the DME darlington grounds the DME pin 1 it very quickly activates and fully saturates the PNP. It's very easy for the voltage at the base to drop 0.7vdc lower than the Emitter's 5vdc and the PNP activates very quickly. In summary: the DME pin #1 coil drive signal never gets to 0vdc it goes to about 0.5 to 0.7 depending on system voltage and that could be problematic.

As far as the turn off time of the PNP the scope trace shows no issue here, it turns off very quickly and cleanly. The only issue I'm working on now is cleaning up a slight negative spike on turn off. Here's the actual scope outputs from the input and output for the circuitry, taken while actually running in car. I'm not sure how much cleaner of an output signal I can get after I cleanup the negative spike where it dips about -2vdc on turn off.

And that output is always clamped at 5vdc when high.

And that PNP circuit is a basic design it does not have every component in it, several diodes are not shown and they have special purposes to help insure quick on and off times for the circuit. So please don't make the assumption that's the complete circuitry as other components are involved (ie: capacitors and diodes) for noise reduction, circuit stability and overcurrent protection if improperly wired.

[scarceller]

And one last serious potential issue with the NPN design as shown in the Bosch spec picture, if you disconnect the DME and leave the circuit connected to the BIM it results in the NPN being turned off and the output goes high and then the coil's are on forever! Very bad! Can easily be fixed with a pull up resistor but just worth noting that you need to pay attention to the details.

With the PNP design this is a non-issue.

I'll be running some more tests tonight on the circuit and trying a few tricks to cleanup that negative spike. I'm actually very pleased with how clean signals are at this time considering all the hi-freq noise potential of a hi-voltage ignition system.

The reason I'm so picky about the signal conditioning and the 5vdc supply is because the next project on my burner is a Wasted Spark setup that will eliminate the entire distributor for the 3.2L cars! And once again everything will be plug and play in the engine bay. Basically this first project will be the building block for the next wasted spark setup. That's why I won't abandon the 5vdc regulator, I'll be using a bit of TTL logic to generate 3 individual coil signals from the single DME coil signal. I need that input coil signal super clean and pre-inverted for wasted spark. But that's a future topic for another thread and another day.

[mysocal911]

Originally Posted by scarceller

The NPN design is also not a good choice because the Darlington within the DME only pulls down to about 0.5-0.7vdc when it turns on. This we felt could result in the NPN not activating (saturating) fully.

I think you don't understand how the NPN circuit functions. When the pin 1 of the
DME ECU grounds (reaches .5-.7 volts), the NPN turns-off with CORRECT resistor
values which can be easily determined if one has a good knowledge of circuit design.
When the DME pin 1 turns off, then the NPN saturates, again based a worst case
circuit design analysis and CORRECT resistor values.

Originally Posted by scarceller

The reason I'm so picky about the signal conditioning and the 5vdc supply is because the next project on my burner is a Wasted Spark setup that will eliminate the entire distributor for the 3.2L cars! And once again everything will be plug and play in the engine bay. Basically this first project will be the building block for the next wasted spark setup. That's why I won't abandon the 5vdc regulator, I'll be using a bit of TTL logic to generate 3 individual coil signals from the single DME coil signal. I need that input coil signal super clean and pre-inverted for wasted spark. But that's a future topic for another thread and another day.

Really?

[scarceller]

Dave,

I understand how a NPN works, when the base drops below .7 it starts to turn off and if it drops to 0vdc it turns completely off. That's my point, have YOU actually measured the output of DME pin #1 to see how low that voltage gets? You may be surprised by the finding. It barely gets below .7vdc and that's not low enough in my book to turn that transistor fully off! The collector of the Darlington within the DME has a few Wire Wound resistors to ground and this is not helping us.

Let me ask, at what base voltage does a 100-200ma driver NPN transistor start to turn on? Think about that for a bit then consider DME pin #1. Sure I could use a voltage divider to further divide the DME pin 1 voltage but why bother? Just use a PNP hi-side and your done.

Not sure what the 'really' comment at the end is? Somehow you don't think wasted spark can be done?

Originally Posted by mysocal911

I think you don't understand how the NPN circuit functions. When the pin 1 of the
DME ECU grounds (reaches .5-.7 volts), the NPN turns-off with CORRECT resistor
values which can be easily determined if one has a good knowledge of circuit design.
When the DME pin 1 turns off, then the NPN saturates, again based a worst case
circuit design analysis and CORRECT resistor values.



Really?

[winders]

mysocal911 (Dave),

I have an idea! Instead of poo-pooing all the stuff Sal is doing, why not show a design that is better than his and provides the desired function?

[mysocal911]

Originally Posted by scarceller

Dave,

I understand how a NPN works, when the base drops below .7 it starts to turn off and if it drops to 0vdc it turns completely off. That's my point, have YOU actually measured the output of DME pin #1 to see how low that voltage gets? You may be surprised by the finding. It barely gets below .7vdc and that's not low enough in my book to turn that transistor fully off! The collector of the Darlington within the DME has a few Wire Wound resistors to ground and this is not helping us.

Let me ask, at what base voltage does a 100-200ma driver NPN transistor start to turn on? Think about that for a bit then consider DME pin #1. Sure I could use a voltage divider to further divide the DME pin 1 voltage but why bother? Just use a PNP hi-side and your done.

Not sure what the 'really' comment at the end is? Somehow you don't think wasted spark can be done?

Again, you need to select the correct resistors to bias the NPN
in the "off" condition for the worst case positive voltage at the DME ECM pin 1
when pin1 is at its highest "low" state value so let's say it's .7 volts.

1. If the NPN needs to pull the 150 ohm resistor to ground with a 5 volt supply,
the current would be 5 divided by 150 or 33 (ma) milliamps. Using what's called a forced
beta of 15, the required base current would need to be 2.2ma.
2. If one uses three 470 ohm resistors:
a. one from the base to ground
b. one from the base to pin 1 of the DME output and
c. one from the pin 1 to the 5 volt supply.
3. When the pin 1 is in its off state, the current to the base of the NPN is 5 volts
minus .7 volts divided by the two 470 ohm resistors or 940 ohms equals 4.5 ma.
But since the base resistor draws current, the actual base drive current is equal
to 4.5 ma minus .7 volts divided by the 470 ohm resistor or 1.5 ma. So the actual
base current is 4.5 minus 1.5 or 3ma which is more than enough drive to pull
the 150 ohm resistor to ground.
4. When pin 1 is pulled to its low state (.7 volts), the NPN base signal now becomes
one half of that voltage (.35 volts) because of the voltage divider of the two 470
base resistors and assuring that the NPN is fully "off".

Again, using resistors to shape the desired signals results in the same number
of components, i.e. two PNP base drive resistors (base to emitter & the base to pin 1),
a PNP collector resistor, and a resistor to ground at the input to the module for proper
module noise immunity & fast turn-off. The same number of resistors in either case.
To eliminate the 5 volt regulator and just use the battery voltage, one would just
use different resistor values.

So using a design approach as indicated above, one can design a very simple twin plug
ignition with just a few resistors and two 964 modules, i.e. not requiring a regulator nor
a transistor, with the first module used as an inverter off of the pin 1 original coil signal
driving the second module and coil.

The above should be easy for anyone to implement.

[scarceller]

Dave,

I don't disagree with you about doing this with an NPN or a 128 BIM as the NPN device as you so describe. But I see 2 issues:
1 - The idea of using such an expensive 128 module as the pre driver and inverter for the next stage is really not a great idea, you simply want to do it this way because you happen to have the device? Not a good reason.
3 - The NPN design also requires more discrete components than the PNP design and it has a possible failure mode if a component fails the BIMs could be left in an on state. The PNP design is somewhat less prone to this type of failure.

I do see your point that IF you have 2 BIM modules in hand you could just use resistors to do the job. However, in a prior post you said 'NO Semiconductors required' this is not really true, the BIM you are using as the pre-module is a semiconductor but you argue it does not count? Instead of using a BIM module why not use a more appropriate lower powered transistor to do the job properly?

Please take the time to draw up what you just explained so we can see the entire circuit at a hi-level, I think that would add to this thread. No?

Also you can't assume that when the Darlington in the DME is off that pin #1 is hi-imp (infinite R) it's NOT. Easy enough to work around but yet another consideration.

I'll stick with my PNP design and using both BIM modules in a symmetrical configuration, they will work just like they do in the 964 cars. Plus it makes future diagnosis of an issue better understood. One BIM for each coil, your design can have either BIM take out the 2nd coil. I think it's easier to understand the 964 setup. I may decide to drop the 5 volt regulator but I'll keep the basic PNP setup.

Originally Posted by mysocal911

Again, you need to select the correct resistors to bias the NPN
in the "off" condition for the worst case positive voltage at the DME ECM pin 1
when pin1 is at its highest "low" state value so let's say it's .7 volts.

1. If the NPN needs to pull the 150 ohm resistor to ground with a 5 volt supply,
the current would be 5 divided by 150 or 33 (ma) milliamps. Using what's called a forced
beta of 15, the required base current would need to be 2.2ma.
2. If one uses three 470 ohm resistors:
a. one from the base to ground
b. one from the base to pin 1 of the DME output and
c. one from the pin 1 to the 5 volt supply.
3. When the pin 1 is in its off state, the current to the base of the NPN is 5 volts
minus .7 volts divided by the two 470 ohm resistors or 940 ohms equals 4.5 ma.
But since the base resistor draws current, the actual base drive current is equal
to 4.5 ma minus .7 volts divided by the 470 ohm resistor or 1.5 ma. So the actual
base current is 4.5 minus 1.5 or 3ma which is more than enough drive to pull
the 150 ohm resistor to ground.
4. When pin 1 is pulled to its low state (.7 volts), the NPN base signal now becomes
one half of that voltage (.35 volts) because of the voltage divider of the two 470
base resistors and assuring that the NPN is fully "off".

Again, using resistors to shape the desired signals results in the same number
of components, i.e. two PNP base drive resistors (base to emitter & the base to pin 1),
a PNP collector resistor, and a resistor to ground at the input to the module for proper
module noise immunity & fast turn-off. The same number of resistors in either case.
To eliminate the 5 volt regulator and just use the battery voltage, one would just
use different resistor values.

So using a design approach as indicated above, one can design a very simple twin plug
ignition with just a few resistors and two 964 modules, i.e. not requiring a regulator nor
a transistor, with the first module used as an inverter off of the pin 1 original coil signal
driving the second module and coil.

The above should be easy for anyone to implement.

[onboost]

Originally Posted by winders

mysocal911 (Dave),

I have an idea! Instead of poo-pooing all the stuff Sal is doing, why not show a design that is better than his and provides the desired function?

^^ Exactly what I was thinking!
There is always more than one way to skin a cat.. and here on Pelican there is always someone with a brighter idea, be it right or wrong.

And although another design using just two 964 modules and a few resistors" has been seen by someone at another shop, the idea nor the design has been shared with others that may be interested.

Sal, carry on, good stuff.. will be in touch in the future about this and some of your DME thoughts and solutions.

[mysocal911]

Originally Posted by scarceller

Dave,

1 - The idea of using such an expensive 128 module as the pre driver and inverter for the next stage is really not a great idea, you simply want to do it this way because you happen to have the device? Not a good reason.

I do see your point that IF you have 2 BIM modules in hand you could just use resistors to do the job. However, in a prior post you said 'NO Semiconductors required' this is not really true, the BIM you are using as the pre-module is a semiconductor but you argue it does not count? Instead of using a BIM module why not use a more appropriate lower powered transistor to do the job properly?

Again you don't understand!

The proposed design has been described many times on this thread and the other
one:

There are only two modules in total being used
and few resistors.

It has less total components and cost than what you have been proposing!
Sorry you lack the engineering insight to understand this, as I've tried numerous
times to explain the design. Based on what's been presented, though, others here on
the forum can calculate the resistors values and design their own twin plug ignition
for just the cost of two modules and a few resistors.

[tshebib]

Originally Posted by mysocal911

The proposed design has been described many times on this thread and the other
one:

There are only two modules in total being used
and few resistors.

It has less total components and cost than what you have been proposing!
Sorry you lack the engineering insight to understand this, as I've tried numerous
times to explain the design.

I can't take this anymore.
Dave, you are being a complete troll.
I'm sure there are many people viewing this and shaking their head with the way you are treating Sal.
Please, everyone chime in if you feel the same.
Sal is contributing a lot to our hobby in a very positive way.
Maybe Dave can do the same if you have it in you.

[scarceller]

Dave,

I get it, stop attacking me! That module is $80! why not use a simple $1 transistor instead? That's exactly what I did. Using a $80 module as a pre-driver is simply not wise. But it will work, have at it if you like.

You are basically proposing what Bosch shows in Post #7 but using a $80 module in the place of the BCX58 transistor, just silly to do that. No?

You can do exactly the same idea you outlined with just 1 module and a much cheaper transistor. No?

Originally Posted by mysocal911

The proposed design has been described many times on this thread and the other
one:

There are only two modules in total being used
and few resistors.

It has less total components and cost than what you have been proposing!
Sorry you lack the engineering insight to understand this, as I've tried numerous
times to explain the design. Based on what's been presented, though, others here on
the forum can calculate the resistors values and design their own twin plug ignition
for just the cost of two modules and a few resistors.