Power Flyback PCB. Other solution!!!

[GintsK]

Hello.
My thought is about unsatisfactory result I got with tiny 320ccm low-Z injectors.

Story. It is n/a Audi 7A 5cyl 20V engine and stock injectors. From factory hitachi use resistors in series. But custom loom was built without them. Wiki says it is easy to manage injector using software. In my case it was not true...

My mistake was flyback wire soldered at wrong pad. So first few days engine was running without flyback circuit. Tuning at low daily loads and idle was real pleasure! Lambda 1.15 with no surge...

 After fried 2 FETs @7000Rpm I found mistaken wire and fix it. Engines operation changes dramatically! No way to tune idle leaner than lambda 0.7-0.8! No control over injectors absolutely ! Black smoke, unburned gasoline, blow-ups from exhaust...

 So i go for Power Flyback PCB. And was really disappointed:  it resolve problem only by 50-60%. Engine idles at 1.0, but not close as smooth as without flyback at all. VE map stays uneven and get lean mixtures at cruise is impossible.

Can anybody suggest  some higher voltage flyback solution than DDFlyback used on Power Flyback PCB? External resistors pack and transient voltage suppresion diode not always nice way out.

Gints
 
Update:
You can find my solution on second post in page 2

[dnb]

It really does sound like you still have an underlying problem with the ECU.   I've not tried lowZ injectors with VEMS yet, but my understanding is that they are difficult as you spend most of the time in the VE table correcting for the long close times...

What is the config data for the injectors?

[GintsK]

Yes. Problem is loooooong closing time. With 0V flyback injectors stays opened unpredictable time. Say normally my injector needs 3ms at idle. With 0V flyback it becomes as short as injector opening time and then injectors sometimes opens sometimes no. Unregulated. If opens idle stays opened more than my 3ms. VE map....no map.

Using power flyback situation is much more predictable. But closing time is still varying without my control. Injection time is about 1.8-2.0ms and VE map have jump.

without flyback wire situation was as usually - no problems manage idle AFR.

I work with efi more than ten years - mainly like diagnostician. And during this time I never see so strange injector oscillogram with so little peak voltage. Usually peak is about twice of supply AFAIR.

[rob@vems.co.uk]

50W 6R8 resistors and the 30v transient supression diode in line with the flyback wire, turn off PWM (duty 100%) and you will find that things work just perfectly.

1000cc injectors running on a 1.8 with 1.2ms PWs worked perfectly

Rob

[GintsK]

50W 6R8 resistors and the 30v transient supression diode in line with the flyback wire, turn off PWM (duty 100%) and you will find that things work just perfectly.

1000cc injectors running on a 1.8 with 1.2ms PWs worked perfectly

Rob

Yes I knew about this solution. But try to figure out something simpler in terms of installation. Power Flyback board requires only cut one wire!

[rob@vems.co.uk]

Please dont consider that as a reason not to do it, 4 resistors and one diode is not a massive undertaking and it will be the solution that works.

[GintsK]

Rob, resistors require separate 8pin connector (for 4cil engine) and own enclosure for clean installation. If I use OE loom where high-Z was used it takes time to design OE-like bullet proof hardware...

OK. Can someone can give guidance how to calculate required power for flyback in single wire like Power Flyback is?

[Sprocket]

I am using a single 611cc low ohm injector with the resistor and transient surpression diode and no flyback board. You have to treat the injector characteristics as those of a high ohm injector. Granted i suspect this is not as good as the PWM, with the resulting fairly low and flat VE map, but it works a treat once its set up right. I am going to try and half the required fuel value and double the VE values to give a little better graduation.

I intend to also try the PWM but im looking for good information on injector characteristics before i do.

Are you sure this is not just an issue with the injector settings, ramp up and open times?? as these are critical to acurate fueling and injector/ FET life.

hope you get it sorted

[Jorgen]

If you are set on using the injectors in peak hold mode you should go with powerflyback. But resistors and the transient suppressor diode is the best way to do it. It gives superior injector control compared to the peak hold solution. The injector is allowed to close much faster then with powerflyback.

You should also be aware that when a FET blows up it often takes out the FET driver as well. The FET driver fails in a way that can make it unable to PWM the injector properly and it can also cause slow closing or an injector that will not close at all after it has been opened.

Jörgen

[GintsK]

As I understand PWMing in peak-hold mode with high voltage flyback should give both: fast opening AND fast closing.

These days I try to prepare two 2L cars with 770-870cc injectors. Must say car using peak-hold mode and high voltage flyback feels better (not VEMS). I can idle it even at 900Rpm and lambda close to 1 and it is with 2 injections per cycle. Other one have resistors and protection diode. If i try go close to lambda 1 feels like one injector sometimes miss to open. True: OE resistors are used - must check it.

I want experiment with this board:  Peak&Hold Injector Driver Board

[GintsK]

Want to continue this topic.
As previous I searching for simple solution for flyback voltage limiting.
I found v3.3 schematic already contain this circuit. But it is not used. Here is even soldpads on V3.3 boards for SMD parts.

Because calculating chocked energy was too complicated to me, I construct some tiny board consisting from 3 parts:

Actually there is additional TVS diode if something goes wrong, but it do nothig. 1.3W zener is 30V because I haven't other.



Basic problem was get some Low-Z injectors for test. Now I have 5 old injectors from Jaguar V12 engine.

Results. With 5 injectors running simultaneous @8000Rpm and 85% DC Voltmeter reads ~27V and 0.47A which give ~12.5W of heat. So heatsink is little too small. It was still touchable with hand after ~4minutes of test.

First I put this board in place of TVS diode: based on supply wire. But get one fried FET together with 1A fuse. It can be because of voltage supply rised by my board. FETs is 60V rated. Original VEMS TVS diode protection works like supply voltage+TVS voltage. As I undertand because of less heat on TVS diode (please correct me if not so). Then I connect my board direct to ground.

My next step is test same with higher voltage zener.

BTW looks like TVS diode easily survive with one Low-Z injector running in peak-hold mode.

[GintsK]

Todays test was using 47V zener. Result is more promising: looks like heatsink is big enough:

Same 8000Rpm and 80...85% duty cycle. DMM reads 37V and 0.2A It is 7.5W of heat.
Heatsink is 6°C/W -> 45°C above environment: touchable by hand even during long test.

So next will be test on real car, where webshop harness is used.

Gints

[rob@vems.co.uk]

Great stuff - let us know how that goes.
You should only really have heat issues when at WOT for prolonged times.

[GintsK]

Make some bench tests: will add cap between across board.
Here is scope shoots - voltage over my board on top (47V zener), injector signal bottom:
without cap transistor tends open to much at peaks:

And using cap voltage across board becomes stable:

Only one injector was actuated during this.
Injector sound changes with cap.
I try even 98%DC test - using 5 injectors no much heat.

[GintsK]

My next prototype will be using TIP107. It is much common darlington than BDW47.