News:

The ORG - No shonky business!

Main Menu

Reverse engineering the Bosch K-Jet ignition module

Started by ptashek, 10 January 2013, 02:56 PM

djenka018

#30
Quote from: ptashek on 22 July 2015, 10:41 PM
Here's the final revision of the schematic, with all parts as they should be, and the entire ignition circuit shown. I'll add an Eagle schematic / PCB / BOM at some point, once my prototype is verified (parts are shipping!). Simulations show this'll work as expected :)

https://www.circuitlab.com/circuit/5ag5z6/w116-ignition-control-circuit/

If you go and make a commercial gain out of this, I want 50%  ;D


How 'bout I give you the whole 100% of the profits?

That's about 100% of the negative profits I made.
Now you owe me money!
;-)
Vitamin C for SL... the SLC

ptashek

Well, I can declare some level of success with the prototype.

I have re-built the original module, using its original parts where possible, and have hooked both units to a 12V power supply, and my scope. While they both trigger as and when they should, neither generates that ~600µs long 70V+ spike on the output. All I see is the "fall-off" 12V signal. So, now I need to test both units connected to a real car. Any volunteers? ;)

This is what it should look like, based on a simulated circuit. The trigger signal is a square wave in the simulation, and on the car. That's what I need to check next, as my current op-amp oscillator leaves much to be desired in that respect :D

Simulated circuit:


Original circuit (channel A is blue, and is the trigger V(7); channel B is red, and is the output V(16)) :




My prototype:



1993 "Pearl Blue" W124 280TE
1988 "Arctic White" W124 200T
1979 "Icon Gold" W116 450SE

djenka018

I'll ask a question that may be overly stupid.
I do not see your coil on the pictures. Is there one in the circuit?
And the coil must be of the correct type (not capacitive discharge type like on D-Jets)
Vitamin C for SL... the SLC

ptashek

Quote from: djenka018 on 04 September 2015, 08:21 AM
I'll ask a question that may be overly stupid.
I do not see your coil on the pictures. Is there one in the circuit?
And the coil must be of the correct type (not capacitive discharge type like on D-Jets)

No, there is no coil and this is what I suspect is causing the "ignition spike" to be missing. It most likely needs that inductive load to fire fully. I have the entire setup spare from my 350 K-Jet, and will attempt to build the entire circuit over the weekend.
1993 "Pearl Blue" W124 280TE
1988 "Arctic White" W124 200T
1979 "Icon Gold" W116 450SE

djenka018

Yes mate, the "rebound" is indeed from the coil. In that case the "ECG" you have is correct.
Just make sure you have a spark plug at the other end (or the spike can be over the limit of your transistor) and you may benefit by using the 0.6+0.4 Ohm resistor in line for the same reason



My diagramme seem to be mismatched with yours. The base of the T3 was not connected to R19/C5/C6 node and I must have overlooked it. I

Vitamin C for SL... the SLC

ptashek

I got it to fire as it should, finally. And yeah, as suspected it won't fire unless fully loaded by the coil circuit.
Here's a short video of my board in action: http://www.youtube.com/embed/qzFwiPDNBnE

Surprisingly enough, I couldn't get it to fire at all with the ballast attached.

I no longer wonder why that module box is so beefy. I've measured the power transistor at 75*C case temperature, running at 62Hz (~1200rpm or so). It can only go hotter from there, so this is definitely one important thing to keep in mind for this project.

Once the board warms up and stabilizes, power consumption is around 60W (5A @ 12V DC).
Peak trigger voltage was just over 130V DC, translating to about 24kV spark voltage in this specific coil.

This is how it looked like on the scope (note the B-channel probe is a 1:10). Perfect match with the simulated circuit.
http://www.dropbox.com/s/q3u5c3h3et0cauu/my_module_proper.pdf?dl=0

Next up - finalise the BOM, make a proper full-size PCB, test in a car.
1993 "Pearl Blue" W124 280TE
1988 "Arctic White" W124 200T
1979 "Icon Gold" W116 450SE

djenka018

Congratulations!

You've now reached the point where a FET may be the next fork (it can dramatically reduce the heat output).

Most of the heat comes from the ripple following the inductive spike (snubber network helps there) and from the silicone transistor (faster switching = less heat, lower switching voltage etc.). It also affects the spark energy, similar to the way arcing of points reduces the spark energy (hence the tuning capacitor which kills the spark).
Vitamin C for SL... the SLC

ptashek

Quote from: djenka018 on 05 September 2015, 05:52 AM
Congratulations!

You've now reached the point where a FET may be the next fork (it can dramatically reduce the heat output).

Most of the heat comes from the ripple following the inductive spike (snubber network helps there) and from the silicone transistor (faster switching = less heat, lower switching voltage etc.). It also affects the spark energy, similar to the way arcing of points reduces the spark energy (hence the tuning capacitor which kills the spark).

Thanks!
I've never done anything with FETs that went beyond replacing those on a board. It would be a rather big job to switch from a current driven circuit, to a voltage driven circuit, something that's likely beyond what I understand well enough.

I've meanwhile redesigned the board to use SMD only components, and that in itself should allow for better heat dissipation. It just needs proper placement and routing - I've given up after way to many tries ;)

Either way, all that I've done for this project is now on GitHub under https://github.com/ptashek/mb
The only thing missing is a full description of how this unit works under the hood.
1993 "Pearl Blue" W124 280TE
1988 "Arctic White" W124 200T
1979 "Icon Gold" W116 450SE

djenka018

Send me the .sch file with SMD's and I will make you a PCB.
I think I got Eagle and gEDA...

I do not think you will be much better with SMD in terms of heat. Most the heat comes from power drive components and that are the 2 large transistors and surrounding components.
Vitamin C for SL... the SLC

ptashek

Quote from: djenka018 on 09 September 2015, 09:36 PM
Send me the .sch file with SMD's and I will make you a PCB.
I think I got Eagle and gEDA...

I do not think you will be much better with SMD in terms of heat. Most the heat comes from power drive components and that are the 2 large transistors and surrounding components.

That'd be great help!
You can get the EagleCAD BRD and SCH files from GitHub: https://github.com/ptashek/mb/tree/master/Ignition%20Module/EagleCAD

As for heat, practically all of the real heat is actually generated by R15 and T5 only. T5 ships over 800W in microsecond pulses (8A @ 100V DC when firing) in theory, while in my unit it actually does closer to 1kW. R15, which runs much hotter than T5, handles 2W continuously, with ~2ms drops when T5 fires. The 100V zener (39V ZD4 + 51V ZD5, which I've replace with just one) is at 5W pulsed, the R22 tops at 1.5W pulsed, and the rest never cross 750mW with most running well below 500mW.

Maybe it indeed makes no difference if they're SMD, or TO-220 mounted flat.
I'll be making another prototype board in SMD at some stage, so we'll have some data to compare :)
1993 "Pearl Blue" W124 280TE
1988 "Arctic White" W124 200T
1979 "Icon Gold" W116 450SE

djenka018

#40
Here's corrected diagramme pdf and a PCB based on your own SMD components.

Some components are undersized for the function and when you decide on exact component I can update the PCB.

TO3 transistor (BU323) is meant to be on the other side of the PCB hence the overlaying components. Normally, this is separated further by about 5mm from the heat sink. Double sided metalised or riveted holes should be used for reliability of contacts.
All the tracks surrounding BSV15 and north from it should not be covered  with an overlay and when the pcb is flowed the solder should increase their thickness.
Vitamin C for SL... the SLC

ptashek

#41
Quote from: djenka018 on 12 September 2015, 07:38 AM
Here's corrected diagramme pdf and a PCB based on your own SMD components.

Some components are undersized for the function and when you decide on exact component I can update the PCB.

TO3 transistor (BU323) is meant to be on the other side of the PCB hence the overlaying components. Normally, this is separated further by about 5mm from the heat sink. Double sided metalised or riveted holes should be used for reliability of contacts.
All the tracks surrounding BSV15 and north from it should not be covered  with an overlay and when the pcb is flowed the solder should increase their thickness.

Thanks! This looks sane, unlike any of my attempts. You clearly have a much better idea of how this thing works :)

As for parts:

T5 is a BUB323, which is a D2PAK-3 packaged version of BU323.
T4 is a ZXTP2041F, which is SOT23 packaged (but anything similar to BC160-16, which is the modern version of BSV15-16K, will do).

Which components do you think are undersized? I'll get them updated.
I've mostly gone with 2W resistors, which is well oversized except for R15. The diodes should all also be well within spec.

BTW, ZD4 is 51V, not 50V. I think it makes sense to replace ZD4 + ZD5 with a 100V, beefier element.
1993 "Pearl Blue" W124 280TE
1988 "Arctic White" W124 200T
1979 "Icon Gold" W116 450SE

djenka018

D1, D8 would have to be 1N400x variety (common rectifier 1A), C7 will have to be a variety of several 100's of Volts and then there are several different types that are qualified for the snubber network, electrolytic capacitors are better in tantalum variation for reliability (alu canister is the least reliable for automotive).

If you want to use SMD BUB323, you will have heat issues and copper thickness (PCB) will be detrimental. If you go with quite thick copper, remaining tracks can be much thinner

wrt to ZD4/ZD5, it is all bout what you want.
With multiple Z-Di the total power is sum of individuals (for same or similar volts) and if 1x Z-Di goes short it does not create a short across. Also, heat is better dissipated.

Same principle can be used for R15. If you were to use say 10x 3.3Ohm/0.5W resistors in series total would be 33Ohm/5W and heat would be better dissipated then from a single resistor.

IMO, for reliability the T5 and R15 should be off the pcb
Vitamin C for SL... the SLC

ptashek

#43
Gotcha. All good points definitely worth taking into account. My next step was drafting out a custom chassis for the PCB, and this will help.

p.s.: Can you post your version of the schematic in Eagle format? It'd save me some time redoing mine :)
1993 "Pearl Blue" W124 280TE
1988 "Arctic White" W124 200T
1979 "Icon Gold" W116 450SE

ptashek

I've updated the files on GitHub, with uprated components and revised schematic.
There's also a BOM file.
1993 "Pearl Blue" W124 280TE
1988 "Arctic White" W124 200T
1979 "Icon Gold" W116 450SE