Diode selection

[xnotx]

Trying to guesstimate which diodes to use in this circuit, they are all zener looking diodes except for 05B which is a big rectifying diode of around 600v/1kv no less than 400 volt as that is the voltage spec for the big metal film capacitor there.
Tried to do a back step trace using chip voltage requirements as a guide, but I'm still a bit confused.

Here are the main components.

Toshiba TC4001BP 14pin
http://pdf.datasheetcatalog.com/datasheets2/19/197001_1.pdf

Toshiba 2SC1815 amplifying transistor (two of these)
http://www.futurlec.com/Transistors/2SC1815.shtml

The SCR thyristor is unknown, but if I had to quess.... something along the lines of a TS820-600T
https://www.digikey.com/products/en?keywords=TS820-600T

Any help with this project would be MUCH appreciated.
All images and files thus far can be download from the link.

.rar https://www.dropbox.com/s/akpasloq4xgd7wz/Project_cdi_images_high_res.rar?dl=0
.zip https://www.dropbox.com/s/h4kytcjstryfa6b/Project_cdi_images_high_res.zip?dl=0
just pics https://web.archive.org/web/20180128151753/https://postimg.org/gallery/tqzm9phy/

Tools (safe).
https://www.rarlab.com/ .rar opener for most operating systems.
http://www.linear.com/designtools/software/ LTspice
https://www.virustotal.com/#/home/upload Online virus scanner for the extremely paranoid folks.


EDITS / Updates:
1-29-2018 -> Fixed routing issue with 2sc1815, cap, changed diodes
2-2-2018   -> Floobydust version LTspice asc
2-3-2018   -> Added Exciter and Pulse waveform images

[Benta]

Sorry, but .rar might be the craze for someone. I can't do anything with that format.

[xnotx]

Yep, .rar was huge back when and I guess it stuck with me. lol
I added .zip to original post.

RAR really whips the llamas *SS... oh wait, that is something else, carry on.

[Benta]

Sorry, but .rar might be the craze for someone. I can't do anything with that format.

Any archiver should be able to open it.
7zip can open it, and under Linux, default archiver can open it through unrar plugin, which is usually shipped with OS.

You misunderstood. I know .rar and .zip but I do not open archives of that kind. I've been burned before.
Post the pictures as .jpg, everyone can handle that.

[Audioguru]

I agree that a RAR file is a nuisance and is rarely seen so my Windows 10 can't open it.
A JPG file type often makes a very fuzzy schematic. A PNG file type is good for schematics.

[T3sl4co1l]

260MB holy shit what is this a feature length film?

I get that it says "high res" but come on, we don't need a freakin' slideshow...

Tim

[xnotx]

You misunderstood. I know .rar and .zip but I do not open archives of that kind. I've been burned before.
Post the pictures as .jpg, everyone can handle that.

A .zip or .rar can't "burn" anyone.
An executable file with malicious content can.
There are no image exploits that I'm aware of, if there were I would shut my computer off and not internets at all, because even facebook is then a vector for infection.
You're being overly paranoid. I take my security, very seriously.....

260MB holy shit what is this a feature length film?

I get that it says "high res" but come on, we don't need a freakin' slideshow...

Tim

This is 2018, I packaged the hi res files so people can zoom in without seeing 1990 level fuzz.

Here is a link to a new picture gallery on postimage.org (God help the poor soul looking for insight, stumbling across this thread in a few years to find the images don't work).
https://postimg.org/gallery/tqzm9phy/
An E-Condom for the paranoid. 
https://web.archive.org/web/20180128151753/https://postimg.org/gallery/tqzm9phy/

I dislike fragmenting this information across several platforms, and do wish this forum did not have 1990 level restrictions .
But it is what it is

Also attached is the asc LTspice file (NOT A VIRUS) LOL
You can use https://www.virustotal.com/#/home/upload if anyone is paranoid....

And again... Help is MUCH appreciated and thank you to those that are willing to help

[T3sl4co1l]

You misunderstood. I know .rar and .zip but I do not open archives of that kind. I've been burned before.
Post the pictures as .jpg, everyone can handle that.

A .zip or .rar can't "burn" anyone.
An executable file with malicious content can.
There are no image exploits that I'm aware of, if there were I would shut my computer off and not internets at all, because even facebook is then a vector for infection.
You're being overly paranoid. I take my security, very seriously.....

Better unplug then...
https://blog.sucuri.net/2013/07/malware-hidden-inside-jpg-exif-headers.html
Probably not a significant vulnerability anymore, and I guess server-side, but they pop up from time to time.  Buffer overflows and SQL injections are still with us today...

Security is always a compromise. I certainly turn a profit, while not spending much time on security.  YMMV.

This is 2018, I packaged the hi res files so people can zoom in without seeing 1990 level fuzz.

Okay but I don't see any case where that many megapixels was deserved (1:1 scale clearly showing "1990 level fuzz" due to optical limitations of the camera), and many of the pictures are redundant. 

Here is a link to a new picture gallery on postimage.org (God help the poor soul looking for insight, stumbling across this thread in a few years to find the images don't work).
https://postimg.org/gallery/tqzm9phy/

Thank you.  Well, Imgur would've been both more accessible and less spammy, but at least I can see the pictures as thumbnails without downloading a huge opaque archive!

Here's a small yet significant optimization, merely saving the line drawing as 256 color PNG:

https://www.seventransistorlabs.com/Images/CDI_Schematic.png

Anyway, those transistors by the SCR don't look right, wonder if they're backwards..?

Tim

[floobydust]

Reverse-engineering something like this is a form of hell. You did a lot of work.
The schematic needs some checking, a few errors around the SCR drive.
Are you planning to make a SMT or through-hole clone of the module? This is a CDI module for Kawasaki KX80?

I had to redraw a schematic to figure it out.

I think you have the pinout wrong for Q1, Q2 (ECB). R9 should connect to Q1-B, R8 to Q1-C. Q2-C connection does not make sense. I think Q1-E goes to SCR gate, as well as Q2-C.
C7 should be flipped, + to D5/R7. Is R6 really 33R?

I think the only zener diodes in the design are D7 and possibly D2.

[xnotx]

~I don't see any case where that many megapixels was deserved (1:1 scale clearly showing "1990 level fuzz" due to optical limitations of the camera), and many of the pictures are redundant. 

~Thank you.  Well, Imgur would've been both more accessible and less spammy, but at least I can see the pictures as thumbnails without downloading a huge opaque archive!

~Anyway, those transistors by the SCR don't look right, wonder if they're backwards..?

Tim

Right, the archive is of all the images I took during the process, they were meant for me to look back on if I had any doubts about anything.
The images that are redundant are for me to be able to read the markings on them... sometimes a little reflection/glare at the propper angle can help me read out part numbers, especially when I can zoom in. Hence the size and many similar images.
I didn't pick and choose which to include in the archive, I included them all 

I think I have fixed the small issue in that area.
The configuration is head scratching, but If you compare to the bare PCB, you can see they are indeed configured in that manner.

Reverse-engineering something like this is a form of hell. You did a lot of work.
The schematic needs some checking, a few errors around the SCR drive.
Are you planning to make a SMT or through-hole clone of the module? This is a CDI module for Kawasaki KX80?

I had to redraw a schematic to figure it out.

I think you have the pinout wrong for Q1, Q2 (ECB). R9 should connect to Q1-B, R8 to Q1-C. Q2-C connection does not make sense. I think Q1-E goes to SCR gate, as well as Q2-C.
C7 should be flipped, + to D5/R7. Is R6 really 33R?

I think the only zener diodes in the design are D7 and possibly D2.

It is hell and extremely frustrating, but fun in it's own way and should be very rewarding once I figure this out. That is the fun of this
I'm still fairly "wet behind the ears", but this is all part of learning.

I went over the issues you have mentioned and there was a small error that was corrected.
The odd configuration of those two NPN transistors are indeed as designed.
I changed over some diodes to 1N4001/1N4007 and chose a schottky for D2 and a zener for D7. Still behaves about the same.

R6 is indeed 33 ohms

I attached a screenshot of LTspice showing the waveform of the ToCoil wire.
Looks a bit weird to me, what do you think?

Also attached, is the updated schematic screenshot, if you could have a look.
And for anyone brave enough to go bareback, the asc is attached as well.

Thanks everyone for the help, keep the positive energy flowing 

[floobydust]

Sorry, to make sense of the circuit, this is my SCH redrawing. It looks like a generation 2 CDI on these dirt bikes.

D2 is still not quite right. Either a small signal or zener diode. Have to carefully confirm this part.
D7 is probably a 9V or 12V zener.
R6 I notice the similar design in this KDX200 CDI.

I did not go into the LTSPice simulation. The pickup and exciter coils actually make a AC blip every crank revolution, not a steady sine wave, so your simulation will need some tweaking. Need an ignition coil load to charge up C9 properly, and to make sure the crank coils are in proper phase. You might have to look at the engine.


The hard part is figuring out how the ignition timing advance is done.

Silicon Chip magazine did a small engine CDI ignition, saying the thermistor is to reduce SCR gate-drive at high temps, I think down 50% at 80C.

Original SCR is probably Toshiba special purpose high di/dt for CDI use. see S6992 and S6A13. 800V non-sensitive gate. I remember Sanken or Hitachi had that green epoxy.

Ignition timing seems to be mainly done by the coil lamination shape? per the SC pic here.

edit: Feb. 4 updated diodes in schematic; corrected wiring error

[xnotx]

Sorry, to make sense of the circuit, this is my SCH redrawing. It looks like a generation 2 CDI on these dirt bikes.

D2 is still not quite right. Either a small signal or zener diode. Have to carefully confirm this part.
D7 is probably a 9V or 12V zener.
R6 I notice the similar design in this KDX200 CDI.

I did not go into the LTSPice simulation. The pickup and exciter coils actually make a AC blip every crank revolution, not a steady sine wave, so your simulation will need some tweaking. Need an ignition coil load to charge up C9 properly, and to make sure the crank coils are in proper phase. You might have to look at the engine.


The hard part is figuring out how the ignition timing advance is done.

Silicon Chip magazine did a small engine CDI ignition, saying the thermistor is to reduce SCR gate-drive at high temps, I think down 50% at 80C.

Original SCR is probably Toshiba special purpose high di/dt for CDI use. see S6992 and S6A13. 800V non-sensitive gate. I remember Sanken or Hitachi had that green epoxy.

Ignition timing seems to be mainly done by the coil lamination shape? per the SC pic here.

No prob, feel free to make any mods you want, sch looks professional 

That KDX 200 CDI design is super simple, but not sure about it... Put it together in LTspice and it didn't do anything?
The difference with that KDX design is the stator coils are arranged at around 90 degrees or so(like a sideways V) and not 180 degrees(opposite sides of each other) so I imagine timing is at a much different point, but compensated through design. (not sure how many magnets on the magneto).

D7 might be a 15 volt as you suspect. It does also supply voltage to the Nor gate chip.

D2 to U2, for me, is the head scratcher... You have the pick-up coil supplying xx Voltage and off that there is an RC Snubber? so then D2 clips negative voltage and leaves positive pulses of around 5 volts pk/pk into one of the inputs(pin 13) of U2(CD4001).
The other input(pin 12) to U2 holds steady at 0 after dropping from 5v, the output of that NOR section (pin 11) is a 5v pwm-looking square wave to C3 which out-puts a weird arbitrary signal... which is rectified by D8 and finally into collector of Q1
And it gets mixed with D5.... It's beyond me why   

So any diode there as D2 produces the same result, so if I had to guess, it is just a rectifying diode?
Another guess, is that most all of these diodes are glass because they withstand temperature changes better?

The pickup coil for this design is just a smaller, normal coil 180 degrees from the exciter coil. No funky head design.

I still need to make a CDI tester to test these things on the bench, that Is another goal of mine.
These are all personal use items.

I guess if we settle on the diodes used, I'll put this away for a couple months.
I still want to grab a newer year version to see what the difference is on them. Who knows, maybe it will be an easier design to work with   
There are a lot of oddities in this design lol
Anddd I grabbed the book "The thyristor book: with 49 projects" so will have some reading to do, and perhaps some small learning projects to keep me busy for a bit. 

moarrr questions
So if you were to place diodes in this circuit, which ones would you suggest for all of them?
And would this thyristor be a good starting point? www.st.com/content/ccc/resource/technical/document/datasheet/bf/65/77/27/f7/59/49/4d/CD00001569.pdf/files/CD00001569.pdf/jcr:content/translations/en.CD00001569.pdf

Thanks!

[floobydust]

For your diode substitutes, vanilla rectifier like 1N4004 for D1, D4, D5, D6. It's OK to use higher voltage parts than the minimum, for reliability. 1N4148 for D2, D3, D8.
D10 and D9 I would use 1,000V 2-3A soft fast-recovery diodes like BYT56M, UF5408 is huge DO-201 and probably too fast.
D7 as 9V? 1/2W zener. It looks like 9V is needed- tried a basic sim. But I can't get more than 5V pulse into the CD4001 at C5. The CD4001 model oscillates 125MHz on the rising edge, so that would have to be fixed.
D2 cannot be a zener as C6 gives a large -30V pulse.

For the SCR, you don't want a sensitive-gate SCR here. The whole circuit (NTC etc.) is set to drive something like a 5-20mA part, not uA of IGT.
I'm using the Toshiba 6A13 which is 800V IGT=30mA max. dV/dt=50V/usec typ. and difficult spec critical rate dI/dt=750A/usec. Most SCR's can't take this high a pulse current.
The 4A ST part, common on chinese small engines and ATV's, is rated 50A/usec and I would use a 12-16A part instead. The SCR is highly stressed and higher rated parts are the same physical size.
S8016 is rated 125A/usec. and the highest I could find.

[xnotx]

Thanks for the suggestions.
Would you be able to get a screenshot of the oscillation you're talking about?

Not sure if there were high di/dt thyristors available in 1980, but the S6A13 came to market 2 decades later lol
Can't seem to find anything with such a high di/dt capability that is an active product.

[floobydust]

This is the LTSPice sim I was playing with, of just the SCR-gating one-shot portion.
It's a kluge, just to see if the teardown schematic is reasonable and for diode currents and time constants etc.

I switched CD4001 Spice models and much tamer, the LTSpice CD4000.LIB gave RF oscillations on the slow edges. I think R2, C2 might be to filter that, although a Schmitt trigger is better suited here.
Running CD4001 in a linear mode, not sure if the (model) gain is accurate. U2 output is not swinging high beyond about 4V.

To go further, you would need an accurate crankshaft coil pulse generator, like a counter using taps at the various crank angles. The ignition timing advance is done with the DC level on C6 moving up with RPM, as I figure.


For the SCR, CDI was used even in the 1960's. Just don't skimp on the SCR, the little 4A parts won't last in motocross.

[xnotx]

Yea it looks good. 
I see what you're talking about with the quirky top of the U1 out waveform.
Your model's waveform at the other end of C3 looks much different, in that it is keeping the waveform more identical on both sides.

I like your simplified layout, much easier to see the flow.
I think I see it now, U1 with c3 and c6 is for the timing. I think it's making some sense.

Ok sweet, thank you for all the help! Uber appreciated 
I still need to find the time to get the drill on that magneto and give it a spin with the oscilloscope to get a better idea of the voltages and whatnot at idle.
I think all the quirks in LTspice sim is just ltspice with incorrect sine wave and pulse and no load/back emf of ign coil to cap charging etc etc.

Almost time to start ordering parts!

[xnotx]

Waveform pictures added.
10x probe @ approx. 2000 RPM
CDI not connected, just scope.

[floobydust]

For both coils, rate of change of the magnetic flux affects EMF (Faraday's law), so faster crankshaft speeds give more output voltage.
It's not a sine-wave because a lone magnet is spinning past each coil.

What did you use to run the motor, with no CDI?

The exciter/charge coil, I'd expect about 200V-300V pulses, which is what you are getting (note the scope is clipping).
The pickup coil, about 20-60V pulses, enough to power the CD4001 and SCR drive.

I see 12msec from end-end of each pulse, and you could figure out diameter/pole length etc. from RPM.

The coil's remanence I think is why the waveforms are not symmetrical, or geometry of the laminations.

Patents for the flywheel CDI ignition explain more. There are flux concentrators in the magnet: http://www.google.com/patents/US3641377

Wow, this circuit goes back to Tecumseh 1970 http://www.google.com/patents/US3524438

[xnotx]

Yup, exciter looks about -80 / +250v easy @ 200RPm ish (speed of drill).
I think the cdi might take advantage of that negative first type pulse to turn the scr off?

The pulse is around 7 or 8 volts. Had it set to .1x5/div(.5v)+10x probe=5v/div. (the exciter was at 50v/div) 

The scope display was clipping a bit, could have moved it down manually but didn't think of it lol
It's approaching about maximum for that little scope.

I hooked up my drill and spun the motor over with plug out.

Thanks for the lesson on remanence, makes sense.
I like that though, it's like someone combined remain and memory to come up with that word 

Nice, that circuit does look like a super simple version.

Oh, just remembered something... The big cap had almost an ohm of resistance, not sure if it was bad or by design.

[floobydust]

I'm suspicious the pickup coil's waveform has zilch for -ve amplitude. That's a problem 
Either the magnetics are wrong/weak, or this CDI circuit is for a different engine? Are you willing to post part number/model numbers.

With the part values we have, I find +7Vpk from the pickup coil is not enough to trigger the IC (in the LTSPice sim) 
Need 2X that to get >1.5V into U1.

Since CD4001 is holding OFF the SCR, it's actually not a problem- the engine would run, even with no CD4001 at all. The SCR will get triggered by the pickup coil's opposite edge.

Without pics of the magneto generator, I can only speculate wtf the CD4001 is supposed to do. Three (timing) resistors in series usually means the engineers were rushed and didn't get a change to debug/polish a design and just rolled it out...

[xnotx]

I'll have to put it back on the scope but at 1x this time.
Haven't had time to play with the new voltage values in LTspice yet lol

Maybe they went for a simulated cd pulse to turn on the cd4001 to enable it for timing control?

I'll get some pictures later today.

[floobydust]

Best to DC couple the scope for the LV pickup coil. Sometimes a scope's low freq rolloff on AC coupling can give a fake DC resting level.
Or try it with a small 1k load resistor on the coil.