SEM cdi

[ty w]

ive been months trying to get find a cheap sloution for sem cdi blown. tried vespa unit and needed too much adjusting to be steady and reliable. tried a chinese racing cdi and2 spark per rev, wont run so i just made one from blueprints and perfect. if you want one let me know. itll work probably any coil. it works the small scooter one with two prongs one black and one green ground, and it works an older yamaha coil that was also used by kawasaki. ive only tried these two and it works both of them. though the old yamaha one sparks bigger and the scooter one doesnt seem to trigger my timing light, making timing trickyer. none the less it still runs the engine. the scooter exciter coil primary is like .2ohm while the yamaha one is about 1.5 ohm, so i suspect the yamaha one is easyer on the cdi, but i dunno that for sure.
and i can build them smaller than one inch cubed, with the small scooter coil it reduced the weight a few g,s

 

[ty w]

tickled pink i am today, my self built cdi works soooo gooood. my eyes got teary, at 110 kph on windy gravel. this seems a solid solution for dead all spark

 

[ty w]

and today ive another idea. ive ordered more pieces and my next design will include 2 cdi in one unit, so for if it burns and leaves me stranded , ill just swich 4 wires and then continue on. with the chinese scooter coil being so small i could have 2 of these with my double cdi all fitting under the gas tank. thats a whole backup ignition except for the stator. so thats my next step, and that way murphys law will prevail and ill never have ignition trouble again

 

[KXcam22]

Can you make it switch over automatically? Then it would truly be a redundant CDI. Cam.

 

[ty w]

i think it could be done. by someone who knows more. ive simply found the schematic for a ktm 4 stroke sem cdi and built one. it worked so good that im doing it again, smaller and 2 in 1. to have it swich auto would that no wires swich and so internally the twowould be feeding the same output. when only one unit working at a time, the woking one feeding the output, but the other connected might be harmed by this power coming in the output when off? im not sure. so im leaving them separate, except for sharing ground wire. a 4-pole double throw switch would do to go from one cdi to the other without swiching wires. would be just a flip of a swich. but what im going to do is to get the pieces out of my old cdi and see if theres any differences from the sem 4-stroke plans and my old 2-stroke cdi. could be the one i built is the same or not, thinking about ignition delay at low rpm. may be a different delay curve, rather than advance curve

 

[ty w]

so the one i built works perfect. i decided to discombobulate the old cdi to check for any differences, and i did find some different resistors, these 2 make timing delay. so the difference in these resistors must affect timing curve. so im going to build one with the original resistances to check for different timing aspects

 

[roberto75015]

so the one i built works perfect. i decided to discombobulate the old cdi to check for any differences, and i did find some different resistors, these 2 make timing delay. so the difference in these resistors must affect timing curve. so im going to build one with the original resistances to check for different timing aspects

that's interesting, can you share the schematics ?

thanks a lot,

Roberto

 

[ty w]

no , because in the beginning i was told its impossibe , it cant be done. so ill tell you the same, its impossible and cant be done. but if you check aroundfor 'ktm Lc 400 " you should find one page that covers the sem cdi. although its the kind mounted to the magneto cover as opposed to internal mounted one the stator may be wound different but the cdi coil is almost identical, and Martin Mikusa did well to show the schematic. its the red cdi for a four stroke, and my home built one of this deesign works great. it still sparks every tdc but has different resistors on the trigger input circut. so i think is a different timing delay at idle, perhaps for engines that idle slower than two strokes. anyways it works great and so i dont ask questions. i had previously butchered apart my old cdi ,so i checked and the resistors being different, made me curious so further inspecting and i discovered the problem with the old one was the diode for the negative half of the charge wave had blown shorted out. this caused the innability of the positive half of ac charge power to build up in the capacitor, it was permanently shorted to ground through tis diode. replaced with a 1N4005 diode and the original cdi now sparks again. i think ill have to test it

the one as shown by Martin Mikusa has the resistors measuring :
R1=100 ohm to ground, then a diode to prevent negative power to scr, then R2 12 ohm to ground.reducing trigger voltage that triggers the scr switch.

on the butchered 2 stroke cdi : trigger signal goes through the diode first , then a 6.8 ohm to ground, and a 100 ohm to trigger scr. ( one diode then two resistors in paralell, one resitor connected to groung and the other feeding to scr). now ive attempted to build a two stroke one from scratch, and it didnt turn out, but it might not be the fault of parts. i may have made an error. also to acheive the 6.8 ohm resistor i tried two 12 ohms in paralell and resulted 6.0 ohms not 6.8 ohms, which could be the culpret as this is resisting some signal power drain to ground. too low of resistance could reduce the signal voltage too much and not trigger the scr . needs more experimentation
another difference is the scr itself. i cannot see for sure the numbers on it, and when i measure the resistances of the new vs old scr they do differ some. and this maybe why the different resistors on the trigger signal circut?
this other site also shows how all the signals interact , on the screen of an oscilloscope, showing how it advances.

just imagine the wave: a=1.5 volt trigger point when wave is +- 6 volt
b=1.5 volt trigger point when wave is +-2volt
point x to point y is one full wave happening 1 time
per engine revolution. so if this wave takes 20 degrees
of engine rotation it can advance or delay almost 5 degrees
with rpm and amplitude increase.


/\
b \
/ \
/ \
a \
--0 volt---------x--------\---------y-------------------------------------------------
\ /
\ /
\ /
\ /
\/



so if the resistors divide voltage to just over 1.5 volt at kick speed or idle (minimum trigger volt required for the nte5419 scr chip, i think) than it will automatically advance this trigger point as amplitude will increase with rpm.
the nte5419 chip is not the original i forget wats the original, something like q4004f42
anywayse theres a bunch of chips that will do
tyn1012 1000v/12A ?
bt151-800 ?
bta151-800 ?
nte5419 800v/12A
nte5438 600v/4A
q4004f42 400v/4A ?
MAC210A8G 600v/10A shape is called TO220AB (AB not necessary)

the diodes can all be 1n4007 1/4 inch long ones, i used 1n4003 on the trigger signal side, and 1n4007 on the capacitor charge side, as long as theyre rated higher than the voltage they stop . like >than 100v on signal and >400v for the charge side

the resistors are the 1/4 inch long ones too. 1/4 watt?

im not sure but the chips power rating might have advers effects on reaction speed of the chip, possibly affecting the max engine rpm ? but im not sure, stats are available and then you could calculate max rpm.

i used a NTE5419 for the first cdi that i made and thats the one im running at this time, as per the 4stroke diagram and resistors.the next ones i built used a comparable chip MAC210A8G and it too sparks for the 4stroke plans, ive not run it yet. and as i mentioned i tried to make the 2stroke cdi but used a 6 ohm resistor where it calls for 6.8 ohm, possibly being the reason why this one didnt work i need to go back over it to verify theres no other reason of fault, maybe i overheated the chip when i soldered it onto the board. they do have a limit of how hot they can get before they cook.

 

[ty w]

my attempt at a sign wave on the previous post wouldnt turn out right. it was to be like a graph but went array
the spacing wont stay constant. it should look like an "S" turned backwards like2 but rounded, and then turned 90 degrees eitherway. the result would be a wave linr first positive than negative. and it you have a cheap analog voltmeter you actually can see it right off of the green signal wire from the stator, the needle will moove quickly +5ish volt then quickly reverse direction and go off scale in neg direction, but only when on dc volt reading. if on ac volt reading, it will then show almost a stable 12volt ac i think. anywayse its the needle going + a couple then quickly reversing direction and showing neg some , thats what youre wanting to see. and if it goes neg first theres a problem, maybe test leads are reversed. neg first will have a much longer delay time and possibly not even trigger the scr. i dont know? , but they say that a clockwise rotating stator(internal) will not interchange with a counterclockwise rotating one. they say it has to do with windings. the ac charge would probably be unaffected but the trigger would be neg than pos putting the timing delay out of order, if it fires. somewhere it says theres a diode in the stator, if so when rotating backwards it wouldnt give a pos half to the wave? it doesnt require the neg half of the wave to trigger as it depends on the scr trigger requirements which happen to be +1.5volt, or maybe 0.5v+. that depends on the exact requirements of the scr chip you get. i do believe their available for different specs of trigger requirements( you can get triggers that swich at a variety of trigger voltages )

 

[ty w]

View attachment 21510View attachment 21510


note the resistor and capacitor marked 6.77kohm and 1k. i dont know what cap could be 1k nor what this does for the charge circut, seems like it would smoothen out the pos half of the ac charge wave or giving longer spark duration

 

[roberto75015]

TYW, thanks for taking care to post all this data which I will try to digest it in the next couple of weeks (christmas break and also lot of food to digest :-)), given I'm not so good as you at understanding electronics circuits.

It's not clear to me if your self made CDI is supposed to have variable advance and by which mechanism and with which curve.

I assume you are referring to this page which explains how the 4t ktm works: http://transmic.net/en/sem.htm.

For you who are definetely able to understand more than me, here a link with a lot of schematics of CDI: http://www.motelek.net/, too bad they don't have a SEM section, but lot of schemas under http://www.motelek.net/andere/cdi/ and other folders.

On my side I'm simply try to see if I can transform the standard SEM I have on my 85 125 into something with variable advance, without replacing the whole system. I tried with the blue coil from Aprilia RS (rotax 123), SEM coil part no 14 04 101 888 00, but this one advances too much and causes the engine to rattle even when starting from a minimal advance: I may decide to open it and try to reverse engineer this model, even if I'm not much sure how you dissolve that plastic/glue that covers all the electronics without breaking everything.

I also would love to try a KTM 2T one (part no 14 07 <something>) but I don't have one yet.

thanks again, will keep you posted if I come to understand more.

Roberto

 

[ty w]

the first link is the one. great info. the other two links dont work for my computer.

i will one of these days put a timing strobe light on it while running to know for sure eitherway
and from what i can tell theres no mechanism to advance timing. its electronic. the signal is a sign wave. the cdi triggers when sign wave hits 1.5 volt. if the sign wave is 1.5 v+ peak then 1.5 volt neg peak, the cdi will trigger at the 1.5 volt peak, about 1/4 of the wave lenth. it the amplitude of this wave increases, to like 6volt + peak then 6v-peak , it will advance the 1.5volt point on the wave. thus triggering spark earlyer. this is all controlled by the R1 and R2. if either are replaced with a potentiometer it will make a variable delay/advance. raise value for R1 or lower value for R2 to delay more , opposite to advance. to acheive a bigger curve , find a scr that triggers at lower voltage, then reduce R2 so just eanough power to trigger at kickover. in doing so you can use the timing difference between 0.5v up to 6v instead of 1.5v to 6v. this will use more lenth of the sign wave. giving more curve. although it will only advance within the wavelenth. i may have read somewhere they thought the trigger point is when the sign wave finishes positive half and crosses the 0v centreline(ground) or half way of the wave length. some non sem stators use two pickup coils at differene timing spot, and ther electronically tuned to have resonant frequency like, so for the late pickup triggers for idle and as rpm increase it triggers from early pickup, advancing timing. after all this cdi torture ive come to like the way sem works, now that i really know it works. the worst of it all is the smoke filling.
somewhere it says like turpentine mixed with acetone or something toxic to dissolve filling. i used heat and pick but risky of stabbing innerds.
now that ive repaired my old cdi aswell, now i need to seal it from weather. i suppose fiberglas resin will do, but i would rather make it so its easyer to open...if i have to. maybe candle wax or silicone, otherwise corrosion and moisture will kill it.

 

[ty w]

the first link is the one. great info. the other two links dont work for my computer.

i will one of these days put a timing strobe light on it while running to know for sure eitherway
and from what i can tell theres no mechanism to advance timing. its electronic. the signal is a sign wave. the cdi triggers when sign wave hits 1.5 volt. if the sign wave is 1.5 v+ peak then 1.5 volt neg peak, the cdi will trigger at the 1.5 volt peak, about 1/4 of the wave lenth. it the amplitude of this wave increases, to like 6volt + peak then 6v-peak , it will advance the 1.5volt point on the wave. thus triggering spark earlyer. this is all controlled by the R1 and R2. if either are replaced with a potentiometer it will make a variable delay/advance. raise value for R1 or lower value for R2 to delay more , opposite to advance. to acheive a bigger curve , find a scr that triggers at lower voltage, then reduce R2 so just eanough power to trigger at kickover. in doing so you can use the timing difference between 0.5v up to 6v instead of 1.5v to 6v. this will use more lenth of the sign wave. giving more curve. although it will only advance within the wavelenth. i may have read somewhere they thought the trigger point is when the sign wave finishes positive half and crosses the 0v centreline(ground) or half way of the wave length. some non sem stators use two pickup coils at differene timing spot, and ther electronically tuned to have resonant frequency like, so for the late pickup triggers for idle and as rpm increase it triggers from early pickup, advancing timing. after all this cdi torture ive come to like the way sem works, now that i really know it works. the worst of it all is the smoke filling.
somewhere it says like turpentine mixed with acetone or something toxic to dissolve filling. i used heat and pick but risky of stabbing innerds.
now that ive repaired my old cdi aswell, now i need to seal it from weather. i suppose fiberglas resin will do, but i would rather make it so its easyer to open...if i have to. maybe candle wax or silicone, otherwise corrosion and moisture will kill it.

 

[ty w]

i think that if it gets more delay at idle, you can redo the timing stator plate timing a bit more advanced, then when it advances for high rpm it advances more. but too much isnt good either, and late timing at low rpm is better than early timing at high rpm as it results in more low end torque not preignition. but im not sure. so i really need to put a timing light on it, to see whats going on as to timing advance. on the other site, the picture of the oscilloscope look like they they show advance below 2000 rpm or 1500 , i forget what the speed was though.

i wonder if you could use an led mounted near the flywheel, and powered by the cdi somehow so it blinks every time it fires coil, this could be useful to do the timing . those scooter ones blink to show spark, i like that feature, and so if that led was on a longer wire that led back to the flywheel , it could blink at a certain position and show timing marks, that way you can check timing without getting the timing light out.

 

[ty w]

ooo nnooo. almost a catastrophy. almost. i foung the ignitor coil ground wire dangling. had it been the only ground, idabeen pushing. but i did have the coil double grounded. it coulda risked blowing the cdi, again.


cdi = can die immediately

 

[ty w]

i need to know more about triacs, and trigger sensitivity

 

[JMang]

I can't thank you enough for your knowledge, it enabled me to build a replacement cdi to fire a yamaha coil for my 400sxc! I had to tweak the voltage divider to trigger the triac on mine, 100/23, but she's running like a dream!

 

[vmxwinn]

I need help, think my CDI/coil has gone out. PM sent.

 

[jo360]

Not hundred percent on all this electrical business but fitted a electrex world stator coupled with a jaycar cdi 2 stroke kit and chinese coil, seems to fire earlier and much brighter than the sem unit.
My thought was to put a variable resistor mounted on the handle bars as a restrictor to lower the exciter output and vary the timing.
The current set up works on a lower exciter voltage and fires just as the magnets come into the exciter coil area.
Any thoughts as to feasibility of such an idea?.