Actually the supply should probably be as least 12V

I have simmed it in LT spice. also 9V is seems OK. You expect a problem with 9V in "real life" ?

 

for others... the final .asc.
Don't mention the file name; it's not correct 'yet'
The only thing I need to do is to find the right TVS's that can absorb sufficient energy in case a spark find its way to the probe.... Any idea on that or maybe a new treat?

 

..................
in the article ref.d in #1 on page 5 they explain the R as "

A second capacitor of about l000pF should be connected as shown. The capacitive divider thus formed divides the input signal by about 1000:1 thus reducing the input signal to a workable 3 - 20 volts. A 1M resistor should be connected across the 1000pF capacitor to provide a dc load. "

The op amp input is a resistor so you don't need the added 1 Meg resistor since the two are effectively in parallel.
So remove the external 1 Meg resistor and change the op amp input resistor to 1 Meg.
Then a feed back resistor of 100k will give you a gain of 0.1.

I have simmed it in LT spice. also 9V is seems OK. You expect a problem with 9V in "real life" ?

9V will give you a positive peak voltage of about 3.5V before it starts to clip, if that's okay.

Since the positive peak voltage is greater than the negative peak, you can add some additional headroom to the positive peak output voltage by reducing the DC voltage to the (+) op amp input (reducing the value of Rb in the Figure 3 schematic).
You can adjust Rb until the negative voltage peak is near zero at the op amp output.

 

The op amp input is a resistor so you don't need the added 1 Meg resistor since the two are effectively in parallel.
So remove the external 1 Meg resistor and change the op amp input resistor to 1 Meg.
Then a feed back resistor of 100k will give you a gain of 0.1.
9V will give you a positive peak voltage of about 3.5V before it starts to clip, if that's okay.

Since the positive peak voltage is greater than the negative peak, you can add some additional headroom to the positive peak output voltage by reducing the DC voltage to the (+) op amp input (reducing the value of Rb in the Figure 3 schematic).
You can adjust Rb until the negative voltage peak is near zero at the op amp output.

Thanks, I 'll take some more samples with the bitscope to check the wave form later on this week. I read that in worst case scen. it can go to 50KV. So input protection becomes very important. You have an idea on the right TVS's or other protection methods? Even for a direct spark pick up?

 

I think just two standard 1W zener diodes of the appropriate voltage connected in series back-to-back will do the job.

 

My brain must have stalled on this stuff since i'm indeed a newbie........ from #18 on, I cannot see the signal mirroring (inverting opamp?) in the output.
The Probe R is omitted as suggested.
2 Zeners 12V added.
The output is much more reduced than calculated 10:1
PWL is adapted for the case of 50KV.

What am I missing???

 

You didn't post you PWL file so I don't know the characteristics of the pulse.
For a gain of 0.1, R5 should be 1 meg and R6 should be 100k as I previously suggested.
Why are you clipping the signal at 12V, when it has a peak value of 50V at that point?

 

You didn't post you PWL file so I don't know the characteristics of the pulse.
For a gain of 0.1, R5 should be 1 meg and R6 should be 100k as I previously suggested.
Why are you clipping the signal at 12V, when it has a peak value of 50V at that point?

You're right about the clipping. I can go up to 50V which I did. Output stil very small for my bitscope... and not inversed. Why???

 

Here's my simulation, which seems to work fine, so I don't know what the problem is with yours. The output is a pulse of about +4.5V.
I changed R1 to 20k to give more positive output voltage headroom for the op amp.
I used an LM324 op amp, which is basically a quad version of the LM358, since I don't have the LM358 model.
Why did you reduce the value of the decoupling capacitor C5?

 

Here's my simulation, which seems to work fine, so I don't know what the problem is with yours. The output is a pulse of about +4.5V.
I changed R1 to 20k to give more positive output voltage headroom for the op amp.
I used an LM324 op amp, which is basically a quad version of the LM358, since I don't have the LM358 model.
Why did you reduce the value of the decoupling capacitor C5?

View attachment 108464

I see U have a good inversion. But.... I have done the same sim with the LM324. Results are not equal. As U can see in my signals as below, not all waves are inverted especially the first and the last are not, while in yr sim all are inverted and the segment after the lage peak should slowly rise since it is falling at the input..... Further on I see a negative peak just before the large peak on the output, where this (inverted) is not present on the input) Exactly those details are important in the interpretation of the waveform from ignitions. What is the diff between our diagrams, I wouldnt know... different spice model or diffrenece in the .TRAN?
Furtheron, if a reduce the voltage to 9V I dont have any inversion any more.....very strange things...

 

You have the inputs reversed on the op amp. Take a closer look at my diagram.
Your connection give positive feedback instead of the negative feedback you want.

 

You have the inputs reversed on the op amp. Take a closer look at my diagram.
Your connection give positive feedback instead of the negative feedback you want.

Many thanks for all yr help!! Guess it was to late at night for me...
I only see one thing I do not like 100% and it is the segment after the large peak which should be rising since the input is falling. Do you see a solution for it?

 

Many thanks for all yr help!! Guess it was to late at night for me...
I only see one thing I do not like 100% and it is the segment after the large peak which should be rising since the input is falling. Do you see a solution for it?

Yes, the time-constant of the 1000pF cap and the 1 meg input resistor is only 1 ms which is too short to see the slow rise in signal, since that time is also about 1 ms.
Change R5 to 10 Meg and R6 to 1 meg, which increases the time-constant to 10 ms.
See results below.
Now the slow slope changes are reproduced accurately at the inverted op amp output.

 

Yes, the time-constant of the 1000pF cap and the 1 meg input resistor is only 1 ms which is too short to see the slow rise in signal, since that time is also about 1 ms.
Change R5 to 10 Meg and R6 to 1 meg, which increases the time-constant to 10 ms.
See results below.
Now the slow slope changes are reproduced accurately at the inverted op amp output.

View attachment 108501

Fantastic!! Thanks to yr explanation I have also the opportunity to learn about opamps!
A few more questions...
-I would like to take the supply from the car battery, where I suppose there is a lot of noise,
Is it necessary to put other C's at the supply of LM358 or do you think the 1µ will be enough.
- I have another question about the location of the zeners and the neon. I read somwhere that the place of the components in case of ESD is important.
- Now the pickup probe with the split tube is connected with a shielded cable of 1meter to a plastic box where I put the Cpick_up.
The split tube is in fact only connected to half of the tube. Is the pick up better if I connect the wire to both halves? or will this also change .
capacity and possibly change the ratio of 1000:1
- Is the localisation of the ground connections important? Now it it only connected in the plastic box (to battery neg). Would it be interesting to have the shield connected at the probe end also (near the split tube but with a gap> spark plug?) . In case of a broken spark plug cable, the spark would then possibly find it's way to the shielding and battery neg instead of to the core (and the LM324 and bitscope and PC...)

Thanks for yr patience ....

 

To minimize battery noise and spikes I would put a 10Ω resistor in series with the battery connection and use at least 100μF to ground at the op amp power pin along with a 0.1μF ceramic cap in parallel.

Place the neon and zeners close to the circuit with short connections between the signal and ground

If you use both tubes, that will approximately double the pickup capacitance and thus double the voltage.
Obviously the length of the tube also affects the capacitance.

A shield ground connection directly to the engine block and to the circuit ground may give you a better signal.
You can try it both ways and see.

 

To minimize battery noise and spikes I would put a 10Ω resistor in series with the battery connection and use at least 100μF to ground at the op amp power pin along with a 0.1μF ceramic cap in parallel.

Place the neon and zeners close to the circuit with short connections between the signal and ground

If you use both tubes, that will approximately double the pickup capacitance and thus double the voltage.
Obviously the length of the tube also affects the capacitance.

A shield ground connection directly to the engine block and to the circuit ground may give you a better signal.
You can try it both ways and see.

I 'll keep U informed after a few tests !!