[TheMonkey]

Originally Posted by Engine-Ear

what do you connect the probes to??

what are the specs on the probes (1X/10X, how my pF load capacitance, operating bandwidth)?

help this gEEk out, will ya?

probes connect to the scope box:


which connects to laptop through USB port.

probes are basic 1x without any attachments. a guy at the MegaSquirt site did some comparisons using these probes and also using some very expensive probes and high end scope, and said he did not see any loss in resolution. not sure on capacitance or bandwidth; i'm a neophyte geek . some tech info on the scope is here: http://www.hobbylab.us/Oscilloscope/...echParams.aspx


also has a 16 channel logic analyzer (or 8, if using the other 8 for logic generator) and attachments that i suppose go into a breadboard, but i have no idea how to use them, or why, or when.


mostly, i got this so that i can take a look at the signal going into my ECU from the VR sensor on the crank, and compare that with the ignition signals coming out. i can make adjustments within the VR conditioning circuit (zero crossing adjustment & hysteresis), and want to make sure that is solid. using the scope to find RFI is a bonus... just need to learn what to look for.

[Engine-Ear]

Hey thanks for the followup...pretty neat stuff.

So are you doing this in prep for your 'ace' ('61)? if not what else are you working on?

[TheMonkey]

yes, it's for my '61 Bonneville bubbletop. it's been tempting to go for a bobcat type of build, but i'm sticking pretty close to stock on appearance, and flavor will be period correct, but the details will be restomod (MPFI, OD, & disc brakes).

check out this similar car on cardomain (i think it's on the cover of Smoke Signals this month). it pretty much captures the essence of what i'm after (except i'll have bench w/ floor shifter & wheels to accomodate discs). put your hand over your heart when you click on this, and ask the kids to be quiet for a moment: http://www.cardomain.com/ride/2954933

[TheMonkey]

well... finally got things set up again. motor fired up. here's a scope shot from the alternator. doesn't look like there's too much variation in voltage but i don't know what's normal?

the red trace is from the alternator output, blue trace is connected to nothing.

i still have something creepy to sort through, where i'm getting ignition misses, but i don't think bad voltage is the problem? what sort of peak to peak voltage variation begins to raise a red flag? this CERTAINLY isn't 32v as originally thought by using the cheap digital voltage meter.

** edited to add: ** i had to kind of 'position' the probe on the output lug, and could not press on it during the capture. i have found that has resulted in small spikes, like at the beginning of the trace, so that may be the reason for the spike at the front end of the trace. i didn't have 3 hands or the reach necessary to press the probe, and capture it at same time.

[LPete]

Here's a guide from a Toyota publication. You need to set your scope to AC, and about 1V full scale. You want to see just the AC component of the signal.

The normal waveform is the top half of three sine waves, one after the other.

[george kujanski]

Doesn't look like much AC ripple there at all, even on that scale. You may not see much anyway since the alt output is connected almost directly to the battery, the battery will filter out darn near all the alternator ripple.

It looks normal to me. What ever is causing your miss, to me it doesn't look like it's the DC power.

I would first check the signal voltage at the computer and check the bat voltage supply at the computer with the ground probe at the computer ground.

George

[TheMonkey]

from the Toyota charts that LPete provided, i'm not sure i can get my frequency reading high enough for that. not sure how fast those examples are, or how high the voltage swing.

but, as George pointed out, AC voltage problems would prob have shown up in the scope shot i provided with bigger voltage swings than seen in the pic.

6AL ignition is grounded and powered directly from the battery, as per MSD instructions.

my engine break-in stand has the electric fuel pump right by the motor. perhaps that is causing some of the problems. can i use my scope directly on the power supply of the fuel pump to look for any noise?

all ignition components are brand new, new plug wires, etc. i have checked all grounding and have new battery cables with negative directly on the aluminum surface on the head.

i did probe the 5v signal inside the ECU (after regulator and ECU filtering) using the ECU ground. it was ROCK steady.

the misses happen when i get my tach signal from a VR sensor and toothed wheel on the crank pulley (through shielded wire to ECU). when i get my tach signal directly from the tach signal provided by the 6AL box (a square wave DC signal), the misses don't happen but i can't control spark, i have to use distributor mechanical advance that way. i have to configure the ECU a bit different between these 2 triggers (both hardware & software), but interesting that misses only happen through VR circuit.

so.... IF the fuel pump is noisy and getting into the VR signal (which is shielded), how do i find that RFI and cure it? also, my fuel pump is grounded through the hard aluminum fuel line. perhaps i should ground directly, but the fuel lines are short and fat, so it seemed sufficient.

thx for any help.

[LPete]

Originally Posted by TheMonkey

from the Toyota charts that LPete provided, i'm not sure i can get my frequency reading high enough for that. not sure how fast those examples are, or how high the voltage swing.

The ripple will be somewhere around .1 to .3 VAC with a good alternator (a lot higher with bad diodes or stator windings). Assuming 2000 RPM alternator speed at idle (rough guess), and 6 "humps" per revolution, that works out to 1 hump every 5 milliseconds, so a 10 or 20 msec/division time scale should be plenty fast.

I agree -- there's not excessive ripple in your pattern. I've spent a fair amount of time trying to look for EMI and RFI using a scope, and I've decided I'm no good at it. It's really hard to tell what's normal, what's not, and what's caused by the leads picking up noise.

[TheMonkey]

any suggestions on how to 'hang a capacitor' on the fuel pump? what capacitor spec should i try and do i just connect it between the pump power lug & ground?

[george kujanski]

Monkey; from your previous post, you show a SIMULATED crank pickup pattern to the ECU. How about getting an actual screen shot of the output from the Vr sensor?

One thing I noticed from your pic showing the mechanical mount of the sensor... The best mount would be on a radial line from the wheel...that would get you a very peaky signal from each tooth, much better than the sig you are getting now.

I'll bet the sig you are getting now has too slow a rise and fall time for the ECU to discriminate where the actual tooth tip is. Take a look at it and post it.

george

[TheMonkey]

George:

both these scope shots are at idle.

the red trace is the VR sensor AC voltage. it goes through a conditioning circuit in the ECU, and converts it to a square wave DC signal (the blue trace). i have programmed the ECU to look at the rising edge of the blue trace as the trigger signal. ECU has some coding to determine if the gap between teeth is a missing tooth. if you look at the VR red trace, that is the steeper side of the teeth and should be a higher resolution than the rising VR signal that sort of comes up from negative, slows down, then rises positive. i don't have much to compare from experience, but in my opinion the signal seems pretty good.

also, second scope shot is the VR signal red trace, and the signal that would be used to fire ignition if i were able to use it without the misses. the blue trace goes from floating and suddenly drops to ground signal when the desired spark is to occur. this signal goes into the white wire on the MSD 6AL box, and the dwell time is irrelevant with the CD ignition. this scope shot looks good too, but surely does not represent a moment that includes a miss. the approximate TDC is 5th tooth after gap, and the ignition occurs right where it is supposed to at about 25° BTDC (10° per tooth).

i think i'm going to try a 0.01 mf 500v capacitor on the fuel pump, and a better ground for the pump. perhaps that RFI is interfering with the VR signal going to ECU, but i would have thought the scope trace would have shown that (not intermittent, but continuous noise).

[TheMonkey]

i'm not sure why the amplitude of the VR signal changes. it goes from low peak to high peak over 12 teeth, and low to low over 21 teeth. this is on a 36 tooth (minus 1) wheel so it's not in any kind of phase with rotations. the installation seems perfectly centered (within a thou on the calipers), round, and flat; without any variations in the individual teeth.

[LPete]

Originally Posted by TheMonkey

i'm not sure why the amplitude of the VR signal changes. it goes from low peak to high peak over 12 teeth, and low to low over 21 teeth. this is on a 36 tooth (minus 1) wheel so it's not in any kind of phase with rotations. the installation seems perfectly centered (within a thou on the calipers), round, and flat; without any variations in the individual teeth.

That is curious. I wonder if it would vary if you look at it with it disconnected (I'm wondering if there's some way the box is loading the signal). Also, does the period of fluctuation vary with engine speed (same number of teeth regardless of RPM?).

[george kujanski]

In looking at the scope traces, the ripple on the red trace is in sync with the time that the MSD would be charging it's internal capacitor.

Make sure the MSD power leads are connected directly to the battery and the ECU also connected to the battery. The alternator should also be running to keep the battery voltage from drooping due to MSD operation.

Question: what type of signal waveform does the ECU expect to see from the crank trigger? If you're not duplicating what the signal condiitoning circuit expects, that certainly would cause incorrect operation.

You also say that the miss occurs when you use the VR sensor to trigger the tach....more evidence that the tach is somehow loading the sensor output as Lpete suggests. Check the Vr signal with & without the tach to verify.

The thing with a Vr (mag pickup) sensor is that as the frequency goes up, the sensor impedance does also, so a loading impedance such as a tach input may be reducing the Vr output voltage as seen by the ECU. I would not use the sensor to drive a tach AND the ECU input at the same time without buffering it first.

george

[TheMonkey]

Originally Posted by george kujanski

In looking at the scope traces, the ripple on the red trace is in sync with the time that the MSD would be charging it's internal capacitor....

interesting. i'll see if that phase is repeatable. also, as LPete pointed out, i'll check out that phase with RPM changes.

Originally Posted by george kujanski

...Make sure the MSD power leads are connected directly to the battery and the ECU also connected to the battery. The alternator should also be running to keep the battery voltage from drooping due to MSD operation.
...

y, as per MSD, i have power & ground directly to battery. ECU gets power through a relay which gets it's power from the main power lug close to the battery. alternator is always hooked up and running as it would be under normal circumstances.

Originally Posted by george kujanski

....Question: what type of signal waveform does the ECU expect to see from the crank trigger? If you're not duplicating what the signal condiitoning circuit expects, that certainly would cause incorrect operation....

what's important, is what's coming OUT of the conditioning circuit. that is what the ECU coding works around. i have the conditioning circuit in the ECU configured to read the VR signal and provide a low to high signal as the VR signal goes through zero crossing from positive to negative, and then provide the high to low signal when a certain amount of hysteresis is read, nearly back to zero volts on the VR trace. that square wave is then read by the ECU to provide a 'tach signal' everytime the square wave goes from low to high. the ECU runs an algorithm to determine if it thinks it has seen a missing tooth where it should have and will create a miss if it thinks it loses sync. i'll fool around a bit with these settings to see if maybe the intermittent miss is a result of too tight of tolerances in the algorithm. problems are so much easier to solve when they aren't 'intermittent'.... ~sigh~

Originally Posted by george kujanski

...You also say that the miss occurs when you use the VR sensor to trigger the tach....more evidence that the tach is somehow loading the sensor output as Lpete suggests. Check the Vr signal with & without the tach to verify.

The thing with a Vr (mag pickup) sensor is that as the frequency goes up, the sensor impedance does also, so a loading impedance such as a tach input may be reducing the Vr output voltage as seen by the ECU. I would not use the sensor to drive a tach AND the ECU input at the same time without buffering it first.

george

i think i'm creating a bit of a misunderstanding with the term 'tach signal'. i'm not actually using a tach on the break-in stand (reading revs on laptop), but i'm using the term 'tach-signal' as the actual signal fed to ECU to determine engine speed and/or crank position to control spark.

so the point i was trying to make about the miss occurring only through VR tach signal.... i can get a tach signal that is already conditioned from the lug on the side of the 6AL box which is a square wave with frequency of 1 pulse per spark event. i configure the ECU to know that 8 pulses is 2 crank revs. my ECU works great when i use this signal, but i can only control fuel with this signal.

now... if i want to control spark (which i do), i put in a 36-1 wheel and VR sensor. i then feed this signal through a conditioning circuit the cleans up the AC sinusoidal wave with varying amplitude (higher revs = higher voltage), and turns it into a clean (conditioned) square wave with a missing tooth. i configure the ECU to know that 36 pulses (with one missing) is 1 crank rev, and i also am able to configure the ECU to know where TDC is relative to missing tooth. changing to this 'tach signal' from the VR sensor rather than tach output from 6AL that causes problems.

from some research, i'm finding that my problem *might* be that my conditioned circuit is sending too low of voltage on the high side of the conditioned square wave for what the ECU is expecting. (2.2v in the blue square trace above). so in a sense, you might be right with this statement: "If you're not duplicating what the signal condiitoning circuit expects, that certainly would cause incorrect operation", but it would be after the conditioning circuit rather than before.

[ponjohn]

Great thread....I would recommend the msd and ecm be be grounded at the same physical point.

Turn each accy on and off to see it's effect on the overall condition of the dc.

John

[Stuckinda60s]

You may need to devise a drive circuit for the signal to the ECU. Use the current crank signal and conditioning circuit to drive a switching circuit with a constant amplitude to be the trigger for the ECU. One method would be to use logic to switch an SCR on and off.

This would serve 2 purposes, isolating the circuits and providing a constant voltage to the ECU despite the higher current flow from higher frequencies.

Rich

[george kujanski]

Another thing: if tyou are getting a true sinusoid from the sensor, that's got a slow rise and fall time and depending on your buffer circuit, at higher speeds, your edge position from the buffer may be delayed in time compared to the desired firing point on the sinusoid. ( a phase delay)

Ideally, when using magnetic pickups as trigger signals you want to monitor the tooth tip so you get a sharp positve and negative pulse from the tip. As you monitor lower in the root of the tooth space, the signal will be more like a sinusoid.

The design of the HEI reluctor and pickup coil is a case in point...as each tooth approaches and leaves it produces a sharp pos and neg pulse and essentially zero volts until the next tooth. Very easy to buffer up with minimal delay.

To summarize, the signal you want shoould have good amplitude at all speeds and quick rise and fall times. That'l l be easier for a buffer circuit to discriminate.

george

[george kujanski]

I just noticed the scope shots in post 31. I didn't see that earlier at work...they block some attachments...nor did I see it when i posted my most recent post.

That said, looking at it now, I definitely would try to move the sensor to monitor the tooth tip, the long axis of the sensor on a radial line from the wheel. That'll give you some dead time between pulses similar to what you see from the sensor at the missing tooth.

Another thought is if the buffer circuit is slow compared to the signal frequency, at some speed the buffer won't be able to follow each pulse and the ECU will get confused.

So my thoughts are: reposition the VR sensor and make sure the buffer circuit has minimal phase delay throughout the operating frequency..I believe awhile back you said the sensor frequency went up to 3 kHz? (@5000 RPM)

Perhaps the wheel has some axial runout (along the crank centerline). The change in gap from the sensor to the wheel would cause the amplitude change.

Moving the sensor as I suggest will eliminate that amplitude change unless, of course, there is also radial runout.

george

[TheMonkey]

i think i'm going to try to tackle RFI first. i suspect that the close quarters on the break-in stand are not good for that. i got a braided strap to ground the fuel pump to the block, and also a capacitor to hopefully absorb some voltage fluctuations from the fuel pump.

George - with 36 teeth at 6k RPM, that is 216kHz.

Originally Posted by Stuckinda60s

...This would serve 2 purposes, isolating the circuits and providing a constant voltage to the ECU despite the higher current flow from higher frequencies.

Rich

Rich - hmm.... gave me an idea... maybe i could use a 4 pin GM HEI module next to the VR sensor to do the conditioning and convert the signal into the square wave. i wonder if that module could handle > 200kHz?