Hairbrained idea for cheap timing meter/onboard display

[needlenose]

Maybe I'm over simplifying it, but for calculating RPM, it doesn't matter how many strokes occur. Since the hall interrupt happens once per rev, all that matters is how many interrupts the device sees in 1s. Value * 60 gives you RPM. If the device receives an injector interrupt *between* hall interrupts, it calculates the time between injector interrupt and the following hall interrupt. (reverse if the probe is on the other side)

If the hall is pulsing at 33.3Hz, then

33.03 * 60s = @2KRPM

30.03ms / 360deg = 0.0834167500834ms. So every 0.0834167500834ms the crank moves 1deg.

So if the 2 preceding hall interrupts were 33.03ms apart and in the next cycle a piezo interrupts occurs 0.750750750751ms before the hall interrupt, then 0.750750750751 / 0.0834167500834 tells you the injector is firing 9deg advance of TDC.

It can also be done using the preceding hall interrupt with the previous interrupt-to-interrupt time in which case you are working from the other direction.

Of course everything should be fed through a hysteresis to smooth the results or your numbers will bounce around so fast you can't read them.

Obviously you would have to calculate in the offset value if the hall is positioned before or after true TDC. This is how I see the software working, first thing in the morning while I wait for grandma to come watch the kids who are spewing diarrhea and vomit so I can leave for work.....

 

[BDCarrillo]

Thanks to riot warrior, we know that the offset is 20 degrees for the mag pickup, which can easily be incorporated. That means we'll see the injector fire ~29 degrees before a mag sensor picks up the zero line.

My initial calculation that included stroke cycles was just to determine what sort of frequency I'd need to see on an oscilloscope to validate sensor output at an approximate idle speed.

A simple way to smooth the data would be to display an average every quarter second or so, otherwise (like you said) an LCD or oled would go bonkers.

 

[ifrythings]

If I were doing this I would use the dampener as a sync pulse for the ip gear, that would be like new vechiles that have a cam and crank sensor, you'd have better resolution this way as you would have (ip gear =108 teeth per 2 revs/2 = 54 teeth per rev) 54 pulses per rev or a pulse every 6.66.... crank degrees.

Also I played a bit with my ferret timing adapter and I could get it to read the number 1 injector timing adapter a bit, this is with zero tweaking so I'm sure if you played with the amplifying circuit in that you could make it read the timing adapters.

Jm2c

 

[laserjock]

I would think it would be pretty easy to filter your piezo signal. You can calculate the range you care about and squeeze the window down.

The mag probe is reading the groove in the balancer. The groove will induce a signal in the probe coil.

Doing it timing meter style would make it easier. Being at a steady rpm you could certainly average to get a more stable reading. It may be that you need to average if the signal from the piezo or mag probe is noisy.

Subscribed.

 

[hce]

Not sure if this helps at all, but I use these piezo switches on my homegrown wheel balancer. May be able to copy its circuitry if strictly an on off input is more beneficial.

 

[david85]

Maybe I'm over simplifying it, but for calculating RPM, it doesn't matter how many strokes occur. Since the hall interrupt happens once per rev, all that matters is how many interrupts the device sees in 1s. Value * 60 gives you RPM. If the device receives an injector interrupt *between* hall interrupts, it calculates the time between injector interrupt and the following hall interrupt. (reverse if the probe is on the other side)

If the hall is pulsing at 33.3Hz, then

33.03 * 60s = @2KRPM

30.03ms / 360deg = 0.0834167500834ms. So every 0.0834167500834ms the crank moves 1deg.

So if the 2 preceding hall interrupts were 33.03ms apart and in the next cycle a piezo interrupts occurs 0.750750750751ms before the hall interrupt, then 0.750750750751 / 0.0834167500834 tells you the injector is firing 9deg advance of TDC.

It can also be done using the preceding hall interrupt with the previous interrupt-to-interrupt time in which case you are working from the other direction.

Of course everything should be fed through a hysteresis to smooth the results or your numbers will bounce around so fast you can't read them.

Obviously you would have to calculate in the offset value if the hall is positioned before or after true TDC. This is how I see the software working, first thing in the morning while I wait for grandma to come watch the kids who are spewing diarrhea and vomit so I can leave for work.....

I think you nailed it there. First we need to measure the phase shift between two pulses; TDC and the #1 fuel injector (offset as required for initial calibration).

But since we're measuring phase shift with time, not angular velocity directly, the phase shift needs to be scaled proportional to the RPM of the engine. Otherwise the faster it spins, the less timing advance you would see. Either pulse input can be used as the time-scale reference since frequencies will be the same. Maybe the mag pickup would produce a cleaner signal though?

For cleaning up the pulse signals, my suggestion would be to look at the datasheets for whatever sensor you're planning on using. Usually they have good suggestions on circuit design for controlling noise. Hysteresis may not be required it you can clean up the signal at the source.

 

[BDCarrillo]

Minor setback on testing the factory adapter... it's MIA on my 91. PO changed injectors and left the old ones in the truck, but didn't install the timing adapter. I'll pull one off my parts truck and get that sorted out.

Had to run out of town on a honey-do mini vacation, so I haven't been able to try the strip piezo yet.

Those "beam" style piezo sensors would pick up noise from the whole engine vibrating.

 

[BDCarrillo]

You must be registered for see images attach

That's an injector pulse from the factory timing adapter. I had one that I butchered making a pop tester prototype... will never seal on a factory line again. Multiple pulses seemed to average out around the 25 mV range, but that's just on the bench with an old injector that pops at 1400 psi.

To make it more practical, I think I'll solder a tab on a replacement one for the truck. I have a piezo sensor board ($6?) that has an adjustable gain to set a trigger threshold. That should smooth out any engine noise and prevent the tail end of the wave from triggering anything.

So, next step: test in-truck, engine running with factory adapter, and bench test strip piezo sensor.

And yes, that's a cheapo DIY $25 oscilloscope.

 

[laserjock]

Interesting. So is that 2 ms per division? Love the cheap scope. Gets the job done. I think there is something there to work with. The question is will that peak width be sharp enough and do you edge trigger on the up side or down or just set a trigger threshold voltage. I think the last one is the easiest but probably least accurate. What you thinking? Is there a way to just measure the two frequencies and come up with a number? I guess not because what you are really looking for would be the phase shift between the two. So if you plotted both pulse trains on your scope, the distance between the peaks would be the timing and if you could average that over several events (say each second) you would probably cancel out a lot of the shot noise.

Sorry just rambling out loud.

 

[BDCarrillo]

It is 2ms. Note that I had it on AC with the ground on the injector line, which is electrically isolated from the copper/brass contact ring on the timing adapter. I think DC would be better, but I was getting a lot of noise from my hand. A hose clamp only made it worse.

A certain peak mV ought to be adequate. To simplify the software side I'll be using a digital output board that takes in analog piezo pulses. That ought give a nice clean signal somewhere on the UP slope.

The o-scope picture isn't terribly good as a representative ... I suspect that it's mostly noise after the injection event. I need to get a better photo with multiple injection events (a full stroke on my tester gives over a dozen shots).

In theory you could average the "above threshold duration" and add it to the up slope trigger, but I'd suspect that the original equipment wasn't that sophisticated.

Better data will be available soon.

 

[mblaney]

I thought I would post a couple of comments / thoughts on this... I have always wanted a better, and live timing measurement system.

Pulse timing - this works but I think it is not a very reliable method.

1. The injection pressure is not really that significant (see the graph posted); from my pop testing I can see the injector line sits at ~ 1400 psi between injections, and (pops at 1800) probably spikes at 1900 during the injection event. The pulse adapter is looking for a 500 psi spike which is not very significant - especially when you are sensing this by expansion of the injector line!

2. Clamp-on pulse timing works but is unreliable for continuous use; I find it hard to maintain a clean signal just in the driveway. The inductive pickups (for your timing light) are also quite delicate and probably won't last for long term use.

3. Piezo tapes (as previously suggested) I think are very good at measuring vibration. The may also pick up line pressure pulses but will be so damn noisy that the data will not be reliable.

4. I suggest there are two options that would be better.
a) The first is easier, install a modified fitting at #1 that replaces the Ford pulse adapter (or modify the ford adapter) to install on in-line piezo pressure sensor. This would directly read hydraulic pressure rather than line flex. This should also be (when compared to a clamp-on) immune to vibrations.

b) The second is much more difficult to - install a piezo to monitor cylinder pressure; this would be similar to the luminosity probe. Sensor spikes would be very obvious!

My shop has calibrated this type of sensors - specifically for measuring diesel combustion for a university project http://www.pcb.com/pressreleases/pressure/175a01_pr I don't know what the cost is for a sensor like this.

I also see that BERU has glow plugs with integrated pressure sensors (for timing). Maybe it is possible to replace an exiting (or all!) glow plugs with one of these http://emissions.borgwarner.com/products/diesel-cold-start-technology/pressure-sensor-glow-plug-psg

http://beru.federalmogul.com/sites/default/files/psg_broschuere_gb_2014.pdf

The title of this thread is "hairbrained" so i think this qualifies!

 

[BDCarrillo]

Right now I'm in favor of the Ford sensor. Keep in mind that the injector opens at around 1800 psi, but the injection pump delivers quite a bit more than that, in the 5-6000 psi range. I don't think any pop tester gets that high of pressure in millisecond intervals. That higher output should produce a significantly stronger output at the stock adapter. Assuming a linear translation from 25mV at a 400 psi differential to a 4000 psi differential, we might see about 2.5v per pulse.

The piezoelectric or other more exotic options are on the table, but if the price hits a couple hundred there's not much point to it.

 

[BDCarrillo]

Quick update: testing in-truck with a factory sensor is on hold until a replacement arrives from @riotwarrior (thanks!)

Got an ADA fruit trinket, OLED screen and piezo sensor on hand. Piezo testing is on hold... I'd rather use the stock fitting.

If injector pulse monitoring is successful I'll press forward with locating an inductive magnetic pickup for the timing tab. I have a couple spares off parts engines to take measurements on.

 

[riotwarrior]

Picking up CSEK and yoke tool along with grease gun tomorrow...so in US...will ship then.

Al

 

[riotwarrior]

no updates what is taking so @#$%£€☆ long?