Welcome Guest [Log In] [Register]
Posted ImageWelcome to the all new Geo Metro Forum. We hope you enjoy your visit.

You're currently viewing our forum as a guest. This means you are limited to certain areas of the board and there are features you can't use and images you can't see. Registration is simple, fast, and completely free. If you're already a member please log in to your account to access all of our features:

Join our community!




Username:   Password:
Add Reply
monster turbo3 build
Topic Started: Jun 17 2011, 05:53 AM (36,537 Views)
t3ragtop
Member Avatar
Turbo3 and Twincam Tweaker

a small update on the megasquirt2 work that continues on the bench:

i've been working on bringing the timing system technology up to a higher accuracy for the megasquirt2extra for sequential operation. i've tested a bunch of methods, mounting techniques, and sensor types along with a number of electrical circuits, some i built and some i bought for testing. i settled on this arrangement for my new project engine and have another set to retro-fit to my current engine.

it probably doesn''t look like much in it's finished state but it represents about $500 worth of stuff selected and purchased whenever i found the stuff cheap.

here are the hall effect sensors and 2 channels of signal conditioning.
Posted Image

the crank signal is generated by the paquet green line gear tooth sensor. it's a hall effect type device that has a magnet mounted behind the hall effect component. when a gear tooth made from a ferrous metal rolls past the head of the pickup, it triggers the hall effect. the pickup sends it's signal to a conditioner board that has a comparator chip on it and a potentiometer for dialing in the signal sensitivity from the pickup. that comparator chip develops a 4.4 volt square wave that gets fed directly to the microprocessor on the megasquirt's main board. it runs 35 pulses per crankshaft revolution with a space for a missing tooth that resets the count.

the hall effect device used for the cam shaft is unipolar, positional type that senses the field from the south pole of a neodymium magnet and switches it's voltage state in the presence of that field. mounted to the cam gear, the "flying" magnet trips the sensor once every revolution (every 2 revolutions of the crank) to tell the megasquirt which phase the valve train is in. again i used the same hall effect conditioning board to generate a second 4.4 volt signal pulse that feeds the 2nd trigger input of the ms2's microprocessor. here's the complete timing set in it's final configuration.
Posted Image

i tested the stratified engine components hall effect conditioning board that uses an optoisolator chip to generate the pulses for the microprocessor on the ms2. it works fine but was pretty expensive to be using 2 units. i also tested 3 conditioning boards that i sourced from china that were built for use on electric vehicles in much different applications. in quantity, those were much more economical and available in smaller lots. 2 of those boards used comparators, 1 just used the analog signal from a board mounted specialty hall effect. one of the 2 boards that used the comparator had the sensitivity potentiometer. that's the one i chose. the other comparator type board has both analog and digital output pins and works great, too. the board is a little bit bigger, has a hole for mounting, and 4 pins instead of 3.

here's a blown up pic of the conditioner board i selected.
Posted Image

you can see that it's 2 boards that haven't been cut apart. this little guy will get installed in an enclosure that's installed right onto the timing wiring harness. the module will have 4 wires running from it, a 5 volt supply from the megasquirt board, a ground wire that runs to the common grounding point under the hood, and a signal wire for each of the crank and cam signals.

that's as simple as i can make this high tech timing arrangement. it has super high rejection mode so electrical noise is minimized. it connects directly to the microprocessor so it bypasses all of the on board signal conditioning on the megasquirt which mean i can leave about half of the component parts off the ms2v3 construction. it's also 100% stable way past 20,000 rpm. with the exception of the sensor sensitivity adjustment which will make it easy to dial in the pickups, there are no adjustments that have to performed to make the electronics work. that makes it "plug and chug" simple, something that gave me fits when i was trying to get my turbo3 running on the megasquirt.

the ms2 using the variable reluctance pickup requires that you fiddle and fart with the sensitivity and hysteresis pots on the board inside the controller to get the timing signals to "sync." that task is different from one megasquirt installation to the next and it's a tedious step which took me hours to perform the first time. also, there's the possibility to get the 2 wires on the v/r sensor backwards which causes more hassles. another problem is that a variable reluctor can generate a low (.5 volt) signal at low speeds and generate a huge signal (70 volts) at 7200 rpm. that big signal has to be conditioned on the ms2 board so that it doesn't exceed the processor input voltage of 5 volts. when you accomplish that, it also puts the low end signal in the "mud" - below the useful level, which in turn gives you false triggers on the timing and those make you lose sync. if you were unlucky, the lost sync and false triggers could fire a coil at an inopportune time and damage the engine.

i'm very sure that my technique is very reliable, 100% stable, and mostly idiot proof in a "set it and forget it" kind of way.

megasquirt is fun! :D
Offline Profile Quote Post Goto Top
 
Tofuball
Member Avatar
Strange Mechanic

Makes sense to me, but I dare ask, what kind of accuracy issues were you running into with your old setup that inspired this?

Offline Profile Quote Post Goto Top
 
t3ragtop
Member Avatar
Turbo3 and Twincam Tweaker

the first thing was the hassle factor with dialing in the sensitivity and hysteresis pots on the megasquirt board. you must remove the top cover, pull the injection and ignition power, apply power to the controller, bring up the logger function, and roll the starter over while adjusting the signal level pot on the controller to establish a trigger signal. the v/r sensor puts out a sawtooth wave (VVVVVVV) and the processor wants to see a square wave. there's a second adjustable pot for hysteresis. between the 2 pots you have to dial in a level that is acceptable to the processor and then use the hysteresis pot to fashion a sort of square wave from the sawtooth waveshape. even if you have it as close as you can get, the rising edge of the pseudo squarewave is higher than the falling edge. also, the conditioned waveshape from the v/r sensor can induce spurious spikes and when the processor sees those it drops signal sync while it fires a coil out of actual sequence. then the count recovers and the ignition fires until the next spike throughs it out of time again.

while i was trying to dial that crap in, i was getting a spurious back fire out of the exhaust and that led me on a wild goose chase on afr and fuel issues.

after i started getting settled into the timing functions with the hardware and firmware i started seeing dropouts at higher rpms. those were caused by the signal amplitude getting higher and higher based on engine speed.

it literally took me days to get things dialed in to the point that i had only occasional sync loss and a sort of stable ignition. after i cogitated on all of that i came to the conclusion that i couldn't take a chance on my spark timing just taking a fit and spitting an errant spark event under load at 20 psi. the pistons and rods are far too precious to run with my pants down. my friend, turbohull, kicked a rod through his turbo3 block which ended his megasquirted g10t project and i think it might have been related to these timing issues.

that's why i started experimenting with the hall effect devices (after i had doped out my first h.e.d. cam angle sensor system.) now, i see that diyautomotive has jumped fully into h.e.d. crank position sensors which makes total sense to me after having had a taste of the level of difficulty of tuning in the variable reluctance type crank sensor.

the v/r stuff might be just dandy for a normally aspirated megasquirt control set with an occasional fart here and there on the ignition but that type of inprecision doesn't work for me as a matter of personal expectations.

in looking at what the lead gurus of the megasquirt world were doing, expanding on the optoisolator type control triggers, is fine for picking a signal off a coil as an input to the megasquirt if you need to isolate a high voltage signal from the microprocessor but it does very little to actually offer any real signal conditioning. the comparator chip i'm utilizing actually converts the analog output from the hall effect device to a ttl compatible, pseudo-digital input for the microprocessor.

my development requires only that you have an air gap adjustment on the sensor that allows it to generate it's output. after that, you can trim the sensor output signal with the pot on the conditioning board which in effect tweaks the h.e.d. sensitivity (and that's not really required as the comparator output is already set by biasing resistors and vcc to the proper ttl output amplitude.)

on the megasquirt main board they dedicate quite a bit of circuitry to allow some flexibility in the choice of engine speed input. the down side to that is that they try to make 2 circuit designs fit every condition so it half assed does several things instead of doing the exact thing i wanted really well. when i assemble my new controller i'll just leave all their signal conditioning components off the board and directly connect my h.e.d. conditioned signal directly to the microprocessor input. the sequential operation requires that you de-populate 3 other circuits, a dozen or so other components including the crystal, off the board as an assembly. i don't need the original coil driver fet on the heat sink or the 2 injector driver fets on the heat sink. i also don't use the tx/rx mux chip for serial comms. that means that i'll only use about half the components from a version 3 kit when i assemble it. ;)

when i had the actual trigger wheel, flying magnet, and hall effect devices set up on a test rig with a motor spinning the wheels, i used my o'scope to verify the signal outputs and they are rock solid, never vary from an output amplitude of 4.4 volts, and the only thing you see is the frequency of the square wave tightening up as rpm increases - all the way to my observed motor speed of 19,200 rpm.

if you want to put shit in your neck, try that some time! i spun the aluminum pulley with the trigger wheel mounted to the shaft of a 1/2 hp single phase electric motor (my polishing wheel rig mounted to a bench in the garage) at it's full speed. if that pulley would have come unstuck from the shaft i would have had a bad assed ninja star ricocheting off the concrete block walls out there. :rocker

anyway, that's why i pressed on with my development. it's one thing to just follow the megasquirt guys' work and another to step out on your own. who knows, those other guys might figure out the same kind of thing on their own. the very fact that diy now sells a hall effect type crank trigger tells me that they're on that path.
Offline Profile Quote Post Goto Top
 
Tofuball
Member Avatar
Strange Mechanic

Sounds logical to me.

I had very good luck with VR sensors, but I'll admit the signal from them and the setup is quite "dirty."

I had clean\reliable spark at 9KRPM on my S5 FC:)

The first iteration (Before I went with the stock setup) I had actually used a GM HEI module and a hacked up crank angle sensor. At 9K something went wrong with the HEI and it turned the whole system into some kind of EMP device. It made no sense but it was affecting everything in the car and the cars around me . . . O_o
Offline Profile Quote Post Goto Top
 
Johnny Mullet
Member Avatar
Fear the Mullet

Moved to project forum.
Offline Profile Quote Post Goto Top
 
t3ragtop
Member Avatar
Turbo3 and Twincam Tweaker

old man winter has moved in and the project car has been tucked into the garage. my new, old 91 vert is in the driveway under tarps awaiting spring for it's makeover. it needs the same floor and rocker work that i did on ol' blue this summer past.

i have quite a few very serviceable spares in the garage so i'll put them to use to build something new.

since i have a hard time not playing with a project, i've been working at building another g10t that takes my development work further. this one will have the hall effect timing triggers from the ground up, balanced crank build, 460cc injectors, turbine tech upgrades for the exhaust manifold with a tdo4 turbo, and most probably a 3 tech performance full house top end (when the tax fairy drops some cash on me.)

i've found a good source for a hall effect trigger wheel sensor that isn't a one off part and the new engine will get this one as a crank sensor.
Posted Image

this time around, the megasquirt wiring will be better integrated. i have a much better idea now of what is involved so running the wiring will go more smoothly. this build will also use a much more simple air to air intercooler, fewer discreet gauges (i can use the megasquirt to display virtual gauges,) and my target hp will be lowered to 130.

here's the raw intake after grinding out the casting peaks from the port runners, port runner matching to the head, polishing up the runners a bit, cleaning up the casting parting lines on the exterior, and bead blasting.
Posted Image

following that work, i painted the intake manifold with a ceramic hi-temp (1000*f) satin black.
Posted Image

the throttle body on this one is probably the best one i've ever had. it was oily and had a bunch of crud built up on it but it cleaned up really nicely. i won't even paint it. i did some pre-assembly, a mock up of the top end to make sure i didn't have any issues with fitment.
Posted Image

Posted Image

the blue parts are leftovers from other projects but those will get re-sprayed in gloss black. basically, the whole engine will be black with small touched of red. i'm using a fuel rail that i had re-conditioned a few years ago and i'll use it as you see it in the pics. i'm thinking that i'll leave the valve cover in it's raw bead blasted form.

one other item i've acquired for my blue turbo3 vert, billet aluminum window cranks. since the car is wedged into the garage for the winter, installing those will wait until the salt is off the roads and i get the car back out on april 2nd.
Posted Image
Offline Profile Quote Post Goto Top
 
idmetro
Member Avatar


:drool (Insert long slow wolf whistle sound of appreciation here) Loooooking Veeeeeeery Nice! :thumb
Offline Profile Quote Post Goto Top
 
bogs
Member Avatar
Duct tape heals all wounds

idmetro
Jan 12 2012, 11:28 AM
:drool (Insert long slow wolf whistle sound of appreciation here) Loooooking Veeeeeeery Nice! :thumb
Couldn't have said it better myself :drool

Offline Profile Quote Post Goto Top
 
t3ragtop
Member Avatar
Turbo3 and Twincam Tweaker

i'm still fiddling with things.

better?
Posted Image

Posted Image

the business.
Posted Image

i've got to play with the new design of the triple ignitor/ triple coil ignition. i'm thinking that i'll make a stainless steel bracket that will serve as the distributor blanking plate, the coil and the ignitor mounting bracket. instead of coil on plug i'll adapt the ford cop units to accept spark cables to make it easier to change the plugs. it will also make the assembly modular but still use individual, discrete components for easier servicing.

another thing i thought of is to mount my hall effect conditioning circuit boards in a little steel box (for shielding) that i'll mount on the intake manifold support bracket. that will make the grounds to the block and reduce the number of wires for the megasquirt timing triggers to just 3.
Offline Profile Quote Post Goto Top
 
Tofuball
Member Avatar
Strange Mechanic

That is what I did, I made a separate bracket and used spark plug wires, rather then having the COP right on top.
Offline Profile Quote Post Goto Top
 
t3ragtop
Member Avatar
Turbo3 and Twincam Tweaker

the main problem i had with the cop arrangement was that when i needed to confirm the engine's base timing there was no room to get the clamp from my timing light around anything on the #1 plug. i ended up using a spare coil with a plug wire on it to check the timing.

secondarily, to change plugs i have to remove the cop units and the supporting bracket to get a socket on them. not that i think i'll be changing plugs all that often.

this engine won't use the water to air intercooler so it will free up the space i needed for that and i'll be able to move the ignition system over to the right side of the engine easily enough.
Offline Profile Quote Post Goto Top
 
t3ragtop
Member Avatar
Turbo3 and Twincam Tweaker

i got some time to play with my hall effect parts for supplying the engine timing information to the megasquirt controls. i set up a little bench test to check the crank and cam sensors and the signal conditioning boards i want to use as my updated and improved trigger signal arrangement.

i used an instrumentation type switching power supply to apply a 5 volt dc source to the conditioning boards. i attached my dvm to monitor the voltage. i removed the 3144 hall effect devices that were attached to the conditioning boards and replaced them with the leads from the industrial package barrel type hall effect sensors that i'll be mounting on the engine. the crank will get the paquet "green line" f58s gear tooth sensor and the cam will get a honeywell unipolar hall effect sensor.

the paquet sensor has a magnet built into it's sensor body so that the magnetic field runs through the rear of the hall effect device embedded into the sensor's body. as a ferrous metal passes the face of the sensor it perturbs the magnetic field and switches the hall effect device. the teeth on my 36-1 crank trigger will pulse the sensor on and off. with it wired to the conditioning board, the sensor's trigger pulse will drive a small comparator which in turn outputs a 0 to 5 vdc signal which will directly feed an input on the megasquirt's processor providing it with 35 pulses per revolution of the crankshaft.

the honeywell unipolar sensor will use a flying magnet embedded into the camshaft gear to operate the hall effect device. the unipolar hall effect sensor will only switch states in the presence of a field from the south pole of a magnet. as the magnet in the cam gear passed the face of the sensor, it will change it's state, again driving the comparator on the conditioning board, outputting a 5 vdc signal to another input on the megasquirt's processor which is programmed to accept the camshaft timing pulse.



this will allow me to use just 3 wires from the megasquirt controller, one as a shared 5 vdc supply for the circuits and one each for the conditioned crank and cam trigger. the circuit grounds will run to the common system ground on the engine's block. i plan on enclosing the conditioning boards in a very small box that i will mount to the intake manifold brace which will make it possible to keep the hall effect wiring as short as i can, provide suitable signal shielding, and give me good access to terminate the megasquirt wiring harness with a weathershield connector.

also by doing things like this i can bypass a bunch of input signal conditioning components on the megasquirt's main board and take the trigger signals directly to the main processor pins.

Offline Profile Quote Post Goto Top
 
Mikemetro
Member Avatar


The ms3 board has the option for wiring hall effect rpm and cam inputs using their VR + pull up option in the wiring on the board. I see how they use one circuit to do the job of two depending on the set up you plan on using (primarily saw tooth or square wave). I was planning on using Hall effect on both cam and crank using their on board conditioning circuit. But you really have me thinking. Does your current test rig do a better job of input signal conditioning than the on board MS3 setup?
Offline Profile Quote Post Goto Top
 
t3ragtop
Member Avatar
Turbo3 and Twincam Tweaker

yep, the comparator type circuits on the conditioning boards i use are way better than the main board stuff on the megasquirt. using the little external conditioning boards lets you delete all of the signal conditioning components from the main board. as a matter of fact, my megasquirt controller set up for 3 channels of injection, 3 channels of ignition, and primary and secondary timing triggers lets you build the board without a double handful of parts including the signal adjusting potentiometers, the timing crystal, the optoisolator i/c, and most of the big driver fets on the internal heat sink.

doing that removes most of the heat generated inside the controller's case. also, the fully conditioned hall effect timing trigger outputs connect directly to the primary and secondary trigger pins on the main processor, are semi-digital, and are fully compatible, pre-conditioned signals. they are also super clean up to 20,000 rpm (the limit of my ability to spin the trigger wheel on my polishing wheel's grinder motor - i'm sure that the signals would remain solid at even higher speeds.)

the problems involved with using the variable reluctance sensor on the crank trigger are, well, problems. the v/r sensor generates a sawtooth wave shape while the processor wants a square wave shape. the megasquirt's onboard conditioning method uses the 2 potentiometers, level and hysteresis, to approximate a square wave. i ended up twiddling those pots for a day, trying to establish a viable signal. also, the v/r sensor acts sort of like a generator and the faster the crank trigger wheel spins, the higher the signal amplitude gets. it can give you a signal that's in the mud at low rpms and a signal with an amplitude of around 70 vdc at 6500 rpm. and. it's not a square wave, it's a misshapened approximation of a squarewave, higher on the leading edge and lower on the trailing edge. the hall effect/ conditioner board outputs don't require any "dialing in," they just work without any adjustments having to be made.

the comparator on the hall effect conditioning board puts out a nice clean square wave at any rpm that never exceeds the 5 vdc supply to the board which makes it inherently compatible with the processor's ttl inputs.

while i'm yappin' about things, i'll tell you that the ms3 control set is way overkill and costly for running a g10 engine. it has 8 channels of injection and 8 channels of ignition, 5 channels of each that you'll never use. the enclosure is bigger and it's already tough to find room under the dash for the ms2's case. the ms2 version 3 is the sweet spot for sequential control of the 3 banger. it's considerably cheaper and the latest revision of code for the ms2 is well suited to the task as well as being very stable. the ms2 v3 kit runs $266 vs. the ms3 kit at a hundred bucks more and if i'm not mistaken, you still have to provide it with a conditioned secondary timing trigger so it won't save you the effort there. also, the stim board that you need for testing as you build the kit is considerably more expensive.

i wouldn't hesitate to suggest the ms3 if you were building a v6 or v8 engine but based on cost vs. performance i don't think that there's any benefit in using it on a g10. as a matter of fact, i think it might be more difficult to deal with in the long run - more hassle in a bigger box. :D if it was what i had to work with, i'd make a go of it but i wouldn't put up the cash for a megasquirt3 just to instrument a 3 banger.
Offline Profile Quote Post Goto Top
 
Mikemetro
Member Avatar


Well I have to agree it is overkill. I actually purchased the ms3 with the ms3x expansion board. At the time my thought process was a bit different. But I assembled it and tested it out with the stim and stim x test boards. only issue i had was c22 was installed reversed. It caused F1 to open up. Replaced c22 (correct polarity this time) and she runs great. I haven't built complex circuits since my electronics courses some 10 years ago. Since then I've been just troubleshooting down to the board level. Only circuits I've built recently are simple logic circuits. Not as easy as it used to be. Used to work on HVPS for Klystron and TWT Amps. A few years back I got hit by an indirect lightening strike. Since then my fine motor skills have not been up to par. When I was building the MS I had to keep a steady supply of one beer every 30 minutes so my hand would be steady enough to solder the smaller components. I think I might have exceeded that rule around the time i soldered in c22.
Offline Profile Quote Post Goto Top
 
2 users reading this topic (2 Guests and 0 Anonymous)
Go to Next Page
« Previous Topic · Project Forum · Next Topic »
Add Reply