Joe Perez

iTrader: (8)

It doesn't exist.

But this recent talk of daughterboards got me to thinking... What if there was a Megasquirt PCA- not a daughterboard but the main board itself, designed under the assumption that not every engine in the whole world is either a chainsaw or a smallblock V8 with a distributor. One designed specifically for people with "modern" cars that have dual-wheel triggers, twin wasted-spark ignition drivers, and so on.

It would have all of the common mods, like a couple of relay drivers, a second (baro) MAP sensor, PWM idle, EBC, a map-switching input and so on all built-in so you just install the components into the holes without having to run jumper wires and solder resistors and diodes on the back. It could even have built-in circuits to drive things like A/C and PCM-controlled alternators.

The injector, IAC and EBC drivers would all have isolated grounds, just like a "real" ECU. And the DB-37 would be replaced with a proper automotive-grade connector- one with a simple snap-fit latch and enough pins to handle every possible I/O you could want. And for those people who still come up with something unique, a prototyping area the size of your hand, instead of one the size of hustler's dick. Maybe even some standoffs and a stacked prototyping daughterboard with a multipin interconnect direct-wired to all the general-purpose CPU pins and some spare pins on the main connector; like an expansion slot, so some future hacker can design, etch and sell plugin cards to automatically do lo-Z PWM in hardware, or four-channel sequential ignition and injection.

You think there'd be a market for that sort of thing?

cjernigan

iTrader: (13)

You're basically talking about the MS2 sequencer but hopefully not 10 bajillion dollars. If you build it they will come. I have a feeling that definitely goes for the miata community. Your efforts might be well accepted with the diy-efi guys. Especially considering the processor they plan to use has much more capability than the MSI chip. You have seen the thread and website about it right?
http://diyefi.org/

elesjuan

iTrader: (11)

I agree, that sounds like what the Diyefi guys are attempting to do. Can't wait to see what comes of it, since I don't have a small block v8 OR a chainsaw in any of my cars...

Might replace my 1.8 with a squirrel cage for my chicken/duck..

Reverant

iTrader: (10)

Joe, everything you describe would not fit the $189 bill that the MS1 does. A nice waterproof 19-way connector from Souriau costs like $70 alone, I would a expect a plain 50-way connector to be about the same cost. The second baro is another $20, etc, it all quickly adds up to more than $500, at which point, you are in fact closing in to the Spectre EMS, which is ready-built, with an AMP connector, etc.

Maybe all I would have asked for is more I/O, a larger prototyping area, AND FOR THE LOVE OF GOD, more spacing-larger pads for Q6, Q7, Q8, Q2, Q19, Q14, Q15.

Mostly everything else can be done in the firmware

Jim

elesjuan

iTrader: (11)

well FWIW I'm pretty satisfied with my MS1 setup even with having to add a huge external proto board into the top of my case. From here as I mentioned in the diyefi thread, all I'm really looking for is a massive, professional, and easy to customize LCD display that I could setup to display ALL of my vital stats for easy view.. Basically I'd add a VR input for a speed sensor then just totally replace my entire gauge cluster with an LCD display. Would be an interesting trick working an odometer into that mixture.. But you get where I'm going.

Reverent: I don't think it would need a weatherproof connector like that. The spirit I get from Joes post is something a little more automotive grade, like the 62 pin connector someone sourced for the boomslang harness, the same connector in the PnP MS. Most auto manufactures that place their PCMs inside the passenger cabin, which is everyone I know of with the exception of maybe General Motors, and likely Toyota.. use standard open element positive locking sockets.. Now, I don't work in the automotive industry, so thats based solely on what I've Experienced. Gm really likes to seal their PCM's and stick them under the hood, for reasons unknown to me. Guess it makes the harness going through the firewall a little less complex, god knows with what PCM's in GM's vehicles control these days... They need SIMPLE! Worked on a 1995 STS Cadillac and the friggin PCM harness had easily 300 pins. Everything in that car from the airbags to the dome lights, wipers, the windshield... all controlled by PCM.

Matt Cramer

What you're describing sounds a lot like the V3.80 board. It's the core board on the eMS-Pro. Except it doesn't have a proto area and is surface mount (and admittedly is pretty expensive at the moment). You might also want to check out the upcoming Megasquirt-III board. They're soliciting ideas about that one right now.

y8s

iTrader: (8)

state of the union of MegaSquirt is up for 2008! why didn't anyone tell me?

http://www.megasquirt.info/viewtopic...d65284eaceafc0

Joe Perez

iTrader: (8)

No, not at all. Sequential output might be a nice side-effect of the design, but that's not the primary focus. In fact, I'd expect things like sequential to be a small daughterboard and handled purely in a simple PLD or FPGA, not another microprocessor.

MS-II sequencer, MS-III, UltraMegaSquirt... To be honest, I haven't been keeping up to speed with what's what these days.

Thing is, I expect that those designs are still going to be pretty "universal." In other words, you'll still have to run jumper wire and tack resistors in different places depending on what your trigger inputs look like, and so on. They're not going to abandon thinks like the VR input because of their precious crankwheel fanatics, but I say who cares about crankwheel triggers?

What I'm talking about is different. This would be nothing more than a change to the printed circuit board aimed specifically at people using modern 4 cylinder engines that came from the factory with distributorless ignitions and two-channel cam/crank triggers with proper open-collector outputs. It would support the existing MS-I and MS-II CPUs with the existing "Extra" code. The only difference is that it would get rid of circuits that don't really matter to that crowd; things like coil(-) and VR trigger, stepper IAC, those LEDs that nobody actually uses as indicators anymore, and so on. Instead, the board layout would favor all of the "common" mods, like dual wheel input, two-channel spark output, a couple of relay drivers, baro MAP, PWM idle, and so on. No more jumper wires- the traces and holes would already be on the board waiting for you to plug components into.

Not that you would be forced to pay $25 for the second MAP sensor for example- but if you wanted to run it, the holes would already be there in the board. Just plug it into the set of holes that have been provided for it and solder it down, no more having to turn it upside down, tack-solder it to the pins on the existing one, glue it to the board, bend one leg up and tack a resistor onto it, then run a jumper wire clear across to the other side of the board. I would expect the cost of the bare board to be similar to that of the R3.0- maybe just a few dollars more.


Yes, exactly! It could even be something a bit more universal, like a couple of Molex "Mini-Fit" connectors- those are rated for 6-9 amps per pin on the high-density ones, and 50A per pin on the big ones, the pins have good strain-relief built in, crimping tools are widely available for them, they're super easy to remove and install, yet have a solid latch on them, and they're very inexpensive. The RoHS-compliant 24 position right-angle thru-hole Mini-Fit JR is $6.00 in qty/10, $4.00/100 (Digikey).


You mean the 3.57? Not even close. In fact, that goes in the exact opposite direction of what I'm talking about. It's essentially the same schematic as the 3.0, and with even less provision for on-board addition of the kind of circuits I'm talking about. There's a reason that DIY had to spin a whole daughterboard for the MSPNP, rather than just a simple harness- they needed someplace to put all the mods!

If you mean the Microsquirt, my apologies. That one is a little closer to what I'm talking about (it's got the right ignition I/O) and a decent connector, but it's too limited- not enough general-purpose I/O (relay drivers & table-switch inputs) and it doesn't even have space for one MAP sensor, let alone two. Sure, you could hi-jack the LED pins (and probably a few others) but then we're right back to the whole science-fair project of external jumpers and resistors and transistors and diodes and such. It ought to all be right there on the main board, ready for you to plug the components into. MicroSquirt is great for motorcycles I'm sure, but this is a car forum. We have enough space for an ECU case that is larger than a paperback version of "The Firm".

Too hackish. I respect what they're doing, but they're just going way too apeshit over processing power and ten thousand I/O pins. By the time they get that code stable we'll have a black Jewish lesbian in the white house.

I'm thinking totally opposite. Simply say that the processing power and software platform of the MS1 and MS2 is "sufficient", and recognizing that the Extra codebase for those two processors is quite mature and stable, build a board for them that makes them roughly equivalent to a Hydra or an EMS in terms of ease of installation.


Obviously this isn't an official B&G project. And I'm sure that Al & Bruce would probably **** kittens if somebody went and designed a "renegade" PCB for their CPU to sit on. There's no way you'd get away with claiming that the design was not influenced by the "prior art" of their 3.0 schematic. But "what if...?"

cjernigan

iTrader: (13)

This is what Matt was talking about: eMS-PRO

Joe Perez

iTrader: (8)

Oh yeah, I kinda forgot about that one. Definitely much closer. Seems from the description that it doesn't run standard code however.

I just keep thinking of a low-buck solution that it is still fairly DIY (ie: available in bare kit form) but that would require only basic soldering skills to pop the parts in, rather than the ability to read schematics, use a scope, fiddle with lots of external components, etc...

In other words, keep the same basic schematic, remain 100% compatible with the regular Extra codebase, don't get carried away with wanting to do bizarre things that consume loads of CPU power and will take months to debug- just keep the same basic concept, cut down (ideally to zero) on the number of hacks required for "average" functionality, and make it cleaner.

Ideally, the finished cost would be within close spitting distance of an equivalently-equipped Rev 3.0 Megasquirt. IOW- all the same components and fundamental circuit topologies that we're already familiar with, just sitting on a different circuit board with a different connector. It's probably a dream...

Any clue how much they will cost and when they will be available?

Matt Cramer

Yes, the eMS-Pro is based on a 3.80 board, which so far is exclusively used for the eMS-Pro. It runs standard MS1/Extra code, but as it uses a surface mount processor it can't use the MS2 based things. And it is only available in ready-made form.

Fitting all the extra circuits on a through-hole board and keeping it the size of the V3.0 would be quite a challenge - particularly since some import guys, instead of wanting the VR circuit gone, would want a second one instead. A second board with the extra circuits on it might work out better.

sotaku

iTrader: (4)

I've wondered about a more modular solution where there's the core and the various modules on separate boards communicating via a bus or something. Packaging and that communication would be the big issues but it'd be sweet to be able to piece together self contained mods to fit your application.

The_Pipefather

iTrader: (5)

Exactly such a thing exists in the form of Cornell University's ECU which they use on their Formula SAE car. See attachment.

They have stacked the following separate boards:

Processor board - this is the "core"
Conditioning board - for all sensor inputs
Interface board - LCD controller for in-dash display
High-current board - containing all the driver components for the actuators

I hope we can see something like this available to us in the near future.

sotaku

iTrader: (4)

Nice! I had thought of the stacking thing a bit as well just due to my experience with Arduino shields. Had no idea someone actually had done it. That's very cool.

Just for the sake of expanding the proto-area on the MS - it may not be too hard to put a proto "shield" of sorts on the back side. However I'm not sure how useful that'd be since I'm not familiar enough with what IO people would want to connect to (esp since adding io or tapping off of it is sorta hackish anyway).

Sorta like this:
http://todbot.com/blog/2006/07/11/ar...dboard-shield/

Joe Perez

iTrader: (8)

Is there a specific reason to try to remain the same size as the 3.0? I think that going to a larger (and altogether different) case would have some benefits. There could be extra space on the heatsink (or simply another one on the opposing side) to build a "complete" 3.0 BOM and still have dedicated spaces left over for the PWM IAC, EBC, WI, a VTEC driver, etc. And the heatsink should screw into the bottom half of the case- I don't see the advantage of the current slip-fit design. Put a bunch of vias through the heatsink "pad" on the PCB, bolt it to the case with either a thermal pad or some grease, and you've got an OEM-quality heatsink assembly.

Plus, you'd need space enough for some proper connectors. I don't think that adding an extra 2 or 3 inches in width (across the face of the endplates) would hurt anything, and moving the PCA down to the bottom of the case would create space for daughterboards on top without increasing the height. Make the case an inch or two longer, and move the RS-232 connection to the same side as the main I/O, and it'll fit in the same length space as the current case. Heck, forget the RS-232 altogether- just put an off-the-shelf RS-232 to USB chipset on the thing. They're cheap.
Well then, a pair of VR inputs it is. If the current 3.0 design is good (I've never used it) then we'll just double it. Put in a dual-comparator system with a tunable filter to handle a pair of hall/optical (open-collector) sensors, and you're good to go. Heck, we could even leave space for the coil(-) optoisolated input so that the one guy who wants to turbocharge his '78 Honda CVCC can use it.

Where I work, we have a saying that applies any time we are debating as to whether or not a certain component will be absolutely necessary on a board design- oftentimes it applies to parts that we'll want in place during proto and pilot testing, but probably won't be of any use for production or the end-user.

"Copper is free."

The gist is that it does not add any appreciable cost to the finished product to put space, pads and holes on the PCB for every component you might possibly want- the cost comes when you actually put the component on the board. If you leave the component off, it costs you nothing.

CANBus = complicated. Just put a couple of stacking connectors on the board, and bring, for example, all of the injector lines and the trigger inputs to one, all of the CPU "general purpose" pins to another, and some back-panel pins to each. Maybe one that just gets power and some back-panel pins to do things that will vary from car to car like A/C and alternator.

Again, one of the primary goals would be to retain 100% compatibility with the MS1 or MS2 CPU (we'd probably have to choose one or the other) running the current-design Extra code.

In other words, Think S100, not PCIe.

Now that's some serious modularity!

I'm thinking not quite that extreme- put all of the "baseline" stuff on one board. Basically, you should be able to take the standard board, assemble it with components populated into the sections that are relevant to your car, and run a turbocharged four cylinder engine using all the factory sensors, ignition, fuel injectors (two injector channels only, but four sparks), IAC, VTEC, and so on, plus all the common "extras" that are hacks now, like some relay drivers, EBC, WI, table-switching input, "Nawz" (God I hate that word no matter how you spell it) and so on without adding any daughterboards, any jumper wires, etc.

The idea being to keep the cost of a "common" build down to a level similar to that of a R3.0 build. Forcing the user to install daughterboards to do basic functions drives up the price and defeats the whole point of the thing- Low cost and simplicity.

The expansion boards are purely for people who want to do hacks. Maybe build a small CPLD / FPGA circuit to split the two injector channels out to four. Maybe one that does PWM lo-z injector control in hardware, maybe one is just a breadboard like the one sokatu found, with another one that's a hole-for-hole identical solder-type protoboard (to transfer your breadboarded circuit into permanent form), and so on. Those are all for others to design and build as they see fit. But all of the daughterboards must be co-planar; the whole thing should be no more than two boards tall to minimize case height:

The_Pipefather

iTrader: (5)

Isn't that more complicated than working with a microcontroller?

Joe Perez

iTrader: (8)

Depends on what you're used to I guess. At Pacific Research, we design audio consoles & routers, and we do all the stuff you'd normally associate with DSPs / uPs, like audio mixing and OC3 switching, with Xilinx Spartan series FPGAs. I like being able to set up ten different pipelines through one and have ten different operations happen simultaneously- fewer timing constraints and such.

But you're right in that microprocessors are probably more common still among the DIY crowd. But frankly, I don't care what people use. They can build a circuit with discrete germanium transistors and mercury delay-lines so long as it fits in the case. I don't want to design the daughterboards or tell anybody else how to do it either. Ideally, the majority of users would be running without any.

The_Pipefather

iTrader: (5)

There's discrete chips now available from Freescale that can actually do the wheel decoding and talk to the main microcontroller via SPI.

Matt Cramer

It's easy to double the size of the V3.0 if you tried adding everything to it, which is why I suggested a daughterboard in the lid. Another thing is that you'd definitely need approval from B&G to sell something that copies the base circuits. I'm not sure what they would say but I am pretty sure they'll expect you to support it throughly at the very least. A daughterboard would require no such authorization.

You might want to see the MS-III discussion for some comments on case size and what might be done about it.

elesjuan

iTrader: (11)

I'm just speaking from my personal experience with this, but adding onto an existing board with daughter boards has too much possibility for problems. While I haven't had any issues with MINE in the respects of that, some of the solder points which I had to utilize to get I/O from the MS main board to the daughter board just leave the door open for some problems. Granted, I have a 2.2 board without the proto area, and that would've helped things out a little.. Still my personal opinion anytime you 'piggyback' onto existing circuits, thats a very good failure point.

I know nothing about engineering electric circuit boards, but agree with Joe it would be nice if the existing circuit design (not layout) for the 3.0/3.57 could be utilized with a larger footprint and TONS of extra I/O's and drivers built in..

What I've pictured in my mind is just like the previous posted controller that uses several "layers" of PC boards. Have the MAIN board for the power supply, then have a HARD connection (NOT WIRES!) to a secondary board for I/O, then a third for processor, etc.. That way you could have, even leaving the physical footprint of the 3.0 board, you'd have that FULL board to sling all of your I/O drivers on, a full board for all processor mods, maybe some redundancy in the power supply.. Maybe a little extra board stuffed in there for some LCD digital dash action.. Thatd be a HUUUUUUGE thing for me, if I could get some Digital dash... Obsessed, I am.


That'd be cool, but not holding my breath!