MS MegaSquirt V3.0 board "issues" revealed

Discuss MegaSquirt, VEMS and other non-free hardware and software here.
Post Reply
User avatar
Fred
Moderator
Posts: 15409
Joined: Tue Jan 15, 2008 2:31 pm
Location: Home sweet home!
Contact:

MS MegaSquirt V3.0 board "issues" revealed

Post by Fred »

Some of this should be in the daily WTF, so I'll just link to this thread from that thread...

This is 128 meg so no 56k and it's as boring as all hell, but I've gone through it for you and though this may not be all, it is some of the important stuff :

http://www.diyautotune.com/videos/megas ... enoise.mov
  • 38:16 - 39:30 I was aware that the +ve feed to the ms was inadequate, thus my unit has a second decent power feed wire to it. I wasn't aware that it was possible to split the high current positive fly back return from the power feed/cpu one easily. That time interval shows you how to do that. (This should NEVER have made it to production...)
  • 45:30 - 46:00 blames faster starting on v drop, and NOT battery internal resistance. This depends on where the cap was put : If it was close to the starter, Bruce is right. If it was close to the battery (and let's hope it was) Bruce was mostly wrong. Yes it's a voltage drop, but one internal to the battery. This seems most likely to me.
  • 50:00 - 52:00 fast idle ground goes to cpu ground!!! bloody hell. If you are switching anything with your fast idle circuit, please lift the transistor leg and ground it separately as it should be. wtf? WTF??
The video was titled "Noise - it's all your fault!"

I guess he was referring to himself :-)

MS was referred to as a baby. When you beat a baby with a stick (who the fuck does that????) is it the babies fault, or the sticks fault that it cries? (how about the dude holding the stick??)

Anyway, basically he said "it's all your fault, but, by the way, I screwed these things up on the v3 board (and much worse on the earlier ones) and here is how you fix them.

This is not common knowledge because the board layouts are hidden and private. This would almost certainly have been uncovered sooner and publicised more if it were reviewable by the public.

He forgot to mention the other grounding design issues that are present with the MS, but as he says, overkill grounding and extra care (you always need some decent amount of it regardless of the board design) can make them a non issue.

Admin.
DIYEFI.org - where Open Source means Open Source, and Free means Freedom
FreeEMS.org - the open source engine management system
FreeEMS dev diary and its comments thread and my turbo truck!
n00bs, do NOT PM or email tech questions! Use the forum!
The ever growing list of FreeEMS success stories!
User avatar
compnut21
DIP8 - Involved
Posts: 22
Joined: Thu May 01, 2008 1:20 pm
Location: Redmond, WA
Contact:

Re: MS MegaSquirt V3.0 board "issues" revealed

Post by compnut21 »

Admin wrote: [*]50:00 - 52:00 fast idle ground goes to cpu ground!!! bloody hell. If you are switching anything with your fast idle circuit, please lift the transistor leg and ground it separately as it should be. wtf? WTF??
that one took me by surprise. they just willingly admitted to a fault in their board that they didn't fix or at least make a note of in the assembly process to fix. oh well, i have a separate transistor there anywho, so ill just move the ground wire for it.
User avatar
Fred
Moderator
Posts: 15409
Joined: Tue Jan 15, 2008 2:31 pm
Location: Home sweet home!
Contact:

Re: MS MegaSquirt V3.0 board "issues" revealed

Post by Fred »

One must wonder whether the timing of the release of this site and it's "why" page and the timing of these announcements were in any way related. "Quick, own up now before we are exposed" or similar :-) There was one thread (now deleted) on a site that is owned by a B&G fan/friend and has other supporters on it that went a little "rough". The guy put the hard word on me to say what exactly I thought was wrong with MS. So after stalling him for a while I told him outright at least some of the things. That guy is on the "design committee" for MS3 so I'm certain that conversation leaked back to B&G etc.

If you don't already have it compnut, I strongly suggest that you add a second positive feed (independently) and run that through a separate relay and fuse back to your supply point. Given this new(? new to me anyway) information it would be prudent to isolate that by cutting the board trace too. It's a shame the two ground planes are so intermingled, for a second there I thought it might be possible to totally fix the V3.0 setups ground faults. Not so unfortunately.

Admin.
DIYEFI.org - where Open Source means Open Source, and Free means Freedom
FreeEMS.org - the open source engine management system
FreeEMS dev diary and its comments thread and my turbo truck!
n00bs, do NOT PM or email tech questions! Use the forum!
The ever growing list of FreeEMS success stories!
gearhead
LQFP112 - Up with the play
Posts: 120
Joined: Sun Feb 03, 2008 9:30 pm
Location: Chicago, USA

Re: MS MegaSquirt V3.0 board "issues" revealed

Post by gearhead »

Admin,

I object to Bruce's tone as well, but maybe that is just his presentation style.

As for Idle PWM being dumped to the digital ground, it is bad, but keep in mind (right or wrong) the board was not designed for this. Ideally, it *should* be a power ground, regardless. I have not really had any problem with this, though. ALso, he doe snot mention that the Zener diode must be removed when making this modification and a 1n4001 needs to be fitted to +12 as well. It would be helpful if this was not the 12V supply for the CPU.

I think you also saw the post where it was claimed that all EFI systems use individual ground returns from all sensors. This is patently not true and an example was given. Personally, I feel a dual ground system is required here. A digital ground (CPU, Sensors), and a power ground (PWM Idle, Injectors) and these should be separate like in most if not all OEM EFI systems.

As for the injection PWM flyback, that was definitely an oversight. What kills me is that this *necessary* fix is not in the megamanual as a required change. Basically, you just go ahead and solder it together and *if* (when) you have a problem, this 'fix' is noted in the troubleshooting guide. I am convinced that PWM current limiting like this is not desirable as it is just more variables to consider and get wrong. Peak and Hold!

I also disagree on general grounds that a large capacitor is a good solution here. If it is on the supply, the micro would not react to true voltage changes quickly enough. It also may mask other noise issues. If it is only on the 12V source to the injectors, it would not be dumping it back to the battery like it should.

Also, pass through traces (to the DB37) could be a bit heavier.

Gearhead
User avatar
Fred
Moderator
Posts: 15409
Joined: Tue Jan 15, 2008 2:31 pm
Location: Home sweet home!
Contact:

Re: MS MegaSquirt V3.0 board "issues" revealed

Post by Fred »

gearhead wrote:As for Idle PWM being dumped to the digital ground, it is bad, but keep in mind (right or wrong) the board was not designed for this. Ideally, it *should* be a power ground, regardless.
Agreed.
I have not really had any problem with this, though. ALso, he doe snot mention that the Zener diode must be removed when making this modification and a 1n4001 needs to be fitted to +12 as well. It would be helpful if this was not the 12V supply for the CPU.
Can you discuss that in greater detail here please?
Personally, I feel a dual ground system is required here. A digital ground (CPU, Sensors), and a power ground (PWM Idle, Injectors) and these should be separate like in most if not all OEM EFI systems.
Absolutely, this is a high priority for us, even if not for them.
What kills me is that this *necessary* fix is not in the megamanual as a required change. Basically, you just go ahead and solder it together and *if* (when) you have a problem, this 'fix' is noted in the troubleshooting guide.
Yes, I wonder if this is justified by the "people make a living from this now, we must look good at all costs" thing that they have mentioned before. It certainly is immoral to leave it like that and let people have problems.
I am convinced that PWM current limiting like this is not desirable as it is just more variables to consider and get wrong. Peak and Hold!
Agreed, the drivers used should be able to handle what is required. Any flyback etc slows the action of a solenoid of any type. hook a capacitor across a relay some time and see how much slower it opens and closes...
I also disagree on general grounds that a large capacitor is a good solution here. If it is on the supply, the micro would not react to true voltage changes quickly enough. It also may mask other noise issues. If it is only on the 12V source to the injectors, it would not be dumping it back to the battery like it should.
A large cap should ONLY be used at the battery such that all components see the same voltage and exposed internal resistance level. Putting it elsewhere in larger amounts than necessary will cause harm for sure.

Admin.

(ps, call me Fred if you wish, I don't mind at all and prefer it. Perhaps I need to create another account or something.)
DIYEFI.org - where Open Source means Open Source, and Free means Freedom
FreeEMS.org - the open source engine management system
FreeEMS dev diary and its comments thread and my turbo truck!
n00bs, do NOT PM or email tech questions! Use the forum!
The ever growing list of FreeEMS success stories!
gearhead
LQFP112 - Up with the play
Posts: 120
Joined: Sun Feb 03, 2008 9:30 pm
Location: Chicago, USA

Re: MS MegaSquirt V3.0 board "issues" revealed

Post by gearhead »

Admin wrote:
gearhead wrote:I have not really had any problem with this, though. ALso, he doe snot mention that the Zener diode must be removed when making this modification and a 1n4001 needs to be fitted to +12 as well. It would be helpful if this was not the 12V supply for the CPU.
Can you discuss that in greater detail here please?
sure. The V2.2 board has a diode to 12V to catch the spike and send it to +12 when the relay opens (or the PWM actuator is released). The V3 board has a zener to gnd. It is not fast enough and usually burns up when it is not removed when the circuit is used as a PWM driver. The solution is to do it like the V2.2 board and put a diode to +12. I solder it on the bottom of the board between the output of the circuit and the +12 feed. Like I said, it'd be better if it were a +12 that was not feeding the 5V regulator, but it works. In the literature, it states that it is best to have a diode in the harness at the PWM actuator, but that is not always practical or desirable, though it does ensure that the spike is fed away from the board, but it is hard to verify that is works on the bench if it is in the harness and not present on the bench.

One thing we may want to think about is to have a constant +12 to the board (could be used as a memory saver... Maybe remembering how long it has been turned off and adjusting startup for that...) so that the injector circuits have a +12 to send their spike to as well as any other PWM outputs that has no potential resistive element like across a relay. The unit would be properly powered up only when it is turned on by switched 12V. (Like the OEMs do it).
User avatar
AbeFM
Post Whore!
Posts: 629
Joined: Sat Feb 16, 2008 12:11 am
Location: Sunny San Diego
Contact:

Re: MS MegaSquirt V3.0 board "issues" revealed

Post by AbeFM »

I'll have to watch some of the video, I'm curious at least. Certainly this idea of hiding from your mistakes gets to me.

Big caps do seem like a cop out - where you have to use them, ok, but don't just throw them everywhere till it works.

As to the flyback, I'd love to see it onboard. I have a couple cases where they don't really work as well as I'd like, and it is pretty annoying. Even burnt out a trace or two.

In general, they could use heavier traces, and not knowing how the board is put together makes it harder to tweak it for sure. More later.... :-)
User avatar
Fred
Moderator
Posts: 15409
Joined: Tue Jan 15, 2008 2:31 pm
Location: Home sweet home!
Contact:

Re: MS MegaSquirt V3.0 board "issues" revealed

Post by Fred »

gearhead wrote:One thing we may want to think about is to have a constant +12 to the board (could be used as a memory saver... Maybe remembering how long it has been turned off and adjusting startup for that...) so that the injector circuits have a +12 to send their spike to as well as any other PWM outputs that has no potential resistive element like across a relay. The unit would be properly powered up only when it is turned on by switched 12V. (Like the OEMs do it).
It's been discussed in various places and if we can make it work in a good and generic way, I'm all for it for the exact reason you state. That combined with MAT AND IAT should mean we can get hot starts pretty much perfect with a good tune and setup (and code).

Admin.
DIYEFI.org - where Open Source means Open Source, and Free means Freedom
FreeEMS.org - the open source engine management system
FreeEMS dev diary and its comments thread and my turbo truck!
n00bs, do NOT PM or email tech questions! Use the forum!
The ever growing list of FreeEMS success stories!
User avatar
AbeFM
Post Whore!
Posts: 629
Joined: Sat Feb 16, 2008 12:11 am
Location: Sunny San Diego
Contact:

Re: MS MegaSquirt V3.0 board "issues" revealed

Post by AbeFM »

Link! Just to keep from polluting the thread. Since, you know, I'm not a fan of constant power.
User avatar
Fred
Moderator
Posts: 15409
Joined: Tue Jan 15, 2008 2:31 pm
Location: Home sweet home!
Contact:

Re: MS MegaSquirt V3.0 board "issues" revealed

Post by Fred »

DIYEFI.org - where Open Source means Open Source, and Free means Freedom
FreeEMS.org - the open source engine management system
FreeEMS dev diary and its comments thread and my turbo truck!
n00bs, do NOT PM or email tech questions! Use the forum!
The ever growing list of FreeEMS success stories!
Post Reply