Puma board for FreeEMS

[nitrousnrg]

Another thing: I find hard to believe that such capacitance exists in an IGBT, but I didn't ask directly that question. Its in the list now.

[jharvey]

I see what your saying, it was more of a caution that this capacitance can turn on the MOSFET erroneously, than a recommendation to add it for bleeding off higher frequencies. I would think that capacitance still exists for IGBT's, however IGBT's are known for slower response times, and it might not be able to react that quickly. It will be interesting to see what he mentions. Perhaps he might recommend a small drive resistor, or inductor to create a filter effect that prevents RF from getting to the top side of the MOSFET. Perhaps some simulation is in order, we certainly don't want some spark RF turning the injector(s) on.

On a good note, the MOSFET drives that have been talked about, don't empirically appear to have this problem, but it's also possible that the setups simply had a good set of plugs.

[Fred]

No idea what has been said since, BUT, I saw mosfet for ignition and went WTF. Marcos, it's important to the health of the spark that the coil is allowed to kick back to a high level unencumbered. IE, you need to not clamp, or clamp to a very high voltage. FETs to do that are uber expensive and uncommon, special purpose ign drivers, on the other hand, do that just fine with clamp levels around 400V. If you run a FET and clamp to 50V or so, you're screwing up your sparks by interferring with the mag flux collapse in the coil. Same goes for injectors, but 50V is fine for them.

Still working on that other post...

Fred.

[nitrousnrg]

Thanks for pointing that out, I wasn't aware of the cost of such FETs.
And, the IGBT has the same base-collector capacitance of a transistor, so I won't use one. I'm going for the ready to use ignition drivers (fairchild probably), isolated from the micro with a 6 channel isolator. This ign circuit will have a separated ground, it can't mess with the microcontroller nor the injector driver; plus, is far away from the analog circuits.

Keep working mate, I'm going to Buenos Aires for a couple of days, I'll be back on friday :-)

[Fred]

By IGBT, I meant dedicated IGN drivers :-) The bosch one that diyautotune sells has proven to be good, and the std one in ms kits till then vb921? sucks arse. There are lots to choose from, though. You should look at the isolation properties of the inputs in these devices before adding complexity. If you're using an opto coupler to transmit the signal, though, that is advantageous as the isolated ground/power rails moving around with load won't affect the relative drive signal level.

Enjoy your trip! Take some photos!! :-)

Fred.

[nitrousnrg]

I'm thinking in Fairchild just because a member of this forum said it worked great for him, but I'm aware that there are lots of ign drivers. Not really sure which one I'll use.

The problem I see with the total isolation is that I can have a separate ground, but I'd need a 5v referenced to that ground, only to supply one pin. I'm working out this one.

I changed the power input circuit:
* 2 fuses instead of 3, one for power hungry devices, the other for cpu and analog circuitry.
* One regulator for digital things, other for analog things.
Quite clear by looking at the schematic


P4 and P5 should be unbridged, and put one for power, and the other for the cpu. That, plus a good isolation between cpu and drivers signals can make this rock solid.

Also, I added a BDM connector. Its just like the adapt board, but I've seen connectors with 6 pins instead of 10... its the same, but I prefer not to have 4 unused pins.

OT: I took only one photo :P I'm waiting for the guy who has recorded the video of the conference.

[Fred]

With respect to ground isolation and 5v signals, you have two things to consider :

(1) how much ground differential there will be, max
(2) what the actual switch threshold voltage is

if (5v - (1)) > (2) then sweet, if not, fail.

I ran a totally lifted ground on the volvo, and suffered no problems at all.

[nitrousnrg]

I'm not worried about digital signals. In fact, I'm considering to use a different ground and the same 5v that the micro uses in both sides of the isolator.
What I want is to totally separate power from analog. Imagine 6A flowing thorugh the ground tracks and cables, I see reasonable to see a few mV of ripple between battery ground and analog ground, wich may -or may not- affect the A/D readings, at least by a few counts from time to time.

Separate the noisy power from the analog circuits is always a good thing. Besides, I'm not adding any component, just replacing the ANDs by an isolator. Everything else is just a pcb re route. Even the power connector allows to have 2 grounds a 2 12v sources :-)

PS: moving your /home partition may lead to a data corruption, which lead to a formatted /home partition, which translates into a wasted weekend fixing your stupid pc. git saved the day for both freeEMS and ecumanager.

[nitrousnrg]

Just a note:
Talking with a mate from Mendoza, he made me realize that I'll need 8 p&h injector drivers. Most of the 4 cyl (big)turbo cars use 2 fuel ramps, with 4 injectors each (and I'm not counting the extreme guys who use 12 low Z injectors for a 4 cyl engine).

This could be the new setup:
* 8 inj drivers (no idea if the microcontroller will handle this!)
* 4 ignition drivers (4 cop, 6-8 wasted spark)
* maybe MAX9926 instead of lm1815; phase detection using the coil is an interesting idea too, my car has it, so it doesn't need a cam sensor to be sequential.

[Fred]

I think that is unreasonable. Most people run just big injectors 1000cc or so for a hard out setup on a 2.0 4 pot. Those that do do a dual setup should be doing it, not for greater flow, but for better idle and emissions. Thus the dual setup will be four low z and four high z. When you do this, it's completley reasonable for fuel from the small close injectors to be cut at higher loads and rpms. Your friend from Mendoza has been smoking too much crack and/or reading too many shitty car magazines with blinged out hondas :-)

When you talk about people with dual rail setups, they are a small part of the market IMO anyway. When you talk about tripple rails, well, I've never even heard of or seen that on any car. I also can not see a good reason to do it. single 1400cc injectors would out fuel virtually any 500cc per cylinder engine with ease. You just can't keep engines reliable at that level, so you're talking about show cars and drag cars anway, and just a few people world wide, not any significant number of them.

Hope that all makes good sense :-)

Fred.