Puma board for FreeEMS

[nitrousnrg]

Hi hamster! Welcome along, its always nice to have new members here :-)

1. As Fred noted a while ago, MOSFETS have a low clamping voltage, which don't let the magnetic field to freely collapse. MOSFTES with higher clamping voltages are more expensive. How did you manage to solve this? 3kv is quite a lot :-)

In this board, MOSFETs will be used for driving ignition drivers, like ford's EDIS, etc, not to drive a coil directly. To drive a coil directly I'll use an ignition driver, probably from Fairchild. The population options depends on the user's setup.

The problem with IGBTs is the capacitance between gate and collector. High frequency spikes pass from the collector to the gate, so extra care is needed to protect the driving circuit. A professor told me this, I think I still don't get it. Isolated should mean isolated.

I care about temperature; switching times, noise and temperatures are the things yet to be tested empirically (soon). Thanks for noting the switching times importance.

2. I'm really open to SMT station advices. This will be my first experience on SMT, and my mates/coworkers insist on soldering each component one by one (aaagh hell no!).

So far, I like the hot plate method, it seems to be a good starting point. I'm taking note of everything btw, so when I get to it i'll as you and jared, if you two don't mind.

Again, we need hardware! Count on me for any help about the board.

I have to go, see u

[hamster]

Well, in my case I was switching the high side and my gate drivers were galvanically isolated. I stacked 3 1.5kV mosfets in series, each one could block 1.5kV and leak the rest, so by the end, no volts were passing.

IGBTs are safer to drive if you do it right. Snubbers are needed to slow down the dV/dT to keep from hurting them. I have a cute driver in a high voltage power supply here that I'll adapt up and draw up.

You just have to be really careful of coulping voltage spikes into the uC, else you get surprise resets that are really hard to debug.

If you need any advice on smt soldering let me know. Maybe I will make a vid of my process the next one I do...

[nitrousnrg]

Sorry, I've been kicked off the university lab were I was connected.

Well, in my case I was switching the high side and my gate drivers were galvanically isolated. I stacked 3 1.5kV mosfets in series, each one could block 1.5kV and leak the rest, so by the end, no volts were passing.

Ahá, makes sense.

IGBTs are safer to drive if you do it right. Snubbers are needed to slow down the dV/dT to keep from hurting them. I have a cute driver in a high voltage power supply here that I'll adapt up and draw up.

Can you elaborate on driving them right?

This design shouldn't have any noise problem, nor risk on any I/O pin, since the power circuits (fets,transistors,stepper driver, etc etc) are isolated from the cpu. Their grounds meets at the battery. Anyway, I'd like to know a little more about driving IGBTs. A link would be just nice :-)

If you need any advice on smt soldering let me know. Maybe I will make a vid of my process the next one I do...

Thanks! I let you know when I'm ready to solder the board.

[jharvey]

IGBTs are safer to drive if you do it right. Snubbers are needed to slow down the dV/dT to keep from hurting them. I have a cute driver in a high voltage power supply here that I'll adapt up and draw up.

The MOSFET's that seem to be most preferred here include OverVoltage protection, such that when the output is above 60v (or 300v for ign) it trickles a bit of current to the gate and prevents the output from getting to high. When using a snubber device, you are primarily relying on the inductive resistance to dissipate the energy which will vary based on voltage created by the inductive load. This results in an aprox exponential discharge curve. An injector will turn off at the bottom of this curve, which is where the tolerance brackets are the largest. Small variations in resistance (due to mild corrosion, temperature, ect) on an injector, with a snubber device will limit the abilities to control extreme precision of an injector pulse. However, when using and OV MOSFET, the discharge curve is held at a constant voltage and the curve is much more linear. This allows the opportunity to have a much more precise control over the injector.

On nitrousnrg version of the board, he's using a P&H design for the injectors with some special extra's. I'm not sure how he can use the MIC4451/4452 chip with that P&H chip. Perhaps it could be used for ignition.

The MIC4451/4452 looks to be great for precision timing and it can do 12 amps. I wonder how handy that could be for a CDI setup. I'm reasonably ignorant about CDI, but at first glance, to me it seems like that chip would be great for such setups.

The general design that has been pushed by Fred and a handful of others here so far has been to drive an OEM ignition driver external to the ECU. This is largely because it keeps the high voltage where it should be and prevents several possible problems in various configurations.

The MIC4451/4452 looks like a interesting chip. I'll enjoy learning more about it.

[nitrousnrg]

On nitrousnrg version of the board, he's using a P&H design for the injectors with some special extra's. I'm not sure how he can use the MIC4451/4452 chip with that P&H chip. Perhaps it could be used for ignition.

The P&H driver (LM1949) drives a transistor, and after the peak, it holds the current by putting the transistor on the linear zone. I'm not sure about how important it could be to have an ultra-fast on/off switch, since there will be a lot of heat being produced by the transistor during the hold. Also, the 1949 have separated grounds for sensing (power) and for the CPU side, so there is no real need to add an isolator.
I imagine it would be right to use the MIC4451 to drive the injectors with a mosfet, or if you use PWM to generate the Peak and Hold. This produces lots of EMI, so I'm not really interested in this method.

For ignition, the isolator I use has 5ns/3.7ns rise/fall time respectively, which seems like a good number to me. It stands more temperature, but it can't deliver more than a few mA. My big reason to use it that it has 6 channels in a single package; the bad thing is that it produces RFI.

The MIC4451/4452 looks to be great for precision timing and it can do 12 amps. I wonder how handy that could be for a CDI setup. I'm reasonably ignorant about CDI, but at first glance, to me it seems like that chip would be great for such setups.

Sure! I'll let you know what my dad says about it. He built and sold many CDI modules back in the 90's.

[hamster]

Well, at least for a MOSFET, the full-on resistance is way low, so even flowing amps of current through one won't heat it up much. The idea behind using a gate driver that can supply a lot of current to get it to switch fast is to minimize the amount of time in the partially-conductive state and thus prevent it from building up heat.

[Fred]

Gate drivers are REQUIRED for basic FETs but NOT required for autofets. In those, it is built in. That is why we love them ;-)

[Fred]

You don't need the stuff in brackets, but its slangish to not use "I'm" and it's more complete/explicit to use "hearing"

mmm, big difference there. In spanish I avoid to make references for myself too much. Thats why I don't use "yo" a lot, it is usually implied, and feels kind of "I'm the center of the universe" if you over use it. It could be a personal impression, but I'm not the only one who has it, tho.
"Tengo un auto" and "Yo tengo un auto" means the same, but I'd prefer the former.
to/for: grrr... little words...

Well, sometimes it's totally fine to omit it, but others it makes less sense. The above case it was needed, really. If somone says "how are you" you can answer "I'm good!" or "good!" both are fine, and the I in that really doesn't sound "centre of the universe" at all. If you were saying "I went here, and I did this, and I know that" then it does, but that's totally different to just being excplicit about who when answering a question like that.

end OT

[hamster]

Gate drivers are REQUIRED for basic FETs but NOT required for autofets. In those, it is built in. That is why we love them

Can you give me a part number for an autofet? I've never seen one with a built in gate driver, I guess. Gate drivers are not always required... If you are not going to need to switch the load fast and often, a gate driver isn't worth adding. But if you are switching fast and have a lot of current to control, it's really desired to keep the MOSFET from melting.

[Fred]

VNP10N07 - they have logic level inputs :-) autofet is a trademark/tradename, omnifet is another, etc. And sure, but if you're doing anything with load and RDS on matters, driving them as designed is a good idea :-)

I thought they actually required more than 5V to switch fully anyway? I'm pretty sure some fet drivers bump it up too.

Fred.