Puma board for FreeEMS

[Fred]

Why so much chatter about a device that can be changed in minutes.

Because too many ms2 units died and had switch on/off trouble from having insufficient current capacity to properly drive external devices (even lousy auto fets!!!). I don't want to see any problems like that on any FreeEMS boards, at all, if I can help it. 1.6k guarantees no issues unless 12v is injected into the output pin somehow, which is super unlikely (however a solder bridge to 5v or ground is much more likely).

[jharvey]

I just checked out the latest on this. I see it still calls for several parts I had originally put in, but have determined as not quite right. The AN protection on the AN ports schematically have two series 1k resistors, for a min of 2k, which is well above the recommended 1.6k min. Also the Digi protection has two 10k resistors for a 20k which is also well above 1.6k recommendation. I seem to recall those values were based on a pencil schematic drawn by someone many moons ago. The cap values will probably change when the resistor values change, and I was expecting some required tunning / empirical data to finalize the protection design. I've seen several claims that this type of protection is often more of a black magic than proven technique. We may be able to do a slightly better job predicting these components, but I expect what ever we end up with, will need to change any how, so having the pads it the key and don't consider those components final at this point. Also many of the digi signals are buffered, so they have a separate driving capabilities, and the 1.6k stuff doesn't apply.

Down around the RPM digi gnd's, you may want to add a net name. Technically the sub schematic grounds them so it's not important for the net list, but it would be a bit clearer at the top level if they show a connection to ground.

The AD597 could have an LED included, I seem to recall the LED is a diagnostic tool. It might be handy to have.

I'm ignorant about the s1846x, I might recommend adding a resistor between each pin and each port pin, such that if you need to add a component you can do so. That component will probably be a 0 ohm resistor for now. If it is, then it can be removed on future designs, but for a first run, it might be handy to have a place to get in on this signal.

I'll try to look at the PCB layout tonight.

[Fred]

http://www.msextra.com/forums/viewtopic ... 01&t=26378

Very old thread on msextra with original recommendations and massive discussion on change. Reality is, low resistance kills s12 cpus. The conservative option is 1.6k such that all 8 pins can be in fault together and it survives. Going with 1k will likely be OK, but it'd be nice to make it that much more bullet proof at no real cost. The ignition drivers never worked right with the higher inline resistance and need a predriver, autofets were ok, that was my confusion.

Fred.

[nitrousnrg]

Jared, about the RPM grounds:
that was a mistake I corrected a couple of days before, not commited yet. The change in the power supplies got the schematic a little messy, I'll make a new commit once the regulator switch is ready.

AD597:
I'm not sure about using this part. (I couldn't order it from digikey). The malfunction alarm is available in the metallic package, not in SOIC8.

0ohm resistor:
I think I'Il put pads all around the cpu in the first try... And that isolator wen out of stock in digikey, so unfortunately, the pads are available.

Fred,
In the device datasheet it says clearly that the 25mA are a maximum for a pin. I'm sure that you can get 25mA flowing from one of it. If the cpu dies, then there is a maximum amount of current the port/microcontroller can tolerate. I'm used to see 25mA per pin, 200mA sunk/sourced by all ports combined (microchip mcus), which goes well below the 25mA per pin rating. The application note says that this is implied in the cpu's power dissipation characteristics.

On my box:

/home/fred/workspaces/home/freeems-vanilla/docs/XDP512-freeEMS-pin-assignment.ods

I'm aware of that, but I wanted to see which channel of the ADC you use for MAP, which for clt, etc. The spreadsheet only say
"All these yellow ones are swap able to suit board layout needs at the designers whim."

No worries, a fast ctrl+F over the firmware should put light over this.

Generic (and maximally efficient) code for firing up to 12 or even 16 sequential injectors.

Oh, thats a good point. my injectors stay in port B, port A will be freed, and fuel pump is going to be moved elsewhere.

[Fred]

In the device datasheet it says clearly that the 25mA are a maximum for a pin.

I think you are misreading it. Please specify the exact page, section, line, cell, etc that I need to know to find this specification in that document. I've looked and found other specifications, but not that one. Perhaps I'm misreading it? Either way, we need to be on the same page here (both literally and figuratively) so post about how I can find this spec.

On my box:

/home/fred/workspaces/home/freeems-vanilla/docs/XDP512-freeEMS-pin-assignment.ods

I'm aware of that, but I wanted to see which channel of the ADC you use for MAP, which for clt, etc. The spreadsheet only say
"All these yellow ones are swap able to suit board layout needs at the designers whim."

No worries, a fast ctrl+F over the firmware should put light over this.

git pull. The column was hidden and I only found out about it an hour ago. I commited the fix a few second after finding out... it wasn't obvious, I forgive you :-) I should remove those comments about it being swappable, I see exactly why you've been doing what you've been doing now. A lot of the docs were aimed at, you guessed it, me! Personal references in public. Those yellow notes were to myself to remind me that the pins were identical and I didn't hve to keep them that way if I didn't want to. etc. Sorry for the misunderstanding and arguments. Woops.

Oh, thats a good point. my injectors stay in port B, port A will be freed, and fuel pump is going to be moved elsewhere.

Good man.

Fred.

[nitrousnrg]

I want to hear comments about this, so here it goes in an easy to read form:
Our cpu (page 1241)


And since at some moment i wasn't sure if it was talking about the pin or port, I checked other datasheet (MCS08QA4DS.pdf), and got

see note 1, for an input the current limiter resistor would be (extra voltage)/25mA.

Also, the application note AN2434 have a few plots like this one:


Still, draining too much mA from too many ports is just bad. This microcontroller, to stand 125° without extra cooling must use less than 80mA (cpu+IO), which is marginal for most applications. The math is on my repo (marcos's notes).

Another recommendation is to tie down or high the unused pins. According to that app note:
"An un-terminated input can oscillate or float to a mid-supply level, causing both of the FET devices to be in a partial on state, raising device dissipation, increasing noise, and drawing additional supply current."

git pull. The column was hidden and I only found out about it an hour ago. I commited the fix a few second after finding out... it wasn't obvious, I forgive you :-) I should remove those comments about it being swappable, I see exactly why you've been doing what you've been doing now. A lot of the docs were aimed at, you guessed it, me! Personal references in public. Those yellow notes were to myself to remind me that the pins were identical and I didn't hve to keep them that way if I didn't want to. etc. Sorry for the misunderstanding and arguments. Woops.

Ahá there is it! Lets see how it looks when routed. btw, I can't pull at uni, blocked port 22 ¬¬

[robertofonte]

Do you know if a LM1815 or MAX9926 can handle digital inputs? VR signals are much more tough than 0-5v, or 0-12v square signals. I was wondering if I can use the same circuit to interface VR or Halls.

The MAX9926 can handle digital inputs of either voltage (actually any voltage will work up to what your input components will take). Just connect the signal to one of the differential inputs and leave the other one floating: one input will give one polarity and the other will inverse it.

This is a very nice chip to use for any input that will be used to measure frequency, time interval or to count the number of pulses. That is what I use on my I/O Extender for those reasons.

Jean

Sorry for my english

I am thinking that could be used Max 9921 for the hall sensor

App from maxim to hall sensor
http://www.maxim-ic.com/app-notes/index.mvp/id/4220


Roberto Fonte

[nitrousnrg]

thanks for the link Roberto!
I ended reading a couple app notes from maxim. Very tough interface sensors apparently.

In a while I'll test my max9924, to see how it handles a square signal. If it works, even with added noise, then I'm sticking with it. If not, I'll check if halls are used here at all -I don't recall seeing anyone with a hall setup-. If they are, then that max9921 is a nice option.

I remember that the hall interface for freeems-hw-1.0 was a TL081, or something like that. The reason being it was deployed in some mazdas (or something like that too), enough reason to know it works. Still, that IC is way more rugged than the TL081, much more automotiv-ish.

[jharvey]

I am thinking that could be used Max 9921 for the hall sensor

I noticed this chip didn't list engine management in it's application list. I think I see why, but perhaps I'm wrong. I see it notes 50khz max hall input frequency. If using a 36 tooth wheel, and assuming a max engine RPM of 10,000 RPM, that would require a min input frequency of 360 Khz. It looks like using this sensor as RPM filter might require a close look to ensure it will work for higher RPM / higher teeth setups. For example, my suby only has 4 teeth, so a max input frequency of 40 Khz, which might work. I don't know if this 50 Khz limit includes Nyquist / ect, so it may still be an issue on lower teeth engines like mine.

[Fred]

Key word, Marcos, instantaneous, not continuous. This is where we were at cross purposes. I want it to be able to be shorted all day, not for a brief transient. What they are talking about there is driving of things like capacitive loads where it might switch, and be at a large current for a short time while the cap charges/discharges, etc. I stand by what I said earlier.