Puma board for FreeEMS

[nitrousnrg]

* Mic2954:
It has 250mA vs 400mA of the LM2937. The last seems like a safer choice.

Updated and no longer in ?

No longer in. Those changes were performed in the master branch, but I created a branch from the Spin1 tag, so now I've to make some corrections to the schematic again.

* low side drive mosfets
totally optional, they're insanely huge (RFP30N06). Perhaps you guys have a better/cheaper/more common suggestion?

I'm not crazy about that choice, I changed to VNP20N07 an OVP MOSFET.

Perfect.

* R151 and R150
Yeah, that circuit was taken from a Bosch sensor datasheet, for the thermistor that my car and quite a bunch of other cars here have. Those resistors aren't part of a normal setup. Ask fred about the values of the bias resistor, it should be calculated with Freetherm afaik.

The issue is that I didn't have an exact match value. I'm guessing you used 39K and 3.3K.

Ah, I see. Yes, I approximated those values, since I had a working setup with them. You can leave 39k and 3.3k, or remove the resistors if you want, only a couple of argies will use that circuit in Spin1, and I'm the one who build the boards.

So 1N4148 is now a different diode We should update the schematic to call for a different diode.

Mmmm, yes, in later versions all 1N4XXX were replaced by schottkys.

O2 has no extra components but the protection circuits, which has some RC filtering. I'd remove the filtering caps at the circuit protection (in all channels), and only filter in the input circuitry. Maybe having diodes so close to the ADC can make the input too noisy, I can check that in minutes.

This being a qty 0, could be a good thing. The diodes and cap should basically not influence the circuit unless you exceed the frequency or voltage of the circuit. I should QUCS this circuit to show it's reactions a bit better. A down side is that I can't figure out how to do a FFT with the copy of QUCS I have.

Still unsure about those caps, most of the input circuits already have a filtering network.
By the way, I'm not a power user of qucs, but if it can throw a table, .csv preferentially, the FFT can be calculated with octave with a couple of commands.

Here's what I see that needs addressing.

R150 and R151 are exact values, but are probably really 39K and 3.3K, what should they be?
1R1W and 1R should these be the same? I probably have a buggered count, the are currently 1R.
1N4448 is replaced by a different diode, but what one?
C21 470pF and C123 1uF doesn't appear in the BOM. What's the digikey/MFG number?

Newest BOM pushed.

* Leave 39k and 3.3k. That BOM was produced from an updated schematic
*1R is R2, in the voltage regulator circuitry. Two 1R1W are for the stepper IC, they are in there for current sensing.
* 1N4448 -> MBR0520L The same as the analog protection schottkys. Digikey: MBR0520LCT-ND
* Further revisions changed C123 from 1uF to 0.1uF, and R111 from 3.9k to 39k, setting the same time constant.
C21 = 470pF should be 330pF, to reuse the cap of the PLL circuit. It was changed later for a similar and now inexistent convenience (fixed in master branch now)

[Fred]

I don't have time to read over those three posts and correct anything right now, but, I have a small correction to the PCB artwork:

VR1 and VR2 need renaming! They should just be RPM1 and RPM2 instead as they can be used with VR or Hall or Optical or damn near anything there is out there (beautiful!!!).

Which reminds me to remind you:

You need to have an optional pull up resistor outside the isolation resistors for each one. And, it would he valuable to have the option of a pull down too, so something very similar to how the XOR gets configured, where options are up, down, nothing, for each input. Leaving the grounds disconnected can be accomplished at the connector card or even loom level, so that's not an issue.

Additionally, it might be nice to have the option of a small capacitor across each VR outside the isolation resistors, just as the toyota ECU that i traced does.

I think allowing for various different conditioning options between connector/loom and max chips is a VERY good idea as it keeps the setup flexible enough to handle almost any situation with minimal board real estate cost.

Fred.

[nitrousnrg]

Good, added those items to the TODO.

Except this one:

[Additionally, it might be nice to have the option of a small capacitor across each VR outside the isolation resistors, just as the toyota ECU that i traced does.

Could you rephrase? I didn't get where the cap should be placed and for what :-)

[Fred]

Across each +/- input pair to kill noise while the output impedance is low and can maintain good signal with minimal shift and rounding etc and still kill induced noise. That's why the capacitor that IS there scares me a bit, it's inside some significant resistance and will shift the phase of the input a bit, in practice it seems fine, but it still scares me a little (unfounded, ignore it). Make more sense now? If not, maybe we can skype when I get back to civilisation tonight? O manyana por la manyana?

[nitrousnrg]

Across each +/- input pair to kill noise while the output impedance is low and can maintain good signal with minimal shift and rounding etc and still kill induced noise. That's why the capacitor that IS there scares me a bit, it's inside some significant resistance and will shift the phase of the input a bit, in practice it seems fine, but it still scares me a little (unfounded, ignore it). Make more sense now? If not, maybe we can skype when I get back to civilisation tonight? O manyana por la manyana?

It is just a language barrier, I don't know what across means, in electric therms.

My guess:
IC->CAP->Resistors-> your CAP (its leads connected to the resistors, like if I'd soldered it to VR+ and VR- of the 40pin header).

I wouldn't remove the original cap, but there is space in that place, so no objections. A while ago someone pointed out the phase shift, and I understand your concerns.

[TonyS]

Marcos, I think Fred is talking about adding TWO caps to each VR input channel. One from VR+ to GND, and one from VR- to GND.

From what I have seen, it is quite common to place a capacitor on every ECU connector pin for ESD and EMC reasons. You do have to be careful to use appropriate values though, as you have to take into consideration the source impedance of signals coming in so that you dont start filtering out the frequencies of interest.

[nitrousnrg]

Hi guys, an option to hack the USB connector and install a mini-B USB:



Based on heavily detailed suggestions:

Tinned 4 locations on the connector, and one on the board; using solder paste helped a lot. Then simply re-heated them, while holding them with my fingers in place. Once the first tack was done and holding it in position, I then put a bit more solder on the diagonal tinned area to really hold it down. Then letting it cool in between soldering tinned locations, I did it one by one until it was done. After that, you can wire up the board, following:

http://en.wikipedia.org/wiki/Universal_ ... appearance
to match Puma's USB connections:
https://github.com/nitrousnrg/puma/raw/ ... _front.png (I thought this one wasn't that heavy)

Marcos, I think Fred is talking about adding TWO caps to each VR input channel. One from VR+ to GND, and one from VR- to GND.

From what I have seen, it is quite common to place a capacitor on every ECU connector pin for ESD and EMC reasons. You do have to be careful to use appropriate values though, as you have to take into consideration the source impedance of signals coming in so that you dont start filtering out the frequencies of interest.

Thanks Huff, it makes sense. Right now VR are the only inputs without a capacitor between the signal and ground. And, if the capacitor is small there shouldn't be any frequency filtering, we're working up to 15khz (although yes, it depends on the values we use).

[Fred]

My guess:
IC->CAP->Resistors-> your CAP (its leads connected to the resistors, like if I'd soldered it to VR+ and VR- of the 40pin header).

Yes, exactly that, exactly as Toyota did.

I wouldn't remove the original cap, but there is space in that place, so no objections. A while ago someone pointed out the phase shift, and I understand your concerns.

No, of course, leave it there, it's easy enough to un-populate it if required. The existing one forms a classic RC filter. My proposed one just kills noise without significant filtering due to the relative impedances at various frequencies of such configurations.

Marcos, I think Fred is talking about adding TWO caps to each VR input channel. One from VR+ to GND, and one from VR- to GND.

Nope, Marcos understood me correctly, for a change :-)

Fred.

[Fred]

Right now VR are the only inputs without a capacitor between the signal and ground. And, if the capacitor is small there shouldn't be any frequency filtering, we're working up to 15khz (although yes, it depends on the values we use).

any >> much - as an engineer, and not a hippy, you should be careful with the language that you use ;-)
15k >> 30k - 10krpm @ 360 slot CAS

[jharvey]

I just pushed a footprint module titled usb-B_u-mini-standard. Has the three common USB B connectors in it.