Preston's Element 14 BOM Attempt

[jharvey]

I see the issue relative to Q8, the sheet names don't match the called for sheets. On page 1, the sheet titled misc output 1, is calling for the schematic general_drive_fuel.sch, instead of the schematic general_drive1.

The input impedance of the A/D should be in the Mohm range, the 1K between the CPU and the cap won't distort the signal, but will help steer RF shunting thru the TVS diode. The 1k between the diodes and the outside world should be chosen to prevent shorts to 12V or gnd and depends on the diodes capabilities. That resistor is again so low that it won't cause resistant voltage drops. The cap should be chosen after the resistors and diodes, such that it's attenuation knee is well above the max frequency we expect on that pin. Such that it's not going to warp or distort the signal via impedance loading. After the cap is chosen, there should be a bode plot done on it to ensure that there aren't phase delays. You can have phase delays even when the impedance isn't attenuating the signal, also remember the signal start attenuating before the knee, so a knee at the max signal isn't sufficient.

In theory, the above sounds good, in practice, ESR variations and layout play a large role in RF and ESD protection, so it's really a load of bunk. One really needs to get a pile of diodes, resistors, and a spectrum analyzer then smack it around until it's right.

About BRV, it wouldn't surprise me if I got it backwards. You keep suggesting the 470 be located in different places. I didn't really look that closely at it. See above for my suggested design approach for selecting these components. The pads are there, but the values/components aren't right.

About TPS, see page 24. Not this page uses an_protect4.sch. Also page 48 uses an_protect4.sch but with different ref designators. This is one place where using KICAD is really handy. Marco did the same thing with the fuel injectors, he discovered you could draw one schematic and use it multiple times, such that you have parametric links and only have to update one schematic. It's a really cleaver approach.

About R257, I'm not sure what's up with that. I have a copy of puma spin1 repo, which I seem to recall was the files sent to the PCB house. It has R257. The posted PDF does not have it on page 19. Check along the top edge of the real PCB, the Gerber files appear to have it there. So I think it really exists. I see it on page 19 as an LED limiting resistor. The lack of this on the PDF schematic tells me that the reverse voltage of the LED's hasn't been tested.

I think the posted PDF pre-dates the PCB spin slightly.

[Fred]

The input impedance of the A/D should be in the Mohm range, the 1K between the CPU and the cap won't distort the signal, but will help steer RF shunting thru the TVS diode. The 1k between the diodes and the outside world should be chosen to prevent shorts to 12V or gnd and depends on the diodes capabilities. That resistor is again so low that it won't cause resistant voltage drops. The cap should be chosen after the resistors and diodes, such that it's attenuation knee is well above the max frequency we expect on that pin.

You need to check the datasheet and/or listen to what I'm saying. Absolute MAX resistance to voltage source is 1k if you want an accurate reading. Absolute MIN capacitance AT/ON the pin is 15nF, past that, tune all you like, but I'd strongly prefer lower resistance and the resistor on the CPU pin CAN NOT be there as the cap MUST be directly on the pin.

About TPS, see page 24. Not this page uses an_protect4.sch. Also page 48 uses an_protect4.sch but with different ref designators.

Cheers. Yep, each of those circuits should be modded to have no resistor on the cpu pin, link there, then 1k or less, preferably 470ohm to test for ravage, to the world, with a suitable cap behind that on the pin directly (through the bridged resistor location)

About R257, I'm not sure what's up with that. I have a copy of puma spin1 repo, which I seem to recall was the files sent to the PCB house. It has R257. The posted PDF does not have it on page 19. Check along the top edge of the real PCB, the Gerber files appear to have it there. So I think it really exists.

Preston, can you see this little bugger? I can't. But mine is covered in components, wires and dust...

I think the posted PDF pre-dates the PCB spin slightly.

That's not good! :-/ Any other differences?

Preston:

http://newzealand.rs-online.com/web/p/p ... /686-8518/
http://newzealand.rs-online.com/web/p/p ... /6868518P/

old ones are even cheaper, in a pack of 50 its 1.6nzd + gst each, nice.

Enjoy! :-)

Fred.

[jharvey]

Per puma Spin1_pcb.pdf, look here for R257

[Fred]

Presto found it, on page 27 of the schem from the site. Cheers Presto :-)

Also, thanks for the pic upload :-)

[jharvey]

You need to check the datasheet and/or listen to what I'm saying.

I see several data sheet's and app notes that don't seem to include that info. Can you reference a specific data sheep, perhaps even a page please. I found this app note, where they use several 10K's in series to an ADC to identify which push button is pressed. It also notes that key 8, is up to 40k.

http://cache.freescale.com/files/microc ... umentation

[Fred]

Page 1253 Section A2, specifically Table A-14. ATD Electrical Characteristics, "Max input source resistance" = 1k, and Section A.2.2.2 Source Capacitance, which states "Cf ≥ 1024 * (CINS–CINN)." where CINN = 10pF and CINS = 22pF, thus Cf = 12.288 which I arbitrarily rounded up to the next highest value of 15nF.

MC9S12XDP512RMV2.pdf = THE datasheet :-)

We do constant sampling, so we require the utmost performance from the device, hence sticking with the manufacturer specifications.

Clearly we get screwed a bit when the source impedance is not so good in the first place, such as TPS which we have no control over and BRV which we can't reduce too far without producing unreasonable heat and v drops in the lines, but never the less, we should allow it to be as good as possible.

Fred.

[jharvey]

Use VNP28N04 for all MOSFET's, as it has a lower Rds, handle all current loads, and limits to 40V such that LED's and such are smacked less.

Higher V limit = faster close. 5 amps is enough for a single injector, 14 is excessive, but fine. and 0.5ohm is sufficiently low to not pose any issues, let alone 0.035 as the VNP14N04 has. Virtually any quality reliable logic level FET is suitable here. Protected ones preferred for obvious reasons, they're almost indestructible.

Many folks use the fly-back diode snubber trick with out issue. Both 40V and 300V OVP MOSFET's will turn off fast enough. I'd like to see a simulation to show how much variation in time we are talking about. Based on gut feel, I believe the variation is in it's in the uS range, as the energy stored in the field of the solenoid isn't that much. All OVP's will have a very consistent off time, so the speed isn't all that important, mostly the repeat-ability. All OVP's will be more repeat-able than the snubber diode approach.

.5 ohm at 1 amp drops .5 volts. This will cause your bat sensing to toss off injector adjustments.

I'll try to look at that datasheet to night, you just saved me reading far more than 1200 pages. I'll see if I can find that PDF on Freescales page.

[Fred]

I'd like to see a simulation to show how much variation in time we are talking about. Based on gut feel, I believe the variation is in it's in the uS range, as the energy stored in the field of the solenoid isn't that much.

Agreed, it probably matters as much as the MF low noise resistors ;-) They're all fine cept the 20v 70amp one without logic input.

.5 ohm at 1 amp drops .5 volts. This will cause your bat sensing to toss off injector adjustments.

No, it won't. All drive circuits affect the dead time of the injector, and the dead time curve used in the tune must match the drive components. For example all of those darlington p&h setups drop 1.2v, always, which is fine, as long as you adjust for it, which you should do anyway. It's all about consistency.

[jharvey]

Page 1253 Section A2, specifically Table A-14. ATD Electrical Characteristics, "Max input source resistance" = 1k

The real page of interest is page 161, which shows it's using one ADC with a sample and hold muxed across 8 inputs. So apparently this is like using a VW bus as a commercial bus. While it can be done, the limitations are that passengers can't be over 4 ft (1.2m) tall, and can't weight over 70 lbs (32kg). It can be done, but the limitations are a bugger. The real issue is the 22pF max capacitance. We can't get down to 22pF with out being buffered by something like an op-amp. So we'll have to survive with some error. I wonder if it's going to be 10 bit accurate, or what it will end up as. I agree, we can minimize the error by keeping the pF minimal, and keeping below 1K on the source resistance. We want resistance between the cap and the pin, such that it can minimize the bounce induced by the switching mux input.

[Fred]

Oh dear Jared! You need more sleep my friend! 22pF is the capacitance of the pin during sampling. It forms part of the equation. That 22pF draws current from the pin to charge up. Thus the cap outside must be MUCH larger to fill the smaller one without discharging. Much larger is 15nF or more. Your bus analogy, while funny, isn't really relevant, but thanks anyway! :-)