PCB layout - DaveBMW

[davebmw]

A bit more of a review of your layout.

General
- I don't see a hierarchy anywhere yet. I don't think it will produce a correct net list this way. I also see the border notes pg X/7, which isn't updated unless it knows a hierarchy. Are you using win or Linux for your KICAD? Do you know what version you have? Perhaps this is related to that error message from XDP_CPU.

Windows XP pro.
I have to find a way to move the CPU schematic from the root and onto its own page, is there an easy way to do this??

3 wire PWM,
- D621 and R621 are hard to read, perhaps spread it out or shrink the text a bit here.

OK will do.

- DRC notes two unconnected nets. That likely indicates, they are driven with incompatable driving pins. Seems like a power out pin (pin at the symbol level) won't connect to another power out pin. Perhaps something like that is happening here.

I only see one warning on the IACV_PWM line that is the PWM idle speed control signal from the CPU.

- The NPN and MOSFET, Those are current limiting right? Yes I see .7/.5 is 1.4A.
They are expected to not work normally right. Mostly short circuit protection. Also they will ease their way into backing off the current, so it might over shoot for a short time. All correct

Any reason why you don't like the current limited MOSFETs?

Yes, I don't like them. they are quite specialised so prone to obsolescence, the discrete method allows greater flexibility in choice of parts and allows the end user to select current limits suited to the application. or leave them out if you don't want or need it

- Are you running two injectors because you have 2 injectors, or is one drive for peak amps and the other hold amps? I think IACV3 is hold amps.

No IACV Stands for Idle Air Control Valve it is the bosch 3 wire type that is arranged as push pull the centre common pin is held at +12V. IACV_1 is non inverted and IACV_3 is inverted e.g. when IACV_1 is at 10%, IACV_3 is at 90% PWM



Any how off to work I go.

Cheers

[jharvey]

Open the existing schematic with CPU in it, save as to the new number, then modify both the orginal and the new schematic to create the interconnects and delete the redundancy. Does that do what you're looking for?

I think KICAD has "save as" if not copy and paste it in windows explorer.

[jharvey]

Was that 8inflacid that had a note about KICAD, something like deleting a "U" or something in the windows version made it work on both platforms. Can't seem to remember.

I'm curious if you plan on the XOR buffer or not? Perhaps it's on the XDP schematic. I might try to encourage you to go that route if you don't currently plan to. The MOSFETs have a much lower isolation voltage vs IGBT's. If anything goes slightly wrong with the FETs, MOV's, or diodes, your TA board is toast. The XOR's give you additional isolation and inversion if desired.

The 600V to 1200V clamps may be a bit high for your harness / connectors. You may want to check your harness / connectors for this. I also suspect foggy days or I guess rainy days in Wales might cause the harness breakdown to lower as well. Perhaps you can connect a high V supply, to test it. Can you borrow a high pots tester from work? Also designing the PCB for 600 + shouldn't hurt anything, It just might not get that high on the PCB.

[davebmw]

Was that 8inflacid that had a note about KICAD, something like deleting a "U" or something in the windows version made it work on both platforms. Can't seem to remember.

Really? I'll have to have a look at it. Thanks thats handy to know.

I'm curious if you plan on the XOR buffer or not? Perhaps it's on the XDP schematic. I might try to encourage you to go that route if you don't currently plan to. The MOSFETs have a much lower isolation voltage vs IGBT's. If anything goes slightly wrong with the FETs, MOV's, or diodes, your TA board is toast. The XOR's give you additional isolation and inversion if desired.

I was going to use opto isolators first off but have found that ones fast enough start getting a bit pricey and sort of specialist so could be prone to obsolescence. I will put an XOR gate in there just as another layer of protection and optional inversion. The chances of the MOV breaking down with a VBO of 680V @ 1mA then on to 1120V @ 25A and 3500A peak clamp current are slim,
the series diode rated at 1200 Vrev and 20A Icont with 250A Ipulse also reasonably unlikely.
Not to mention the MOSFET's I am using are heavy at 50A Icont, 60V, 200A Ipulse. I run these now on MS2 driving coilpacks with 120V MOV and no series diode and they are holding up well.

Go to http://www.rswww.com for a look

MOV = ERZV10D681 RS No 226-8107
Diode = 20ETS12PBF RS No 541-1326
MOSFET = STP55NF06FP RS No 486-2335

Pretty much though, if all the measures fail... you have just been hit by lightning and you really don't care about the XDP chip any more!


The 600V to 1200V clamps may be a bit high for your harness / connectors. You may want to check your harness / connectors for this. I also suspect foggy days or I guess rainy days in Wales might cause the harness breakdown to lower as well. Perhaps you can connect a high V supply, to test it. Can you borrow a high pots tester from work? Also designing the PCB for 600 + shouldn't hurt anything, It just might not get that high on the PCB.

Yes i see what you mean I have looked at the cable to the COP and they look more than adequate for upto 2.5KV and have thicker insulation than the rest of the cables and are double insulated back to the ECU in a conduit also. The OEM doesn't have any such countermeasures for overvoltage clamping it just has pretty damn big transistors with quite a high voltage rating I have no doubt. If anyone can find data on Bosch 30014 transistors that would be handy, as it would be nice to know how or what they are.
The harness on BMW's is double or triple insulated and covered by plastic covers pretty much all the way, so again it has to be seriously detrimental circumstances to get the cables wet. (By which time you have drowned) oh you get the idea!

[jharvey]

Ah found that note, here it is

http://www.diyefi.org/forum/viewtopic.p ... a&start=70

It was the last post, but also note a bit later on he noted he updated both version to the same revision, and it worked fine from there on. Seems as thought there are some incompatibilities between KICAD revisions.

[davebmw]

I have been tinkering most of the day with the Type_B getting the Hierarchy right with the root sheet leading off to all the sub sheets.
I have had to change all the net labels to global to work across the hierarchy but everything seems to be gelling quite well at the moment.

To do sunday (in a few hours):

Sort out the custom lib files.
Start the analog, Digital and VR sections.
Plop out a new Type_B release (0.51)

Monday, Away on business, will have some time in the hotel room in the evening and if there is enough time:
Design a new component 144 pin LQFP for 144 pin XDP chip.
Produce the CPU schematic for the 144 Pin XDP.

Tuesday through friday:
Route most of the PCB in PADS. may have to bring the injector and IGN pins out onto headers until final locations set in stone.

Must beat Fred!! (not with a stick or anything), this is a race you know!

Dave, (Tired) going to bed now

[davebmw]

OK I have been really neglectful recently after blowing my head gasket and I have been really busy at work.
So I have cobbled together what I have got and attempted to tidy it up a bit by sorting the hierarchy and sectionalising things.
I have added some future development stuff in the form of knock sensing, EGT, WBO2 and USB comms. All of which I have been researching recently.
I have yet to finish the analogue, VR and digital input conditioning and protection sections in favour of the more interesting and complex world of knock sensing and WBO2 control.
USB should be relatively easy and could be a useful add on or permanent fixture on the second serial comms port of the XDP chip.

I have skipped a version and gone straight for 0.52, 0.51 was really pointless and quite messy.

Jared could you take a look and see if I have resolved the library problems you were seeing before and let me know again?

FreeEMS_0.52_Type_B_DJB_WIP.zip

(240.98 KiB) Downloaded 738 times

Any comments and questions are more than welcome

[jharvey]

XDP_CPU now opens, however FreeEMS_TypeB flopps. Noting lib then failed structure. Likely really only a library issue. Appears that this typeB schematic is where the hierarchy is located as it's pg1 of 16.

It looks like I can guess fairly accurately what the hierarchy is by the page numbers and bus names. So I won't let the pg1 hold me back from taking apeak.

[jharvey]

In analog conditioning, appears you plan for more conditioning, there is a cap floating in space, also I'm assuming the lower trace is connected to GND. It's not labeled. The 39K is current limiting, yes, perhaps you want to move that to the right past the BPF, also the 5V6. If so the caps will be better protected against voltage spikes. However it changes the BPF, so perhaps you don't want to move that.

In LPDrive, he he, low power output for nitrous... Nice. A thought that might apply to all the driver circuits, the current limiting NPN's may have a fair gain tolerance to compensated for. These .1r's might want to be pcb traces, a mesh of traces that can be cut and trimmed to calibrate the current limit. Either that or you may want to measure B before you purchase your resistors.

In power, some blue text is upside down... Inductor size for L1 and L2? I'm assuming big. Are you matching the inductive and capacative props to make this look more like a resistive input?

In TA header, the bus isn't currently required, the global pins will link to the others just fine. However there is some extra space, so perhaps you're adding to this schematic.

In XDP_CPU, some blue text wondered over into the border. They also make a max233 chip, that includes the caps, however it costs more and takes the same footprint. P2 connection to PE? I don't understand this, I'm guessing for future stuff. I like the decoupling caps, I've thought of doing more of this on the one I've been working on, but have figured it's really only needed at the CPU, and close to the chip, so I'm hoping it's taken care of on the TA card.

General notes, I can't do DRC with out the hierarchy, so I don't know what the flags are about. Once I get the hierarchy thing working, I might have some notes about the file structure. Also the border notes R.51 not .52.

All in all, a very nice job of it. Keep it up, well after the head is back on that is.

[davebmw]

In analog conditioning, appears you plan for more conditioning, there is a cap floating in space, also I'm assuming the lower trace is connected to GND. It's not labeled.
Yes lots more conditioning for MAP, AAP, TPS, MAT, CLT, AUX1, AUX2 etc etc etc. and yes lower trace is analog GND

The 39K is current limiting, yes, perhaps you want to move that to the right past the BPF, also the 5V6. If so the caps will be better protected against voltage spikes. However it changes the BPF, so perhaps you don't want to move that.
No the 39K is part of the potential divider for Vbatt monitoring. the BPF cleans up any spikes which plauged my MS2 install this may not be needed of subject to experimentation and adjustment to suit the application. have also been looking at MOV's for input protection.

In LPDrive, he he, low power output for nitrous... Nice. A thought that might apply to all the driver circuits, the current limiting NPN's may have a fair gain tolerance to compensated for. These .1r's might want to be pcb traces, a mesh of traces that can be cut and trimmed to calibrate the current limit. Either that or you may want to measure B before you purchase your resistors.
Yes the real purpose of the Low Power drive is to control things like relays, camshaft actuators, nitrous and any other accessories you can think of. The current limiting "foldback" transistors are to protect the MOSFET's against short circuits and may be omitted if the end user is confident that their wiring is sound. You can calculate the current limit selection quite easily, I will put a table of values in the schematic for ease of selection

In power, some blue text is upside down... Inductor size for L1 and L2? I'm assuming big. Are you matching the inductive and capacative props to make this look more like a resistive input?
Thats for our Austrailian friends;) No I'm at a loss to why the text is up the wrong way. the LC is just for filtering top clean up the rails.

In TA header, the bus isn't currently required, the global pins will link to the others just fine. However there is some extra space, so perhaps you're adding to this schematic.
I was thinking of loads of tranzorbs and protection for the TA card

In XDP_CPU, some blue text wondered over into the border. They also make a max233 chip, that includes the caps, however it costs more and takes the same footprint. P2 connection to PE? I don't understand this, I'm guessing for future stuff. I like the decoupling caps, I've thought of doing more of this on the one I've been working on, but have figured it's really only needed at the CPU, and close to the chip, so I'm hoping it's taken care of on the TA card.
You can never be too careful with decouplers

General notes, I can't do DRC with out the hierarchy, so I don't know what the flags are about. Once I get the hierarchy thing working, I might have some notes about the file structure. Also the border notes R.51 not .52.
Yeah I'm not too sure why the hierarchy is playing up i'll have to get a fresh installed PC to try it on. and well spotting i forgot to change the borders.

All in all, a very nice job of it. Keep it up, well after the head is back on that is.

The right (allegedly) head gasket has arrived today and the weather is good so this afternoon will more than likely be putting it all back together and getting the bitch running!