[info] Frankenso - ECU shield compatible with OEM case / official

[kb1gtt]

If we use a harness like this http://dynamic.focalprice.com/CA0548B/0.3_meter_Dual_USB_2.0_Motherboard_Panel_Mount_Cable_Black.html?utm_source=CS&utm_medium=GM_US&utm_campaign=CS_GM_US_CA0548B&source=CS_PLA_US&gclid=CLWioqmF-L4CFU5nOgodlF4AgQ, you wouldn't need to open the ECU to get access to the USB. And you could mount the USB just about any where on any ECU.

usb_harness.PNG (64.54 KiB) Viewed 79773 times

I wonder if SD can be done on a harness connector like the above noted USB. That USB thing on PC's is very common.

Mouser seems to have vertical SD cards http://www.mouser.com/ProductDetail/Yamaichi-Electronics/PJS008U-3000-0/?qs=nO3wRANIIvnp3f0XwG2X3A%3d%3d however for that price, you might as well put a SD on a little PCB and then solder that in at a 90 degree. Something like this is nearly the same thing at nearly the same price. https://oceancontrols.com.au/MSD-005.html

Look in my "project_specific" KICAD library for "usb_u-mini-standard-header" Perhaps we can add the vertical USB to that footprint as well. I have not built this connector yet, but it should work for a variety of connectors. The only concern I've heard was about pick and place machinery. I choose only connectors that allowed for thru hole mounting as I feel that's a more rugged design. If something goes into qty production that would not be as friendly to low cost assembly as it could be. Other than that it should work fine.

Multi_USB.PNG (14.61 KiB) Viewed 79773 times

[AndreyB]

If we use a harness like this...

That's the wrong side of the USB - this this harness we would need to use a very unusual USB cable. But that's an option

[kb1gtt]

Ah, more like this you say http://www.datapro.net/products/usb-panel-mount-b-to-socket.html

I'd think that lower cost and mini / micro can be obtained with more than a minute of shopping around. Seems panel mount is the hot word for that search.

[AndreyB]

Please join me in celebrating first Frankenso traces!

http://i.imgur.com/sBV9muT.png

[puff]

if it's not too late - please, make a provision for a bunch of simple level shifters (for 5V ignition modules like mine or logic-level controlled COPs) + for lm1815 vr inputs?

[AndreyB]

This board will be using MAX9926 and TC4427 same as Frankenstein, the question of component availability in Russia would need to be addressed separately

[puff]

the first one is not available from Russia's major electronic parts suppliers (although LM1815 is not available either), the second one costs an arm and a leg (in comparison to BSS138 - a handful for a coin). there's still some space on the board, and there are free legs on the microcontroller… what about egt and lambda controllers? knock sensor circuit?
i guess it would be somewhat a premature release…

[kb1gtt]

For outputs, there are 6 channels of high / low drivers. They can be wired to drive 12V or 5V, as well they can sink when they are not driving. Those are generally intended for driving an external ignitor. The AN inputs are also capable of translating a 12V digital signal into a 3.3V level signal.

Can you post a suggest circuit or something like that? If time permits I may find a way to include it, but I'm also very interested in drawing a line in the sand and getting this to a certain level of completeness. Also worth noting, this is a modular design. There have been several efforts made to allow additions and changes of modules.

[Number-One]

Very good.

[Sudo]

I haven't looked at the boards thoroughly, but the USB differential pairs need some work. And I don't recommend having 2 USB connectors in series. Those extra stubs can cause reflections. Another important thing for diff.pair is parallelism. Although you are not using High Speed, it is still a good idea to follow these guidelines:

http://www.usb.org/developers/docs/hs_usb_pdg_r1_0.pdf
http://www.electronic-products-design.com/geek-area/electronics/pcb-design/general-pcb-design/pcb-track-impedance

[kb1gtt]

Good feed back, I'll read those PDF's. I understand the concerns about the stubs hanging out there like that for the secondary vertical mount USB connectors. I would expect only one to be populated at a time, so the cross talk and inductance/capacitance issues should be fairly small. However when dealing with MHz signals I can see how small variations can cause warp-age and potential issues. I don't know if the warp-age issues would cause the eye diagram to drop below the tolerance thresholds or not. I've seen some really dirty USB designs, so my gut feel is that it wold be tolerant of such issues. The reason for that funky USB is because we are looking for flexibility. I tend to like this approach for flexibility http://rusefi.com/forum/viewtopic.php?f=4&t=569&hilit=usb+harness&start=11 Those harnesses can allow for the flexibility we are looking for, but they are kind of pricey and I don't like adding extra connections that can fail. So I'm not sure what is a better approach.

Keep up the feedback and feel free to tear it apart. What doesn't kill it will only make it stronger. All I ask is that you try to be content focused, and so far it's good content, keep it coming.

[kb1gtt]

The PDF suggest using 7.5:7.5 while I was using 8.5:8.5. KICAD claims the impedance for 7.5 is like this

7.5-7.5_trace-impeadance.PNG (64.5 KiB) Viewed 80010 times

while the 8.5 is like this

8.5-8.5_trace-impeadance.PNG (59.54 KiB) Viewed 80010 times

To me they seem close enough. I'm assuming the ZOe and ZOo are some how even and odd, and I have no idea why they don't match 90 ohms. Should I go that extra step for 7.5? I use 8.5 as it's 6 mill at a 45 degree angle. The PCB machines often do that 6 mill in the X and Y, which often makes your angles 8.5. By using the 8.5 you can often avoid headaches with failed angle traces. I think I'll leave that at the 8.5 as it appears to be only a 1 ohm difference.

[AndreyB]

From http://forum.chibios.org/phpbb/viewtopic.php?f=8&t=1337&p=16429#p16429

[Sudo]

I normally use Saturn PCB Design tool. http://www.saturnpcb.com/pcb_toolkit.htm What you are looking for is Zdiff. The Kicad tool doesn't show the result for that. Why is your "H" value so high? Assuming we will use OSH, It should be 6.7 mils according to their stackup. http://support.oshpark.com/support/articles/122263-4-layer-board-stackup-and .

Both 8W:8S are within 10% tolerance according to the Saturn tool, more so than the 7.5W:7.5S. 9W:8S seems even better.

My concern wasn't the spacing or width, it was more so the parallelism and stubs. USB spec can tolerate some inconsistency in the routing, but I am just nick picking on it. At the end, if it works either way, it's all good to me.

[AndreyB]

Assuming we will use OSH

Probably not osh - osh would be ~$600 for three boards. Cannot afford this even for a test run

[Nobody]

Assuming we will use OSH

Probably not osh - osh would be ~$600 for three boards. Cannot afford this even for a test run

Try http://www.4pcb.com/66-each-pcbs/index.html?gclid=CNKVrNXjtL8CFUKCMgod9jwAaw , I've been very satisfied thus far and yes really a US company and not farmed out to China. They even toss in a bag of pop corn - no joke LOL. I've even got an extra board before.

[kb1gtt]

Didn't know the OSH park stackup, thanks. That's much better. Following the USB guidelines much better, I checked in a better revised layout. I also added a pile of breakout vias such that you can solder wires to pins if you have the discovery board installed. Also many thanks about the link to pcb trace calculator.

About the screw, the screw I took out of the case is .165 major dia across the threads. and .265 inch across the head. I have put in a drawing that marks the PCB such that you know the screw head clearance. The hole in the PCB is very loose.

About the spacing of the chips in the top right, I expect there will be a piece of alum above it which will function as a heat sink. I expect them to each generate up to 5watts. I don't know or trust the datasheet's 27C/W thermal resistance, as I expect heat to flow via heat sink/case not just the PCB copper. Off of memory I seem to recall 10C/W to 15C/W, but won't really know until it's tested.

Time for some ZZZZ's. Keep up the good feedback.

[AndreyB]

Try http://www.4pcb.com/66-each-pcbs/index.html?gclid=CNKVrNXjtL8CFUKCMgod9jwAaw

It would have been a great deal, but their $66 option is "Max. Size 30 sq. inches" and this board is 39 sq. inches

[kb1gtt]

Here's a picture that helps show the basic feature set found in this board.


I wanted to thank the folks who offered feedback. The feedback made the design that much better. To address the feedback that was provided by this forum thread, see below comments.
-- This design expects the use of the 90 degree USB connectors but has an optional standard USB connector. If the standard USB connector is used, there will be some "stubs", but they are less than the guides recommended max .2 inches long, so hopefully they would work good enough. As well parallelism is much better now. Attached prediction indicates very close to the suggested 90 ohm impedance.
-- Screw clearance checked and drawn on silk screen, as well a 100% scaled paper print out was done and checked in the enclosure. Several other clearance issues were avoided by this print out validation process.
-- Thermally I'm looking for an app note I once found that indicated if you put on a heat sink, you can get the Rth much lower. I'm working on creating a better answer, but that will take a bit. I expect we'll be OK thermally but we'll need / want a heat sink. The on-board thermocouples will allow us to measure the Rth as well we'll be able to use them for long term diagnostics and field thermal damage prevention.

Again thanks for the feedback, and please feel free to continue the feedback. I'm more than willing to add items to the TODO list or known issues list.

[kb1gtt]

Gotta love links, I mean, I don't know what you are talking about. That picture looks fine to me

[AndreyB]

That picture looks fine to me

What picture? No idea what you are talking about

[kb1gtt]

Now that we have crossed this line in the sand, how do we get GPS and accelerometer data into this board? I think those are two features that people will be looking for. What modules should we start considering. We could tap in the the SPI. Are there SPI GPS and SPI accelerometers we should consider? Please note we left via's for several IO on the STM board, so we can build an add-on card if we need to.

[kb1gtt]

This SPI GPS module seems to be reasonably low cost and read-ably available. http://www.mouser.com/ProductDetail/Maestro-Wireless-Solutions/A2200-A/?qs=sGAEpiMZZMsjLMBIknjmktYVBsqsIotyqEY%252bXiYuHqE%3d

802.11 can be obtained with this http://www.mouser.com/Search/ProductDetail.aspx?qs=Ad2LRJtGXDs1eLIWb4m46Q%3d%3d

[rus084]

why we need 802.11?

[AndreyB]

Now that we have crossed this line in the sand, how do we get GPS...

On the one hand we have integration with Ublox NEO-6M, and hopefully they all use the same text protocol since it's a standard. Not sure if binary protocol is the same but anyway, let's not get deeper into offtopic.

On the other hand, right now we are totally messed up: we have HIP sensor without any firmware support, we have LCD screen with minimal support, we also have a joystick and thermocouple input. Oh, and we have SD card with pretty lame logs.

What I am hinting is... We need software developers to breathe life into this hardware

[AndreyB]

Ops, do we not have W* jumpers in the VR input circuitry? If we have 2A-2B-2C-2D hard-wired, that would limit OEM harness applications.

[kb1gtt]

The lack of W jumpers for VR is due to voltage concerns.

802.11 and other such features are handy as you can connect to a laptop with out wires. Some people have 802.11, others have bluetooth.

I figure I can do the hardware thing much better than the software thing, might as well make it interesting for software people, so might as well give them extra options, and hopefully someone will develop any piece of the pie.

[AndreyB]

I figure I can do the hardware thing much better than the software thing, might as well make it interesting for software people, so might as well give them extra options, and hopefully someone will develop any piece of the pie.

There is a point in this

[kb1gtt]

About the spacing of the injector drivers and the expected heat, I couldn't find the spread sheet I've done before, so I re-created it here. https://daecu.googlecode.com/svn/Hardware/trunk/simulations/Injector/Injector_heat.xls

The short version is that at 9kRPM I expect under 3W of heat per drive chip. The datasheet specifies I should expect between 27C/W to the leads and to 90C/W if you use the PCB only as your heat sink with heat conducted via the leads. That datasheet also specifies the Tj max allowed temp of 150C, with a shutdown at 175C.

I once found an app note that specified that if you heat sink the the nonPCB side of the SO-8 package, instead of just relying on the Cu and bottom PCB, you can get well under that 27C/W specification. However I can't seem to find it now. I do expect we'll have a heat sink above the chip, so I expect we'll be able to at least reach that 27C/W spec. Off of memory I seem to recall that app note claimed the heat sink could get you down around 10C/W. In reality we don't know the exact make up of this chip, so we won't know the value until we build it and measure it. This is part of why I pushed for the thermocouples in the layout. For now I'll assume the 27C/W is what we'll be able to achieve, and the 90C/W is the worst we can expect with no heat sink.

Per my suggested environment noted here http://rusefi.com/wiki/index.php?title=Manual:Hardware#General_suggested_environment I expect an ambient up 50C. So we allow a 100C rise.

I expect "normal" operation to be more like what is shown in this graphic. Basically I expect a max RPM of about 7.5kRPM, and with no heat sink, it could be 2.5kRPM.

40V-12V-1ohm-100Crise-Rj25C.PNG (35.6 KiB) Viewed 30326 times

However if you really let it heat soak and you let the junction temp get up to that 150C, this is how it would look at that temperature. Effectively the max RPM drops to between 4.5kRPM and 1.5kRPM. So keep the ECU cool, if you expect to be full RPM for long periods of time.

40V-12V-2ohm-100Crise-Rj150C.PNG (36.06 KiB) Viewed 30326 times

[Nobody]

Just a side note - low clamp voltage in theory will limit have fast an injector closes.