Speeduino is now on Github Sponsors (Rather than Patreon): https://github.com/sponsors/noisymime
For discussion of Speeduino compatible boards designed / built by other members of the forum and for guidance around making such a board
By klotzy_550
#48562
Hi guys,

I am making a new ecu that is going onto my three wheeler project. I would really appreciate it if a few of you could take a look at my schematic and routings. Criticism and questions are welcome, I am by no means an expert in this arena.

The area that I have the most concern around are the IGBT (FGD3050G2), IGBT Driver (IX4310T), Crankshaft / Camshaft Inputs, Magnetic reset (DRV5021A3), and general routing. I have three main grounds on my system: AGND, DGND, and PGND. DGND and AGND short internally at a precise location and DGND and PGND short together at a common chassis star ground. My concern with the IGBT driver is if it will be quick enough to react in a timely manner. My engine at most will spin 9000rpm down hill with the wind to its back. With a 36-1 encoder I am looking at around 18.51us / degree. I think the turn off time will be more than sufficient since it is nano seconds for the turn off time which really matters for acurracy. The IGBT's turn off time is 15us worst case. +/- 1deg should be acceptable correct? Will the rise time for the IGBT driver be a concern? Its metrics state 15ns worst case @ 1nF, however the selected IGBT has 22nF of gate capacitance. From what I have seen this will cause the rise time to increase, however even if it rose by a factor 100, that would still be acceptable since it is the falling edge that is important, correct? I have used the magnetic reset in the past with mixed results, not sure if it is really worth it since you need to have a pretty strong magnet. On top of this you still need to time the reset pulse from arduino just right to actually flash the controller. I have also just accomplished this by turning the ignition key off / on at the right time. Do you think with the onboard IGBT and during higher switched current operations I could induce a magnetic field strong enough to trigger a reset? Opinions would be appreciated.

Thanks again for the time and effort,


klotzy_550

Image
edit: routed the small Airwire on the ATMEGA2560's PIN No: 2
Image
Link to SharePoint hosted PDF Schematichttps://550performancecom-my.sharepoint ... g?e=tRabb8
Last edited by klotzy_550 on Tue Feb 23, 2021 3:05 pm, edited 1 time in total.
User avatar
By PSIG
#48572
Good to see you! I have to run so must be short - but cool project. 8-)

All random: Can't say about most of it, as IGBT depends on coil and required power, inductance, etc, and the driver depends on the IGBT (though most have a dedicated cap especially when regulated supply is limited), don't know your CK/CM sensors, etc.

Reset should not be necessary, as I described elsewhere for triggering reset for upload, and now included in code to enter "U", so next characters initiate a reset (when the upload starts). Works with Bluetooth and maybe others.

IGBT rise/fall time is not random, so may be compensated at high-Rs if it drifts. More reason for the dedicated driver cap. Without reset switch needed, the flyback EMR is not a big concern (for that). I'll try to look closer when I get some time to myself. :roll: Have fun!
By fram
#48582
Hi

Before any "deep" investigation of your schematic and PCB design, 2 main feedback:
have three main grounds on my system: AGND, DGND, and PGND. DGND and AGND short internally at a precise location and DGND and PGND short together at a common chassis star ground.
1/I'm afraid to say it's not a good idea. You ought to have a single ground plane for all the sources, making separation between grounds may lead to hazardous potential difference between the separated planes, creating ground loop with noise issue, and big return ground path with "high impedance". That is not good.

Except for some special design, the best ground is a single and full ground layer with the shortest and the lowest impedance for ground connections of the components.

2/ 45° orientation for Mega 2560 is nice, but i's a waste of space, and not the best choice to have a good placement especially for small PCB board. Go back to "regular" 90° orientation for the 2560 and work again your placement.

;)
By klotzy_550
#48584
PSIG wrote:
Tue Feb 23, 2021 2:56 am
Good to see you! I have to run so must be short - but cool project. 8-)

All random: Can't say about most of it, as IGBT depends on coil and required power, inductance, etc, and the driver depends on the IGBT (though most have a dedicated cap especially when regulated supply is limited), don't know your CK/CM sensors, etc.

Reset should not be necessary, as I described elsewhere for triggering reset for upload, and now included in code to enter "U", so next characters initiate a reset (when the upload starts). Works with Bluetooth and maybe others.

IGBT rise/fall time is not random, so may be compensated at high-Rs if it drifts. More reason for the dedicated driver cap. Without reset switch needed, the flyback EMR is not a big concern (for that). I'll try to look closer when I get some time to myself. :roll: Have fun!
David,

These are the measurements I captured from my coil. Primary resistance: 2.9 Ohm, primary inductance: 6.3mH, Normal running dwell: ~4ms, cranking dwell: ~8ms. Using this information I was able to calculate my coil energy to be 38.5 mJ when running and 53.0 mJ while cranking. Ideally the 300mJ that the IGBT is rated for should be sufficient, correct?

I am not currently running a Cam sensor, just providing a provision for it, the crank sensor is a hall sensor similar to the ZF cherry made by honeywell. Followed their recommendations for pull up resistor, I added a low pass cap to filter some unwanted higher frequency noise.

Regarding the reset command, if I send a "U" to the controller, it should reset programmatically? Currently I talk to this via a FTDI interface, but this would be nice to know!

Thanks for the input!
By klotzy_550
#48585
fram wrote:
Tue Feb 23, 2021 9:16 am
Hi

Before any "deep" investigation of your schematic and PCB design, 2 main feedback:
have three main grounds on my system: AGND, DGND, and PGND. DGND and AGND short internally at a precise location and DGND and PGND short together at a common chassis star ground.
1/I'm afraid to say it's not a good idea. You ought to have a single ground plane for all the sources, making separation between grounds may lead to hazardous potential difference between the separated planes, creating ground loop with noise issue, and big return ground path with "high impedance". That is not good.

Except for some special design, the best ground is a single and full ground layer with the shortest and the lowest impedance for ground connections of the components.

2/ 45° orientation for Mega 2560 is nice, but i's a waste of space, and not the best choice to have a good placement especially for small PCB board. Go back to "regular" 90° orientation for the 2560 and work again your placement.

;)
What would you recommend, just a continuous ground plane with predefined "slits" to best direct current flow? I guess my biggest concern is noise from the onboard IGBT causing issues with the rest of the system like brown outs or system resets.

I used this document as reference for deciding what to implement regarding the planes: https://www.analog.com/en/analog-dialog ... nded.html#

Doesn't say one way is the best, but did seem like the author did prefer a ground plane for mixed signal use.

Regarding the orientation of the 2560, your point is definitely valid, however I do like this approach and it is driven more by function than aesthetics. Yes by placing it at 45° results in an X-Y box of 14.14mm instead of the 10mm that it normally would occupy. However, this angular placement positions the analog inputs, crystal, and digital outputs in pretty ideal locations. These locations result in shorter trace runs. The orientation also space claims this area to promote more room for routing and a continuous ground plane. I am also land locked by physical size and connector pinout to be drop in compatible with my previous designs driven by an OEM design.

I cannot recall if you had any boards with an IGBT on board or not, but if you have , would you have concerns about EMR on a shared ground plane? Also with the IGBT, it has an internal gate and gate pull down resistor. I did not place any external ones, that should not be a concern correct?

I appreciate your feedback, likely will convert my AGND Plane to DGND or short them at the 2560 pins, not sure yet. I am nervous about the PGND plane though, I think converting it to DGND and then inserting a slit in the ground plane to prevent interference in the other circuits may be the best approach, thoughts?

Thank you for the input, it is much appreciated!
By klotzy_550
#48586
Updated layout:
Image
Image

Everything shares DGND plane, schematic still has aesthetic labels for AGND though. The plane is naturally "split" segregating high current and analog signals which should help mitigate induced noise.

Rerouted the VSS signal just to look cleaner.

Secondary plane for IGBT Collector as a heat sink.

Link to SharePoint hosted PDF Schematic: https://550performancecom-my.sharepoint ... g?e=tRabb8
Last edited by klotzy_550 on Tue Feb 23, 2021 9:32 pm, edited 1 time in total.
User avatar
By PSIG
#48598
klotzy_550 wrote:
Tue Feb 23, 2021 1:20 pm
Regarding the reset command, if I send a "U" to the controller, it should reset programmatically? Currently I talk to this via a FTDI interface, but this would be nice to know!
Follow this if you have any UART, or read the whole page if you want to update your 16U2 for reset command.
By klotzy_550
#48599
PSIG wrote:
Wed Feb 24, 2021 2:54 am
klotzy_550 wrote:
Tue Feb 23, 2021 1:20 pm
Regarding the reset command, if I send a "U" to the controller, it should reset programmatically? Currently I talk to this via a FTDI interface, but this would be nice to know!
Follow this if you have any UART, or read the whole page if you want to update your 16U2 for reset command.
I was aware of the 16U2 custom bootloader, but not the control pin / U method. I will add a wire to my RST circuit! That is actually pretty awesome. Thanks for pointing it out.

-Maverick

Great, I might buy one now

Simple diagram for 2ch Bosch 200 ignitor

CLT sensor wrong reading

Once again as above. You need to find and fix you[…]

Kawasaki Concours 2.0

Going bonkers trying to wire up a Bosch LSU4.9 to […]

Still can't find what you're looking for?