For discussion of Speeduino compatible boards designed / built by other members of the forum and for guidance around making such a board
By fram
#48606
Hi Klotzy_550

-You can improve some traces, eg direct connection for 5V close to ICSP connector (and remove 1 via)
-Remove all copper antennas (maybe you have to change the setup for copper generation) they are noise catcher from HF point of view, and nice antennas from RF point of view. (maybe we can catch the bluetooth! ;) )

Also, at the end, you have to ground the ground :D , adding a lot of vias from top to bottom.
(Try to have a copper ground ring all over the board outline, if you can, with vias)
Attachments
k_500_2.JPG
k_500_2.JPG (274.09 KiB) Viewed 6037 times
k_500.JPG
k_500.JPG (236.36 KiB) Viewed 6037 times
By klotzy_550
#48609
Fram,

I tried to implement most of your recommendations to the best of my ability.
I wasn't able to find a great way with Eagle to reduce the little antennas, so I increased my ground plane polygon's isolate value, essentially this increases the isolation between the plane pour and components / traces. While it greatly reduced the little antennas, a few still exist. For the rest I grounded them out with vias. I then implemented a via ground ring around the perimeter where I could. Rerouted VCC, great catch by the way! Rerouted Tacho output to skirt along the high current components instead around the perimeter of the board which I will admit was pure laziness originally.

Also I changed out my IGBT driver to MCP1416T.

Image
Image

Thanks again for the help, I appreciate it.

-Maverick
Attachments
(126.74 KiB) Downloaded 239 times
By fram
#48628
Hi Maverick

Sounds better :)
Anyway, you have some ways of improvement:
-Change LDO package with a smaller one and improve the placement for PSU.
-Decrease copper size for PGND, on bottom right, and replace with GND copper to improve ground ring.
Attachments
BRcTkcg.png
BRcTkcg.png (68.01 KiB) Viewed 5930 times
By klotzy_550
#48635
fram wrote: Fri Feb 26, 2021 10:25 am Hi Maverick

Sounds better :)
Anyway, you have some ways of improvement:
-Change LDO package with a smaller one and improve the placement for PSU.
-Decrease copper size for PGND, on bottom right, and replace with GND copper to improve ground ring.
Fram,

This iteration includes the recommendations you suggested. I swapped out the LDO D2Pak with a SOT223. I really do not want to go much smaller than that to ensure good package thermal coefficients. The thermal resistance between the junction and case increased from 5C/W to 10.9C/W assuming I have enough thermal copper mass to sink to, which in this case I should have a pretty good mass. Junction to ambient remains roughly the same though, however this value shouldn't matter as much since it will end up potted. I rotated the electrolytic capacitor so the ground tab faces the ground ring. I was able to further reinforce the ground ring with this iteration. I also reduced the copper size around the IGBT, but not as much as your suggested. My reason for this is that I need some thermal mass to sink heat to from the IGBT. The ground ring currently meanders around this IGBT thermal plane. I did my best job to reduce little ground antennas, but a few still exist due to my pour isolation.

Attached are updated images of the board routings.

Image
Image

Thanks,


Maverick
Attachments
(126.6 KiB) Downloaded 253 times
By fram
#48641
Hi Maverick

As you said, you should be ok with SOT223 package for LDO, according to voltage drop , thermal copper mass and thermal resistance of the casing. (i measured a static current of 70mA for speeduino in this kind of small design, so not big deal for the LDO)
Ona last thing you can do is to shield the crystal with ground vias all over the cristal.

This design should work, the only thing i don't know will be the behaviour of the IGBT. (think about adding a top heatsink)
;)

BTW, you will not use any sensor interface IC for Cranck/Cam signals?
By klotzy_550
#48657
fram wrote: Sat Feb 27, 2021 9:36 am Hi Maverick

As you said, you should be ok with SOT223 package for LDO, according to voltage drop , thermal copper mass and thermal resistance of the casing. (i measured a static current of 70mA for speeduino in this kind of small design, so not big deal for the LDO)
Ona last thing you can do is to shield the crystal with ground vias all over the cristal.

This design should work, the only thing i don't know will be the behaviour of the IGBT. (think about adding a top heatsink)
;)

BTW, you will not use any sensor interface IC for Cranck/Cam signals?
Fram,

I added some ground vias around the crystal. The crystal housing itself is grounded to two of the four pads. You are correct in estimating around 70mA, with a very similar design I was consuming 0.5 - 0.6 mA according to my supply.

My only real concerns with the IGBT are that I am not placing any gate resistor or gate pulldown on the board. However, the IGBT has both already inside, a 110 ohm gate resistor and a 10k-30kohm pull down. Should be good, according to the datasheet, but it is a point of concern for me. A bigger concern is general noise from the IGBT causing the controller to lock up. I would like to do my best to avoid this at all cost. However, the way the circuit is designed, I can pull the IGBT off the board and run the gate drivers logic output directly to the main connector.

The crank and cam signals are intended to be logic level (open drain NPN). On my project I am only using a Hall sensor for a crank signal. I know this limits the type of sensors that can be used for motion detection, but I have accepted that.

Thanks again for all the help and suggestions, I will be sure to continue implementing your recommendations in future designs.

Updated design:
Image
Image


Thanks,



Maverick
By klotzy_550
#48945
Update,

Received my PCB's from Oshpark and assembled one. Turned out really good. I did end up finding one issue with the routing and not the PCB Mfg. On my header I had listed pin 6 as a 5vcc supply. However, Eagle did not include this in the 5V0 net. However, if you click show on the entire 5V0 net the pin 6 pad lights up, just doesn't have an air wire to it. I have since fixed this issue in ECAD. For this board I just ran a 5V0 wire from point to point to fix the issue. Other than that last evening it accepted a bootloader. If time allows I will try to flash Speeduino onto it. In the meantime, I am attaching some images of the board assembled.

-Maverick

Image
Image
Image
Image
User avatar
By PSIG
#48952
Image
A suitable case and it's a win for testing. I want one for my key chain and another for a trans controller. :lol:
By klotzy_550
#49051
fram wrote: Fri Mar 12, 2021 8:22 am Nice device! ;)

2 things:
- try to clean the PCB and remove flux residues as much as possible. (Isopropyl alcohol)
- think about adding 2/3 dots of hot glue or silicone at the edge of electrolytic cap and PCB and.
Thanks for the Tips Fram, I did end up cleaning it up with Iso Alcohol. I did use some hot glue to secure the aux data lines then silicone conformal coated the PCB. Now it is potted so everything should be adequately held in place.

-Maverick

you could set 'skip revolutions' to 3, then it wou[…]

Hi, I am trying to assign Signed values to the x-a[…]

blitzbox

I've finally figured out why MAP and Lambda weren'[…]

Hello, I bring news!! Let me tell you that after […]

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