Author Topic: 250V 10A + MCU PCB layout  (Read 839 times)

0 Members and 1 Guest are viewing this topic.

Offline oschonrockTopic starter

  • Regular Contributor
  • *
  • Posts: 67
  • Country: gb
250V 10A + MCU PCB layout
« on: August 23, 2020, 11:16:22 am »
Wanted to share a half finished draft of a PCB layout I am working on for this project:
https://www.eevblog.com/forum/projects/mains-switching-research-break-out/msg3170280

I want to focus on the PCB layout and system issues here, not the application / schematic (current HV part of schematic is attached), which has been discussed elsewhere. Suffice to say the application is to switch 110-230VAC  @ 10A (peak and more typically 3-4A) using power MOSFETs triggered by an MCU. The approximate anticipated power dissipation is 20-30W max (basically 2x Ron x I^2 - because only 2 of 4 MOSFETs conduct at one time + switching losses up to 10-20kHz).

Mains voltage, so a secure, servicable enclosure required (intended usage is in a power electronics Lab). I spent some time researching this, since I am not super experienced at choosing cases fort this sort of thing. In the end I chose this Hammond heat dissipating extruded case (bit expensive) -- better ideas welcome!
https://www.hammfg.com/files/parts/pdf/1455NHD1201BK.pdf

I also spent some time experimenting with designs around SMT (probably D2PAK) MOSFETs vs TO220 style packages. I ended up concluding that for this sort of power, trying to get rid of the heat through the board with thermal vias etc, was not a great idea. So I went for TO220FP style package (the plastic wrapped 2.5kV isolated tab), and bolting them to the floor of that case on the provided channels. The "FP" package has a higher termal resistance, but avoids need for silpad/laird. So the TO220FP MOSFETs are below the board with holes in board to access the mounting bolts.  -- if there are good ways to do this SMT, please let me know.

I kept the HV side all THT and the LV side will be ~100% SMT for compactness. Power entry and exit is via Bugin PCB IEC connectors (male and female pair -- note that the 3d model for the near side / female one is wrong (not available) in the screenshots, the footprint is correct).

Other considerations were:
  • Carrying the 10A - I went for "zonal island" style tracks for the high current areas - I did some calcs and I am anticipating less than 1W of heat in all high current tracks combined at 10A. That's at 1oz copper; no need to go to 2oz?
  • Creepage / clearance -- Weakest link on AC side are the TO220 legs / pads -- but we have isolating slots in board to protect LV side of the opto isolators/SMPS/DC/DCconverters
  • Initially considered having a separate 2nd board for MCU, but decided I have space and might actually have less noise on signals with short tracks vs interboard cables. I "may" still have a tiny daughter board in the top of the case to mount the control elements (rotary encoder POTs / switches + indicators, possibly an LCD - but concerned about noise on the board-to-board interconnect)

The SMT LV MCU side is not done/shown. This is the "easy" part for me. This is more about the HV side, layout, the isolation and system / heat choices I made.

Screenshots / schematic below

Any comments much appreciated. Any obvious blunders? Thanks
« Last Edit: August 23, 2020, 08:34:12 pm by oschonrock »
 

Offline uer166

  • Super Contributor
  • ***
  • Posts: 1026
  • Country: us
Re: 250V 10A + MCU PCB layout
« Reply #1 on: August 23, 2020, 07:19:30 pm »
Looks like you're shorting out phase and neutral on J1 if you were to turn all the FETs on, I can't see how that is safe if intentional.
 

Offline oschonrockTopic starter

  • Regular Contributor
  • *
  • Posts: 67
  • Country: gb
Re: 250V 10A + MCU PCB layout
« Reply #2 on: August 23, 2020, 07:23:56 pm »
Looks like you're shorting out phase and neutral on J1 if you were to turn all the FETs on, I can't see how that is safe if intentional.

Thanks.

Yup, that's intentional. Necessary for flywheeling of inductive loads.

And you're right that it requires care. It is the same problem as "shoot-through" on a half/full bridge motor driver.

I have been debating if I should interlock this in hardware, eg drive the LV side of the 2 optocouplers with a single signal and its (hardware) inversion.   

This may end up being the final solution, but I lose control of the precise timing then.



« Last Edit: August 23, 2020, 08:29:05 pm by oschonrock »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf