jmbw
Newbie level 3
- Joined
- Feb 7, 2012
- Messages
- 3
- Helped
- 0
- Reputation
- 0
- Reaction score
- 0
- Trophy points
- 1,281
- Location
- Monson, MA, USA
- Activity points
- 1,331
This has probably been done to death but I got nowhere searching the archives...
I'm trying to build my first off-line SMPS (+24@4A, +5@2A, -8@.2A, universal input) and I want to do it as safely as possible. If this sees the light of day it'll be super low volume (like a dozen total sold) so getting UL certification is out of the question, but I'd like to design it *as if* it were going to pass.
I'm too stupid to fully understand the actual design (my brain is digital-only) so I'm relying on the MC34262 data sheet examples for the PFC part and PowerInt's software-generated schematics for the actual circuit and design of the transformers (separate PWM for +24 and +5 so they're both regulated). But the safety/RFI aspects are all greek to me, and what I've seen of the UL language is very hard to understand. So I'm trying to match it up with what I've noticed in actual commercial SMPSes that I've opened up:
- The PCBs seem to normally be single-sided (with wire jumpers -- 1970s all over again!) and made out of something other than our old friend FR4. Is this for safety/reliability reasons (dielectric strength of .062" board too low for tracks to cross?), or is it just to cut costs? I got my design to route with just one jumper but some of the nets had to get a bit long, which makes me uncomfortable (it would have been nice to send them over the top, but it seems as if it can't be a coincidence that commercial SMPSes avoid that).
- Somewhere I read that you need a min of 4mm spacing between the primary side tracks and AC earth, and 8mm spacing between pri and sec. Sounds reasonable and of course I have I have a nice obvious 8mm stripe of nothingness across the board under the xfmrs/Y-caps/optoisolators (although I can't believe that simply bending the leads on a DIP-4 optoisolator out to .4" spacing really counts, when the package is still the same size so its own creepage is unchanged), but what about spacing between tracks within the primary-side circuit? If I understand the online PCB trace calculators correctly, 50 mils is enough spacing between tracks with a 380VDC difference on a single-sided PCB with solder mask (right???) and I guess that's just about believable but I'd like to be sure. It sort of seems like no big deal -- if the PFC stuff shorts out it'll just fry components and/or blow the fuse but that's much less of a big deal than zapping the user.
- What about grounding? Since the pri and sec are totally isolated (except for Y caps) it's obviously not cool to tie *both* the AC earth and the DC ground to the case (or in this case to the mounting holes -- it's an open-air PCB that will mount in a user-supplied case), but it seems like *one* of them should be grounded to the case.
- Not relevant to this project but I'm curious for the future: does an offline SMPS always have to be on a separate PCB from the rest of the device? This seems to be true in stuff that I've disassembled, but I don't see why it has to be, as long as proper clearance/creepage from the primary side is observed everywhere. In that case is it OK for the non-SMPS part of the PCB to be double-sided? (If it wasn't already OK for the SMPS.)
Thanks!
John Wilson
D Bit
I'm trying to build my first off-line SMPS (+24@4A, +5@2A, -8@.2A, universal input) and I want to do it as safely as possible. If this sees the light of day it'll be super low volume (like a dozen total sold) so getting UL certification is out of the question, but I'd like to design it *as if* it were going to pass.
I'm too stupid to fully understand the actual design (my brain is digital-only) so I'm relying on the MC34262 data sheet examples for the PFC part and PowerInt's software-generated schematics for the actual circuit and design of the transformers (separate PWM for +24 and +5 so they're both regulated). But the safety/RFI aspects are all greek to me, and what I've seen of the UL language is very hard to understand. So I'm trying to match it up with what I've noticed in actual commercial SMPSes that I've opened up:
- The PCBs seem to normally be single-sided (with wire jumpers -- 1970s all over again!) and made out of something other than our old friend FR4. Is this for safety/reliability reasons (dielectric strength of .062" board too low for tracks to cross?), or is it just to cut costs? I got my design to route with just one jumper but some of the nets had to get a bit long, which makes me uncomfortable (it would have been nice to send them over the top, but it seems as if it can't be a coincidence that commercial SMPSes avoid that).
- Somewhere I read that you need a min of 4mm spacing between the primary side tracks and AC earth, and 8mm spacing between pri and sec. Sounds reasonable and of course I have I have a nice obvious 8mm stripe of nothingness across the board under the xfmrs/Y-caps/optoisolators (although I can't believe that simply bending the leads on a DIP-4 optoisolator out to .4" spacing really counts, when the package is still the same size so its own creepage is unchanged), but what about spacing between tracks within the primary-side circuit? If I understand the online PCB trace calculators correctly, 50 mils is enough spacing between tracks with a 380VDC difference on a single-sided PCB with solder mask (right???) and I guess that's just about believable but I'd like to be sure. It sort of seems like no big deal -- if the PFC stuff shorts out it'll just fry components and/or blow the fuse but that's much less of a big deal than zapping the user.
- What about grounding? Since the pri and sec are totally isolated (except for Y caps) it's obviously not cool to tie *both* the AC earth and the DC ground to the case (or in this case to the mounting holes -- it's an open-air PCB that will mount in a user-supplied case), but it seems like *one* of them should be grounded to the case.
- Not relevant to this project but I'm curious for the future: does an offline SMPS always have to be on a separate PCB from the rest of the device? This seems to be true in stuff that I've disassembled, but I don't see why it has to be, as long as proper clearance/creepage from the primary side is observed everywhere. In that case is it OK for the non-SMPS part of the PCB to be double-sided? (If it wasn't already OK for the SMPS.)
Thanks!
John Wilson
D Bit