Another SMD lebowski PCB

bobc

10 kW
Joined
Jan 20, 2011
Messages
993
Location
Knutsford England
I couldn't find anything that did just what I wanted so I spent a few hours this week on kicad
It is a "re-imagining" of lebowski's controller board using SMD components, Other features are
1) all "power" stuff is assumed to be on the associated power board; that's current sensors, gate drivers and some feedback resistor strings.
2) all interface ports from the original board are retained - changes are 3 pin RS232 plug and a simple CAN circuit to allow daisy-chaining.
note version1.2 of the board (latest) has reverted to the original rs232 4 way pinout
3) the power supply and "drive" leds from the original are retained
4) a single 24way connection to the power board also supports the board. On the vehicle it will have a 3d printed cover on the inverter to further support the PCB and provide strain relief for throttle and hall interfaces.
5) main power supply is only OK up to 100V. On a bigger battery either horrid volt dropping or an external other 15V supply can be wired in - either would be on the power board
6) the board is about 1" x 2" on 2 layer PCB (35mm x 53mm)
7) power supplies, the 15V and 5V both have about 1/2 an amp spare.
8) this board will fit, unchanged to any current or voltage power module.
the layout is shown below
View attachment 3
schematic is attatched
I'll get a few made by one of the chinese PBC places
In the fullness of time a power board design will appear on here at about 10kW for an emax scooter upgrade using that ixys 6-pack; I have a prototype spinning nicely now using the old PTH lebowski control board. This board will plug onto that, it is way smaller than just one of the DC caps on the other board!
If it works OK I'll obviously bung the gerbers on ES into this post for the community
 

Attachments

  • buildmap12.pdf
    40.6 KB · Views: 332
  • smdcontrollerBOM.xls
    11.5 KB · Views: 236
PCB quotes down to 77p each for 10 off (that's about a dollar) Not bad for prototypes!
Order submitted - paid £2 each in the end to get faster delivery :)
 
graphic screen dump has made signal tracks look narrower - tracks/gaps are 0.25mm min, pretty narrow :)
Note schematic & BOM have had a couple of edits - V1.1 has the 5V PSU set to 5.1V
V1.2 has a change to the PSU supervisor/reset circuit to enable subsequent re-flashing
 
FWIW I started on an associated power module.
assocpow.GIF
That's the ixys 6pack(mti145wx100gd) so targtting 80V or so and towards 100A
Still waiting for the first PCBs.....
 
So you went with the one power board with sensors on that, rather than power board, sensor board and brain board. Does the powerboard still have the mounts for the water cooling heatsink?
 
Hey Andy
This board IS usable with external current sensors - simply don't fit the ACS jobs, replace with 3 links; the 3 small connections are ready to wire off to external sensors. This board design is just to get an idea of the scale of things, I'll do a 3D render to look at heatsink, housings etc. There's inrush (powerup) controls to go on there still - will use external relay for that & an external main fuse. In general though, looks feasible and fairly compact. No not thought about water cooled version at all :)
Here's a render of the power PCB with the control PCB in place on an air heatsink. The yellow (hex) pillars are the motor terminals.
assembly.jpg
assocpow.GIF
 
In the words of Montgomery Burns "excellent"

Sent from my ALE-L02 using Tapatalk
 
bobc said:
Hey Andy
This board IS usable with external current sensors - simply don't fit the ACS jobs, replace with 3 links; the 3 small connections are ready to wire off to external sensors. This board design is just to get an idea of the scale of things, I'll do a 3D render to look at heatsink, housings etc. There's inrush (powerup) controls to go on there still - will use external relay for that & an external main fuse. In general though, looks feasible and fairly compact. No not thought about water cooled version at all :)
Here's a render of the power PCB with the control PCB in place on an air heatsink. The yellow (hex) pillars are the motor terminals.
View attachment 1

Have you considered adding a groundplane (for the return of the digital currents) around the area where it connects to the processor board ? Maybe some wider than minimum lines ?
 
Thanks Bas,
yes = some of the routeing on the power board is less than optimal and will be altered before I spend money on it; keep notes and suggestions coming!!!
The gate drive signal route as well as those analog tracks off to the current sensors might be re=routed too (experimenting with the "push aside" router on kicad, a "view" selection)

I might look at a "little" one too, say using D2paks or smaller SMD discrete MOSFETs - should be able to make that, with this control PCB, into a nice little "cube" of power, round 2 - 3 kW. No - I should finish one thing before starting the next..... :)

Fingers crossed, I'll get one of the 15V switchers going today, ordered some 16301s to do the 5V supply at the weekend.
Bob
 
Happy to do the work on the box but the stl you sent is not quite right. Would like to have a go myself at a design but need a model of the internals or a built up powerboard and some Vernier calipers :)
 
If I read that right you haven't shelled out for the pcbs of the power board yet, I think we should have a look together before you hit go on that order from the perspective that there may be better compromises on layout to get a better fitting box which doesn't hurt the electrical design. who doesn't want a better fitting box?
 
The 15V supplies are outputting 16V (so quite close...) from 18V upwards. OK up to 35V unloaded (untested higher).
(note a 3MEG in the R42 position would bring it down to dead on 15 if I can be bothered....)
DSC01706.JPG
Andy I'll run off a "step" file of the whole assembly, see if that's any more usable.... Should be reasonable size.
I don't want to alter parts placement, but the routeing is definitely up for adjustment!
 
For the power board, why do the hex standoffs need to be so long, 20mm makes them have a big moment about the pcb if the wires are yanked and makes the box less tidy, could we use shorter ones? also the rear hex standoffs near the caps are way to close to make a decent wall for the printed box to slip on so can we space them a bit further from the caps?

And the upright control board could also benefit from being further from the caps. If that board came forward a touch we could make some nice circular cap covers. Plus we should think about better water proofing that those holes for sliding the pins up through the box. I have a few ideas which i will make into a step file and send back to you.
 
smd lebowski powerboard revision.png I was thinking along these lines. Cause if we are going to print it we can be a bit creative and not be restricted to things that needed to come out of a mould. The tops would be covered obviously and it would be printed in bits and could be glued or bolted. I figure the lid for the brain section could be removable but you could have a single waterproof gland coming out of it for the connections rather than those weird slits which allow it to be slid on but also allow water in. I mean with ribbon cable you could lie it flat to get the construction to be flatter.
 
All good points. reason for 20mm hex pillars? we have a stack of them at work that we'll never use - yeah 10mm would be better, lifts the cover clear of the current sensors (tallest things on there)
I'm now thinking maybe lay the control board flat - makes plugging stuff in & viewing LEDs easier. I'd leave average 5mm or so space round the Al caps - they do get heated by ripple currents.
More SMD control board prototype buildup over the next few days, I have most of the bits now. The 1210 inductor for the 5V switcher is good for 310mA if memory serves so that's my 5V power budget; I'll get that going next. When the 5V is right I can start to build up the microcontroller circuits. Hand building SMD stuff with an iron is no big deal, you need a good fine(ish) iron, good tweezers & I need a good magnifier now.
I typically dump some solder on just one of the PCB pads for a device, then place the device with tweezers in my left hand and the iron in my right. Make sure the part is reasonably aligned and has "sat right down" on the PCB. After that it's pretty straightforward to solder all the other pins/pads. The "thermal pad" under the LM5017 is a special case; I put a big hole in this board so I can solder it with an iron, usually such a device would need to be "reflowed" on a production line or vapour phase reflow oven. The 3 boards pictured above all seem to have gone on OK.
(I'll open a thread specific to the power board when I get to it, it was only really mentioned as an example application..... :)
If you're building one: same advice as lebowski on his build thread really: make sure the 5V supply is right before you solder on the micro, that's about it really. I'd make the 5V work (apparently you actually want 5.1V for reliable parameter storage) adjust by adding R39, probably in the several Mohm region.
Then make the 15V supply work. Again a trim resistor placement is available R42, and again it will be in the Mega-ohms.
An approximate calculation for R39 is 500k/Verror (where Verror is 5V supply voltage - 5.1)
An approximate calculation for R42 is 3.1million/Verror (where Verror is 15V supply voltage - 15)
example - you've built the 5V supply with no R39 and measure its output voltage as 5.18V Verror is 0.08V so R39 should be 500000/0.08 = 6.25Mohm (use 6.8M)
 
5V supplies (OK 5.1V) are built but not yet tested. A minor build issue became apparent, I had not left enough width for the 47uH inductor in the 5V switcher - I've upgraded the layout to resolve in future builds. meanwhile on this board things are together but a bit squashed :) I'll fit the last few bits & the LEDs tomorrow & see if it powers up right.
 
amberwolf said:
Now if you could just get them pre-stuffed/soldered with all the parts for that price.... ;)

Awesome thread....

As to Amberwolf's comment:

$20 for a solder paste mask (laser cut vuegraph in selected weight)
Solder Paste
Tweezers
Toaster Oven from Good Will

A little patience...

Sell the 6 boards that are left after hay wires, testing, and your project.
And - that's about right.

For beefing up traces that handle power - our sole source is Advanced Laser
Just convert the traces you want over to DXF
Spec your material and thickness
Bake it on before any of the sensitive stuff

(Note... more is less ... go thin or risk baking your silk screen all to hell)

-methods
 
I'd like to patch any "holes" in the system & I note there is no interlock from the 15V supply. If this is working badly and (say) only generating 7.5V, the 5V supply will be right as rain and the CPU none the wiser, so the power stack will run with insufficient gate voltage & set fire to things.... The gate drivers have a UV lockout which I guess acts as a backstop..
Anyway, I could run a signal from the 15V LED into the 5V switcher ENable pin - that should cover this hole - what do we think? If the 15V goes down by 2V the 5V supply will disappear - should work? Just a couple of SMD Rs needed. Worth the bother?
 
bobc said:
I'd like to patch any "holes" in the system & I note there is no interlock from the 15V supply. If this is working badly and (say) only generating 7.5V, the 5V supply will be right as rain and the CPU none the wiser, so the power stack will run with insufficient gate voltage & set fire to things.... The gate drivers have a UV lockout which I guess acts as a backstop..
Anyway, I could run a signal from the 15V LED into the 5V switcher ENable pin - that should cover this hole - what do we think? If the 15V goes down by 2V the 5V supply will disappear - should work? Just a couple of SMD Rs needed. Worth the bother?

Would it be possible to design something like that such that you can choose when populating the PCB ? Either you put these components in and don't have this functionality, or you put those components in and you do ?
 
Lebowski said:
bobc said:
Worth the bother?

Would it be possible to design something like that such that you can choose when populating the PCB ? Either you put these components in and don't have this functionality, or you put those components in and you do ?
Yeah, easy - I'm still debating with myself if it's worth it.
In my mind, if the drive stops working, that's no big deal - get a tow home. But if the drive bursts into flames it is a big deal. So only switching on when you know you can switch on properly seems important to me.
Locking out the gate drive enable with the inrush relay would also fall under this category (reminder of previous topic) :)
Doing some background reading, the UV lockout on the integrated drivers (IRSxxxx) is in the 8.5V region and is there for both upper and lower channels. 8,5V on a gate turns on most mosfets pretty well; I think I'll give this a miss.
PS operation of the gate driver UV lockout will cause a large current loop error which should raise a fault in the control chip.
 
Building up one of the circuits now. I was testing the 5V supply by back driving the 15V from a bench supply & this seems to have done in 2 of the 3 15V supplies. So when anyone else does it... build and test the 5.1V supply first then do the 15v....
I hope I'll get as far as the rs232 setup screens this week but I can't promise (then I'm away from home for a fortnight)...

PS - now got to order some spare lm5017s.....
 
Back
Top