36 Pin QFN IC prototyping

Thread Starter

gautam023

Joined Apr 26, 2021
19
Hello everyone,

Does anyone have an idea of where I can buy a 36 pins QFN-DIP adapter with 2 exposed bottom pads? The pinout of the IC is attached below. Closest I could find was a 38 PIN QFN-DIP adapter and that is not working.




Screenshot 2021-06-21 173145.pngScreenshot 2021-06-21 173145.png
 

Irving

Joined Jan 30, 2016
2,308
Why not make one? easy to do and cheap.... I did my own TSSOP 20/30/40 0.5mm/0.65mm pad spacing, 0.3/0.4/0.6 adapters - cost me a couple hours of time and under $25 from a chinese pcb shop...
 

ErnieM

Joined Apr 24, 2011
8,196
Check eBay but the shipping time will be long. Sorry I'm on vacation and all my searches come back in spanish so I can't help search.
 

Thread Starter

gautam023

Joined Apr 26, 2021
19
Did you find anything? What IC is this for?
Did not find any pre-made adapters. Took your suggestion and designed my own. The IC is LT3958 and it is a power converter IC that can be used as Boost, Flyback, and SEPIC. I am using it in SEPIC mode to get +15V,-15V, and +5V from a 72V battery pack
 

Attachments

Irving

Joined Jan 30, 2016
2,308
Looks good, but I suggest you extend pads 37 & 38 as per datasheet to minimise track inductances. I'd make it a smaller DIP and consolidate pins eg 1,2,4,23, 36,35,34 are all NC and connected to SGND so don't need to be brought out to individual DIP-pins, similarly for SW (38), 8, 9, 10, 20, 21 and the GND pins 12 - 17... , but may need a couple of pins depending on current to/from external components...

Looking at the SEPIC circuit you actually only need to bring out nominally 8 pins from the carrier, Vin (27), SGND, RT(33) , SS (32), VC (30), FBX (31), GND, SW - I'd put the Vcc bypass capacitor on the carrier from pin 28 to SGND and extend SGND and SW on the back of the carrier to provide additional heatsinking

1624453573041.png
 
Last edited:

Thread Starter

gautam023

Joined Apr 26, 2021
19
Looks good, but I suggest you extend pads 37 & 38 as per datasheet to minimise track inductances. I'd make it a smaller DIP and consolidate pins eg 1,2,4,23, 36,35,34 are all NC and connected to SGND so don't need to be brought out to individual DIP-pins, similarly for SW (38), 8, 9, 10, 20, 21 and the GND pins 12 - 17... , but may need a couple of pins depending on current to/from external components...

Looking at the SEPIC circuit you actually only need to bring out nominally 8 pins from the carrier, Vin (27), SGND, RT(33) , SS (32), VC (30), FBX (31), GND, SW - I'd put the Vcc bypass capacitor on the carrier from pin 28 to SGND and extend SGND and SW on the back of the carrier to provide additional heatsinking

View attachment 241979
Left it that way because it will be easier to orient and solder the IC onto the board. I don't have any hot air guns or Reflow machines and need to do them by hand for the prototypes at least. So having those pins separate will help with alignment.
 

BobaMosfet

Joined Jul 1, 2009
1,896

Irving

Joined Jan 30, 2016
2,308
OK, but I think you'll have problems with those long thin tracks...
You can still retian the individual lands, but join them up to each other, e.g.

1624465522680.png
 

Irving

Joined Jan 30, 2016
2,308
This is better than my design. Thank you for your help
You're welcome. It's not perfect, I still don't like the long runs. Try turning the QFN 45 or 90degrees anticlockwise so the individual pins end up on both sides at top and SW and GND end up on opposite sides at bottom..
 

Thread Starter

gautam023

Joined Apr 26, 2021
19
You're welcome. It's not perfect, I still don't like the long runs. Try turning the QFN 45 or 90degrees anticlockwise so the individual pins end up on both sides at top and SW and GND end up on opposite sides at bottom..
Yeah. Will do something like this (pic attached below). The screenshot is from my complete design of the final product. Wanted to test it on a breadboard before sending it for PCB manufacturing and ran into this trouble
 

Attachments

Irving

Joined Jan 30, 2016
2,308
Hmm, why not just make up a small test pcb with just the buck converter on it to verify the layout? You could make a board with a few candidate layouts so you can compare and contrast them. That's what I would do rather than mess with a dip carrier which may introduce other side effects...

What was the problem you were encountering?
 

Thread Starter

gautam023

Joined Apr 26, 2021
19
Hmm, why not just make up a small test pcb with just the buck converter on it to verify the layout? You could make a board with a few candidate layouts so you can compare and contrast them. That's what I would do rather than mess with a dip carrier which may introduce other side effects...

What was the problem you were encountering?
I want to test the chosen values of my passive components. Including them in the PCB design will defeat that purpose.
 

Irving

Joined Jan 30, 2016
2,308
I want to test the chosen values of my passive components. Including them in the PCB design will defeat that purpose.
And you can't swap parts on a PCB? Breadboarding an SMPS rarely gets the results you expect, too many stray inductances/capacitances and lack of ground planes - modern SMPS chips are quite finicky in this respect. Also lead lengths of through-hole passives on a bb don't translate well to SMD parts on a PCB.

In your position I would make a few small test PCBs, using larger SMD parts so easy to hand-solder so I could do proper evaluation of the PSU or other circuit characteristics if I wasn't confident enough about the design before releasing to manufacturing - especially if its going on a multilayer board. In one example I had something like 15 'modules' each on their own mini-board., which were then stuck down on a wooden baseboard. That allowed me to test relative positions/distances/screening/etc between the subsystems and identified where we might have had issues on the final assembly. Simply being able to rotate one particular module 90 degrees identified and removed a massive interference/crosstalk problem that would have been a real issue if it hadn't been caught at that stage.
 

Irving

Joined Jan 30, 2016
2,308
Yes, that's the right size but I think we've moved on from that. Breadboarding an SMPS with all those thin tracks isn't going to go well...
 
Top