Worklog N64 Tríos -- a 3x3 inspired motherboard redesign.

Joined
Jul 9, 2020
Messages
227
Likes
498
I enjoyed spending the countless hours exploring the nintendo 64 on my last extreme trim attempt. Doing deadbug clock relocations, ramswap and manual ram termination was fun and rewarding... especially when the thing would boot and play games stably. @YveltalGriffin offered much inspiration for these undertakings. Thanks mate, for showing me the way!
IMG_9021.jpg


Once i tried to trim the mav chip section off the board, the trim went down for the count and stopped booting. I spent a solid day trying to get it back in order with no luck.


1653125229487.png


I even had designed an RCP breakout board and cart slot connector with FFC connection.

CART SLOT TOP.png
RCP:PIF Breakout for cart.png


Rather than go through all of that again I figured better to spend that time designing a 3x3-esk custom mobo.

Spent the first day or two working on the symbols and schematic.

1653125572784.png


Another day or so measuring and dialing in my footprints. (thanks to @Gman for helping me verify my rcp footprint). I discovered the magic of the Kicad Footprint wizard to get things populated in the footprint before fine tuning everything. Stoked about the learning thats been happening. Ended up needing to extend the pads of the footprints for hand soldering purposes...

After dialing in the footprints i got started working on the layout. I was heavily referencing and consulting with gman about his 3x3 design. Thanks for the help and inspiration with that Gunnar @Gman. Legendary homie.

Getting started:
1653126107683.png


RamBus routing:

1653126153888.png


Building the termination and clocks:

1653126239531.png


Pif circuit and VCC regulator:
1653126341755.png


Bit by bit...Edge cuts with round corners. Mount holes, FFC connectors and breakout pads. Decoupler caps... still need to work on some routing.

1653126519387.png
1653127452670.png
1653127596361.png


After lots of nudging and verifying and checking back and forth from schematic to compendium to board I got to a place where I felt i was very close to a first revision. At this point @YveltalGriffin kindly offered to look it over for a quick design check. He noted how I could work on my routing to not break up the ground plane so much on the back and to avoid so many ground islands in my design... Also how to leverage use of the grid to make drawing fill zones a breeze. I also got some insight into the general layout work flow and how next time I will be able to do it more efficiently. I had to do so much nudging and tweaking because of the order I laid things out. Learning learning learning... Ground plane integrity brooooh. Big thanks to Yveltal for pushing me to learn these skills myself and still lending his knowledge to help me hone the craft more clearly. Epic friend.


I took the time to tuck my routing in as best i could. Its a fun puzzle but i think it turned out way more clean and crisp after taking Yveltal's advice. Anyway here is where I ended up this evening:
1653127056110.png
1653127214188.png
1653127330355.png


After the edits I did another DRC verification and a visual pass over all the systems before building the gerber files and ordering the first print from JLC. I built the BOM and ordered the parts on mouser tonight too! So once i get back from the music tour at the end of june I'll be able to assemble and test this baby out. Super excited to see if it turns out.

I think i am gonna name the project either Tríos or Três. Spanish or portuguese for "Threes" as a play on 3x3. We shall see...
 

Attachments

Last edited:
Joined
Apr 12, 2020
Messages
128
Likes
242
Beautiful! Could we be seeing a purchasable bare PCB in the near future from SparkleBear Technologies? Hehe
 
Joined
Jul 9, 2020
Messages
227
Likes
498
Have not gotten the board from JLC yet but I already integrated a digital audio amp based on the 64 amp designed by Gman. I also did a bit more tidying of the busses on the back side of the board as well as some silk tidying too :)
N64TriosFront with audio amp.png
n64trios back with audio amp.png


I will probably just wait for the other set of boards to be verified before ordering another set with the amps on it... just incase i need to revise the core... But this is ready in waiting :)

Dang, its not going to be until the end of june that ill be able to test it out because i'll be heading out on a 3 week music tour! Yahoo!!

Beautiful! Could we be seeing a purchasable bare PCB in the near future from SparkleBear Technologies? Hehe
I don't think i'll be selling this board as it uses ideas derived from Gman's circuit schematics of the PMS1 and 64amp. If selling it was a goal of mine i'd want to design a version of that circuit myself... and im not quite there yet :)
 
Last edited:
Joined
Mar 14, 2018
Messages
36
Likes
14
What size did you use for your trace width? Most look the same minus the vterm line I think
 
Joined
Jul 9, 2020
Messages
227
Likes
498
Smallest traces were 0.2mm. Most of them are 0.25mm... Vterm is .6mm. Internal layer trace for 5v to buss across the board is 2mm. The bottom of the rambus lines i used 0.5mm.
 
Joined
Jul 9, 2020
Messages
227
Likes
498
Its been a moment since my last post. Heres an update :)

Maaaaany many weeks ago, back in may, these boards arrived. I took these photos and played with my new microscope for a quick moment knowing i wouldn't have the time for a while. So pretty...

BCA7319E-5A0C-4866-BE73-C5E697B7B711_1_105_c.jpeg
.
2BC21414-38B7-4E5F-9171-8B7DCA7BCFB9_1_105_c.jpeg
.
6559AE3A-5CF5-4A4F-B8B2-8DE5764C29A3_1_105_c.jpeg


I finally was able to set up a workshop and assemble my first revision of the redesign. It was almost torture to carry around the parts and pcb for a few months but life has its own way sometimes... well pretty much every time.

I knew the first thing I needed to do was test the Vterm LDO. I had never personally calculated the resistor values and had some doubts about what value to use as the Vref. (formula Vref= (R1 + R2 ) / R2 ). Turns out I nailed it... Its a few milivolts higher than I had aimed for but I was so pleased to see it function as intended.

IMG_9508.jpeg
.
IMG_9509.JPG

Next assembled the cartridge slot ffc breakout board.

IMG_9515.jpeg
IMG_9513.jpegIMG_9514.jpeg

Looks nice and tight. checked pins with cable inserted and made sure all connections were properly showing contuinity.

Time to assemble. What an exciting moment it was. I had never pasted a board this big before... this being only my second time pasting. The first being the week before and only a small little codec circuit for a pi synth. I had some loose n64 chips from my RE research. I didn't want to start with those as they'd been kicking around in my project box semi protected; some of the pins were a bit bent. Rather than go through that i decided to just start with a freshy!

Ok Gotta test out the N64 chipset before removing any components! :) No better cartridge to have as your only non everdrive game ...Thanks Mario.
IMG_9516.jpeg


IMG_9517.jpeg
IMG_9518.jpeg

I think I'll try to straighten out those loose chips and stick em back on the board i used as a donor for the redesign pcb. New life... That or save them to sand down @Redherring32 style. We shall see..

IMG_9520.jpeg




I didn't have many flat pieces of anything. I ended up doing the pasting on the back of a fancy Malaysian acacia wood tray. Used the technique taught to me by brother @YveltalGriffin but I'm sure many of you are familiar with. The illustrious 4 pcb pasting frame jig!! :) All tight and ready for a pasting.

IMG_9521.jpeg


I used the thermally stable chipquik which was actually really smooth and nice to work with. I was all set and sorted to paste when i realized that my stencil was significantly warped. I tried to tape it down but it just wouldn't lay flat. Unfortunately, I didn't take any photos to document; i was kind of in the zone. I ended up scoring the middle of the stencil between the halves and broke it so it was more manageable to get a successful pasting. It was a big mess on most of the QFP pins but all of the small passive pads turned out! So I didn't think that I would get a better pasting with the condition of the stencil; GoopyGoop gobs splooged through the warped stainless stencil slits.

I figured it would need rework anyway.

IMG_9522.jpeg
.
IMG_9523.jpeg


After an hour or so of placing tiiiiny 0402s and aligning the chips as best I could, I started the reflow with my hot air station. It's a beautiful thing to see this come together... I had not yet discovered that there was still a long way to go though. I started reflow air thinking my blue mat could handle it... NOT it started to bubble up slowly and the expansion of the air just made a huge bulge. It was entertaining........ anyway, I ended up using the back of a muffin tin.

IMG_9524.JPG


So the small passives would not blow away, i ended up using 360c with 10% airflow using the widest nozzle. Success for the most part. Only blew away 1 of the many... Here she is after the hot air.

IMG_9526.JPG




Up next is pasting the back. Gotta figure out how to do that -- would not like to hand solder all those tiny caps and resistors. Hopefully, with eithe standoffs or other random PCBs. Time to search. Top will also need a good reworking... But that will be for another day and another worklog post. Stay tuned... more to report shortly. Good days work though. I can already feel the dopamine pooling... ;P

CHEERS! :)
 

Attachments

Last edited:
Joined
Jul 9, 2020
Messages
227
Likes
498
After working on the top paste I discovered there was a 3v3 short on the board somewhere. I inspected all of the pins closely and it didnt appear to be bridges or whiskers. I decided to pull off the cpu to check. Luckily it was the source of the short. I went and lifted every pin power pin then every ground pin metering in between each lift to see if i could find it.

IMG_9529.jpeg

IMG_9531.jpeg
IMG_9532.jpeg
IMG_9533.jpeg
IMG_9534.jpeg

After some decent head scratching and going through soldering and desoldering the cpu a few times and not fixing short; I just pushed each row of pins down individually and discovered what side of the chip was making the short and then referenced schematic. Turns out I accidentally swapped three of the VCC and VSS lines on the schematic. Whoops. All this play on the pins weakened one of the corner vcc pins and it ended up falling off... This one will get the @Redherring32 treatment.. hehe


NO!
1663214839741.png
. YES!
1663214903342.png


I lifted the erroneous power and ground pins and bodged them together with some magnet wire. I mistakenly scratched up the solder mask with my new set of helping hands. I even put shrink tubing to stop that but it bit through it. More tubing for next time..... Now that the short is gone I can move on to pasting the back!

IMG_9555.jpeg


The next day I woke up stoked to figure out how I was gonna paste the back. A few ideas had been kicked around... maybe something with standoffs? or random pcbs? hand soldering wasn't my first desire so I felt intent to make this work. I ended up just carefully stacking pcbs under the edge of the pcb so the chips would have clearance. it was a puzzle and i enjoyed figuring it out. It ended up like this... mish mash of bare pcbs for louii, αsnes, norns synth, and the 4 spare trios boards. This time i realized the flattest surface in the house was actually the glass stove top! A bit of degreaser and the tape stuck really well :)

IMG_9543.JPG
.
IMG_9544.jpeg


This round, the pasting turned out really nice! only 0402 on the back so no big QFN pin holes to ooze through

IMG_9546.jpeg
.
IMG_9549.jpeg


Time to place! This part is really zen for me. I was using this AMAZING application of HTML that is called iBom. It lets you see the layout and highlights the parts in red on a graphic of the PCB. Very helpful for not making mistakes and also speeding up the time between placing parts. No more hunting for designators. Im in love :)

IMG_9550.jpeg
1663213878383.png


Hot air reflow went really well. I used 360c at 10% airflow. Only a few tombstones to deal with but no parts blew away :)

IMG_9556.jpeg



All of the joints look great; nice and shiny! I inspected all the traces where the scratches were made... all intact. Time to recheck for shorts and try to plug her in! I had to wire up my vga board and bust out the test screen. Flex cable between board and cartridge slot. Best game ever... aaaaand.

IMG_9557.jpeg
IMG_9560.jpeg


No video. The board does not boot the game or show any video. All the voltage rails are showing 5v and 3v3 appropriately. The ram chip are getting warm and the Vterm is 2.56v. The clocks are firing off, I checked them with the scope. Even though severe anti-aliasing was occurring I could tell that the clock signals were matching behavior from a stock unit. The RCP and CPU were not getting hot. Maybe they were getting slightly warm but im used to them cooking. So perhaps something is wrong with the power rails around the chips... Something to investigate.

I took a good look under the microscope as well as photos of the board close up. I intended to stitch them together in photoshop but lost steam when they showed up on the computer in a distorted aspect ratio. Aaaand i spent like an hour trying to figure out a stitching program...to no avail. Ill put them below incase you want to check out the gore of the debuging battle.

IMG_003.JPGIMG_004.JPGIMG_007.JPG
IMG_013.JPGIMG_052.JPGIMG_043.JPG
IMG_016.JPGIMG_055.JPGIMG_040.JPG
IMG_019.JPGIMG_049.JPGIMG_037.JPG
IMG_022.JPGIMG_025.JPGIMG_028.JPG

Man that is FUBAR. jeezum. What a man does in private is between him and his soldering iron...

Oye vey.

I updated the Schema/PCB layout to add some bulk capacitors in any free space i could and as close to the power pins for the chips as I could manage... Cleaned up the silk a bunch -- including moving the bear and title text. I nudged some components around to fit the caps. I moved the 3v3 test pad to the bottom of the board and made it a more robust through hole pad because the other delaminated and vanished. Ignore the uamp section for now. (and just for fun I played around with Mitxela's track rounding plugin)

1663220632410.png
1663220684792.png
1663222821340.png


All in all this was a big push of progress even though we are not booting yet. I sat there and verified both big chip's pin traces against the board scans and everything matched between schema and scans as far as i could tell. I think i will also want to redesign the Stepdown regular section to utilize a chip that is in stock. That is pretty much the subcircuit from the PMS1...

One day at a time.
 

Attachments

Last edited:
Joined
Nov 4, 2019
Messages
24
Likes
169
Oh no, 2 name drops in 1 thread.
I have become infamous.
 
Top