Lets build another CTPCI

[dml]

Continuing my CTPCI adventure which began towards the end of this older thread:

viewtopic.php?p=127072#p127072

....

I set everything up this morning to start assembling the board.

I have a stack of blank PCBs, two PCI controller ICs and enough parts to build one complete unit plus 90% of a second unit.

Screenshot 2025-04-10 at 12.06.55.png

The pics are just the carefully aligned IC - no soldering yet. Wish me luck!

Screenshot 2025-04-10 at 12.07.22.png

[dml]

Before....

Screenshot 2025-04-10 at 13.01.37.png

Got the worst of the two ICs - the PCI controller - tacked down at the corners after a bit of a battle keeping it aligned.

Screenshot 2025-04-10 at 13.02.03.png

That IC was difficult not just because it is the bigger of the two but it's also not brand new so the pins were not taking solder immediately. Not ideal. After a few round-robin passes with lots of flux it went ok.



Got the whole thing hand-soldered in!

Screenshot 2025-04-10 at 13.02.25.png

Screenshot 2025-04-10 at 13.02.41.png

Nudged every pin with the tweezers, found a few tricky ones which were not stuck down on the first pass. Sorted those so it just needs a continuity check and it's done.

[dml]

The CPLD was a bit easier.

I did a better job with the initial alignment this time and held with a bit of kapton tape (with the other IC I used a small heatsink but wasn't heavy enough).

Got it 100% on first try and it didn't move when I added the tape. I stuck one end of the tape to the board and just let it kinda fold-over the IC on its own then pressed it gently with the tweezers.

Screenshot 2025-04-10 at 13.15.34.png

Screenshot 2025-04-10 at 13.15.12.png

Got it tacked down..

Screenshot 2025-04-10 at 13.16.02.png

Both sides, after a bit of a wash....

Screenshot 2025-04-10 at 13.16.14.png

Screenshot 2025-04-10 at 13.16.37.png

[mikro]

Impressive soldering skills, wow.

[dml]

Impressive soldering skills, wow.

Thanks - I'd like to think I could have been equally fast with the passive components but it's been painful so far. I lost two capacitors to the floor - one pinged out of the tweezers, another left the capsule strip when I wasn't looking - eventually found them (somehow) and then managed to stick one accidentally to a nearby/wrong pad.

Fixed that and have made quite a lot of progress but it is extremely slow :/ I should maybe have invested in a hotplate or something. Although it is some kind of progress...

[dml]

Beginning to realise the more of this I do, the more annoying it's going to be if I blow it up :p

Screenshot 2025-04-10 at 15.45.50.png

[foft]

Seems you are making good progress. Sorry didn’t reply to the inner layers, seems you got there though:)

The main issues I normally have are with the little 8 pin resistors having a missed connection. Last time I managed to put the wrong resistors round the clock distribution ic, that was annoying to find!

If you figure out the best way to solder the 40 pin headers let me know, there is not quite enough space to fit the iron without accidentally melting them a wee bit…

[dml]

Seems you are making good progress. Sorry didn’t reply to the inner layers, seems you got there though:)

:)

working on PCBs isn't my day job - and my electronics background has been more home-made and thru-hole than dealing with board manufacturers, grains of sand and microscopes - but its going ok so far (I think!)

The main issues I normally have are with the little 8 pin resistors having a missed connection. Last time I managed to put the wrong resistors round the clock distribution ic, that was annoying to find!

I think these are my new least favourite SMD devices. Small enough to get lost or stick to the iron tip, just enough pins to need aligned and make bridges and enough of them to be super tedious and annoying.

At least they are simply straight-thru - and not resistor networks with a common pin - those would need orientation as well!

If you figure out the best way to solder the 40 pin headers let me know, there is not quite enough space to fit the iron without accidentally melting them a wee bit…

I was going to ask about that step - I noticed a trap for new players here :) how to get the headers aligned on the surface pads without pre-drilled holes to help.

I was thinking the headers need to actually be pushed into the CT6x and placed upside-down on the CTPCI as a sort of alignment jig, then tack the headers on without melting everything? I guess I'm soon going to find out how realistic that is...

[dml]

Oh yeah - almost forgot!

R18 is listed in the BOM .xls (ctpci_dev.zip) as a 22R resistor. But R18 on the PCB is a resistor network.

Are there other errors in the BOM ctpci_totalparts.xls? There are two values of plain resistor so it would be super easy to place them incorrectly. Which I think is something you just pointed out in your post.

[edit]

...seems there's another copy of the BOM in the form of ctpci.xls in the dev .zip and the listing is different. R18 is a resistor network, matching the PCB... also the component counts in the first BOM don't match the list sizes. Second BOM looks correct.

Well I don't think I made any mistakes so far with the other BOM since the only potential mixups are with the resistor values and I haven't started on those yet. Not really possible to confuse the resistor networks or the caps.

Would be better to confirm against a schematic but I don't think there is one.

[dml]

Finished populating the components and done with the most tedious parts I think, except maybe for the headers :/

Screenshot 2025-04-11 at 19.25.55.png

Screenshot 2025-04-11 at 19.26.15.png

The resistor array thingies were extremely annoying. I still need to check everything for bridges/shorts and good connections but I don't think I can face any more of this green board today. Will return to it at the weekend.