DickeOlga Limited Batch build

  • Author Author suavek
  • Create date Create date
  • Last update Updated
  • Blog entry read time Blog entry read time 5 min read
What I am describing here is DickeOlga accelerator build I decided to execute after a long wait in a line for accelerator version built by 8-Bit Dreams. The accelerator designed by Sven Hamag and published at GitLab (project ID: 64402317).


glue_20251025_074503.webp


* BUILD IDEA

When I joined the waiting line in this thread
Dicke Olga - 030 CPU 64/128MB Accelerator card for A2000 Amiga

declaring interest, I was in 78 place. After a couple of months my position hasn't changed much (to 69) and I figured I would be waiting probably over a year before I get DickeOlga (DO) in my hands. After checking this is open-source project documented well I decided to pull the trigger and start my own little limited batch build.
I need only one board for my A2000 but with minimum order of 5 at most FAB houses including JLCPCB and PCBWay, I'm hoping remaining four accelerators end up in other community members hands sold non-profit just to cover parts cost and FAB house service. That's the plan at least.

Being EE for a living and having past experience dealing with PCBWay it was not a brainer to go for it. After checking parts availability from DO bill of materials it looks like PCBway should not have the problem with getting those parts either.

* GERBERS AUDIT

Next when reviewing PCB Gerber files I noticed the paste layer Gerber has issue which is paste presence on edge connector goldfingers. We don't want any solder covering gold fingers, do we? So, I removed it in CAMtastic before submitting to PCBway.

Next day I received quote for bare FABs: $15/fab (for batch of 20pcs) plus shipping, tariffs and taxes TBD. In next step the plan was to quote birdseeded board (SMT assembly), as with PCBway it is cheap providing reliable results. There is no point of saving $29 per board and spend half day per PCB manually populating tiny smt parts.
For this I need a BOM.

* BOM DEBUG

I went through GitLab BOM (v1.8.2) and found to be unusable for submission to fab house. So, I had to go through schematic and BOM in order to make a BOM that would be usable for CM house (pcbway) because the one on the GitLab gets rejected due to insufficient information about components.

Here are several issues spotted and fixed in the BOM
(a) J4, J5, J6 listed in BOM but not present on SCH neither PCB; what are these? I assumed these are leftovers from different board version.
(b) Capacitors have no voltage rating in the BOM, had to figure this myself from schematic analysis. Unnecessary time consuming, should be part of the designer documentation (BOM).
(c) Inconsistency in component reference designators naming - component functions used as ref.designators, makes it very difficult to organize BOM to prep for manufacturing, example: JP4, POWER, etc.
(d) missing ref. designators on the silk, example: CN1 (edge connector, gold finger).
(e) most parts have no part numbers, instead generic names describing parameters, like F1 through F6 fuses. What is missing is reaction time (fast, slow, else) parameter which is critical for board reliability. That information can be deducted from the design schematic, which again.... takes time, while it should be listed in the BOM.
(f) Insufficient information about all resistor network arrays, except value there's no example part number or at least IPC footprint name or guide. Without that it is again.... analysis of the PCB layout, dimensioning of the existing footprints to find resistor network in the package that would fit the board footprint.
(g) 74LVTH16543 comes in two package flavors, BOM doesn't say which one applies. Had to dimension the footprint, again.
(h) XC95144XL-TQ144 is incomplete part number, had to find correct one via datasheet and supplier availability.
(i) Incomplete X1 part number, had to figure voltage out of the Gerbers how it is powered because schematic does not show it either.
(j) and I have not even touched through hole parts yet. This will follow.

* ORDERING

Due to 50% tariff for PCBAs from China, and because this is the 1st time I am engaging Olga build, I reduced the initial 20 pcs batch to handful of sets only. The cost of the birdseeded PCBA turned out to be around $190 including shipping. All remaining through hole parts are to be added to PCBA in my lab using professional grade PACE solder/rework station.

I placed order with PCBway next day.
After a about later I received shipment from PCBway with ordered boards. Customs didn't hold it to pay tariffs but with Fedex invoicing these charges later, I can still expect customs fees related invoice to come and be paid.

1000037151.webp


Close up inspection showed quality of SMT assembly is excellent as with previous work I ordered from this FAB house.

1000037147.webp


1000037146.webp


While I was waiting for the PCBAs, I ordered one outstanding smt component that PCBway could not procure and it was ICS570A. It's arrived day after boards delivery.

1000037148.webp


* THROUGH HOLE PARTS BOM REVISIT

Update: 2025-10-27
I have to go back to the bill of materials and verify through hole parts manufacturer info and part numbers. I found almost all of that information is missing in the BOM, just a breadcrumbs information for what needs to be soldered. So it took me all Sunday evening to review schematic and board layout again to collect all the part numbers for remaining parts and there were 27 line items to search, identify and order.
1000037441.webp


Green columns contain information I identified and added to the BOM. Once this project is complete I will create it's fork at Gitlab and upload that information so next time someone wants to build this board this BOM may come handy.
Parts were ordered this morning from Digikey. While the obsolete 1Mb Flash AM29F010 had to be sourced from eBay being non stocked at major electronics component stores. I managed to find 45ns version which may play important role in Olga performance. Again, the original BOM calls for part nunber with "S" suffix but I was unable to find reference in chip datasheets about such marking. The best bet was to find fastest Flash derivative and that is what I did. The 45ns it is.

It's going to take few days to receive all of them.

* FINAL PCB ASSEMBLY

Next, I need to assemble through hole parts which I describe further in this blog. Prior that a basic test needs to be done with power check just on bird seeded board to ensure no shorts are present

...Update coming up as soon as I receive outstanding parts from vendor. Stay tuned.
Next page: OLGA PCB Bring up

Comments

There are no comments to display.

Blog entry information

Author
suavek
Read time
5 min read
Views
360
Last update

More entries in Member blogs

More entries from suavek

Back
Top Bottom