The Wombat ADB-USB input converter is now back in stock at the BMOW Store. Thanks for everybody’s patience while we waited out multiple delays on delivery from my manufacturing partner. There’s now plenty of Wombat stock at BMOW HQ, so they won’t be selling out again any time soon.

What’s a Wombat? The Wombat is a bidirectional ADB-to-USB and USB-to-ADB converter for keyboards and mice, and was developed by Steve Chamberlin here at Big Mess o’ Wires. It can connect modern USB keyboards and mice to a classic ADB-based Macintosh, Apple IIgs, or NeXT, or connect legacy ADB input hardware to a USB-based computer running Windows, OSX, or Linux. For more details, please see the product description page.

 
Global Shipping is Back!

After ten days of confusion at the hands of US Customs and Export Control, BMOW global shipping is back! They’ll be celebrating from Calgary to Cardiff, Melbourne to Madrid. I never did get a clear explanation of exactly what went wrong, but some post office processing facility at the San Francisco International Distribution Center screwed up in a big way, resulting in thousands of international packages being incorrectly returned to sender from shippers all over the US West Coast. What a mess.

I’ve taken this opportunity to adjust the international shipping rates to reflect the latest USPS First Class Package International service rates. Pop quiz: what do France, Germany, the Czech Republic, Gibraltar, Hungary, Iceland, Lithuania, Poland, and Portugal all have in common? They’re nine countries whose USPS rate group offers cheaper shipping than most other international destinations outside North America. Why those nine countries? I really have no idea.

Can you guess what distinction is shared by Australia, New Zealand, Brazil, Hong Kong, South Korea, and Japan? Shipping to those six countries is more expensive than most of the rest of the world. I can sort of understand Australia and New Zealand due to their relative geographic isolation, but why single out Brazil or Japan? The lesson I’ve learned from these past ten days is never question the post office.

I’m looking forward to shipping lots more BMOW hardware to enthusiasts everywhere.

Good news: the International Shipping Fiasco of May 2021 appears to be resolved, although the underlying causes are still unclear. Early indications are it was some kind of USPS screw-up with electronic collection of customs data at the San Francisco International Distribution Center. Shippers all over the US West Coast have been affected. There is blood in the water. Stay tuned for more details.

BMOW’s most recent Canadian test shipments have passed through customs successfully, so I’ve re-enabled Canadian shipping for the BMOW Store. Assuming there are no further surprises, I expect to re-enable international shipping for the rest of the world within the next few days.

The Yellowstone version 2.0 PCBs are here! They sure look shiny. I wonder if they work.

This is the second generation of my FPGA-based disk controller design for the Apple II. In comparison to the version 1.0 PCB, you can see there’s more going on. While it’s nothing in comparison to any professionally-designed device, this will be the most complex thing I’ve ever designed or assembled.

Update: It looks like this problem may have been resolved by USPS, with international shipments sent May 11 or later proceeding normally. But there’s still confusion about the source of the problem, the date it was resolved, and whether it’s truly fixed. It appears it’s necessary to buy new postage to reship any returned packages – reshipping using the original postage label reportedly does not work. Unfortunately, all requests for postage refunds have been denied so far.

Help! This is very bad. Since sometime about April 26, nearly every international shipment I’ve sent has been returned by US Export Control, for unknown reasons. The packages come back with an export compliance sticker that says something is “missing”, but doesn’t explain further. I discussed the problem with the staff at my local post office, but they couldn’t explain it either, and said the customs information on the package looked OK to them. They also said that when an outbound package is rejected by US Export Control, there’s basically no recourse: the customs inspectors don’t respond to calls or emails about specific packages. In the staff’s words, “You just have to guess”.

All of the packages were shipped by US Postal Service’s First Class Package International service. I’ve not recently changed anything about the packages, their contents, the address labels, or the customs information printed on the label. I’ve previously shipped thousands of substantially identical packages internationally without incident. But not anymore.

TL;DR – I am completely unable to ship any international packages, and I don’t know why, or who to ask for help. The shipments never get out of the USA.

To put it mildly, this is bad.

The specific package shown in the photo contained a ROM SIMM and an empty plastic case for one of my electronics products. But most of the rejected packages contained a BMOW Floppy Emu – a disk drive emulator for retro Apple computers.

None of the returned packages had been opened, so there was never any time that an export compliance person looked at the contents inside the package, looked at the item description on the label, and decided whether they matched. The decision to return the package was 100 percent based on the information printed on the label, perhaps combined with information submitted electronically when the postage was purchased. But nobody at the post office seems able explain anything more than that.

There are some clues. Here’s a list of every international shipment I made between April 19 and May 6.

Almost all of the packages shipped prior to April 26 have been delivered, or have at least made it outside the US. The three from before the 26th that show “in transit, in USA” are misleading because they all have tracking numbers starting with the letters UC, which indicates a destination country where tracking normally isn’t available beyond the US border.

But almost every package shipped After April 26 is either still stuck in the US, or has been returned.

Something changed within a few days around or after April 26. The date of change was probably May 1 – roughly when packages shipped April 26 would arrive for customs processing. A new customs inspector employee? New inspection procedures or policies? New export control laws? Or maybe something changed with Shippo, the service that I use to print the address and customs labels?

Notice that the return label says more information is available at https://postalinspectors.uspis.gov/ if you click on “Export Issues” under Consumer Help. But that section and that heading do not exist. I couldn’t find anything on that site to help explain what’s going on. Edit: see this 2018 archive.org backup of the uspis web site.

With no information and apparently no recourse, I’m starting to panic a little. For the time being I’ll have to disable all ordering for customers outside the USA, which is a large fraction of the total for BMOW. And I’ll have to contact all the customers whose shipments have already been impacted, or who’ve ordered and paid but whose packages haven’t yet shipped, and try to determine what’s the next step.

If you know anyone who works in any type of export manager capacity, who might be able to advise or consult with me on this, that would be fantastic. Or if you know anybody affiliated with the postal inspection service who might have insight into what’s going on, that would be great too. I’ll take all the help I can get to resolve this crisis.

The Yellowstone 2.0 disk controller PCB has been sent to the fab. Now we wait. It took longer than I planned to finish it, first because I wasted a couple of days trying to shrink the board by 7 mm, then because I was obsessing over tiny layout tweaks. Now I can breathe a sigh of relief… except for the bad news. And the problem. And the other bad news.

Bad news #1 isn’t really a surprise. Thanks to the global IC shortage, the prices of components have increased. When I compare the IC prices from 2017’s Yellowstone 1.0 bill of materials to today’s prices, they’re up anywhere from 15 to 300 percent. Doh! I’m hoping this won’t affect the retail price too much, since the cost of materials is only one element of the total, along with cost of assembly, programming and testing, parts shipping, and general business costs like order fulfillment and customer support and other overhead.

The second problem is also cost-related. The twin DB-19 female adapters and cables for connecting the two disk drives are going to be more expensive than I’d like. The raw parts cost for a Yellowstone board with two DB-19F adapter cables is like 60 percent more than just the Yellowstone board alone! How can that be? Mechanical parts like connectors and cables just seem to be inexplicably expensive compared to most ICs except the FPGA. The worst offender is the DB-19 female, which isn’t manufactured anymore and is only available as new-old-stock from surplus suppliers.

As a result, I’m seriously considering selling Yellowstone in two versions: one that’s the board alone, and another that’s the board plus DB-19F adapter cables. With the board alone you can attach Floppy Emus, or Disk II drives, or internal Macintosh drives, or anything else with a ribbon cable and rectangular connector. But you’ll need DB-19F adapter cables if you want to connect drives like the Apple 3.5 or Unidisk.

Bad news #2 is my biggest concern now. The supply of the FPGA chip that I selected has shrunk dramatically in just a few weeks. I assume this is also due to the global IC shortage. It’s close to the point where I wouldn’t be able to manufacture any boards even if I wanted to. Only a few weeks ago, several different suppliers each had thousands of parts available, but as of today only a few hundred remain. I’m strongly considering buying them all right now, even though I don’t need them yet and the board design isn’t even finalized, just so I’ll have something on hand to do at least an initial run of manufacturing.

It took longer than expected, but the PCB layout is finished for version 2.0 of the Yellowstone disk controller. Phew! There’s still a bit of cleanup work to be done before I send the board for manufacturing. Hopefully I can finish that off within the next few days. I think it’s getting close.

Quite a lot has changed from version 1.0. The board now has two disk ports instead of one, with some DIP switches to control how they’re used. A 32KB RAM is now paired with the FPGA, reducing the amount of internal block RAM that’s needed and hopefully (to be confirmed) enabling me to use a lower-end member of the FPGA family that’s cheaper and more widely available. Level shifters and open drain buffers are now present on all of the disk I/O signals. In normal use none of the FPGA’s pins will be directly connected to anything off-board, which should improve robustness. The 3.3V regulator is now rated for 800 mA instead of 300 mA. I’m not convinced this was really necessary, but it’s something many people suggested upgrading from the 1.0 design. All of the chips except the FPGA are now using SOIC packages with 1.27 mm pin spacing, to make soldering easier and reduce the chances of assembly problems. Version 1.0 mostly used TSOP packages with a much finer 0.55 mm pin spacing.

Too bad there are no 5V FPGAs anymore. Half the chips here are dedicated to level shifting, and could disappear if the FPGA used the same voltage as the disks and computer.

The board routing is improved too, although you can’t tell from looking at the image. The edge connector on Apple II peripheral cards only has one pin for each power supply voltage, including ground, so it’s important to make those PCB traces nice and wide and to minimize the number of vias between the supply and the load. Yellowstone version 1.0 did a poor job of this.

I added a very simple bus sense input that could be used to support batch testing. It’s just a voltage divider connected to the 12V supply. If the FPGA sees this input high, it can assume the card is installed in an Apple II. If it’s low, it can assume it’s on a test bench and alter its behavior accordingly. I’ll figure out the rest later.

I can’t resist looking into the future and observing that chips U1 through U7 could be used as-is for many purposes other than disk control. Those chips form a generic design of an FPGA plus 32K RAM with level shifters connected to the Apple II bus. Only chips U8 and U9 are really specific to being a disk controller, because they determine the direction and I/O voltage of the signals on the disk ports.

Parts availability may be a problem here. Most readers have probably heard about the global IC shortage that’s affecting everybody right now, and BMOW is no exception. The prices of many ICs have increased significantly, and some types of ICs have become difficult or impossible to find anywhere.

A question for PCB designers and assemblers: when a board like this one has dozens of passive components, how do you decide on a numbering scheme? Is it better to number them in order according to their physical location on the board, making them easier to locate quickly? So for example the column of resistors on the right could be numbered consecutively R1 through R11. Or is it better to number them so that components with the same values/types have consecutive numbers, regardless of where they are on the board? So for example R1 through R6 might be all the 10K resistors, then R7-R10 could be all the 6.8K resistors, and so on.