I wanted to address the recent departure of Perifractic and talk about some of the change in direction. This could be a long one, so settle in.
The absolute first thing I want to clear up just in case anyone is thinking this, Christian (Perifractic) and I are not mad at each other, we didn't have a fight or a falling out or anything like that. We are still friends. This was a mutual decision and it is really mostly to do with his personal issues as he has already stated.
That being said, Perifractic and I often disagreed over certain aspects of this project. And because I respect his opinion and experience, I often gave in to his advice on matters. There were two issues specifically that we disagreed on, which was the inclusion of a case on phase-1 systems, and whether or not the Commander X8 should ever see the light of day. So, now that he's stepped back from the project, I want to talk about both of these issues and a general change in direction for the entire project.
Dropping the case, moving to kit sales.
One huge issue that Kevin and I have been dealing with (mostly Kevin) is trying to figure out how we can get this manufactured. In order to get the cases we needed a minimum order of 1,000 units. And that in an of itself isn't a problem, but assembling 1,000+ motherboards would be. As I've shown in some recent videos with kit computers, it can take an entire day just to solder one together. So, 1,000 units (or more) means 1,000 days of work. So, the obvious answer would be to have them commercial built and wave-soldered. PCB-Way even offers this. But, this becomes more of a nightmare than most people can possibly imagine. I'm not going to go into the nitty gritty details. It wasn't until Kevin spent hours discussing the issues to me, that I was able to comprehend the enormity of that challenge myself. Here are just a few of the issues involved. Getting them the parts. They have to be shipped to China. Tens of thousands of dollars worth of parts. And they have to get through customs and actually arrive at the factory. If even one piece is missing, then the whole project stalls. Shipping to China is often a crapshoot with packages mysteriously disappearing. Many of the parts are custom or rare parts that cannot be sourced easily in China. And of course, factories like this require parts to be in reels and various forms designed for their machines to do. When you actually look at the BOM on the X16 and realize how many parts we're talking about, it becomes a nightmare to find all of the right parts in the right form factor for their machines, and get them those parts. And everything has to be right the first time. Every single part.
And then there's the issue of where to store 1000+ cases, keyboards, and motherboards while it is all being assembled. We'd probably have to rent a space to do that. Somebody has to pay for that. And then somebody has to actually assemble all of this into a box. Thousands of boxes. Incredible amounts of labor here.
And to top things off, there is a chip supply shortage right now. Some of the chips we need aren't even available at the moment. And those that are have gone up in price by 50%. This is hopefully a temporary problem.
Then of course, we had a sample case from the manufacturer. When Kevin mounted the X16 board inside the power supply died after about 10 minutes. And our board is only pulling somewhere like 10% of the max load of that power supply. So, it was definitely a bad power supply. This made us very concerned about buying 1,000 of these cases and potentially having a bunch of dead power supplies. That was sort of the straw that broke the camel's back.
And one last thing I'd like to mention. All of us on the X16 team have spent uncountable number of hours developing this product. And many of us have also invested quite a bit of money. In fact, most anytime an expense comes along, all eyes and fingers point to me. I know Perifractic has already invested a 4-digit number of Dollars towards this project. Myself, that number is 5 digits. And up to this point, none of us have made a dime back. Moving forward with the original plans of full case and fancy packaging was going to require another large infusion of cash (most likely from me) that I just don't have. I have no desire to mortgage my house to fund this project. The risk is just too high, and despite what some people think, YouTubers like me are not millionaires. So, bottom line is we're in over our heads. So, unless there is some wealthy benefactor out there that would be willing to finance this project and take on the high risk, things needed to change.
So, after much discussion, Kevin and I decided the fastest and easiest way to bring this system to market would be to take a new approach. As you all know, we had always planned a stage-2 and stage-3 of this product. I always knew those would be the mass market versions because they would be cheaper. But I wanted to have the "real thing" with DIP style chips as the basis for the design. And I wanted it to be available along side the phase 2/3 products for those that wanted it. And we have that now. So, what we'd like to do is just start selling the product we have in two ways.
One option is a kit that you solder together yourself. It would include no warranty and no tech support, other than community support here on this forum or facebook. Obviously, we'd be able to supply individual replacement parts if somebody fries or somehow destroys some part in the process of assembly. We just won't be able to help every Tom, Dick, and Harry to troubleshoot why their kit isn't working after they assembled it.
The second option would be to build-to-order. So you place the order, we build the computer and add a substantial mark up for the 8-hours of labor to assemble it. But you get a fully working and tested board, which would come with some minimal warranty. I don't have specific numbers for cost. But I'd imagine a fully assembled kit would have a markup somewhere in ballpark of $100 to $200 over the kit version. I know that sounds like a lot, but when you see how much time is involved, it only makes sense.
Also both the kit and pre-built machine will come with the custom keyboard. I've already paid 50% down for the PS/2 keyboards, so it makes sense for me to pay the rest and have those included with the computer.
So, the bad news is, you don't get a case with phase-1. But on the bright side, this change means the kits could be available relatively soon. However, as you saw in the last video there are still some Kernal bugs that need fixing and our primary Kernal developer has taken a small hiatus due to some other large project that is consuming his time. We're not sure at this point when these bugs would be fixed. However, we could start shipping limited number of "development" systems to people who are already writing code. And hopefully all we'd need is a ROM update to fix these boards when the kernal bugs are addressed.
Phase 2, Phase 3, and the Commander X8
So, once phase-1 gets underway, we will consider a few future options based on demand and popular opinion.
Phase 2 would be an all surface-mount product. It would still have discrete CPU, RAM, ROM, etc. But it may drop some logic chips in favor of a CPLD to reduce size, complexity, and cost. It should be substantially cheaper. Because it is surface mount, assembly actually becomes MUCH easier on a mass-production scale. In fact, Kevin said he could possibly even assemble them at his own place of business (TexElec) which would simplify the situation a lot. This product could be sold as a board or possibly include a custom case, unlike Phase-1. But, we could potentially skip Phase-2 all together if it seems like people would be more interested in Phase 3.
Phase 3 would be a small board like a raspberry Pi where the entire system is basically in an FPGA. I suppose if there were enough demand (like tens of thousands of units) we could maybe get a custom ASIC produced.
The Commander X8 - Believe it or not, this product already exists. I've had one sitting on my desk the last 6 months. This is entirely designed by Frank. It's a 100% FPGA implementation. It is sort of a subset of the Commander X16. It has mostly the same architecture, but it has minor differences. There is also already an emulator for it. It's about the size of a Raspberry Pi.
So what is the deal with the X8? Frank and I were in favor of bringing this product out 6 months ago due to the delays of the X16. But some team members didn't like the concept, saying it would dilute the image of the X16. And they made some good points. So, we decided not to release it at that time. But now that things are changing, I thought at minimum I should explain what it is and see what kind of interest people have in it. On the bright side, it is a product basically ready to be released. But does fall short of some of the cool things on the X16. So let me explain how it differs. Most of these concessions and incompatibilities boil down to using a smaller, cheaper FPGA design.
It has 64K of base RAM and 64K of VRAM. It does not have any banked RAM beyond that.
BASIC works essentially the same and should be compatible with most X16 programs that are coded in BASIC.
VRAM access is fundamentally different. There is a 256 byte window into the VRAM which is mapped to a section of base RAM. You can move the window around. This is actually more efficient than what we do with the X16 and is only possible because it is all inside an FPGA. This does mean software written in assembly language will need to be tweaked to be compatible.
The Vera is more or less the same. All of the same registers. Same PSG sound features too. But, programs that use more than 64K VRAM would need to be modified.
There is no Yamaha sound chip. However, as we've seen already. The 8-voice sound system in the Vera is pretty darned capable!
Uses a USB keyboard instead of PS/2. and USB for controllers (so, no SNES ports)
Runs at 12 Mhz instead of 8.
So, the big question is, if we were to release the X8, would that essentially replace Phase-3 of the X16? How would this product live along side the X16? Would it have an effect that people would simply code software for the X8, thus making the X16 be sort of like the Commodore 128 or the Plus/4, where all of the software is written for the lowest compatible system and therefor never taking advantage of the full system? So, to explore that, I would counter and remind people of what features the X16 has that the X8 would never have and see if that is enough to justify writing software for it?
X16 has a TON more RAM
X16 has twice the video RAM
X16 has 4 expansion slots
X16 has a Yamaha sound chip
X16 has an IEC disk drive port (although admittedly that is not implemented in the kernel yet, but should be working at some point)
X16 is infinitely more "hackable"
X16 has SNES ports.
So this product is "really close" to what I envisioned the X16 phase 3 to be like, but not quite. But on the bright side, it could literally be available almost immediately (pending getting a batch produced) if people are interested in this.
So, at this point I'm looking for feedback on the future. Should we continue as planned to phase-2, then phase-3? Or should we drop phase-2 and move straight to phase-3? Again, phase-1 would still be available concurrently, and indefinitely for those that want it. In other words, Phase 3 doesn't replace Phase-1. Or, should we scrap phase-2 and phase-3 and release the already finished Commander X8? Or, should we do some other combination of things?
