toot_toot wrote: ↑Mon Apr 04, 2022 9:37 am
But it is their own fault. Instead of making the second Kickstarter a quick turnaround of the original Next hardware, which was what was originally promised, they started to tinker with it and make “improvements”, adding in things like 2MB of memory as a stretch goal.
Well no -- it is clearly stated in the KS2 campaign text that the pcb would be improved. This was actually a necessary step because after Brexit, there would be border controls in place and lacking a certified CE mark may have meant we could not export into the EU from the UK. We did not think the KS1 board would pass and from the testing we did during KS2, it's clear it likely wouldn't have because one of the major causes of emissions is the ribbon cable connecting the main board to the daughterboard. The noisiness of the KS1 board in general was also a good indicator there might be problems.
So we knew the pcb would have to be redone and we took the opportunity to correct the deficiencies we found in the first board. Even without the CE issue, I would have been pushing hard for this because it would have been a huge missed opportunity to correct the issues we knew about. Around 20 of them were identified from minor to user-noticeable.
Except as a backer of the first kickstarter, I remember the problems they ran into sourcing the memory chips because they had to be very specific timing and IIRC were already limited to one supplier.
The problems (it turned out) were in the memory interface implementation and synthesis. At the time, specific brands of sram chips were working whereas others were not. That issue was discovered when the KS1 board-only was coming off the assembly line and were found not to work properly. Or maybe it was when people tried to upgrade to 2MB with additional chips that the issue was discovered -- I can't quite recall. These chips are commodities in that they are intended to be interchangeable. There is no longer any problem with using equivalent parts in all boards.
That was the “2C” board, then they added in even more hardware features with their update in Feb 2021, remember this was supposed to be shipping by August 2021.
2A and 2B were the boards produced in the first KS. The 2C was the first KS2 board designed in-house (but was not produced). An engineering company was then hired to get it to pass CE, to improve the board and to incorporate changes not yet part of the 2C. This led to the 2D and 2E. The 2E was the last Spartan 6 board and it was ready to go in September. This company was subject to covid shutdowns like everyone else as were everyone else involved in making the test boards. Nevertheless being ready by September was not missing the mark by much IMO given the circumstances.
The Xilinx FPGA they’d chosen for the Next was already in the process of not being manufactured when the original Next was eventually released.
Where are you getting this information from? The Spartan 6 is considered an active part by Xilinx even today. Here's today's
fpga product selection guide. Xilinx moves a product into EOL when it decides to discontinue it. This gives a further two years until the last product is produced. The Spartan 6 has had a long life because it's very good value and there was no indication that it was about to be EOL'd. Xilinx has another fpga, the Spartan 3, that has been around for even longer (6 years longer) and that continues because it has a number of large automotive customers.
What happened with the Spartan 6 is not that it was discontinued. Xilinx suddenly lost access to the foundry where it was made and couldn't make more, at least not for regular customers. This is not the same as the situation with its other fpgas where the problem is simply backed up orders and plugged up supply lines. I do think Xilinx's inability to make Spartan 6s does mean customers are being forced to redesign with different parts and that's very likely going to impact on demand for the Spartan 6 in the future.
This is from a Kickstarter that raised £1.8m in funds, it’s not as if they’re having to raise capital to buy the components. So why weren’t those items bought at the start? Why is it that 18 months after it was supposed to be launched that they still haven’t bought them? I think it’s down to the team thinking there’s still going to be hardware changes and they want to wait until the last minute to lock down those parts.
One common characteristic of failed kickstarters is that they spend their money up front without knowing what the product is or how it's going to be produced. The KS2 board is not the same as the KS1 board. The BOM is very different. The fpga is one of a few things in common. Ordering the fpga well ahead means you can't return for refund if defects are found. This is why nobody does this. It's simply not a good idea to gamble when you have a fixed amount of cash on hand to spend.
Your assumption that we would have received the fpgas had they been ordered immediately is also iffy. People in the queue last year received 1-2% of their orders.