With everyone winded up around recent The A1200 announcement, some of us including myself forgot about new C64 resurrected by Peri and his team. What concerns me is not delay but lack of news and updates from the company. What worries me is possible technical challenge these guys had to go through. the last video on the subject from Chinese PCB factory was over a month ago. Since then, it's awfully silent. I have two theories: it's either legal issue with Italians or technical issue Peri showed on his YT video. Which BTW was a serious red flag to me as it proved someone who designed the PCB for new C64 either has little experience with routing of high speed signal busses or missed important item during design which is a need to (1) match traces length within each differential pair of HDMI signals and (2) make sure they are 10% within characteristic impedance of 100 ohms.
Although the (2) could not be verified based on what they showed in the video (snippet from the video in the picture below), but (1) issue was very clear with length difference to be above recommended 2.5mm (3mm max) when you take this photo and scale it down to 1:1 using HDMI footprint as reference and then simply measure adjacent traces within each pair. IMPO the fix they showed also in the YT video - btw done by CAM engineer - which was weird because the guy was not designer but employee of PCB fab house. Who simply shifted signal path impedance to operating zone that eliminate bad pixels by changing inline components and flooding with ground. Which might have masked the issue - not fixed it. And the V&V was to plug the board back and test if those bad pixels are gone. That's not enough. And test done on flawed board design with masked issue.
I commented under this video right after it aired about possible issue being masked not really fixed so it may backfire later with new batch of boards or parts having bit different parameters, but Peri said to me his engineers fixed the problem and I should not worry about length matching between pairs. Trouble is there was no length match within each pair something they have not noticed/address. As much as I like Peri and cheer for this project to be success (I ordered one of those C64 myself), to me as pro EE this is serious concern, and I am curious if that may be the reason they delayed shipping because they scrapped the boards. I will know if that was the case once I get the new C64 in my hands where I will compare HDMI routing with what we see on the YT video and it will becomes clear "who was right". Not that I care that much but I Am just curious. The fact of the matter is there's mismatch within each pair and I hope to see this being address on the physical PCB inside the C64 I hope to receive.
The proper fix was to add single loops on each shorter route for each pair (yellow marked below) to achieve match.
With Altium Designer they used (I use it also daily for work), designer can bring up and see all routing lengths and can match within 0.000001 inch no problem.
Like on example from one of my projects below, it's all easily manageable, all it takes is to set proper rules and adhere to HDMI standard or whatever interface standard you are designing in.
Example of length matching for high-speed video interface, where evidence for (1) inter pair matching are serpents and (2) within each par length match is a little facing up loop right non the left from each serpent. That (2) was missing in new C64 routing. Hope they reviewed it and acted accordingly.
Of course, I could be wrong about the reasons for the silence and delay so time will tell. Cheers.
Although the (2) could not be verified based on what they showed in the video (snippet from the video in the picture below), but (1) issue was very clear with length difference to be above recommended 2.5mm (3mm max) when you take this photo and scale it down to 1:1 using HDMI footprint as reference and then simply measure adjacent traces within each pair. IMPO the fix they showed also in the YT video - btw done by CAM engineer - which was weird because the guy was not designer but employee of PCB fab house. Who simply shifted signal path impedance to operating zone that eliminate bad pixels by changing inline components and flooding with ground. Which might have masked the issue - not fixed it. And the V&V was to plug the board back and test if those bad pixels are gone. That's not enough. And test done on flawed board design with masked issue.
I commented under this video right after it aired about possible issue being masked not really fixed so it may backfire later with new batch of boards or parts having bit different parameters, but Peri said to me his engineers fixed the problem and I should not worry about length matching between pairs. Trouble is there was no length match within each pair something they have not noticed/address. As much as I like Peri and cheer for this project to be success (I ordered one of those C64 myself), to me as pro EE this is serious concern, and I am curious if that may be the reason they delayed shipping because they scrapped the boards. I will know if that was the case once I get the new C64 in my hands where I will compare HDMI routing with what we see on the YT video and it will becomes clear "who was right". Not that I care that much but I Am just curious. The fact of the matter is there's mismatch within each pair and I hope to see this being address on the physical PCB inside the C64 I hope to receive.
The proper fix was to add single loops on each shorter route for each pair (yellow marked below) to achieve match.
With Altium Designer they used (I use it also daily for work), designer can bring up and see all routing lengths and can match within 0.000001 inch no problem.
Like on example from one of my projects below, it's all easily manageable, all it takes is to set proper rules and adhere to HDMI standard or whatever interface standard you are designing in.
Example of length matching for high-speed video interface, where evidence for (1) inter pair matching are serpents and (2) within each par length match is a little facing up loop right non the left from each serpent. That (2) was missing in new C64 routing. Hope they reviewed it and acted accordingly.
Of course, I could be wrong about the reasons for the silence and delay so time will tell. Cheers.
Last edited:
