ANN.lu archives - Pegasos "too cheap" rumours debunked (comments,flat view)

26-Apr-2024 01:19 GMT.

UNDER CONSTRUCTION

Anonymous, there are 68 items in your selection [1 - 50] [51 - 68]

[Rant] Pegasos "too cheap" rumours debunked

Posted on 24-Jul-2004 12:37 GMT by Johan Rönnblom

68 comments
View flat
View list

For some time now, rumours that the Pegasos is sold at a cheaper price than production cost have been frequent in some circles. Recently, these claims were brought out into the open and could quickly be shown to be based upon incorrect assumptions about the Pegasos hardware. For some time now, rumours that the Pegasos is sold at a cheaper price than production cost have been frequent in some circles. In this thread well known AmigaOS4 contributor Stefan Burström brought the rumour out into the open claiming: "The USERS of a cheap, subsidised mainboard are happy because they have cheap hardware. However, they did not pay the actual cost of the hardware. [...] The Pegasos users may be happy for a short while the Pegasos is cheap, but the truth is that it doesn't finance itself."

He later clarified himself to speak only about the Pegasos 1: "Well, I was refering to Pegasos 1 since that is the only board I have made any homework on, so don't put any words in my mouth I didn't speak.", "I brought it up simply because this 'subsidised' discussion has been here before so I decided to do some homework. On _that_ board. I have no information on the Pegasos II so I decided not to discus it. Simple eh?" and "I insist on it because I am not claiming that the Pegasos 2 is subsidised. Simple eh? This whole subsidised story started with the Pegasos 1 and back then I supported it and did some homework."

Stefan supported his claims by stating that he had experience in the field: "Oh, btw, a part of my professional job is to design cost effective consumer electronics, so I think I have a fair amont of knowledge of the actual costs associated with PCB manufacturing."

He then claimed that based upon his calculations of the Pegasos mainboard PCB cost, the machine must be too expensive to make: "I started out with the PCB to get a starting point of the discussion. [...] I find it hard to believe that a board like this would have a PCB with a cost of 1/3th of the total BOM [Bill Of Materials]."

He explained that his guess was based mainly of his estimate of the PCB cost: "I started building a BOM way back yes. I guess I still have the draft somewhere on my old A4K. I never got that far as checking prices for the more advanced chips though." and "But fwiw, I calculated the PCB cost now just because it was the easiest one to do with most chance of getting accurate prices even 2 years back. For the NB, SB, Ethernet Phy, AC97 etc. it would have been much harder to find the accurate numbers which I started to look up way back."

Stefan's estimate of the mainboard PCB cost: "Right but it is still a ~100 sq inch PCB. 6 or 8 layers I'd guess. Microvias between layer 1-2 and 7-8 to be able to route the BGA's. A small scale production run of such a PCB easily reaches 100 USD per board. And that is before the startup costs for the PCB fab is distributed on the boards." and later clarified that "The expensive part is the micro via layer, not the actual # of the layer it goes through."

The inclusion of a the cost for a micro via layer did not come from knowledge about the Pegasos 1 board, however: "I havn't seen anything but pictures of a Pegasos so I havn't been able to inspect the boards."

Instead, he motivated it by referring to his stated knowledge about PCB design: "Nope, since I know that the Artica is a 492 pin BGA with a ballpitch of around 1.27 mm. Further more, the southbridge is is most likely as similar package as the VT82C686 (I have the datasheet here) which also has ballpitch of 1.27mm. Given a track width of 5 mils and clearance of 5 mils that would make it impossible to route using only through hole vias. Convinced yet?"

However, the fact is that the Pegasos (1 and 2) boards have six layers, that the area is 63 square inches rather than 100, and that they do not have any expensive micro vias.

Thus, it seems that the rumours that the Pegasos 1 (and Pegasos 2, even if Stefan is not among those making that claim) is based on incorrect assumptions about the Pegasos hardware.

Finally, I'd like to give Stefan some credit for having the guts to bring this up in public, rather than keeping it "behind the scenes" where these claims are seldom questioned and are quickly accepted as facts by many people.