The human factor and PCI SIG
It has also happened to us that we think the Astron plug is already at the limit and fully plugged in. You can see very clearly in the photo below that the remnants from the injection molding can not only make inserting (“threading”) the plug extremely difficult, but they also push themselves into the very narrow space between plug and socket as a kind of wedge. The further you insert the plug, the greater the resistance until you think the plug is fully inserted. The manufacturing tolerances of the Astron connector seem to be quite large in some cases, because the locking mechanism is virtually non-existent on some adapters, or at least tactilely imperceptible. The quality of the injection molding must definitely be questioned here.
It is more appropriate to use nylon 66 (PA66) rather than nylon 6 (PA6) because it is much more temperature resistant. However, PA66 is somewhat more granular and therefore more difficult to process. From my point of view, it is a serious defect that the edges of the connectors were not chamfered or that they did without any post-processing. I also informed NVIDIA about this yesterday. Last week, I had a conversation with a supplier about the problems with injection molding, and he definitely believes that PCI SIG is also responsible for this. What applies to the beveled inner contacts of the socket must also apply to the outer edges of the connectors.
Then it also works vertically and without force. In the end, as always, it is also a question of the right leadership. Because what happens if you don’t get the plug inserted vertically? That’s right, you look for the path of least resistance, angle the plug and look for the two most lateral contacts of the socket to start here. That, in turn, is deadly to any Tulip design, especially if it’s double-slotted! And if you pay attention: Depending on whether the user is left- or right-handed, it hits the other outer side with the scorched contacts.
This may explain why one side is particularly hard hit. If, in rare cases, center contacts are also affected, the user has certainly angled the connector on the long sides. The cases in the Founders Edition are rarer, which is probably also due to the fact that the socket is freely accessible. But more on that in a moment when we talk about the coolers.
A revision of the connector by the PCI SIG has already been discussed in the media. Here, too, Gabriele Gorla was able to personally confirm once again what I had already learned from the third-party suppliers last week. But double is better. The currently planned changes will only affect the four Sense Pins, but they are quite a real solution. Due to the shortening of the contacts, the sense pins only become contactable when the plug has been fully inserted!
Conversely, this means that the graphics card will no longer start without the first two sense pins being assigned or recognized. Only PCI SIG itself knows why this was not planned from the outset. If, in a second step, the shape of the connector housings could be corrected by specifying beveled or chamfered edges, a large part of the problems on the customer side would automatically disappear.
Insertion and removal with artificial obstacles
But not only the connector and its production can mislead the user, but also the environment of the connector! Unfortunately, many board partners did not think along with the graphics card design and they put design and optics above functionality. To understand my criticism, I will now show you two cards, both with 12VHPWR connector, but with different coolers. I had already addressed this sore point in one of the linked articles and even cut my own fingers bloody.
You can see very clearly on the upper solution that the fingers find enough space to be able to grip the the plug properly and also push it in far enough without just pressing on the plug end or the cables. It is also still possible to grip or stabilize the connector from the side.
In addition, releasing the lock when pulling out the plug is still very easy to do later. With the lower cooling solution, the plugging in is already connected with hurdles, because a few millimeters of space are missing (also on the side) and you cannot get the plug out again without the risk of a finger injury.
Here, too, I see the PCI SIG as having a duty. What has been mandatory for CPU sockets for decades is called “Keep out Area” and describes the areas next to the socket that must not be covered or used by backplates or brackets. What is the problem with defining such an ergonomically necessary zone for this connector as well? A simple bounding box could be used here to ensure that the user does not have any problems when plugging in or unplugging such a connection. I also communicated this to NVIDIA yesterday, because if the PCI SIG is too inflexible as a cumbersome consensus apparatus for such basics, NVIDIA could simply give this to its own AICs as part of the Green Light Program. Personally, I think this is even more essential than the Pantone-accurate proofing of the box art and its colors.
Further articles and basics
321 Antworten
Kommentar
Lade neue Kommentare
Veteran
1
Urgestein
Urgestein
Urgestein
Veteran
Veteran
Urgestein
Mitglied
Mitglied
Veteran
Veteran
Veteran
Veteran
Veteran
Urgestein
Veteran
Veteran
Urgestein
Alle Kommentare lesen unter igor´sLAB Community →