How essential is gold plating for PCIe edge connectors

[Nogtail]

I have a product with a PCIe card form factor. The product is intended to be relatively low cost, so a cheap PCB + BOM is essential. With an ENIG surface finish the cost is acceptable, even for small quantities. The quotes I've got for hard gold plating have more than doubled the price of a fully assembled board. It also limits where I can get the board made, as a lot of the lower cost shops don't offer it as an option.

How bad is it to use an ENIG finish for PCIe edge contacts? I know it's a lot thinner than hard gold, but for a product that is only likely to be plugged in/unplugged a small handful of times will this matter? I can't seem to find any good information on the durability of an ENIG finish for edge connectors. Does anyone know of products that use an ENIG finish for edge connectors?

Thanks!

[ataradov]

I have no scientific information on this, but in this post https://www.eevblog.com/forum/kicad/card-edge-connectors/msg2003468/#msg2003468 I have pictures of ENIG PCIe connector after more than 100 insertions. There is a visible tarnish, but the connector remains operational.

I don't have any data on whether it would be a problem for a long term operation. This tarnish may be enough to cause oxidation or something like that.

[Kenneth Rosenstroem]

I do not understand why people use gold plating. Silver is a lot more conductive than gold. It is also harder than gold, so the wear and tear should last longer with silver. Of course, there is oxidation, but if the connection is made (connector), it should not oxidise where there is contact. Maybe I am wrong, if so, please enlighten me. I know silver is used for speaker cables, and cable connectors in high-end HiFi systems, so why not in electronics? As a bonus, it is also cheaper than gold

[tszaboo]

ENIG is probably fine. Bare copper is terrible, you touch it once and your fingerprint will be visible etched into it forever.
Though I don't think hard gold plating is very expensive step if you order larger quantities. Unless you want to order the chinese special online pricing. You absolutely can get better pricing if you ask quotes. Talk to your fab.

[Hawaka]

I do not understand why people use gold plating. Silver is a lot more conductive than gold. It is also harder than gold, so the wear and tear should last longer with silver. Of course, there is oxidation, but if the connection is made (connector), it should not oxidise where there is contact. Maybe I am wrong, if so, please enlighten me. I know silver is used for speaker cables, and cable connectors in high-end HiFi systems, so why not in electronics? As a bonus, it is also cheaper than gold

Silver is indeed more conductive, but the main goal of gold plating is corrosion resistance. I do believe that it could corrode even if it is inserted, just like a screw in a nut. A contact does not stop corrosion in-between.

Silver might be cheaper as raw material, but all PCB supplier already have gold plating process. I do not know any with silver process. Therefore cost will be surely higher if you really want silver.


From my understanding hard gold is used for longevity. If only ENIG is used, then the plating would be slowly pushed away from the contact point as it is soft. This could result in bad contact after a long insertion period. At least that’s what my colleague once.

[strawberry]

I do not understand why people use gold plating. Silver is a lot more conductive than gold. It is also harder than gold, so the wear and tear should last longer with silver. Of course, there is oxidation, but if the connection is made (connector), it should not oxidise where there is contact. Maybe I am wrong, if so, please enlighten me.

silver is in good quality switches and it is x100 better wear resistance than solder plated contacts (solder turn black after some connect/disconnect operations(solder shavings and dust paste))
silver oxidize after some ~10y. cleaning and applying fresh grease/oil will do another ~10y

[dmills]

Silver oxide is conductive, it just looks ugly. The problem with silver is the sulphate which is an insulator and forms due to sulphur in diesel exhaust.

The trap with gold is that it only really works well when the mating contact is also gold (In particular, gold on tin is a nightmare, an effect called 'Tin fretting', it is real and fucked me once.

 

[E-Design]

I have a product with a PCIe card form factor. The product is intended to be relatively low cost, so a cheap PCB + BOM is essential. With an ENIG surface finish the cost is acceptable, even for small quantities. The quotes I've got for hard gold plating have more than doubled the price of a fully assembled board. It also limits where I can get the board made, as a lot of the lower cost shops don't offer it as an option.

How bad is it to use an ENIG finish for PCIe edge contacts? I know it's a lot thinner than hard gold, but for a product that is only likely to be plugged in/unplugged a small handful of times will this matter? I can't seem to find any good information on the durability of an ENIG finish for edge connectors. Does anyone know of products that use an ENIG finish for edge connectors?

Thanks!

Since you state that it will only cycle a small number of time (say 10 or less) in the life of the product, ENIG only (no hard gold) should be fine. Ive done it before and has proven reliable. It really is all about the cycles the connector will see in its life.

[TimFox]

Silver does not oxidize under normal conditions.
The black tarnish that happens in normal indoors operation is sulfide.
In good switches with silver contacts, the surface is cleaned by the wiping operation of well-designed sliding contacts.

[Ground_Loop]

I think I would evaluate my client's wants and needs and whether a better plating adds enough value to them that they are willing to pay for it.

[langwadt]

I have a product with a PCIe card form factor. The product is intended to be relatively low cost, so a cheap PCB + BOM is essential. With an ENIG surface finish the cost is acceptable, even for small quantities. The quotes I've got for hard gold plating have more than doubled the price of a fully assembled board. It also limits where I can get the board made, as a lot of the lower cost shops don't offer it as an option.

How bad is it to use an ENIG finish for PCIe edge contacts? I know it's a lot thinner than hard gold, but for a product that is only likely to be plugged in/unplugged a small handful of times will this matter? I can't seem to find any good information on the durability of an ENIG finish for edge connectors. Does anyone know of products that use an ENIG finish for edge connectors?

Thanks!

Since you state that it will only cycle a small number of time (say 10 or less) in the life of the product, ENIG only (no hard gold) should be fine. Ive done it before and has proven reliable. It really is all about the cycles the connector will see in its life.

afaik a regular PCIe connector is only rated for 50 insertion cycles

[strawberry]

Silver does not oxidize under normal conditions.
The black tarnish that happens in normal indoors operation is sulfide.
In good switches with silver contacts, the surface is cleaned by the wiping operation of well-designed sliding contacts.

heed someone to exercise switches regularly otherwise oxide thickens too much
old style switches are better than miniature (thicker silver and more contact pressure)

[Nogtail]

In my case I'm mostly using the connector for power/low speed data. The power requirements are also very low so I'm not too worried about slightly higher resistance in the connectors unless it gets crazy high. I'd imagine the connector will get 2 mating cycles for 90% of customers - once for production test then again for installation into a system. I'd imagine around 10% of customers might be "enthusiasts" who give it a few more cycles. I might try and do some accelerated life testing when I get the prototypes to see how well ENIG holds up.

[Terry Bites]

Its just a way of moving gold into landfill never to be revovered.
It looks cute but given the very few inserions cycles (usually 1) it makes little difference.
Once you gave a gas tight mating any contact finish will be relaible.
We managed for most of the past 100 years without gold plating on connectors and the world did not end.

[tszaboo]

Its just a way of moving gold into landfill never to be revovered.
It looks cute but given the very few inserions cycles (usually 1) it makes little difference.
Once you gave a gas tight mating any contact finish will be relaible.
We managed for most of the past 100 years without gold plating on connectors and the world did not end.

Have you tried a vintage headphone recently? Constant contact issues with the SCART connector? Internet disconnecting because of the terrible connectors on the POTS line? CRT TV loosing picture that get suddenly resolved if you smack it from the side?
Don't act like this is a non existing issue.

[tooki]

It looks cute but given the very few inserions cycles (usually 1) it makes little difference.
Once you gave a gas tight mating any contact finish will be relaible.

Well that’s true, except for the part (namely, the entirety of it) where it’s actually completely incorrect.

https://www.digipas.com/documents/Effects-of-Fretting-Corrosion-on-Au-Sn-Plated-Contacts.pdf

[tooki]

https://eprints.whiterose.ac.uk/4940/1/repository_liskiewicz_paper09.pdf

[tooki]

https://suddendocs.samtec.com/productspecs/fretting-corrosion.pdf

[MisterHeadache]

Silver oxide is conductive, it just looks ugly. The problem with silver is the sulphate which is an insulator and forms due to sulphur in diesel exhaust.

The trap with gold is that it only really works well when the mating contact is also gold (In particular, gold on tin is a nightmare, an effect called 'Tin fretting', it is real and fucked me once.

 

What he said.  Mixing plating in electrical interconnects is a very bad idea (I know, that's off topic from OP).  Silver's sensitivity to sulfur also gets pooh-poohed too often because it depends on the exposure PPM over time.  And there's lots of sneaky sources for sulfur exposure (e.g, nearby rubber gaskets, cardboard packaging, and greases, to name a few).

[niconiconi]

I do not understand why people use gold plating. Silver is a lot more conductive than gold. It is also harder than gold, so the wear and tear should last longer with silver. Of course, there is oxidation, but if the connection is made (connector), it should not oxidise where there is contact. Maybe I am wrong, if so, please enlighten me. I know silver is used for speaker cables, and cable connectors in high-end HiFi systems, so why not in electronics? As a bonus, it is also cheaper than gold

From what I've read, immersion silver just looks horrible after oxidation so clients and consumers don't want it.

Immersion silver is also a very appealing finish. It is being used successfully on consumer electronics in China. Some
fabricators in the United States have installed electroless silver plating lines and are using them on production PCBs. To date, the results have been good. There are some reports that when soldering the first side of a double-sided assembly, the silver on the second side corrodes rendering it difficult to solder.

My view on this is much like the experience we had with ENIG. In the early days of its usage, everything looked great. Slowly, we began to experience failures from black pad, until one day, many of us had disasters on our hands. As a result of this experience, I choose to wait and see how this works for a while before jumping on to the immersion silver bandwagon. After all, most of us still have to go through the lead free solder transition and don’t need a surface finish problem on top of everything else.

Figure 4.38 is a picture of a PCB with immersion silver. A number of network equipment manufacturers have tried immersion silver on complex multilayer PCBs such as this only to find that after a few months the silver plating in press fit holes, test points and other places where the silver is exposed has corroded to the point that the PCB looks like it should be thrown away. In fact, the test points on the back side of the PCB in Figure 4.38 are already tarnished and the PCB has not been used yet. While the PCB may still be functional, it looks so bad cosmetically that customers who pay large sums of money reject them. For this reason, immersion silver is not a good choice for anything but consumer electronics where the PCBs are never seen by the customer.

- Right the First Time, Volume 2 - A Practical Handbook on High-Speed PCB and System Design, Lee W. Ritchey.

[vidarr]

I am very serious about my computers and I always do gold and I have never had a problem ----ever. Many of my machines get taken apart and re-arranged dozens of time. Most of my very best machines I even keep on test benches, because of this.

Re: Gold on tin:

"... an effect called 'Tin fretting',..."

I don't think you will run into this problem, if you buy from quality manufacturers. I don't even know if they do this method now, but I am not an expert on this exact thing. I own tinned copper cookware and know how to retin them. That is the extent of my tin knowledge. I don't doubt this other guy had this problem, but I never heard of it before.

[dmills]

I had to have the office recall over 100 units of a fairly expensive broadcast industry box because on a ONE character error on the BOM that made one half of a mating set of connectors tin not gold.
These were Samtec parts, so not junk, and were failing in fly packs where the kit was moved by air freight. 

You could actually see the tin fretting but needed a decent microscope to identify it, lots of red faces over that one.

[Psi]

Silver plating is a bit of a nightmare for PCBs, sometimes it will tarnish super fast and other times it wont.
It can be quite unpredictable.

[Haenk]

I think there is a reason, why memory modules and computer "cards" switched to gold plating many years ago - and never went back. Even though those are hardly ever moved around.
And honestly, there is almost no gold involved at all.

[vidarr]

I think there is a reason, why memory modules and computer "cards" switched to gold plating many years ago - and never went back. Even though those are hardly ever moved around.
And honestly, there is almost no gold involved at all.

This.