Compelling reasoning for NOT using chip sockets?

[+OLD CS1]

Huge bugaboo for me.

 

Any time I see projects which talk about bending pins on ROM, RAM, or other manufactured chips (especially difficult to find ones,) or soldering directly to the same, I cringe. I practically rage (being still at the age when doing so does not necessarily have effect on my ultimate longevity) when I see things like this:

 

eBay Auction -- Item Number: 231487612877

 

 

Notice the JiffyDOS ROM in the bottom-left is soldered directly to the board. As well, it appears the selection switch is soldered to a pin bent outward from the ROM.

 

When I built my stereo SID I read to solder capacitors directly to the SID chip, bend out pins, solder wires to pins, and I though, "but I'll ruin the damned thing for sure!" That was back in 1992 when it was easier for come across SIDs. But 20 years later, not a friggen chance (although Dave Haynie had something like 19 of them up in his moving sale.) Anyway, instead of performing all that surgery on the delicate silicon masterpiece I did it on a socket, then placed the SID into the socket. Worked like a charm (probably still does, but my 128D is not in place right now.)

 

Other than potential clearance issues with other components or enclosures, is there any compelling reason for not using sockets on projects like this?

 

 

[Keatah]

I built a high reliability Apple //e Platinum machine. It uses no sockets. Of course it is much more tedious to do work on and you can't easily swap chips to troubleshoot. And same for upgrades, no easy removal.

 

What is gained is reliability and vibration resistance. Less chance of intermittent contacts too. So the reliability cancels out the tediousness of frequent repairs.

Edited March 4, 2015 by Keatah

[Rybags]

In the day probably knocked $10-$20 off materials costs and simplified manufacturing at the expense of more labour for any failed machines.

 

I guess by the mid 1980s the chips were more reliable and market more competitive to the point where it became more economical to just go most/all non-socketed and set aside failed machines for the more intensive repair process they'd require.

[+OLD CS1]

I can see in terms of manufacturing, I just cannot see a reason not to do it for home projects. I know about the "walking chip" problems, but I always took that as an acceptable risk.

[+InsaneMultitasker]

When I repair or refurbish cards, I always grimace at finding directly-soldered chips. For example with Myarc cards, the most commonly replaced chips are the bus buffer chips (LS244/LS245) and 75% of the time they are not socketed. Socketing these chips allows for easier testing and troubleshooting. I would think the same is true in home-brew projects especially when the circuit or idea is in-process.

 

Personally I use either a good double-wipe socket or machine pin socket. Do not use single-wipe or cheaply made sockets. Many of the Horizon RAMdisks were built with these sockets and as some people here can tell you, the future pain is not worth the few cents saved.

[+save2600]

I built a high reliability Apple //e Platinum machine. It uses no sockets. Of course it is much more tedious to do work on and you can't easily swap chips to troubleshoot. And same for upgrades, no easy removal.

 

What is gained is reliability and vibration resistance. Less chance of intermittent contacts too. So the reliability cancels out the tediousness of frequent repairs.

Not talking specifically about your case, but I bet a bit of that reliability in soldering directly is negated when (not if) someone leaves the iron on too long and weakens/overheats a part of the I.C.

 

I'm with the OP... whenever performing an upgrade, best to always use a socket. No chance of damaging a chip and simple replacement when the time comes. Can only remove an I.C. so many times before you make a total mess out of a board.

[Keatah]

My hi-reliability //e doesn't sit around at home all the time.

 

For a home/hobbyist environment sockets are just fine.

[Keatah]

Not talking specifically about your case, but I bet a bit of that reliability in soldering directly is negated when (not if) someone leaves the iron on too long and weakens/overheats a part of the I.C.

 

Thought of that too. I used a clip heatsink for each pin when I soldered. Then we conformal-coated the board and plated the expansion slots. The basic standard peripherals were soldered in too. Disk Controller, 64K/80col card, SuperSerialCard..

Edited March 4, 2015 by Keatah

[+acadiel]

When I originally designed the 16K cart board, I made a decision to just solder the 74LS379's. After a handful of them went south (probably due to IL winter static), I started socketing later 16K and then the 64K boards, and we have put sockets ever since. Makes sense for ease of replacement.

[+save2600]

Thought of that too. I used a clip heatsink for each pin when I soldered. Then we conformal-coated the board and plated the expansion slots. The basic standard peripherals were soldered in too. Disk Controller, 64K/80col card, SuperSerialCard..

Isn't that conformal coating a total PITA to deal with when you need to replace components though??

 

I've worked on refrigerator logic boards with that crap on 'em and ugh... guessing there's a compound to clean/eat it away if you need to desolder something?

 

Have seen some kind of a coating on computer motherboards too lately.

[+Ksarul]

Conformal coating is great to protect the boards in rough environments (from chemical exposure to being underwater). It can be a pain to remove, especially the old epoxy-style stuff that covered the board to a depth of about 5-10 mm. That stuff buried the components completely and you had to remove it carefully with a Dremel tool to get at the component you wanted to replace--and then fill it all back in when you were done. It was usually a block covering the whole board. . .