To socket or not to socket

JulesWallet

Member
Joined
Aug 8, 2009
Messages
331
Reaction score
11
Location
Glendale, Arizona
I just picked up the 74ls chips for my Berzerk ZPU. The original chips are soldered right to the board. Should I put sockets in and use them or just solder the chips directly to the board again?

Any input out this would be appreciated.

I have attached at photo of the chips on the board.

Thanks.
 

Attachments

  • 103_2277.jpg
    103_2277.jpg
    102.5 KB · Views: 48
Last edited:
I just picked up the 74ls chips for my Berzerk ZPU. The original chips are soldered right to the board. Should I put sockets in and use them or just solder the chips directly to the board again?

Any input out this would be appreciated.

I have attached at photo of the chips on the board.

Thanks.

I prefer sockets myself.
 
Always socket. They are cheap and will save you or someone a LOT of work later.

I don't really understand this point of view. What work later on are you saving? I can see socketing certain kinds of IC's like RAMs, ROMs, PROMs, and other devices that are more prone to failure in the (relatively) short-term. But socketing standard TTL logic gates seems like a waste.
 
I don't really understand this point of view. What work later on are you saving? I can see socketing certain kinds of IC's like RAMs, ROMs, PROMs, and other devices that are more prone to failure in the (relatively) short-term. But socketing standard TTL logic gates seems like a waste.

There are plenty of 74LS chips that are common failure points (244, 161, ....). And repeated failures do happen (sometimes due to poor board design)... and it's great when you find that the failed chip is already socketed. There's no good reason not to socket them. Are you saying you've never found a failed logic chip that has been replaced before? I certainly have, and I'm not in the repair business (I only fix boards for myself and friends). Finding a failed chip that has been replaced already and is soldered directly to the board is very aggravating....
 
There are plenty of 74LS chips that are common failure points (244, 161, ....). And repeated failures do happen (sometimes due to poor board design)... and it's great when you find that the failed chip is already socketed. There's no good reason not to socket them. Are you saying you've never found a failed logic chip that has been replaced before? I certainly have, and I'm not in the repair business (I only fix boards for myself and friends). Finding a failed chip that has been replaced already and is soldered directly to the board is very aggravating....

Actually there IS a good reason not to socket them - the sockets can become a failure point! On larger boards (for instance, Missile Command), you can get a LOT of flex when moving that thing around. Depending on your sockets, and where the chip in question is on the board, you can easily cause the the chip to move around by flexing the board.

I'm not saying that it's going to happen - most of the time it won't. But I do guarantee that the next time you have a problem, the first thing you're going to have to do is to re-seat all the socketed parts.

It's a choice between having to solder the same part more than once (with the attendant risks of damaging the board) and having to deal with socket problems. I'm not saying don't socket, I just don't like to hear people saying that there's no downside to socketing.

Personally, I wouldn't socket TTL logic unless it dies a LOT. And then I'd have to start to try to figure out if there isn't some way I can fix the board, rather than repeatedly replace the part.
 
Ok, clearly there is a reason that they don't socket EVERY chip on the boards. I agree, that would be problematic. But in general, having one new socketed chip isn't likely to cause problems if done right.

And it can save time later and, as mentioned before, reduce the odds of damaging the board doing repeated repairs.
 
Ok, clearly there is a reason that they don't socket EVERY chip on the boards. I agree, that would be problematic. But in general, having one new socketed chip isn't likely to cause problems if done right.

And it can save time later and, as mentioned before, reduce the odds of damaging the board doing repeated repairs.

Send it out! :) I understand where you're coming from but I'm just saying, if you reworking twice on one IC might damage the board, just send it out for repair.
 
Actually there IS a good reason not to socket them - the sockets can become a failure point! On larger boards (for instance, Missile Command), you can get a LOT of flex when moving that thing around. Depending on your sockets, and where the chip in question is on the board, you can easily cause the the chip to move around by flexing the board.

If you use a good quality socket then that really shouldn't be an issue. Friends don't let friends use single-wipe.... :D
I guess I'm a bit spoiled... I can buy most dual-wipe and machine-pin sockets for a dime a piece....
 
There are plenty of 74LS chips that are common failure points (244, 161, ....). And repeated failures do happen (sometimes due to poor board design)... and it's great when you find that the failed chip is already socketed. There's no good reason not to socket them. Are you saying you've never found a failed logic chip that has been replaced before? I certainly have, and I'm not in the repair business (I only fix boards for myself and friends). Finding a failed chip that has been replaced already and is soldered directly to the board is very aggravating....

I've never had to replace the same TTL logic gate IC twice, even with the boards that come broken from other repair houses. I'm not saying that socketing every replaced chip is 'bad' or 'wrong', but from past experience it just seems unnecessary since they don't fail that often. Now I will socket if I know I'm going to be troubleshooting the same chip, i.e. lifting a drive or load pin. But otherwise I skip it. If there's one reason I can think not to socket it's because the socket would obscure the land pads on the component side. If there's a case where a microfracture exists between the signal trace and the pad annulus (which does happen!) then you'd have to lift the socket in order to find it.

EDIT: I take that back. There was one time where a new counter or adder failed within an hour while on the bench. Must've been a oddity at the very front of the bell curve. It was a bitch to find too! All the rest in the batch were fine. Still, after about 7-8 years and a couple hundred boards that's the only incident I can think of.
 
Last edited:
I don't really understand this point of view. What work later on are you saving? I can see socketing certain kinds of IC's like RAMs, ROMs, PROMs, and other devices that are more prone to failure in the (relatively) short-term. But socketing standard TTL logic gates seems like a waste.

+1

Brian
 
Why weren't all the ICs originally socketed to begin with? To save on costs?

Because aside from the cost, that's just BEGGING for trouble. Socketing a few ICs is OK if you want to (as Matt mentioned, good sockets don't let loose that often), but if you socket EVERY IC, then the chance of something going wrong gets much higher every time the board flexes. Can you imagine something like the Missile Command PCB with all those chips socketed? You'd find a couple of chips in the bottom of the cabinet every time you shipped the game!
 
Quality of the sockets is the key issue. Using single leaf sockets is always asking for trouble. Some people only use twin leaf sockets...which is usually ok. I always use machine pin sockets and socket every IC I work on.
When we made military boards with DIP type IC's, we socketed every single IC. These didn't have the lightweight plastic DIP's but the heavy ceramics instead. The boards went through shake-and-bake testing with shaking that far exceeded anything any arcade game would go through and have never had an IC pop out of a socket. With 80 or more ceramic IC's per board, these puppies got pretty heavy.

Ed
 
So basically, what it boils down to is....socket the chip if you want. But either way someone on here will think you are either smart or stupid no matter what you do.
 
Back
Top Bottom