To socket or not to socket? Iron or chip? Thoughts?
« on: January 09, 2010, 02:34:09 PM »
Hi people,

Perhaps this is my OCD reassurance seeking but... I'm wondering people's thoughts on two things...

1. IC sockets - good or bad?  I have heard that IC sockets increase parasitic weirdness and other such unpleasant side effects (not that I can say I have ever heard it my self).  But... hot solder can damage ICs.  Is there a consensus as to which is better - direct solder on the IC (be quick) or socket?

2. In an active summing stage (into the inverting input of the opamp) the output from that opamp results in inverted polarity.  Should I run it through a second a DOA (inverting input) to flip it (about $20 with a DIY opamp) or use an output transformer to flip it (similarly about $20ish from Classicapi.com).  Or, I guess... a cheaper IC at unity gain.  I gather it matters a little bit if the output is going outside of the unit to another unity.  The transformer there acts as a balancing stage.  But if the signal needs to remain unbalanced which is best?

THanks.

CC


tv

Re: To socket or not to socket? Iron or chip? Thoughts?
« Reply #1 on: January 09, 2010, 03:00:13 PM »
I socket stuff on my perfboard prototypes ...

Usually not on my pcb builds.
If you sprinkle when you tinkle, please be neat and wipe the seat.

keefaz

Re: To socket or not to socket? Iron or chip? Thoughts?
« Reply #2 on: January 09, 2010, 03:54:47 PM »
direct solder on the IC (be quick)
The datasheet of OPA604 says the DIP version can stand 10 seconds lead soldering at +300°C temperature(absolute maximum rating)
I think you need only 2-3 seconds maximum or so to solder an IC lead on the PCB.
The same datasheet also advises to solder the IC directly to achieve best heat dissipation

But all your questions depend on the final product requirement.. For exemple a console input module could benefit from IC on socket, if the console is used a lot of time and will be subject to various abuses, it could be convenient to replace fried ICs if they are on socket, or in the case of IC upgrades

PRR

Re: To socket or not to socket? Iron or chip? Thoughts?
« Reply #3 on: January 09, 2010, 06:50:54 PM »
> hot solder can damage ICs.

No, it won't. Not if you know how to solder as keefaz says: get in and get out in a few seconds.

FWIW: power transistors can be so hot the solder on their legs is molten, and they still do their job. The Silicon is not damaged. The leakage current is out-of-spec while hot but that's temporary.

Most sockets can be damaged by careless soldering.

Use sockets whenever possible. Sane and sanitary Audio systems don't have "socket weirdness". If you think you hear some, apply C-37.

Re: To socket or not to socket? Iron or chip? Thoughts?
« Reply #4 on: January 09, 2010, 09:02:22 PM »
Good to know.  I don't know where I picked up that sockets cause parasitic stuff.  Also good to know that solder heat doesn't hurt (well if you know how to solder - which I do - getting in and out).

I apply both suggestions.  I'll use sockets from here on in.

Iron or Opamp for polarity reversal - any takers?

CC

jdbakker

Re: To socket or not to socket? Iron or chip? Thoughts?
« Reply #5 on: January 09, 2010, 10:05:02 PM »
For a bit of a counterpoint, the majority of all broken equipment that ends up on my desk for post-mortem or repair has electromechanical 'issues' either causing or contributing to their failure. Switches, pots, relays, connectors and sockets: I've grown to be weary of them.

(context: re-capping is seen as maintenance not repair here, or I'd likely see more of that too)

Use a socket if you think it's likely you'll need one. I would use a socket for:

- chips that connect to the outside world. This includes line drivers and receivers.
- all chips in subsystems with loose standards for voltage rails and I/O. In audio that would be 500-series modules.
- anything that runs at or close to voltage and/or power dissipation limits.
- prototypes or anything else that you want to experiment with (obviously).

On the other hand, for anything that gets taken on the road or otherwise sees a lot of vibration or temperature cycling I'd recommend against sockets.

Consider how hard it is to get the part out without damaging the board. Double-sided boards with plated through holes are better candidates for sockets than single-sided ones.

I prefer to use the more expensive sockets with machined/turned contacts; I've had less trouble with them.

Good to know.  I don't know where I picked up that sockets cause parasitic stuff.

They do, but you'll need something faster than a 5532 to bring it out. A 6172 might be troublesome, depending on the rest of the circuit.

JDB.
« Last Edit: January 09, 2010, 10:07:20 PM by jdbakker »

tv

Re: To socket or not to socket? Iron or chip? Thoughts?
« Reply #6 on: January 10, 2010, 02:22:10 PM »
6172 is good if you mount the PDIP part similar to the soic part - like a "bigger" smd, on traces (but you must design the pcb in such a manner).

this gives you space to place the bypass caps tru-hole right on "top" of the chip (on the other side of the pcb), rigidly and in the closest proximity to the chip == no caps hanging on the bottom of the pcb.

plus it actually looks badass.
If you sprinkle when you tinkle, please be neat and wipe the seat.

blue_luke

Re: To socket or not to socket? Iron or chip? Thoughts?
« Reply #7 on: January 10, 2010, 02:54:50 PM »
JDbakker sums it up quite nicely but I will add my two cents here..

Most mass assembled circuits are assembled by robots (pick'n place) inserting a chip or a socket is the same work, but then I think the chip must be inserted in the sockets by hands afterward.
I am not sure but I think there is no way to mechanically insert the chips in the sockets.

Concerning the 'funnies' sockets may provokes, it is true in RF, fast logic and video circuits.
Most of the time you will not see sockets in those departments.
Hi-speed ADC and DAC falls in that category!

For my own audio stuff I always use sockets, and the cheap one to boot!!

Here is a question... What is C-37??  I can't figure that one! :) a chemical or an expression?

s2udio

On the end of a Rural Twisted Pair.

blue_luke

Re: To socket or not to socket? Iron or chip? Thoughts?
« Reply #9 on: January 10, 2010, 03:10:55 PM »
Seriously???   :o


blue_luke

Re: To socket or not to socket? Iron or chip? Thoughts?
« Reply #10 on: January 10, 2010, 03:12:41 PM »
Now I understand the title of a thread somwhere else here...
'forget C-37! this is the real stuff!''
For me a C-37 is a vintage sony microphone... and a good one!
:)

Andy Peters

Re: To socket or not to socket? Iron or chip? Thoughts?
« Reply #11 on: January 10, 2010, 04:28:43 PM »
1. IC sockets - good or bad?  I have heard that IC sockets increase parasitic weirdness and other such unpleasant side effects (not that I can say I have ever heard it my self).  But... hot solder can damage ICs.  Is there a consensus as to which is better - direct solder on the IC (be quick) or socket?

This engineer, and the company for whom he works, hates sockets. Unreliable, lousy signal integrity, extra cost, extra assembly step, yadda yadda yadda.

Then again, I haven't done a design with a socketable through-hole component in years, with the exception of a CMOS image sensor from Micron which came in a 241PGA. (And the sensor doesn't work, so the design was scrapped.)

And now that we have low-cost in-circuit programming you don't need sockets for memory devices.

-a
"On the Internet, nobody can hear you mix a band"

tv

Re: To socket or not to socket? Iron or chip? Thoughts?
« Reply #12 on: January 10, 2010, 05:10:11 PM »
..she's got a socket in her pocket..
If you sprinkle when you tinkle, please be neat and wipe the seat.

JohnRoberts

Re: To socket or not to socket? Iron or chip? Thoughts?
« Reply #13 on: January 11, 2010, 10:31:36 AM »
+1 to what Andy said. Most new chips don't use sockets.

We stopped using sockets at my old job when their cost effectiveness diminished. There is no value in making chips easy to replace if they don't fail.

IC reliability has improved so much over the last few decades that ICs don't just fail randomly any more. IC failures are usually caused by some external stress and it is cheaper to protect the IC from that stress and avoid the service call entirely.

----
To answer blue-luke  yes you can machine insert IC and Sockets, we did for many years at my old day job (I think the machines were called HPDI but not sure on that).
-----
+1 to JDB, yes chips that touch the outside world are the most susceptible to failure, but that too can be mitigated with decoupling and clamps.

When we did the sharp pencil analysis before eliminating sockets from our thru hole designs we did see the trend JDB describes.  But anything you can measure you can improve.
 
JR
Don't only half-ass tune your drums. Visit https://circularscience.com to hear what properly "cleared" drums sound like.

owel

Re: To socket or not to socket? Iron or chip? Thoughts?
« Reply #14 on: January 11, 2010, 01:28:50 PM »
For production, IC sockets are not necessary. ICs can be soldered directly to the board.

But for DIY projects, IC sockets allow for easy experimentation. (and repair)

If you're going to spend extra for sockets, just don't use cheap IC sockets.

Some cheap ones squeeze the IC's out (with repeated heating and cooling)... and requires reseating of the chips. Or the socket terminals develop oxide and requires the ICs to be pulled out and reinserted again to "clean" the terminals.

Re: To socket or not to socket? Iron or chip? Thoughts?
« Reply #15 on: January 13, 2010, 10:15:00 AM »
I recently fixed a Marshall Shredmaster distortion pedal by putting in sockets.  The pedal was working well, other than making a loud ocean surf sound behind the signal.  I first retouched the solder on the existing TL072 opamps, with no improvement.  I then desoldered the opamps and popped in sockets thinking that I might want to play around with different replacement opamps for fun.  When I put the original opamps back in the sockets, the pedal worked perfectly.  Go figure.

Andy Peters

Re: To socket or not to socket? Iron or chip? Thoughts?
« Reply #16 on: January 13, 2010, 10:31:33 PM »
I recently fixed a Marshall Shredmaster distortion pedal by putting in sockets.  The pedal was working well, other than making a loud ocean surf sound behind the signal.  I first retouched the solder on the existing TL072 opamps, with no improvement.  I then desoldered the opamps and popped in sockets thinking that I might want to play around with different replacement opamps for fun.  When I put the original opamps back in the sockets, the pedal worked perfectly.  Go figure.

The sockets didn't solve the problem.

Removing the part, cleaning the holes and doing a decent soldering job on the socket solved the problem.

-a
"On the Internet, nobody can hear you mix a band"


 

Related Topics

  Subject / Started by Replies Last post
5 Replies
1768 Views
Last post January 14, 2010, 09:16:20 PM
by pH
7 Replies
1891 Views
Last post September 23, 2010, 04:19:54 PM
by zayance
0 Replies
1323 Views
Last post February 27, 2011, 10:01:29 PM
by kambo
2 Replies
1021 Views
Last post June 09, 2011, 07:57:59 PM
by njm