Discrete Opamp Open Design

[linuxguru]

Originally Posted by grhughes

There are a host of footprints out there; The Deane Jensen 990, The Pacific Recorders 990, the John Hardy 990, the API2520 and DIP8. And what about FM Acoustics modules? I think there probably needs to be an adapter for each case. And in each case I don't think this is gona be plug and play...

I don't know if this is a workable option, but here is a first-pass layout of an API2520 to DIP8 single-opamp adapter. Of course, the opamp will have to be chosen to be able to handle high rail voltages of +/- 24V if necessary, which limits the available choices (the alternative is to put two low-noise low-dropout +/- 15V regulators on the adapter board, but that opens up another can of worms about dirty/clean grounds, input signal clamps, etc.).

R1 may be a zero-ohm resistor, in which case the adapter ground-plane is directly hitched to the dirty ground, or it may be a lift resistor of a few ohms to isolate it from the dirty ground. The C1,C2 rail-to-ground bypasses are optional.

It should be possible in principle to use a DIP8 discrete (like the folded Kaneda I posted earlier) in place of a monolithic opamp with the adapter.

[grhughes]

"..... here is a first-pass layout of an API2520 to DIP8 single-opamp adapter." linuxguru
Thank you! Ray

[dchisholm]

Originally Posted by linuxguru

. . here is a first-pass layout of an API2520 to DIP8 single-opamp adapter . . .

Regarding your capacitor C2, is there a consensus about passing a trace between the legs of a part in 0805-package? As I recall, for quite a while the manufacturers said "Don't do that!" and the manufacturing engineers were solidly behind them - but circuit designers and PWB layout folks said, "Hey, it makes the routing easier and if production has their feces amalgamated the failure rate isn't enough to worry about.".

Are there any recent thoughts or statements about this?

Dale

[VivaVee]

Originally Posted by dchisholm

Regarding your capacitor C2, is there a consensus about passing a trace between the legs of a part in 0805-package? As I recall, for quite a while the manufacturers said "Don't do that!" and the manufacturing engineers were solidly behind them - but circuit designers and PWB layout folks said, "Hey, it makes the routing easier and if production has their feces amalgamated the failure rate isn't enough to worry about.".

Are there any recent thoughts or statements about this?

Dale

It is still bad practice. The PCB trace going between the pads is covered by solder mask but the pads are not. Therefore, the 0805 component fails the coplanarity test i.e. it does not sit flat and rocks, end to end. What might seem acceptable to a diyer or a design engineer that only ever builds one unit, is not acceptable to production/manufacturing engineers who take responsibility for building hundreds and thousands of units.

[bcarso]

Originally Posted by VivaVee

It is still bad practice. The PCB trace going between the pads is covered by solder mask but the pads are not. Therefore, the 0805 component fails the coplanarity test i.e. it does not sit flat and rocks, end to end. What might seem acceptable to a diyer or a design engineer that only ever builds one unit, is not acceptable to production/manufacturing engineers who take responsibility for building hundreds and thousands of units.

What if one left the solder mask off the trace under the 0805 part?

[rsavas]

What if one left the solder mask off the trace under the 0805 part?

Introduces other problems, if component is off center could cause a short, under the component (hard/impossible to see) or a possible place for a solder whisker to form.
Few options, first is to use a via, of course. Use a 1206 part. Also if/when they wave solder backside, they usually use glue to hold the component in place. Wave soldering will do a better job in this case rather than a paste & re-flow process.

[bcarso]

Originally Posted by rsavas

Introduces other problems, if component is off center could cause a short, under the component (hard/impossible to see) or a possible place for a solder whisker to form.
Few options, first is to use a via, of course. Use a 1206 part. Also if/when they wave solder backside, they usually use glue to hold the component in place. Wave soldering will do a better job in this case rather than a paste & re-flow process.

Well I can see that --- but based on the sort of contemporary geometries being used routinely, and the "huge" dimensions of an 0805, and the fact the the trace could be skinny in that vicinity... isn't that concern fairly minor?

Now if it is diy and I'm the one placing and soldering, I'd heartily agree But as a general consideration for automated assembly I'm surprised it would still be a prohibition.

Of course things have gone to H E double hockey sticks since ROHS...

[VivaVee]

Originally Posted by bcarso

What if one left the solder mask off the trace under the 0805 part?

The solder mask is typically 15-20um thick so if you specified 1oz copper weight (35um thickness) and pulled back the solder mask clear of that nasty traces you would be ok. But the copper will corrode. You will have a short shelf life for the PCB unless you store it carefully - a humidity cabinet is then required. So you then decide to plate the copper with a finish that does not itself corrode. So you go for ENIG (Immersion Gold over Electrolysis Nickel) - it looks pretty too although your PCB is now more expensive. And you need a silver bearing solder paste... which is more expensive

Or, you just NOT route that pesky trace under the 0805!

[VivaVee]

Originally Posted by rsavas

Also if/when they wave solder backside, they usually use glue to hold the component in place. Wave soldering will do a better job in this case rather than a paste & re-flow process.

It really depends on the assembly house. The 'backside' could be done using reflow with a higher temperature solder paste and then the 'frontside' is done with a lower temperature solder paste. Through-hole components (e.g.connectors) could use selective wave solder where a mask is used.

For something like this issue, the assembly house often just looks at the design, curses the bleeping designer and adds a surcharge to cover the rework and failure rate for this and other 'nice' layout features. It is only when the design is properly reviewed/redesigned that manufacturing and field failures 'magically' reduce.

[S.A.G.]

Alan,

What about using a 1206 part instead of a 0805? Would it be acceptable to route a trace between the pads of a 1206?

Thank you

Giorgio