Does socketing components affect quality?

[amb]

Quote:

Originally Posted by PinkFloyd So "Analogue Devices" are talking bollocks when they say: "IC sockets should be avoided, since their increased interlead capacitance can degrade the bandwidth of the device." ?

No, I didn't say that. Datasheets often make remarks of a general nature and not specific to audio circuits. Opamps are used in applications where bandwidth and speed requirements are far higher than in audio. Excessive bandwidth is actually a liability rather than a benefit in audio, and this is why many circuits employ bandwidth-limiting compensation capacitors. When operating in this fashion, using a socket should not be a problem, unless the pcb is poorly designed and things are already at the verge of instability.

 

[PinkFloyd]

Quote:

Originally Posted by amb Excessive bandwidth is actually a liability rather than a benefit in audio.

Could you expand on this if possible?

 

[amb]

Quote:

Originally Posted by PinkFloyd Could you expand on this if possible?

Bandwidth in excess of several hundred KHz to 1MHz range is not necessary in audio, because audio signals live in a range far below that and we can't hear any higher. A circuit with too much bandwidth is susceptible to stability concerns, RFI, and myriad other sundry issues, which would degrade performance within the audio band. A good amp design should have enough bandwidth so that there is no roll off in response or phase shift within the audio band, and that musical signals cannot be slew-limited in any way, but not more than that.

EDIT: The "several hundred KHz to 1MHz" bandwidth remark I made refers to the frequency response of the amp. This is not to be confused with the GBP (gain-bandwidth product) spec of opamps and transistors. The GBP spec is the frequency at which the device stops being an amplifier. Since real amplifiers have gain, we need a GBP higher than the intended frequency response by at least the amount of gain. Take, for example, a gain = 10x amp, and our target frequency response is up to 10x the top of the audio range (200KHz), then a device with GBP of at least 2MHz should be used. This is a bare minimum and I would be happier with a device that has GBP of 20MHz. Devices that have even higher GBP would be of academic interest and of dubious real-world benefit.

 

[robzy]

Okay - so i take it that socketing transistors etc does not affect sound quality, i just have to watch out for mechanical problems such as oxidisation which will affect sound quality.

Thanks for that guys

Rob.

 

[dgardner]

You can always solder the parts into the sockets when everything is done and all the tweaking and debug is over.

 

[chillysalsa]

One other consideration is that socketing will add some thermal drift which could result in DC offset changes.

I don't think that will be an issue for the Dynalo though, because there is a DC servo to correct for drift due to heating of the transistors (which has a bigger effect on DC offset).