Surface Mount Technology - Is it better?

Posted by: John Bailey on 16 December 2001

I was just admiring the internals of the NAP6-50 in the Autumn News brochure but it got me wondering why it is that Naim haven't gone down the surface mount route when so many others have. Linn especially made a big play of it.

Any ideas?

Posted on: 16 December 2001 by JohnS
I remember when surface mount technology came in, in the late 80s, it was cheaper for larger production runs, more reliable and better lent itself to automated assembly. Also the PCBs could be made smaller, obviously. Here is an excerpt from an article on Designing Better Surface Mount Boards.

quote:
In the period before Surface Mount Technology became common, most PCB designs were created by carefully trained designers, using complex, expensive software that often required specialized computer workstations. Electrical Engineers designed the circuits and PCB Designers designed the boards. Most companies manufactured their own products in-house and developed their own unique manufacturing guidelines and rules. Many of these companies even manufactured their own raw printed circuit boards. Although this method of product design and manufacturing may be viewed as expensive and slow by today's standards, at least everyone involved in the product design cycle understood their own part of the process. Each person was an expert in his or her particular specialty.

Today however, the corporate structure of an electronic product development company is changing. The in-house PCB design departments are being phased out. The PCB fabrication department is gone and PCB assembly is out-sourced. Only the Electronic Design Engineer is left, but the job description has changed. Now the design engineer is not only expected to design the circuit, but frequently has to design the PCB as well. Alternatively, the PCB design work may be out-sourced to an independent PCB design company or a temporary contract employee. While each of these parties may know their own specialty well, what has been lost is the intimate understanding of the rules and manufacturing process requirements. The Institute for Interconnecting and Packaging Electronic Circuits (IPC) has struggled for years to develop and distribute guidelines for good PCB Design and Design for Manufacturability. They were very successful in the era of pin-in-hole or thru-hole PCB technology. Unfortunately, as Surface Mount Technology started to become more popular, their job became more difficult. This is because thru-hole PCB technology was relatively stable, mature and well understood. Surface Mount Technology, on the other hand, is relatively new, very dynamic and continually evolving.


So it could be argued that Naim PCBs are the equivalent of Bristol, Morgan or Bentley motorcars. Expensive, mainly hand-made, but not necessarily inferior to production line cars.

Just my $0.02 worth smile

-John

Posted on: 16 December 2001 by bam
SMT makes unit cost lower in high volume production both through component cost and benefits of reduced PCB area. A PCB can be populated on both sides too, making track routing more dense. Some reliability improvements result from smaller mass parts and a combination of solder and glue fixing. The whole thing is designed for automation so any manual intervention such as repair is really tricky: components are tiny and markings, if any, are hard to see.

Audiophile pros: smaller sizes and shorter tracks are good for HF performance and reduced RF emissions. In audio this really matters most with digital circuits.

cons: I'm not sure. I would not expect SMT to sound better as components have been designed purely with cost and size in mind.

Do Naim products not use SMT? I would be surprised not to find it in the CD players. Perhaps Naim do not manufacture SMT PCAs in house. A good reason for Naim not to have widely adopted SMT yet (if this is true) is production tooling upgrade costs and the need to redesign circuit boards.

Posted on: 16 December 2001 by Martin Payne
At the barbeque we were told that the major problem with using SMT was the difficulty of adding pair-matched components to boards when the machines are fed from ribbons of components containing thousands of identical items.

cheers, Martin

Posted on: 16 December 2001 by Andrew L. Weekes
Bam's comments hit most of the nails on the mark.

The cost of upgrading production facilities is huge, and would need a lot more space than the current conventional robotic assembly machines.

The big benefits are speed, the speed of Naim's and any other conventional robot assembly is pedestrian by modern standards, a modern machine can assemble complex PCB's in the blink of an eye, it's a very impressive spectacle to behold.

It's often not economical to invest oneself, but find a good company to outsource. I have call upon a 3/4 million ukp SM production facility that is modest by modern standards, it's owned by a local company we have a good working relationship with.

Linn also have some superb SM production facilities eek

SM brings huge benefits at high frequencies, indeed it almost becomes mandatory for some work, reduced lead lengths bring reduced stray inductance, stray capacitance and give more idealised component performance.

Size of the finished item can reduce too, although in Naim's case technologies used elsewhere wouldn't allow this benefit to be realised much.

Power dissipation is a problem with such tiny components, whole working environments have to change to accomodate new procedures, and everyone needs good eyesight wink

Naim use SM technology where they have to, CD player chipsets are SM for expample, as is much of the microcontroller based output.

The AV2 is almost exclusively SM.

I wonder if they've ever tried SM technology to replace the existing leaded stuff - I'd doubt, with careful application / component choice that performance would suffer.

They will eventually be forced down this route, as leaded component availability in newer devices becomes scarce or non-existent. They will have to find a way to make it work, or they will find themselves unable to use newer, higher specification components - many are only available in surface mount form.

It brings other nightmares though, in low volume environments, and requires, for guaranteed supply, investment in stock. Manufacturers of passive components, for example, have a habit of repackaging devices, with the same spec, into smaller packages, obsoleting old parts. With leaded components a smaller component can almost always be fitted with longer leads - with SM the component may not bridge the pads it is intended to be soldered to, meaning a PCB redesign.

Martin's comments about pair-matched components is valid too - jigs for testing them are difficult, as is handling, and assembly would have to be done by hand. To be honest though, unless every semi is pair tested, the speed of assembly improvments would offset hand assembly time, which can be done easily.

Rework / repair techniques need to change - IMO SM rework, with the right tools, is quicker and easier than leaded, but training and different skills are required.

I love it, but the obsolescence issues cause me regular headaches in the small volume production facility I support.

Andy.

FWIW I feel Naim could benefit enormously if carefully applied, in key areas, to augment or support the existing tried and trusted technologies, sort of hybrid devices, similar to the op-amp designed for the CDSII, and reused in the NAP500 and NAP150. I have just such an idea in mind - I'll leave you guessing as to what it is for the present though wink

Posted on: 16 December 2001 by Martin Payne
quote:
Originally posted by Andrew L. Weekes:
I'll leave you guessing as to what it is for the present though wink


A power supply!! ;-)

cheers, Martin

Posted on: 16 December 2001 by David Dever
quote:
The AV2 is almost exclusively SM.

Only the decoder board and front panel (as on other Series 5 units) is exclusively SMT-based. The rest is double-sided through-hole construction.

Dave Dever, NANA

P.S. No bop-amps in the 150; input buffer only in the 500.

Posted on: 17 December 2001 by Andrew L. Weekes
quote:
P.S. No bop-amps in the 150; input buffer only in the 500.

My mistake - I'd assumed the black resin-potted 'things' were the amps.

What are they then, I've noticed them in NAP500's and 6-50's

Regards,

Andy.