why are boss pedals so complicated?

[greaser_au]

I have a 'budget' plastic case phaser - the effect is entirely usable (as are the delay, & the tuner that I have from the same range), just a bit noisy. However it will remain anonymous for reasons that will become clear in a moment...  

It's pretty much all all surface mount parts on a multi-layer board, except for the pots, sockets, & transistors/diodes - has a row of TO-92 packaged J2XX (I cant remember) FETs. I'd be inclined to hotrod it with quieter opamps, but my last foray into repairing another pedal of this particular range was more than a bit of a disaster due to the (in my opinion) really poor board substrate quality - maybe if I still had access to the surface-mount gear I used to use when I used to do PCB repair...

[R.G.]

What about reels of surface mount parts?  Or reels of through-hole parts for that matter?

Suddenly your pick and place robot becomes horribly inefficient if you need to manually place parts on a PCB after hand-testing them...and of course your manufacturing cost jumps way up.

I have no doubt that there are situations where you can't easily use a minimum wage "robot". I was illustrating the opposite side of the discussion.

If you have a surface mount operation, you're already into heavy computer automation of parts placement, and there are other avenues open to you to automate part sorting in a way that is even faster and a lower per-part cost for sorting.

I figure the large scale manufacturers must be able to get JFETs pre-sorted within a certain tolerance.

I figure that too; however, for a part you sell at retail for $5.00 per hundred, you can't afford to spend much on presort. Many semiconductor makers have abandoned making "jellybean" parts like discretes entirely. The industry MBAs refer to this as "the dog food market". But probably the folks that still make discretes have a sorter somewhere that could run a batch of sorted parts, given sufficient financial inducement to set it up. The Japanese and to some extent their imitators have already factored this in by producing transistors which come as B, O, Y, G, etc suffixes, corresponding to black, orange, yellow, green, whatever color groups which imply a gain or Idss bucket. Not matched exactly, but sorted into a few buckets that aren't too dramatically different. This is a holdover from the days when the Japanese semiconductor industry was struggling to get good results from highly variable germanium devices and pre-sorting was a necessary way of life.

Manufacturers can test surface mount resistors that are .4mm across, so surely they can test other parts.

Yep. Actually, they'd probably sort transistors at the die/wafer test where they winnow out the non-functional ones before going to the expense of wasting bonding and encapsulation materials on defective dies. The cost of test and matching is almost entirely the cost of physically manipulating the device into/out of the tester and buckets, tester time, and the administrative cost of there being different buckets to choose from.

That being said, I haven't ever found pre-sorted JFETs from Mouser.

And you won't. It's too expensive for Mouser, too.

My point?  That companies that mass-produce phasers for $29.99 retail probably don't care if its "dialed in". 

I'd put it another way. The companies that make $29.99 retail phasers have to come in with parts and labor for under maybe $7.00 each for the finished goods, including parts, labor, testing, packaging, shipping, warranty, overhead, insurance, taxes, and profits, as well as amortizing their engineering/development costs and sending the bosses and salesdroids to Hawaii for sales conventions. They *can't* add in $0.25 for matching. That could be their entire profit margin. The rest of the 2x to 4x multipliers are to get distributors and retailers to sell the goods. Until they've had their noses rubbed in it hard, techies can't often appreciate how the distribution channel works. It's worse than watching sausage being made. 

[Ronsonic]

The Japanese and to some extent their imitators have already factored this in by producing transistors which come as B, O, Y, G, etc suffixes, corresponding to black, orange, yellow, green, whatever color groups which imply a gain or Idss bucket. Not matched exactly, but sorted into a few buckets that aren't too dramatically different. This is a holdover from the days when the Japanese semiconductor industry was struggling to get good results from highly variable germanium devices and pre-sorting was a necessary way of life.

The story goes that back in the early days, before Sony was permitted to spend the outrageous sum of $25,000 for the rights to produce transistors, Japanese pocket radio producers were buying barrels of-out-of spec transistors from the US and hand sorting them into useful sets.

I haven't done something needing matched JFETs, but have found that for boosters and gain stages almost everything can be sorted into three buckets near the fat part of the bell curve and bias up perfectly well. That bell curve has very steep sides these days and I almost never see parts out on either extreme.

[Mark Hammer]

...companies that mass-produce phasers for $29.99 retail probably don't care if its "dialed in".  I haven't used any of those pedals but I have a hard time imagining that they can even begin to compare with their more carefully manufactured cousins, except by accident.

Anyone ever bought one of the cheap ones?  If so, did it sound good?

I did buy one a month ago or so, and it actually doesn't sound too bad.  I think the one I made with "matched" JFETs of the same transistor number sounds a bit better, but then I'm the guy who put in all the work so I think I'll leave the blind A/B judgment to somebody else.