39
\$\begingroup\$

I found a PCB that has had some rework done. When I saw it, I thought that someone had actually repaired it after purchase:-

bad circuit

It's got additional wiring (white & brown) and under the hot glue are two ceramic capacitors. The capacitors have been sleeved with insulation (difficult to see). They smell like decoupling capacitors.

I can understand altering my own boards, but this is a commercial speaker from a successful company (Labtec):-

speakers

The work seems to be based around the main amplifier chip (TDA2005). I've seen other post design changes to boards, but that's only been a wire here or there snaking across a PCB. Why does a commercial board get sold with this state of rework?

PS. Also note the hand thickened tracks in the top left.

PPS. The two circles also show hand soldered surface mount capacitors added to this side of the board which is clearly a through hole design.

\$\endgroup\$
12
  • 6
    \$\begingroup\$ I can't think of anything but simply the fact that its cheaper to have some chinese guys do it than re-design, re-run everything. Especially if the error was discovered after n units were made. \$\endgroup\$
    – Wesley Lee
    Commented May 5, 2017 at 12:17
  • 6
    \$\begingroup\$ Note that if you watch some teardowns of good quality lab equipment (low qty runs), you may also find some bodges (although way better executed). Sometimes it's enough and more cost-effective. \$\endgroup\$
    – Wesley Lee
    Commented May 5, 2017 at 12:19
  • 3
    \$\begingroup\$ Labtec may well be "a successful company", but those look like LCS-1040 speakers which are basically cheap junk - and they were designed 15 or 20 years ago. Usually, you get the quality you pay for! \$\endgroup\$
    – alephzero
    Commented May 5, 2017 at 13:56
  • 5
    \$\begingroup\$ Is it a single-sided board? I wonder if these were the only connections that couldn't be made on a 1-side board, so instead of shifting the whole things to 2-sided they just made this kludge? \$\endgroup\$
    – JeffUK
    Commented May 5, 2017 at 15:48
  • 5
    \$\begingroup\$ It's not a question of whether the tracks would fit, but whether they could physically get there in 2 dimensions, or if they needed a third dimension to hop over another track. \$\endgroup\$
    – JeffUK
    Commented May 5, 2017 at 16:08

5 Answers 5

93
\$\begingroup\$

This looks like it's due to a combination of bad engineering, bad management, and cheap labor.

Circuits don't always perform as expected when first designed. Even experienced designers make mistakes occasionally. Usually these are caught in the first prototypes, then the board respun.

Not all engineers have the experience and skills to get a circuit mostly right the first time, and the discipline to test it properly before committing to production.

Couple that with management that doesn't understand the engineering process, and a junior engineer hired into a position over his head who won't or can't stand up to management. It's exactly those types of managers that hire a junior engineer for such a role in the first place. "After all, it's just engineering, and all engineers are plug-replaceable, so I might as well hire the cheap one right out of school that won't give me all this crap about can't this, and test that."

You've got 10,000 populated boards and someone finally discovers that this thing oscillates when the volume knob is turned to 60% and you use one of the wall warts from the last shipment of 5,000 you just received. Your junior engineer doesn't understand what's happening, but determines that the wall warts are within spec. Now you've got a big problem, so you hire a consultant to look over the design and fix it.

The consultant shakes his head, tells you the whole design is a mess. You don't want a new design. After all you already paid for one. You tell the consultant you absolutely need a fix for this design. He comes up with the kludge you see above. You have the same factory in the far east that made the boards rework them. The labor cost is cheap, so it's better than scrapping 10,000 finished boards.

Good engineering is expensive. Bad engineering is even more expensive.

\$\endgroup\$
9
  • 14
    \$\begingroup\$ +1 great answer. I might also add, sometimes the original circuit works great all the way through design and qualification and the product manufacturing begins. Then it is discovered that some key feature, or state, that nobody had thought about or tested for is missing. e.g. What happens when the users presses these two buttons at the same time.... You can't always predict what an idiot will do... a better idiot always appears. When you have already invested 500,000 on your initial production run, fixing them can be cheaper then scrapping them. \$\endgroup\$
    – Trevor_G
    Commented May 5, 2017 at 13:38
  • 33
    \$\begingroup\$ @Trevor: Yes, idiots can be very clever. \$\endgroup\$ Commented May 5, 2017 at 13:40
  • 7
    \$\begingroup\$ Related: "There's never time to do it right, but always time to do it over" \$\endgroup\$
    – Shamtam
    Commented May 5, 2017 at 14:09
  • 9
    \$\begingroup\$ +1 Also if you take more time the shipment of speakers is going to miss the September 4 sailing for the Port of Long Beach and the customer will cancel if they can't get them in time for Xmas and you'll be stuck with 3 40-foot container loads of garbage. Get the folks living in your dormitories to work double shifts and the shipment goes out in time. \$\endgroup\$ Commented May 5, 2017 at 14:56
  • 6
    \$\begingroup\$ I hate to admit. I was both the young engineer who did the mess and now I'm the old engineer who fixes the mess. Times never change. \$\endgroup\$
    – Janka
    Commented May 5, 2017 at 21:50
11
\$\begingroup\$

You also have to consider things are normally manufactured in batches.

Sometimes, especially for things like this, VERY LARGE batches.

That is a huge investment cost.

When an issue is discovered during production the design is usually fixed in the next run, but that can leave you with half a million dollars in inventory that is defective.

At that point the bean counters calculate whether the loss associated with throwing them away is worse than the loss incurred in reworking them. With cheap labor, the latter can often be the better alternative.

As for how a rework like that can end up being needed, See Olin's answer and my comment therein.

\$\endgroup\$
4
\$\begingroup\$

Most commonly this is a result of a new revision of the design and they simply didn't want to spend the extra time or money to completely re-do the board layout and have it respun. It's much cheaper to bodge extra components or wires onto existing boards.

Generally new revisions may come up after the previous revision has been out in the field, and reports of failures, sub-par performance, or undesired operation are given by the customers.

\$\endgroup\$
4
\$\begingroup\$

Sometimes product is returned because its defective. Sometimes those defects are repaired or reworked. Sometimes those repaired boards pass test and end up in shipped product. BestBuy offers lots of notebook PC's that are refurbs.

\$\endgroup\$
1
\$\begingroup\$

Wire lengths for decoupling capacitors can be very critical, so soldering them straight to IC pins like that might be the easiest solution to place them very neat the active parts without having to use a double sided PCB.

Alternatively, various brands of the same IC can have differences in how they deal with decoupling capacitors missing or unsuitably mounted - the original design might have been tested with a more tolerant type and without the capacitors mounted that way, then a change to a different IC supplier was done that needed the kludge to work reliably. Another possibility is that a nearly but not completely pin compatible IC type was fit to an old board design due to the originally specced type becoming unavailable or too expensive.

\$\endgroup\$
4
  • 1
    \$\begingroup\$ Is (was) there a predecessor to the TDA2005 which is now itself obsolete? \$\endgroup\$
    – Paul Uszak
    Commented May 6, 2017 at 15:44
  • \$\begingroup\$ The TDA2004 for example? \$\endgroup\$ Commented May 7, 2017 at 3:05
  • \$\begingroup\$ So there was a TDA1 then? \$\endgroup\$
    – Paul Uszak
    Commented May 8, 2017 at 11:37
  • \$\begingroup\$ There seems to have been a TDA100 in the 1960s :) The oldest commonly found one seems to be the TDA440. These are all kinds of ICs, many of them not audio amplifiers - but the TDA2004 is :) \$\endgroup\$ Commented May 8, 2017 at 11:52

Not the answer you're looking for? Browse other questions tagged or ask your own question.