Flaw direct from Tektronix factory

[Jwalling]

While working on a TDS7104A I noticed that the 1M and 50Ohm LED indicators for CH1 were not working.
Figured it was a couple of bad LED or the driver.

Nope. It shipped like this originally from the factory. I can't believe they didn't catch this! 

[T3sl4co1l]

Yeah but now the warranty is void. 

Odd that that missed inspection though.  Humans, whaddya gonna do?

Tim

[tautech]

xrunner found another here:



https://www.eevblog.com/forum/repair/repair-of-tektronix-pg501-pulse-generator/

[Jwalling]

Yeah but now the warranty is void. 

Yeah! I got a repair quote for $6950. 

[Jwalling]

xrunner found another here:

I could sort of see how that one could get by... For a while anyway. No vibration testing, and it probably was making contact OK during test.

[xrunner]

xrunner found another here:

I could sort of see how that one could get by... For a while anyway. No vibration testing, and it probably was making contact OK during test.

Yea that was on a PG501 plug-in. The output wasn't working when I got it, but it had a cal sticker on it. It may have been touching just enough a long time ago. But yea, somebody was anxious to get off work that Friday at the factory and get to happy hour! 

[hendorog]

xrunner found another here:

I could sort of see how that one could get by... For a while anyway. No vibration testing, and it probably was making contact OK during test.

A good example showing Testing and Quality are not the same thing. Quality happens when something is built.

I wonder what percentage of scopes are tested off the production line?

[Wuerstchenhund]

Quality happens when something is built.

Indeed.

I wonder what percentage of scopes are tested off the production line?

LeCroy runs every scope through an extensive test process, and as far as I know Keysight and R&S  do the same.

But don't forget that for many years Tek had to suffer under Danaher, who's main method of revenue generation was excessive cost cutting (DBS, Danaher Business System).

[Paul Moir]

Doesn't matter.  As T3sl4co1l alluded, if it's humans doing the testing they will fail sometimes.  We just aren't cut out for that kind of work.

[Wuerstchenhund]

As T3sl4co1l alluded, if it's humans doing the testing they will fail sometimes.  We just aren't cut out for that kind of work.

Which is why reputable manufacturers usually have processes and systems in place to mitigate for the 'human element'.

Such processes are also part of ISO9000 certification.

[coppice]

As T3sl4co1l alluded, if it's humans doing the testing they will fail sometimes.  We just aren't cut out for that kind of work.

Which is why reputable manufacturers usually have processes and systems in place to mitigate for the 'human element'.

Such processes are also part of ISO9000 certification.

As long as you put in writing "we do really shitty half baked things", and you actually do shitty half baked things, you get ISO9000 certification. I thought the myth that ISO9000 was about achieving high quality had faded away years ago. Its only about achieving consistent results, whether they be consistently good or consistently horrible.

[bd139]

True. My father's company is ISO 9001 cert and he's a total cowboy two man band IT outfit. It means nothing other than you prepared documentation to their standards and can fudge some lies on audit.

ISO 27001 is the same turd in the security space.

[Wuerstchenhund]

I thought the myth that ISO9000 was about achieving high quality had faded away years ago.

I never said ISO9000 was about highest quality, it's as you say a tool for making sure results are constant with little variations. But part of that is the understanding that avoiding errors in the first place will reduce subsequent costs which would be spent on rectifying stuff. Which is why most of the reputable manufacturer try to get end control right.

[T3sl4co1l]

Yeah but now the warranty is void. 

Yeah! I got a repair quote for $6950. 

No warranty for obvious factory mistakes?  Ouch!

A good example showing Testing and Quality are not the same thing. Quality happens when something is built.

Statistics, it's a wonderful thing, but must be applied properly to be beneficial.  (Or, inappropriately, as befits an agenda, as is so often the case...)

It's true that some things can't be tested into quality; it's interesting that some things can.  That was what they said about early semiconductors, but lo and behold, a production line making 2% yield still made good parts.  Just not many!  On the other hand, tests that aren't triggered by the underlying failure (or which are unreliable, like human inspection) don't improve quality very much.

A lot of production errors (like soldering defects) can similarly improve quality by inspecting for them; but the inspection needs to be thorough and reliable (e.g., machine AOI versus semi-automated human inspection).  On the other hand, an electrical test helps a lot, but is not complete (for one example, I've seen a board with D2PAK head-on-pillow solder defects -- enough to make contact, but no good thermally, so they cooked).

Tim

[T3sl4co1l]

As long as you put in writing "we do really shitty half baked things", and you actually do shitty half baked things, you get ISO9000 certification. I thought the myth that ISO9000 was about achieving high quality had faded away years ago. Its only about achieving consistent results, whether they be consistently good or consistently horrible.

The Dilbert effect.   (Well, one of them.)

Tim

[PartialDischarge]

Yeah but now the warranty is void. 

Yeah! I got a repair quote for $6950. 

Not bad, recently I asked for a quotation of all 15 knobs in that scope. 7 of one type + 7 smaller ones + the small trigger one. 25€ each + taxes. Nearly 500€ all of them 

[Keysight DanielBogdanoff]

I wonder what percentage of scopes are tested off the production line?

We definitely run 100% of our oscilloscopes through hardware tests and performance characterizations. I can't fathom not testing every unit, but I haven't seen their production lines so 

[alm]

Humans are fairly bad at staying alert at things that only go wrong only 0.1% of the time. Which is why they introduce fake weapons at security checkpoints to keep the X-ray operators alert. I imagine having the LEDs blocked by the case is quite rare, so it would not surprise me if it did go through QA inspection but the inspector failed to catch it because they were more focused on other parts that fail more often, like performance issues. Also the failure would not be obvious to an inspection for external defects while powered off.

[Paul Moir]

Which is why reputable manufacturers usually have processes and systems in place to mitigate for the 'human element'.

I had three levels of verification miss an obviously incorrect label in a very ridged quality system.  Now a computer scans a barcode.  You do anything enough and you'll find that humans are at best 99%, and are far worse than that if they don't expect to see the problem.  Their strengths are elsewhere.

[Jwalling]

Yeah but now the warranty is void. 

Yeah! I got a repair quote for $6950. 

Not bad, recently I asked for a quotation of all 15 knobs in that scope. 7 of one type + 7 smaller ones + the small trigger one. 25€ each + taxes. Nearly 500€ all of them 

I bought some in the last few months from Tek. $12USD each. Still pretty painful.

[Electro Detective]

It pays to 'Take It Apart' especially nowadays, on any gear you rely on,   

or fingers crossed and pray yours didn't come off the assembly line on Friday afternoon   

or Monday morning   


I've seen and sorted out some unbelievable whoppers   

and the times I contacted the companies direct about it to point out the issue/s, AND offer a simple fix suggestion at no cost to them or production delays..
well, talking to their rear brick wall would have gotten better results   

One large lot of recalled electrical fittings I suspect went into landfill rather than just simply undo 2 housing screws,
and reverse the neutral and earth/ground wires on the terminal block (2 screws) so it would comply with electrical codes it got approval with in the first place    and not trip RCD/GFCIs

i.e. with a small power screw driver at the correct torque/clutch setting, probably get it done in under a minute per unit

Instead it was easier for them to landfill the lot   


FWIW, I kept and fixed mine with a 'no tech' screwdriver in 2 minutes   




 

[coppice]

I wonder what percentage of scopes are tested off the production line?

We definitely run 100% of our oscilloscopes through hardware tests and performance characterizations. I can't fathom not testing every unit, but I haven't seen their production lines so 

Do you soak test 100% or your scopes? Even low cost consumer equipment is typically given a 48 hours soak test, with a recheck after the soak. I can't imagine even a low end test equipment maker doing less.