Test Equipment Anonymous (TEA) group therapy thread

[Robert763]

Wow didn't think it was that bad... so we laugh at the audiophools that want to use NOS tubes of a particular brand and manufacturing plant, but now we are doing pretty much the same with transistors ! 

I just checked my Matra missile 3 phase 400Hz inverter. Made in 1982 and all of its x6  2N3055 are made by RCA.

So if I can't sell the thing for a decent price, I will sell the transistors at 15 Euros each instead ! 

In a similar vein, I remember a story about some company that designed ages ago a board with a 68HC11 or something MCU. running at 4MHz or something, slow clock of back in the day. Their board was working just fine. Then years later suddenly their board was losing its mind, didn't work anymore.  Turns out the manufacturer for their MCU moved on to a newer fabrication process, as you do. The chip was internally then much faster, faster rise times. Of course the manufacturer didn't deem it necessary to inform their customers of this change !
So the board would act up, and had to be redesigned with "high-speed" techniques, then it would work again.

I've seen exactly that with a Allegro serial in parallel out driver chip. We had been using them in production for many years without issue but then started getting random failures. The wrong outputs would turn on. Unfortunatly on one machine a failure of this type could damage thousands of dollars worth of the customers material so they were not happy. The field service people had been struggling with these apparently random failures for a while before it came to me. On investigation I found that the timing of the clock and data input signals did not meet the timing specification of the chip. Or anything actually becuase the clock changed just before the data    The signals came for a I/O card in a PC. The PC software subroutine had never been changed in the several years it had been in use without issue.  I also noticed that the same data was sent about 30 times. Talking to the now "head of Sofware" who wrote the subrotine he admitted that it didn't work at first but as repating the dats many times seemd to work he left it like that.
So the timing ment it should never have worked, but it had for years  Now the data rate was close to the maximum specification for the Allegro chip becuse the time to send data had an significant effect on prcess times.
Getting the subroutine re-written so the timing met specification and only sent the data once fixed the issue but why did it only show up after years and numerous machines including different types? Talking to Alegro revealed that they had changed this chip to a different process with smaller features. Testing old nad new chips showed the the new chips were an order of magnitude faster than the old ones and thus much more sensitive to poor timing of the input signals. Faster PCs over the years didn't help.

[Robert763]

Packages from the Skegness auction arrived yesterday. Poor packing and contents damaged 

[bd139]

Oh great. Fingers crossed you can extract a suitable refund from them.

[McBryce]

Are you a Tek lover ? Are you German ? Do you have germoney ? Then the answer is this:

https://www.ebay-kleinanzeigen.de/s-anzeige/kennlinienschreiber-tektronix/2111929863-168-13379

You forgot a few questions: Do you have too much space and a concrete floor? Do you ignore the fact that modern devices that cost a fraction of this can give you much better measurements and the results can be stored?

McBryce.

Is there any one in particular you can recommend to buy or build?

I don't think any of the modern ones will offer you the voltages that the Tek can achieve, but for jellybean type devices, the Peak Atlas DCA 75 Pro produces good results when connected to a PC.

McBryce.

[TERRA Operative]

Today I pressure washed the balcony, cleaned out the barbeque (and now have a list of spare parts to order from Amazon) and got the place ready for the mother-in-laws friends to come tomorrow and fill their faces with a good amount of burnt meat.

Also, car comes back from Toyota tomorrow too, went in to be repaired after some stupid old woman tried to pit-manoeuvre herself against my car a few months back (She was deemed 100% at fault, rare here in Japan, thank myself for installing a dashcam ), so I bought some Gazoo Racing pedals to celebrate.

Now, I think I'll peruse Yahoo Auctions, there's a TDS794D 2GHz scope I have my eye on......

[DC1MC]

Are you a Tek lover ? Are you German ? Do you have germoney ? Then the answer is this:

https://www.ebay-kleinanzeigen.de/s-anzeige/kennlinienschreiber-tektronix/2111929863-168-13379

You forgot a few questions: Do you have too much space and a concrete floor? Do you ignore the fact that modern devices that cost a fraction of this can give you much better measurements and the results can be stored?

McBryce.

Is there any one in particular you can recommend to buy or build?

I don't think any of the modern ones will offer you the voltages that the Tek can achieve, but for jellybean type devices, the Peak Atlas DCA 75 Pro produces good results when connected to a PC.

McBryce.

I think the correct answer is "NONE of the modern ones that cost less than a Porsche will offer voltages/currents that came even close to the old Tek beast".

DC1MC

[Vince]

Are you a Tek lover ? Are you German ? Do you have germoney ? Then the answer is this:

https://www.ebay-kleinanzeigen.de/s-anzeige/kennlinienschreiber-tektronix/2111929863-168-13379

You forgot a few questions: Do you have too much space and a concrete floor? Do you ignore the fact that modern devices that cost a fraction of this can give you much better measurements and the results can be stored?

McBryce.

Is there any one in particular you can recommend to buy or build?

I don't think any of the modern ones will offer you the voltages that the Tek can achieve, but for jellybean type devices, the Peak Atlas DCA 75 Pro produces good results when connected to a PC.

McBryce.

Even for jellybean stuff it's no good. It does only very low currents and very low voltages. So it's limited to small signals jellybean... and even then, a humble 2N2222 can do 800mA, can this tester even achieve that ? I haven't looked but it feels unlikely... would drain the batteries at the speed of light. Maybe you can hook an external power supply to it ? But then that's more cost, and more wiring and mess on the bench...and what about all the many beefier packages, TO220 and TO3 that can handle hundreds of volts and 10 or 20 amps ? How do you test these ?
Also the Tek is a stand alone device, less messy and more practical. It's to the point.

I would love to have a modern Tek curve tracer equivalent for cheap, but I am not sure it can even exist. A decent quality cabinet and all the controls required, that costs money. A beefy linear power supply that can supply "clean" high voltages and high currents costs money. Decent quality materials and build quality costs money too. A good comprehensive service manual to "protect your investment" in the long term by helping you keeping it running, is valuable to.

I guess the only part that could be cheap in a modern version of a Tek, would be of course a large colour LCD panel to display the traces, and some CPU to process the data and show you useful stuff on the screen. Like measure hFE, slopes, turn-on voltages, XYZ, printed on screen, store curves, print, whatever a modern instrument could do easily/cheaply in S/W. That would be nice. But 90% of the stuff, mentioned in previous paragraph, I just don't see how it can be made dirt cheap.

I think there is a reason why these old Tek curve tracers are so sought after. Of course you pay for the nostalgia and speculation sadly, but the instrument also does have inherent merits / qualities, which are still valuable and relevant today IMO.



Yes, I should be outside gardening, but I gave up, too hot / sunny. I am staying inside, sorting my transistors instead.. will go gardening later in the evening when the sun will be going down.

[dl6lr]

Time for the Tek 1240 again.

After replacing the Rifa stinkers, I measured all the voltages according to the service manual.

Rail	Tolerance	Max. Ripple	Voltage	Ripple
+13V	12.1 ... 12.9	0.5Vpp	12.26	200mVpp
+12V	11.4 ... 12.6V	1Vpp	12.00	20mVpp
+5V	4.85 ... 5.15V	100mVpp	4.94	10mVpp
+3V	-1.9 ... -2.1V (Ref to 5V)	100mVpp	-2.00V	10mVpp
-5V	-4.75 ... -5.25V	200mVpp	-4.92V	<10mVpp
-12V	-11.4 ... -12.6V	1Vpp	-12.16V	<10mVpp

External monitor that uses a composite video signal showed a jitterless picture, so the display board is fine. Problem must be on CRT driver board A06. Started to measure the jitter in the horizontal deflection system. See schematics.

HDRIVE from display board is fine, the same on Pin 3 of the MC1391P. Oscillator of MC1391P runs fine, so the combination of R513, R532 and C522 is fine. Voltage at Pin 6 (shunt regulator) is fine at slightly below 9V and without ripple. Sawtooth input at Pin 4 is fine and within spec, it jitters around (as the picture is) and is unsync with the HDRIVE signal. +55V from the deflection transformer is within spec and it does not leak through VR435 (which would cause the frequency to be way off to reduce the output voltage). Phase detector output on Pin 5 is stuck and I can not see any changes while the frequency changes. So my conclusion is the MC1391P is toast. Have to get a replacement.

[Vince]

Today I pressure washed the balcony, cleaned out the barbeque (and now have a list of spare parts to order from Amazon) and got the place ready for the mother-in-laws friends to come tomorrow and fill their faces with a good amount of burnt meat.

Also, car comes back from Toyota tomorrow too, went in to be repaired after some stupid old woman tried to pit-manoeuvre herself against my car a few months back (She was deemed 100% at fault, rare here in Japan, thank myself for installing a dashcam ), so I bought some Gazoo Racing pedals to celebrate.

Now, I think I'll peruse Yahoo Auctions, there's a TDS794D 2GHz scope I have my eye on......

Wow a 794D my dream !!! 

Rare as henth teeth, so I am not even dreaming of it actually, I would only be disappointed never to be able to find one.
Hence why I am rather set on the 784D as it is, at least in comparison with the 794D, much, much more widely available.

Keep us informed and post pictures of the thing once you get it, because of course you will get it !

[TERRA Operative]

Keeps us informed and post pictures of the thing once you get it, because of course you will get it !

Hahahaha, shutup you! 

I think once I get this one, fingers crossed, I'll stop buying them (at least so much)... Next I'll have to find an AWG710B arbitrary func gen and then sell off my AWG510.

Too bad postage is such a killer these days, I have a choice of a TDS620B, 3x TDS754D's (one is in for repair though), a TDS714L, a TDS784C and a TDS784D so I could sell you one for a good price... 
I also have a spare HD option (Upgraded to CF card) too with some Tektronix apps preloaded on it to sell one day too.

[mnementh]

@mnem be careful what you knock yerself out with. Quadral Montan Mallet 46 kg, Vulkan 70kg, Titan 140 kg, all for Mark 3, single unit.

Yeah, I knew that would give you a chuckle.  Sounds about like what I needed to knock my arse out.

Did some more stripping of the nasty paint on the Dynaco bookshelf speakers I mentioned in the 'what did you buy today' thread... 

      -Pat

Dammit, Pat! Between you and the Valkyrie up there, you actually making me think about going back for these I saw at the Thrift for $20 the other day...

   

You have any idea how long it's been since I actually replaced the surround in a driver...?   

mnem
Obviously, the Q-mallet wasn't enough to knock any sense into me... 

[Vince]

Keeps us informed and post pictures of the thing once you get it, because of course you will get it !

Hahahaha, shutup you! 

I think once I get this one, fingers crossed, I'll stop buying them (at least so much)... Next I'll have to find an AWG710B arbitrary func gen and then sell off my AWG510.

Too bad postage is such a killer these days, I have a choice of a TDS620B, 3x TDS754D's (one is in for repair though), a TDS714L, a TDS784C and a TDS784D so I could sell you one for a good price... 
I also have a spare HD option (Upgraded to CF card) too with some Tektronix apps preloaded on it to sell one day too.

That's very nice of you ! 

Do keep at least one 784D all optioned up in stock (hell a 794D of course, if you find another one !)... after the war and covid, shipping might become more reasonable... gonna take at least a couple years though I guess, so in the meantime  do think of giving it 5 minutes use every other week to keep it healthy ! It's like a car, it degrades if you let it sit for too long ! 

[Cerebus]

It's worse than that. A lot of "legacy" transistor types being listed by manufacturers don't even try to match the original silicon. They just put a die from a "better" spec current process. They just list the headline specifications and don't worry about the detail stuff. One thing that stands out is the physical size of the die. They shrink over the years.
... snip ...

As with the transistor you mention the modern process 2N3055s very often have a significantly smaller SOA than the originals. So if one encounters lots of original 2N3055s ganged up for current sharing in something old it's a big hint that someone needed to do it to meet the SOA requirements and a modern replacement may well just go 'fhut' on first power up. Also the original 2N3055 is much more secondary breakdown resistant than modern versions, another pit to fall into.

[Cerebus]

I'm curious as to what the glass capsule parts are in the protection circuit; they are not GDTs, you can see a solid slug inside them, and they are marked "VRxx", and have a (in circuit) resistance of ~190MΩ.

My guess would be "varistor". Does the slug inside look like it was sintered out of powder? That would be a clincher.

[Cerebus]

...
On investigation I found that the timing of the clock and data input signals did not meet the timing specification of the chip. Or anything actually becuase the clock changed just before the data    The signals came for a I/O card in a PC. The PC software subroutine had never been changed in the several years it had been in use without issue.  I also noticed that the same data was sent about 30 times. Talking to the now "head of Sofware" who wrote the subrotine he admitted that it didn't work at first but as repating the dats many times seemd to work he left it like that.
...

Ah! I think those software guys need educating with something heftier than a cluestick. Time to get out the metastability mallet. 

Actually I'm quite surprised that something that randomly hit a stable timing margin only one time in thirty managed to work at all.

That's actually quite a good example of the chewing gum and string quality of a lot of software that you'll always hear BD and me moaning about. Programming is hard, proper hard, and there's a tendency for programmers to develop hubris as well as code. Programmers have to learn a lot about each application area that they have to develop for as well as the actual computer technology that is their basic stock in trade, and I've noticed a tendency for a lot of then to acquire a delusional self belief that "If I can learn to do all this, I can do anything.". This has the obvious result of lots of Dunning-Kruger moments.

[bd139]

Oh fuck don’t get me started on that 

[mnementh]

NEW TOY For my OLD TOY:



While I was shopping for a hedge trimmer... No, seriously; they're getting wild & woolly and I also have some trees need pruning right up against the house... I saw multiple reviews raving about this pruning blade. People saying they just use this instead of a chainsaw most of the time, because it's less assache. On multiple sites. 

After having it in my hand, I can see why; 3TPI and aggressively offset razor sharp carbide tips just like a proper crosscut sawblade. 

This one blade cost $12; that's as much as the whole kit of metal and demolition blades in the pack above, which have given yeoman service for over a year.

While I don't expect it to trim the shrubbery worth a damn, I do expect it will make very short work of the brush I need to cut to 3 foot lengths for pickup... 

mnem
*runs thumb across cutting edge*

"Holy fuck that's sharp! Well, I guess if I'm gonna gut my leg off by accident, I wanna get it over quick..."

*laugh from store stocker two aisles over*   ~fer reallz

[Kosmic]

...
On investigation I found that the timing of the clock and data input signals did not meet the timing specification of the chip. Or anything actually becuase the clock changed just before the data    The signals came for a I/O card in a PC. The PC software subroutine had never been changed in the several years it had been in use without issue.  I also noticed that the same data was sent about 30 times. Talking to the now "head of Sofware" who wrote the subrotine he admitted that it didn't work at first but as repating the dats many times seemd to work he left it like that.
...

Ah! I think those software guys need educating with something heftier than a cluestick. Time to get out the metastability mallet. 

Actually I'm quite surprised that something that randomly hit a stable timing margin only one time in thirty managed to work at all.

That's actually quite a good example of the chewing gum and string quality of a lot of software that you'll always hear BD and me moaning about. Programming is hard, proper hard, and there's a tendency for programmers to develop hubris as well as code. Programmers have to learn a lot about each application area that they have to develop for as well as the actual computer technology that is their basic stock in trade, and I've noticed a tendency for a lot of then to acquire a delusional self belief that "If I can learn to do all this, I can do anything.". This has the obvious result of lots of Dunning-Kruger moments.

I would say that software quality is directly related to development practice and not necessarily individuals. Poor or inexistent development practice lead to low quality even if your staff are really good. When I say "development practice" I mean coding guidelines, code review, automated tests and submit procedure. All that is expensive though (in time and infrastructure) and you need to decide what level of quality you are looking for. If the question was never raised the quality level will probably be really low.

[bd139]

...
On investigation I found that the timing of the clock and data input signals did not meet the timing specification of the chip. Or anything actually becuase the clock changed just before the data    The signals came for a I/O card in a PC. The PC software subroutine had never been changed in the several years it had been in use without issue.  I also noticed that the same data was sent about 30 times. Talking to the now "head of Sofware" who wrote the subrotine he admitted that it didn't work at first but as repating the dats many times seemd to work he left it like that.
...

Ah! I think those software guys need educating with something heftier than a cluestick. Time to get out the metastability mallet. 

Actually I'm quite surprised that something that randomly hit a stable timing margin only one time in thirty managed to work at all.

That's actually quite a good example of the chewing gum and string quality of a lot of software that you'll always hear BD and me moaning about. Programming is hard, proper hard, and there's a tendency for programmers to develop hubris as well as code. Programmers have to learn a lot about each application area that they have to develop for as well as the actual computer technology that is their basic stock in trade, and I've noticed a tendency for a lot of then to acquire a delusional self belief that "If I can learn to do all this, I can do anything.". This has the obvious result of lots of Dunning-Kruger moments.

I would say that software quality is directly related to development practice and not necessarily individuals. Poor or inexistent development practice lead to low quality even if your staff are really good. When I say "development practice" I mean coding guidelines, code review, automated tests and submit procedure. All that is expensive though (in time and infrastructure) and you need to decide what level of quality you are looking for. If the question was never raised the quality level will probably be really low.

Good staff evolve good processes. Bad staff evolve no processes or evolve them and circumvent them.

[Kosmic]

...
On investigation I found that the timing of the clock and data input signals did not meet the timing specification of the chip. Or anything actually becuase the clock changed just before the data    The signals came for a I/O card in a PC. The PC software subroutine had never been changed in the several years it had been in use without issue.  I also noticed that the same data was sent about 30 times. Talking to the now "head of Sofware" who wrote the subrotine he admitted that it didn't work at first but as repating the dats many times seemd to work he left it like that.
...

Ah! I think those software guys need educating with something heftier than a cluestick. Time to get out the metastability mallet. 

Actually I'm quite surprised that something that randomly hit a stable timing margin only one time in thirty managed to work at all.

That's actually quite a good example of the chewing gum and string quality of a lot of software that you'll always hear BD and me moaning about. Programming is hard, proper hard, and there's a tendency for programmers to develop hubris as well as code. Programmers have to learn a lot about each application area that they have to develop for as well as the actual computer technology that is their basic stock in trade, and I've noticed a tendency for a lot of then to acquire a delusional self belief that "If I can learn to do all this, I can do anything.". This has the obvious result of lots of Dunning-Kruger moments.

I would say that software quality is directly related to development practice and not necessarily individuals. Poor or inexistent development practice lead to low quality even if your staff are really good. When I say "development practice" I mean coding guidelines, code review, automated tests and submit procedure. All that is expensive though (in time and infrastructure) and you need to decide what level of quality you are looking for. If the question was never raised the quality level will probably be really low.

Good staff evolve good processes. Bad staff evolve no processes or evolve them and circumvent them.

Haaa that would be great but it's not always like that. We got some good programmers at work and every time we add a new procedure, they can't stop complaining. Some people just don't want to follow the rules.

All that is going to be solved when AIs will be able to code 

[Vince]

OH NO!!!! 

Sorting/organizing my transistors....so I just dug out my cool Brother label printer that I just bought earlier this year... and the freaking thing is already broken ?!   

Well I think it might the cassette tape on its own, rather. Well I hope anyway...
Problem seems to be that it won't feed properly any more, at all really. When I ask it to feed the tape so I can get my label out... I can hear the printer activate the feed motor just fine, however the tape just won't come out ! It just gets mangled inside the cassette instead !!

I tried pulling the mangled tape out, then try again.. no luck.

Paid freaking 12 or 15 Euros IIRC, for that one single 9mm cassette, because genuine ! That will teach me ! Now I will only buy cheap aftermarket cassettes, can't be worse than this crappy genuine one !

Zero experience with these things... if anyone can look at that cassette and tell me what's wrong and if it can be fixed... please speak up.
I printed only 2.5meters worth of labels and a cassette IIRC contains much more than that, 10meters or something... so I have wasted lots of expensive tape here !

[xrunner]

Paid freaking 12 or 15 Euros IIRC, for that one single 9mm cassette, because genuine ! That will teach me ! Now I will only buy cheap aftermarket cassettes, can't be worse than this crappy genuine one !

I have used these for years and never bought any OEM cassettes - no problems at all.

[Vince]

I would strangle you but you live too far away, the pond is protecting you...

[Kosmic]

I had 1 OEM cassette doing something like that. I tried fixing the problem but eventually went in the garbage.

Something I didn't know about those cassettes though is that they keep a negative of everything you print inside. So if you print something sensitive make sure you burn the used cassettes 

[bd139]

I learned that sticky labels and sharpies are a better investment