Test Equipment Anonymous (TEA) group therapy thread

[Specmaster]

Some of the neon decimal points also don't work, and there's a problem with digits 1 and 2 in that they don't give a reading, except sometimes a momentary one while switching ranges. It's giving incorrect readings in the period settings, though they are linear up and down the ranges, which might be a helpful clue. The general crustiness is indicative of several years in a damp shed or cellar, to my mind.

*sighs* another item for the repair queue...   

Looking at the level of rust on the transistors, many of whom are highly likely to be unobtainuim, I doubt that it would survive for long in that growing repair queue of yours, best leave it out on the bench for immediate surgery I think. 

[nfmax]

I would argue that this is best stop-motion done for many years.
https://youtu.be/41BX9zn_qZo

"Electronics for dogs" and "Spare Track" does it for me. W & G deserve a very special place in the Pantheon of moving pictures, indeed.

(One should perhaps think about the resources available to Rymdreglage and Aardman, respectively -- the former is a rural 2-man hobby operation, while someone paid Aardman. )

Re W&G, we have a dog who reminds me a lot of Grommit, so that makes me.........
😂

I'm over-TEA'd. When I saw W&G I wondered what the picture had to do with Wandel & Goltermann

[Cerebus]

No guesses so another image.

Must be a rubidium frequency standard from a military navigation system.      I wouldn't be too surprised if there is an Efratom or Ball Aerospace-Efratom  rubidium clock inside the box.

Almost right. It's for a secure frequency hopping radio. I connects to the Have-Quick enabled PTR-1751WWH
It contains a Frequency Electronics FE-5650A-M
The output is 800kHz divided down (by 16) from 12.8MHz generated from the interal DDS. As the 12.8MHz is crystal filtered it is not easy to change to 10MHz even though the 5650A has DIP switches internally. It can be done but 800kHz will easilly divide to 100kHz and 1Hz so still useful.
Anyway worth £20 each anyday.
Note I've not put the number etc. I got burnt a bit on these years ago. Some turned up on ebay I wasn't placed to buy more than one. Someone else bought one and asked about it on the time nuts mail list. I told him what I knew and net thing he had bought them all, taken the Rb's out and sold them at a large markup.
Scotttrade12 has done similar recently. The first listed them complete at silly monery then dropped the price to £20
www.ebay.co.uk/itm/224434964973? but that is the EMPTY CASE. He sold the Rb's for atound £100
www.ebay.co.uk/itm/324574645606.

I saw the FE-5650A-Ms that scottrade listed and mused about buying one. Problem is that I don't regard him as a reliable, knowledgable vendor. Had I know that they'd just been freshly extracted from inside one of those I might have chanced a punt. Oh well, c'est la vie.

Ah, but did you know that it was fixed 800kHz output ?

That was another thing against it, there was no option marked so the frequency was "who knows?". As it is my interest is in "stable" rather than any specific frequency, my hat needs a third corner.

[Cerebus]

Gahh,
NLP AI conference.

mah head hurtz ...

Your head will soon be replaced by NLP AI. Pointy-haired bosses won't be able to tell its output is rubbish, and will resort to "because the computer says so".

Read comp.risks to see how such things appear to "work" for spurious "reasons", e.g. https://catless.ncl.ac.uk/Risks/32/80/#subj4.1
e.g.   
  "Some AIs were found to be picking up on the *text font* that certain
  hospitals used to label the scans. As a result, fonts from hospitals with
  more serious caseloads became predictors of covid risk."

I was first aware of that in the early 80s, when Igor Alexander's WISARD correctly distinguished between cars and tanks in the lab, but failed dismally when taken to Luneberger Heath in north Germany. Eventually they worked out the training set was tanks under grey skys and car adverts under sunny skies.

Heh, heh. I remember that. More than that I knew Igor at the time. Happy bunny he was not when that happened. Jokes may have been made at his expense, people may have pretended not to recognise him on a sunny day (Brunel, where he was at the time was famously wet and grey).  

Igor's over-enthusiasm at what he could do with "black box" models, and the subsequent failures, was one of the reasons that, back when I was doing AI, I chose to work on the "knowledge engineering" and "automated inference" end of things. My personal understanding of how hard that was to do, and why we discounted neural nets as "the answer" is why I am a hyper-sceptic in the face of the current fashion for fools believing that "AI" will solve all our problems.

[Saskia]

cannot help it, it's my job.

[bd139]

And it's my job to scrape up all the corpses when it all turns to shit 

Talking of which was going for a nice walk out on the Ridgeway this evening but some cunt smashed their car on the M25 and caused a tailback about 5 miles long and some bearded sky cunt turned the rain on. Decided to do a 180 and come home and play with test gear instead.

[Saskia]

that's ok. if I can program an AI to pretend to do my job well enough to keep my bosses happy, I can start playing with my toys while the AI handles the rest ...

[Cubdriver]

Well the HP 5245L arrived today, and HOLIO-FUCK this thing is heavy! Must be 12kg if it's an ounce!
I imagine the conversation between the salesman and the customer when it was new...
Customer: Bloody hell, this thing weighs a ton!
Salesman: There are carrying handles on the side, what more do you want? Anyway, it's only 12kg.
Customer: Are you going to cover my chiropracter bill?
Salesman: As long as you supply NIST traceable documentation.


Since it has one of those stupid proprietary HP power connectors, for testing purposes I had to make a Heath Robinson lash-up involving Safeblok, flex, and croc leads.
Naturally I checked the main fuse before giving it any beans, and I'm glad I did:



It's a nice vintage fuse and all, but a 7A 150V fuse in a device that requires a 1A 250V one?   



There's some issues with the display:



Digits 3 and 4 are absent, and digit 5 has problems lighting the entire number, with 8 not appearing at all. This may or may not be failed tubes, as looking at the state of the boards it's a wonder anything works at all:



Some of the neon decimal points also don't work, and there's a problem with digits 1 and 2 in that they don't give a reading, except sometimes a momentary one while switching ranges. It's giving incorrect readings in the period settings, though they are linear up and down the ranges, which might be a helpful clue. The general crustiness is indicative of several years in a damp shed or cellar, to my mind.




Must have been fun* hand soldering the loom onto the plug-in carrier.



On the plus side, the accurate and stable frequency readings imply that this little beauty is working as specified:





*sighs* another item for the repair queue...   

Bear in mind that it replaced these:




which tipped the scales at 48 lbs/22 kg, so at the time it was relatively svelte for the breed!

-Pat

[Robert763]

No guesses so another image.

Must be a rubidium frequency standard from a military navigation system.      I wouldn't be too surprised if there is an Efratom or Ball Aerospace-Efratom  rubidium clock inside the box.

Almost right. It's for a secure frequency hopping radio. I connects to the Have-Quick enabled PTR-1751WWH
It contains a Frequency Electronics FE-5650A-M
The output is 800kHz divided down (by 16) from 12.8MHz generated from the interal DDS. As the 12.8MHz is crystal filtered it is not easy to change to 10MHz even though the 5650A has DIP switches internally. It can be done but 800kHz will easilly divide to 100kHz and 1Hz so still useful.
Anyway worth £20 each anyday.
Note I've not put the number etc. I got burnt a bit on these years ago. Some turned up on ebay I wasn't placed to buy more than one. Someone else bought one and asked about it on the time nuts mail list. I told him what I knew and net thing he had bought them all, taken the Rb's out and sold them at a large markup.
Scotttrade12 has done similar recently. The first listed them complete at silly monery then dropped the price to £20
www.ebay.co.uk/itm/224434964973? but that is the EMPTY CASE. He sold the Rb's for atound £100
www.ebay.co.uk/itm/324574645606.

I saw the FE-5650A-Ms that scottrade listed and mused about buying one. Problem is that I don't regard him as a reliable, knowledgable vendor. Had I know that they'd just been freshly extracted from inside one of those I might have chanced a punt. Oh well, c'est la vie.

Ah, but did you know that it was fixed 800kHz output ?

That was another thing against it, there was no option marked so the frequency was "who knows?". As it is my interest is in "stable" rather than any specific frequency, my hat needs a third corner.

If you want one of the 5650A's drop me a PM.

[25 CPS]

Bear in mind that it replaced these:




which tipped the scales at 48 lbs/22 kg, so at the time it was relatively svelte for the breed!

-Pat

Are both those 523 counters yours? They're beautiful.  I've got a 523D that I really should get moving on restoring.

[mansaxel]

I'm over-TEA'd. When I saw W&G I wondered what the picture had to do with Wandel & Goltermann

I might, just might, have set that up on purpose....

 

[AVGresponding]

Bear in mind that it replaced these:




which tipped the scales at 48 lbs/22 kg, so at the time it was relatively svelte for the breed!

-Pat

Yes, the manual states a mass of 14.4kg (sans plug-in), and describes that as "lightweight".   

Rather more annoyingly, the only manuals I can find (from the usual suspects) are for later versions. I can find a 628 prefix and a 716 prefix manual, but this has a 425 prefix, so apparently I need a 544 prefix manual.   

Because of this it's not clear to me whether the display assemblies are all the correct ones. The available manuals state 6x 600kHz, 1x 5MHz, and 1x 50MHz modules, but this has a 5-2-1 layout.



Need to pull all the modules out, one way or another. Do the display modules come out without removing the front panel? I removed the retention bar across the top but they still don't want to come.



EDIT: The 628 prefix manual has the 5-2-1 layout, I just hadn't looked that far when I last posted.   
          Currently reading on...

[Cubdriver]

Bear in mind that it replaced these:




which tipped the scales at 48 lbs/22 kg, so at the time it was relatively svelte for the breed!

-Pat

Are both those 523 counters yours? They're beautiful.  I've got a 523D that I really should get moving on restoring.

Yes, they're both mine - took a few years to find them.  Need to finish cleanup and checkout of them and get a good set of photos one of these years.  Cool old pieces of gear.

-Pat

[Cubdriver]

Bear in mind that it replaced these:




which tipped the scales at 48 lbs/22 kg, so at the time it was relatively svelte for the breed!

-Pat

Yes, the manual states a mass of 14.4kg (sans plug-in), and describes that as "lightweight".   

Rather more annoyingly, the only manuals I can find (from the usual suspects) are for later versions. I can find a 628 prefix and a 716 prefix manual, but this has a 425 prefix, so apparently I need a 544 prefix manual.   

Because of this it's not clear to me whether the display assemblies are all the correct ones. The available manuals state 6x 600kHz, 1x 5MHz, and 1x 50MHz modules, but this has a 5-2-1 layout.



Need to pull all the modules out, one way or another. Do the display modules come out without removing the front panel? I removed the retention bar across the top but they still don't want to come.

I think you should be ok with the installed modules, as they're all at or above the minimum operating frequency for their positions.

Once the retaining bar is removed, they should just pull out - no need to remove the front panel.  That said, if they've been in place for a long time (as they likely have), they will not come out willingly and it'll take some pulling on them to really convince them that you're not kidding and want them out.  Removing the smaller boards behind them, then getting under and gently prying with a small pinch bar will help break them free.  When removing them, be careful to pull them slowly so they don't suddenly pop free and smash the nixie.  Controlling the extraction once you get them moving is the biggest challenge.  Once I get them moving in the edge connectors, I normally keep a firm grip with one hand at each end and rock them gently fore and aft while lifting, so as to walk them out of the edge connector sockets.

-Pat

[AVGresponding]

Yeah they lift at the back, but not at the front. Having said that, the plug-in screw was stubborn af too. Will persist in a non-Big Clive manner.

[factory]

Yeah they lift at the back, but not at the front. Having said that, the plug-in screw was stubborn af too. Will persist in a non-Big Clive manner.

Be careful those early boards are quite fragile (one in my 5243L was cracked across the connector area), I think the cards are the correct ones, would be very obvious if later ones had been fitted.
I can probably help with transistors for the 5245L-4B card, I bought a lot of cards from Sphere (minus tubes) for the decoder assemblies.
You will probably need to replace all the decimal point neons, as they remaining working ones will snap off (copper plated steel rust wires) when the assembly is removed.

Would it not have replaced the 524 counter? as that is the only older one with an eight digit display, I'd post a picture but I think they are trapped on my old HDD*

*Got to catch-up again, computer is now rebuilt, minus transferring the stuff from the old HDD's, currently 55% of the way to format the new 4TB HDD.
The build got delayed by a day, due to me not noticing the new Mobo was larger & needed more hex spacers, plus the gorrilla that built the machine in 2008 split one of orginal spacers forcing a longer screw in, as well as another one that had a metric screw forced into #6 UNC hole , thankfully the local PC shop in Nantwich had some hex spacers in their spares box  . 

David

[med6753]

Wow...does this bring back memories. IBM SLT logic.

What Fran doesn't realize is that there was a follow on to SLT called MST (Monolithic System Technology) which used the same case style and footprint but rather than discrete transistor chips and resistors used integrated circuits. Most of S/370 mainframes used MST.

[Cubdriver]

Yeah they lift at the back, but not at the front. Having said that, the plug-in screw was stubborn af too. Will persist in a non-Big Clive manner.

Be careful those early boards are quite fragile (one in my 5243L was cracked across the connector area), I think the cards are the correct ones, would be very obvious if later ones had been fitted.
I can probably help with transistors for the 5245L-4B card, I bought a lot of cards from Sphere (minus tubes) for the decoder assemblies.
You will probably need to replace all the decimal point neons, as they remaining working ones will snap off (copper plated steel rust wires) when the assembly is removed.

Would it not have replaced the 524 counter? as that is the only older one with an eight digit display, I'd post a picture but I think they are trapped on my old HDD*

*Got to catch-up again, computer is now rebuilt, minus transferring the stuff from the old HDD's, currently 55% of the way to format the new 4TB HDD.
The build got delayed by a day, due to me not noticing the new Mobo was larger & needed more hex spacers, plus the gorrilla that built the machine in 2008 split one of orginal spacers forcing a longer screw in, as well as another one that had a metric screw forced into #6 UNC hole , thankfully the local PC shop in Nantwich had some hex spacers in their spares box  . 

David

I suppose that could be argued, but I had pics of the 523s handy and had to excavate the 524 to get something.  That one's REALLY a big 'un!  (And arrived pretty banged up; haven't even tried to apply power to it yet)





That and with the gazillion tubes, it's more a winter counter....

-Pat

[factory]

Well the HP 5245L arrived today, and HOLIO-FUCK this thing is heavy! Must be 12kg if it's an ounce!
I imagine the conversation between the salesman and the customer when it was new...
Customer: Bloody hell, this thing weighs a ton!
Salesman: There are carrying handles on the side, what more do you want? Anyway, it's only 12kg.
Customer: Are you going to cover my chiropracter bill?
Salesman: As long as you supply NIST traceable documentation.


Since it has one of those stupid proprietary HP power connectors, for testing purposes I had to make a Heath Robinson lash-up involving Safeblok, flex, and croc leads.
Naturally I checked the main fuse before giving it any beans, and I'm glad I did:



It's a nice vintage fuse and all, but a 7A 150V fuse in a device that requires a 1A 250V one?   




Must have been fun* hand soldering the loom onto the plug-in carrier.

*sighs* another item for the repair queue...   

The mains inlet connector is not made by hp but Belden/Volex & another make I can't remember, see the links from the first page under "Electronics Miscellany", the power lead/cord has a PH-163 connector, as used for lots of US made TEA & other equipment at the time.

I'm sure I'll have lots of fun modding my 5243L, to allow use of the full range of plug-ins, I did end up with a 5245L carcass from chipwoods* that will provide the loom, *less said about that the better.



David

[factory]

Yeah they lift at the back, but not at the front. Having said that, the plug-in screw was stubborn af too. Will persist in a non-Big Clive manner.

Be careful those early boards are quite fragile (one in my 5243L was cracked across the connector area), I think the cards are the correct ones, would be very obvious if later ones had been fitted.
I can probably help with transistors for the 5245L-4B card, I bought a lot of cards from Sphere (minus tubes) for the decoder assemblies.
You will probably need to replace all the decimal point neons, as they remaining working ones will snap off (copper plated steel rust wires) when the assembly is removed.

Would it not have replaced the 524 counter? as that is the only older one with an eight digit display, I'd post a picture but I think they are trapped on my old HDD*

*Got to catch-up again, computer is now rebuilt, minus transferring the stuff from the old HDD's, currently 55% of the way to format the new 4TB HDD.
The build got delayed by a day, due to me not noticing the new Mobo was larger & needed more hex spacers, plus the gorrilla that built the machine in 2008 split one of orginal spacers forcing a longer screw in, as well as another one that had a metric screw forced into #6 UNC hole , thankfully the local PC shop in Nantwich had some hex spacers in their spares box  . 

David

I suppose that could be argued, but I had pics of the 523s handy and had to excavate the 524 to get something.  That one's REALLY a big 'un!  (And arrived pretty banged up; haven't even tried to apply power to it yet)





That and with the gazillion tubes, it's more a winter counter....

-Pat

I'd definitely argue that, as the 523x series has time interval functions, that aren't included as standard with the 524x series or 5245L (unless the correct plug-in is fitted of course).
Pics of my 524B are trapped on my phone and I didn't reinstall the PC drive bay panel with the USB's, as it needs more repair/bodging, or better still wildarizing. 

David

[AVGresponding]

I need to check the OP to be sure, but I don't think we've had "you know you're a TE addict when you use a new toothbrush to clean rusty transistors rather than your teeth".

Tesco website is a bit fooked atm, DDOS? Needed more toothbrushes...   

Had all the topside boards out of the 5245L, gently cleaned (not gently enough, a rusty transistor leg broke, plugging the Metcal in right now), adjusted some clearances (some of those TO-5 case were, shall we say, inadvertently thermally coupled?).
Swapped some display boards around, and as soon as I've fixed that tranny leg on the A18 display board I'll give it a burl.

[tggzzz]

Gahh,
NLP AI conference.

mah head hurtz ...

Your head will soon be replaced by NLP AI. Pointy-haired bosses won't be able to tell its output is rubbish, and will resort to "because the computer says so".

Read comp.risks to see how such things appear to "work" for spurious "reasons", e.g. https://catless.ncl.ac.uk/Risks/32/80/#subj4.1
e.g.   
  "Some AIs were found to be picking up on the *text font* that certain
  hospitals used to label the scans. As a result, fonts from hospitals with
  more serious caseloads became predictors of covid risk."

I was first aware of that in the early 80s, when Igor Alexander's WISARD correctly distinguished between cars and tanks in the lab, but failed dismally when taken to Luneberger Heath in north Germany. Eventually they worked out the training set was tanks under grey skys and car adverts under sunny skies.

Heh, heh. I remember that. More than that I knew Igor at the time. Happy bunny he was not when that happened. Jokes may have been made at his expense, people may have pretended not to recognise him on a sunny day (Brunel, where he was at the time was famously wet and grey).  

Igor's over-enthusiasm at what he could do with "black box" models, and the subsequent failures, was one of the reasons that, back when I was doing AI, I chose to work on the "knowledge engineering" and "automated inference" end of things. My personal understanding of how hard that was to do, and why we discounted neural nets as "the answer" is why I am a hyper-sceptic in the face of the current fashion for fools believing that "AI" will solve all our problems.

Accessible references would be welcome; I haven't found any.

[mnementh]

that's ok. if I can program an AI to pretend to do my job well enough to keep my bosses happy, I can start playing with my toys while the AI handles the rest ...

Until they hire the AI directly for half your salary in cloud bandwidth... 

mnem
 

[bd139]

I need to check the OP to be sure, but I don't think we've had "you know you're a TE addict when you use a new toothbrush to clean rusty transistors rather than your teeth".

Tesco website is a bit fooked atm, DDOS? Needed more toothbrushes...   

Had all the topside boards out of the 5245L, gently cleaned (not gently enough, a rusty transistor leg broke, plugging the Metcal in right now), adjusted some clearances (some of those TO-5 case were, shall we say, inadvertently thermally coupled?).
Swapped some display boards around, and as soon as I've fixed that tranny leg on the A18 display board I'll give it a burl.

Good luck with that one. May the odds be ever in your favour.

Rust is one of those things which is an absolute bastard though and I’d consider it cursed now for eternity. That would consist of turd polishing it and throwing it back on eBay for spares repair.


Bad word in these parts: HR. They strike again hence my lack of spamming for most of today. Apparently they didn’t read the contract they drafted poorly, issued to me, I joyously signed for while laughing maniacally and that they are now bound by which they violated and I am now ensuring is enforced properly  . Be careful what you sign as they say 

[bd139]

Gahh,
NLP AI conference.

mah head hurtz ...

Your head will soon be replaced by NLP AI. Pointy-haired bosses won't be able to tell its output is rubbish, and will resort to "because the computer says so".

Read comp.risks to see how such things appear to "work" for spurious "reasons", e.g. https://catless.ncl.ac.uk/Risks/32/80/#subj4.1
e.g.   
  "Some AIs were found to be picking up on the *text font* that certain
  hospitals used to label the scans. As a result, fonts from hospitals with
  more serious caseloads became predictors of covid risk."

I was first aware of that in the early 80s, when Igor Alexander's WISARD correctly distinguished between cars and tanks in the lab, but failed dismally when taken to Luneberger Heath in north Germany. Eventually they worked out the training set was tanks under grey skys and car adverts under sunny skies.

Heh, heh. I remember that. More than that I knew Igor at the time. Happy bunny he was not when that happened. Jokes may have been made at his expense, people may have pretended not to recognise him on a sunny day (Brunel, where he was at the time was famously wet and grey).  

Igor's over-enthusiasm at what he could do with "black box" models, and the subsequent failures, was one of the reasons that, back when I was doing AI, I chose to work on the "knowledge engineering" and "automated inference" end of things. My personal understanding of how hard that was to do, and why we discounted neural nets as "the answer" is why I am a hyper-sceptic in the face of the current fashion for fools believing that "AI" will solve all our problems.

Accessible references would be welcome; I haven't found any.

Better to ask for any cost-benefit analysis of ML research. I’m not going to use the term AI because that’s just a lie. It’s not artificial and it’s not intelligence.

I always like to point out that actual meat shaped intelligence usually can’t resolve all facets of an algorithmic problem or any decision making process so why should an ML model be able to do it with any level of reliability or determinism that doesn’t put your business at mortal risk.

A fine example is the risk management system I was involved in a few years back. They fed their entire corpus of good and bad risk outcomes into an ML platform by a normal consultancy we’ve all heard of. The first test suite they ran through it, written by some actual actuaries, suggested that they should quite happily lend £200m to an opium farm in Afghanistan (it wasn’t that but something monumentally as stupid from a risk perspective).