Test Equipment Anonymous (TEA) group therapy thread

[Robert763]

Back to TEA.  My £4 Permameter turned up last week.

Pretty much as described but sensor and meter serial numbers don't match. Basically works but won't zero properly and very sensitiv to movement.  So I opened up the probe. Fortunatly it was on a flat surface as it filled with thick damping oil (feels like silicone) 
It's really basic, a metal strip with a magnet at one end and a single strain gauge at the other. Put in on a magnetic surface and magnet is attracted to it straning the gauge. Probs are is its not mechanically balanced so affected by gravity and how level it is. Gauge is a 3 wire connection so that's not great. It uses AC and has both resistive and capacitive zero controls.
So no practical use but parts wise not ttoo bad. Sifam 100uA meter with plain markings, two pairs of 3 pin Lemo connectors, 10T pot, Dual stator variable capacitor, couple of decent switches and some collet knobs.

That's neat. Which axis is it supposed to measure from? What is the original application? The only magnetic permeameter I've ever worked with was a big bruiser like this thing, which really only worked on flat stock of very limited dimensions.

mnem
 

Sits horizontal on the flat sample being measured. I think it was used to check heat treatment of stainless alloys. it's an aerospace unit.

[Robert763]

More TEA. I mentioned previously that the muppets in Skegness broke the screen on a fuel test set I bought.  The thing is basically a precision capacitance meter and dual decade box. Both these are guarded. All controlled by a small PC104  PC running a dedicated program on FreeDOS.
So the screen is a Planar Electroluminescent 640x480 monochrome (amber). These are rare and very expensive even used. It is driven by a PC104 graphics controller card. The basic card supports CRT/LCD and EL. So I thought no problem stick a 310x240  EL that I have (it’s just text and CGA text graphics)  or a cheap LCD in there. However having got to the PCB (there is a six card PC104 “stack” in a frame so not easy to get to) the makers have bought a stripped down version that only supports the specific EL 640x480  display.
The software is on a CF card. Plugging that into another PC with standard VGA showed it ran OK. So I dug out a 8 bit VGA card and made a PC104 to ISA adaptor. Unfortunately that didn’t work. Probably a BIOS issue, I checked the card in another motherboard.

The design of the unit meant that the CPU card had to sit in the middle or top of the stack of cards. This requires the CPU board to have male PC104 connectors as well as female. None of the PC104 cards I had in “stock” had both VGA output and male connectors.
So I embarked on a complete re-casing into a ½ rack 3U case with a uITX +PC104 motherboard with new power supply. First thing was making up 64 pin and 40 pin Male to Male extension cables to connect the motherboard to the custom PC104 stack. It actually all works.
The one thing it never did was allow you to measure the decade outputs with it’s own meter.

Even more surprising, this unit, last calibrated in 2015 but re-built, matches my ESI 252 last calibrated in 1995 and GenRad 722D built in 1948 and last calibrated in 1980. The 252 and 722D also agree. Overall a useful instrument if a little large for what it does.
The capacitance decades are binary sequence (+2000pF on high range) each relay switched section has a custom precision fixed capacitor in parallel with a high quality piston trimmer capacitor. Currently I use it with an external monitor but there is space on the panel for a 5.6” LCD if I start using it a lot.

[Cerebus]

And software wranglers wonder how the hell it ever worked while watching it fail 

The hardest problem in software engineering is failing well.

Anyone can be monkey see, monkey do -trained to do things that work in nice conditions, but add a printer, a user or some packet loss, ie. reality and things start to get quite hairy.

Well that was interesting timing. I was awoken at 2AM by an alert where a piece of software had failed and decided to tell everyone about that by throwing 180 megabits at the logger from all runtime threads in a tight loop until someone poked it 

I presume you mean 180Mb/s? If there's one thing that's as bad as "no logs" it's "logs with so much in them that you can't find the problem"; analogous to trying to find a 200ps glitch in several seconds worth of scope trace.

[Cerebus]

I gather the reason why we got done when we did (apart from qualifying for DCMS/government funding) is that our local part-time Celeb ordered FTTP on demand for her country residence. Price rumoured to be around £30,000 to £50,000. I suppose she can afford it!

You mean "The taxman can afford it". If it's so that the part-time Celeb can take part in broadcasts then it's an allowable business expense. The only question then is whether they get to write the install cost down over time or manage to finagle it in one go as operating expenditure.

[Cerebus]

As of this morning it appears that Mouser USA is still shut down. Last e-mail I got from them said parts would be shipped this coming Monday. Turns out there's a bunch of guys on the Facebook Tek group waiting on parts too. Mouser doesn't operate Saturday/Sunday so I'm a bit skeptical Monday will happen. In my mind if they are back up it would be smart for them to offer overtime to their staff and work Saturday/Sunday to clear the backlog. Especially for their high volume corporate customers. I'll bet the employees would be willing to do that if they lost income earlier in the week. But we'll see what happens and in reality I'm in no rush for the parts.   

The boy on the mule would be there by now. 

[tggzzz]

Just down the road from BD, SW19 3QG 

https://www.i-bidder.com/en-gb/auction-catalogues/wignall-brownlow/catalogue-id-wi310082/lot-d1f0713a-0aa2-4052-9c46-acd400a422b8

[Cerebus]

So what, you're going to use it to count complaint tickets on the Kubernetes deployment you're trying to unfuckerize...? 

mnem
 

In that case, UHF won't cut it. 

[Specmaster]

Just down the road from BD, SW19 3QG 

https://www.i-bidder.com/en-gb/auction-catalogues/wignall-brownlow/catalogue-id-wi310082/lot-d1f0713a-0aa2-4052-9c46-acd400a422b8

Just down the road it might be but, there are a few things wrong with it I think, firstly its Tektronics and secondly, more importantly its got a FAN

[25 CPS]

...mentioned previously that the muppets in Skegness broke...

Those three words explain everything that went wrong.  Granted, it could be shortened down to one with just "Skegness" but I digress.

[Cerebus]

Two: If it had been anybody but me saying these things, you'd have dropped it 3 pages ago.

Or you could have stopped digging yourself a hole the first time someone correctly countered your claim that " no manufacturer in the history of ever has done this [used the same oil for the engine and automatic gearbox]".

The engine oil became dirty, and the same oil was used for the automatic gearbox. Dirt confused the hydraulic logic

To my knowledge, no manufacturer in the history of ever has done this. ever.

You did what you know always annoys Robert and went on as if your central thesis hadn't been demolished, acting as if  you'd said something different, dissembling instead of giving a straight up "I stand corrected".

It takes two to argue...

[bd139]

Just down the road from BD, SW19 3QG 

https://www.i-bidder.com/en-gb/auction-catalogues/wignall-brownlow/catalogue-id-wi310082/lot-d1f0713a-0aa2-4052-9c46-acd400a422b8

Ooh that isn't far at all

[bd139]

And software wranglers wonder how the hell it ever worked while watching it fail 

The hardest problem in software engineering is failing well.

Anyone can be monkey see, monkey do -trained to do things that work in nice conditions, but add a printer, a user or some packet loss, ie. reality and things start to get quite hairy.

Well that was interesting timing. I was awoken at 2AM by an alert where a piece of software had failed and decided to tell everyone about that by throwing 180 megabits at the logger from all runtime threads in a tight loop until someone poked it 

I presume you mean 180Mb/s? If there's one thing that's as bad as "no logs" it's "logs with so much in them that you can't find the problem"; analogous to trying to find a 200ps glitch in several seconds worth of scope trace.

Yes 180Mb/s. We have several hundred Gb a day of logs but that's taking the piss 

[med6753]

Just down the road from BD, SW19 3QG 

https://www.i-bidder.com/en-gb/auction-catalogues/wignall-brownlow/catalogue-id-wi310082/lot-d1f0713a-0aa2-4052-9c46-acd400a422b8

Ooh that isn't far at all

It looks rough but appears to be in one piece. BUT.....are the vacuum tubes present?

[bd139]

On topic for once. Racal pr0n attached. Seems fairly clean and no sign of any buggery. Will do a full check out later and power it up but back to Kubernetes now 

At first glance it looks untouched by Gorillas. Hopefully applying power proves that. 

Well all is not well in the house of Racal. The low frequency input is absolutely A ok fine. In fact it looks like it's roughly on cal at least as well.

The high frequency input is quite frankly fucked. No response at all. Usually in these circumstances I instantly suggest that someone plugged a transmitter into it and keyed it. I shall investigate. I'm not really that bothered as I've got something to fix now. As long as it's not an unobtanium part it will likely be fairly trivial.

[med6753]

On topic for once. Racal pr0n attached. Seems fairly clean and no sign of any buggery. Will do a full check out later and power it up but back to Kubernetes now 

At first glance it looks untouched by Gorillas. Hopefully applying power proves that. 

Well all is not well in the house of Racal. The low frequency input is absolutely A ok fine. In fact it looks like it's roughly on cal at least as well.

The high frequency input is quite frankly fucked. No response at all. Usually in these circumstances I instantly suggest that someone plugged a transmitter into it and keyed it. I shall investigate. I'm not really that bothered as I've got something to fix now. As long as it's not an unobtanium part it will likely be fairly trivial.

Shit, a dump and run Gorilla. 

[Ero-Shan]

There is a Fluke 8350A (with Nixie tubes!) in the German ebay ads for 100 Euro:

https://www.ebay-kleinanzeigen.de/s-anzeige/fluke-8350a-digital-multimeter-nixie-amateurfunk-messgeraet/1664660248-168-3483



No affiliations with the seller.

Nice one. He also has a R+S NGK 35 for 110 €.

[bd139]

On topic for once. Racal pr0n attached. Seems fairly clean and no sign of any buggery. Will do a full check out later and power it up but back to Kubernetes now 

At first glance it looks untouched by Gorillas. Hopefully applying power proves that. 

Well all is not well in the house of Racal. The low frequency input is absolutely A ok fine. In fact it looks like it's roughly on cal at least as well.

The high frequency input is quite frankly fucked. No response at all. Usually in these circumstances I instantly suggest that someone plugged a transmitter into it and keyed it. I shall investigate. I'm not really that bothered as I've got something to fix now. As long as it's not an unobtanium part it will likely be fairly trivial.

Shit, a dump and run Gorilla. 

Not yet.

Did some debugging. There's no 24V supply to IC36 which is a "thick film amplifier" Plessey SP8635B. Poked around and found that Q30 in the power supply which drops 31V to 24V is duff. There's no volts at the collector. It's a ZTX550.

But why did he die?

Well it's one of three things:

1. The IC is duff. Unfortunately possible. There is a PIN diode attenuator and cut off relay which should prevent that, but it's Plessey who are the Lucas of components.
2. The decoupling network in second attachment has a shorty mctant in it.

Place your bets!

More diagnostics coming later. Need to go for a walk first.

[AVGresponding]

Rapid parcel with X/Y caps arrived yesterday evening so today I removed the smoke bomb from the HP 6633B.
The RIFA was a 220n, and the caps I ordered are 100n (and some 3n3), so I had to make a Frankencap:



Fucking Murphy tried to spoil my day by stealing the nuts from the M2.5 screws that hold the IEC plug, fuse and filter board to the chassis, but little did he know I have a box of small stainless fixings, so FU Murphy!


It now sits in the pile, and I've retired the 7065. Had to remove the stupid rubber baby buggy bumpers off the TTi arb to make things fit neatly.



Even on minimum brightness the TTi display is so bright there's no way to get an exposure setting that suits everything and it at the same time.   
It's actually outputting a 100Hz 4vrms sine wave, being measured by the counters and the Keithley 197.

[mnementh]

Two: If it had been anybody but me saying these things, you'd have dropped it 3 pages ago.

Or you could have stopped digging yourself a hole the first time someone correctly countered your claim that " no manufacturer in the history of ever has done this [used the same oil for the engine and automatic gearbox]".

The engine oil became dirty, and the same oil was used for the automatic gearbox. Dirt confused the hydraulic logic

To my knowledge, no manufacturer in the history of ever has done this. ever.

You did what you know always annoys Robert and went on as if your central thesis hadn't been demolished, acting as if  you'd said something different, dissembling instead of giving a straight up "I stand corrected".

It takes two to argue...

So how many more are we going to add to this list of cars that had a really, really bad design? Showing more examples of a bad idea does not make it a good idea; it just shows there were more than one person who had that bad idea. 

Like that giant-scale UAV built from hobbyist-grade quadcopter parts and being flown over spectators with firmware that clearly didn't have any failsafe programmed.

That said... obviously I was wrong in that nobody ever did it. I stand (well, sit) corrected. Just because i have trouble believing anyone would make such a poor design, does not necessarily make it so.

I guess this proves what my grand-dad used to say... "There is no idea so bad that someone hasn't tried to make a buck off it."

mnem
 

Except when I did exclaim exactly that, with deliberate emphasis, 3 pages ago.

And not only that, I explained why I didn't let it go up to that point in simple, reasonable language. The fact was I simply couldn't believe anybody would make such a bad design choice into an actual product.

Now... can we please let it go, or do we need to pummel me some more with quotes out of context...?

mnem
 

[mnementh]

...Did some debugging. There's no 24V supply to IC36 which is a "thick film amplifier" Plessey SP8635B. Poked around and found that Q30 in the power supply which drops 31V to 24V is duff. There's no volts at the collector. It's a ZTX550.

But why did he die?

Well it's one of three things:

1. The IC is duff. Unfortunately possible. There is a PIN diode attenuator and cut off relay which should prevent that, but it's Plessey who are the Lucas of components.
2. The decoupling network in second attachment has a shorty mctant in it.

Place your bets!   More diagnostics coming later. Need to go for a walk first.

LOL... toldja so!

On topic for once. Racal pr0n attached. Seems fairly clean and no sign of any buggery. Will do a full check out later and power it up but back to Kubernetes now 

At first glance it looks untouched by Gorillas. Hopefully applying power proves that. 

Ahhh... The TEA anthem: "Off we go, on another wild RIFA hunt."

mnem
Uh oh... now all those tants feel left out...

mnem
Capacitors are Murphy's footsoldiers.

[Ero-Shan]

Since I cleaned up my workbench last weekend, and all this talk about Rohde & Schwarz PSUs with BU508A, I decided to finally pull out the NGPV 40/3 and have a look at it:



It worked a whole lot better than expected, albeit being dusty and full of nasty stickers on the front panel. 30.00 Volts set, 29.96 measured. Not too far off.
The device has a fan, but totally silent. Maybe it is temperature controlled, 10 MΩ is not exactly a heavy load.
After a few minutes however, a well known smell seeped out of the thing. Something inside seemed to rise its temperature and heat the dust.
Maybe the fan wasn't temperature controlled after all?

The ubiquitous porn shot:



Here's the fan, blowing into a heat sink tunnel.



Did the rotate-with-finger thing. Boy, that needed some torque! No motor in the single-digit watts class would turn it!
Removed the whole power transistor assembly and the fan (which actually does have some temp control; a bi-metal switch puts it into full power when the heat sink gets too hot):



At first I went to my motor department to see whether I might have something to replace it with. Of course not. It's a rather small department anyway.
There was just one thing left to dot: try to fix it.
By now you should know that I'm not the greatest friend of riveted ties, but sure as hell the brackets are riveted to the stator. Grrr.
After drilling them off, the front bracket and bearing would slide off the axle without a hitch. Holding the back bracket and turning the rotor did turn the bearing in the bracket instead of the axle in the bearing! It took quite some force to get the axle out of its bearing.
I cleaned it all with IPA and reassembled the thing. To my amazement, the motor now ran smoothly.



Putting all together again, the PSU now runs with a blowing fan and without that distinctive smell.



Now it really deserved to be de-stickerized and cleaned.



One fine day I will have to check the GPIB functionality. This PSU is not really meant to be used by the front panel controls.

[Cerebus]

Now... can we please let it go, or do we need to pummel me some more with quotes out of context...?

mnem
 

No, not when you claim a quote is 'out of context' when it is provided complete with context. 

[Cerebus]

Well it's one of three things:

1. The IC is duff. Unfortunately possible. There is a PIN diode attenuator and cut off relay which should prevent that, but it's Plessey who are the Lucas of components.
2. The decoupling network in second attachment has a shorty mctant in it.

Place your bets!

More diagnostics coming later. Need to go for a walk first.

I'll bet you a pint of your choice that it's the evil little tant.

[bd139]

I have checked the tants  and they are fine. Think Q30 just died.

Edit; going to throw 24v at the decoupling network off a bench supply with current limit and check that the IC isn’t toast first. That could have gone bad and caused the SOA to be exceeded of the transistor.

[Cerebus]

Bugger, looks like that's a pint I owe you. Unfortunately you won't be collecting it for a while, and even then you might have to dress like this:



I guess though it'll make a change from this: