EEVblog #1334 - Mystery Dumpster Teardown

[EEVblog]

Mystery dumpster teardown time!
With the most amazing mechanical mains power switch you'll ever see!

[SilverSolder]

Loved the "ribbon actuator", it looks like something you could buy by the foot (or meter)!

[PA0PBZ]

I think you missed 2 important things:

1) You can change the KIND of input, it's now set to TC, I bet it can do voltages too.



2) You put the ribbon back wrong, it no longer goes under the printhead.



Interesting device, that's for sure.

[eti]

What kind of dunderhead ignoramus philistine would throw equipment of such high calibre, with such a thick metal casing, into a skip with sufficient force to dent it?

People lack sense, the consumer mindset is a disease. This is a machine of great engineering and built to last 100 years, I'd like to find the person and slap them.  People lack any sense or integrity, the "just bin it and get a better one" attitude is extremely irksome.

[tszaboo]

OK, so what do you use an equipment like this as an EE?
I had to do temperature testing of parts recently. The goal was to record worst case temperatures of components on a PCB. Ambient temperature had to be increased in a thermal chamber, and the temps had to be stable for an hour to take a reading. And had to be repeated on 10 samples. Those are the rules, it is for safety, so you dont blow up an oil refinery (worst case) or a harbor.

If I do the measurement sequentially, one by one, the test would take me about a week to complete, I dont think I could've fit the testing of a third card into a day. Instead, I bought a scanner card to our DMM6500 and used that. That had 10 TC inputs, so I could finish the test in 2 days, by testing 5 PCBA at the same card. Honestly, I should've bought a logger like this, that would've made my job even simpler.

[The Soulman]

I've seen those switches in plenty of consumer gear (many) years ago.

[thm_w]

What kind of dunderhead ignoramus philistine would throw equipment of such high calibre, with such a thick metal casing, into a skip with sufficient force to dent it?

People lack sense, the consumer mindset is a disease. This is a machine of great engineering and built to last 100 years, I'd like to find the person and slap them.  People lack any sense or integrity, the "just bin it and get a better one" attitude is extremely irksome.

Its worthless old equipment, how are you going to download the data to a PC? How much data can it store, if any?

OK, so what do you use an equipment like this as an EE?
I had to do temperature testing of parts recently. The goal was to record worst case temperatures of components on a PCB. Ambient temperature had to be increased in a thermal chamber, and the temps had to be stable for an hour to take a reading. And had to be repeated on 10 samples. Those are the rules, it is for safety, so you dont blow up an oil refinery (worst case) or a harbor.

If I do the measurement sequentially, one by one, the test would take me about a week to complete, I dont think I could've fit the testing of a third card into a day. Instead, I bought a scanner card to our DMM6500 and used that. That had 10 TC inputs, so I could finish the test in 2 days, by testing 5 PCBA at the same card. Honestly, I should've bought a logger like this, that would've made my job even simpler.

Here is a 32 channel device for $1,500: https://www.hgsind.com/product/pmd-mxt-8-16-24-or-32-channel-digital-temperature-scanner-universal-thermocouple looks kinda cheap.
There is also the DAQ6510, up to 80 channels. I'd rather have the DMM or DAQ as you can use it for other purposes.

[Vgkid]

On the Aromat relays there is the Matushita "M" so they are Panasonic.

[Burt]

The roller assembly at 15:14 is pretty clearly the door latch.  Mates with arrow shaped plate mounted on door.

[Tom45]

That instrument is probably quite common for chemical engineers.

[tocsa120ls]

Pioneer A-339 has them too.

[SilverSolder]

Can these things still be sourced?  Or has it gone the way of the dodo...

[tocsa120ls]

Can these things still be sourced?  Or has it gone the way of the dodo...

I think when monolithic bilateral switches got cheap enough (CD4066) these pretty much went out of fashion.

[SilverSolder]

Can these things still be sourced?  Or has it gone the way of the dodo...

I think when monolithic bilateral switches got cheap enough (CD4066) these pretty much went out of fashion.

They would be cool for a mains switch...

[free_electron]

those switch connector cables were made by alps. i've seen them in hi end audio stuff as well as test equipment ( advantest , yokogawa , mainly japanese stuff )

[VK3DRB]

A nice product, and a beautiful power switch but the layout guy let the side down with his sloppy component designator work. PCB layout is artwork and should be treated as such; created by craftsmen, not tradesman. Nothing worse than inheriting sloppy work, or trying to debug a board with poorly laid out designators (as Dave found!). It says a lot about the layout person. No excuses, it does not take much extra time to do a good job with designators.

There are a number of systems around, but I believe I have an excellent one, especially for high density circuitry which I have used for years. You might find some handy hints there:

Designator Guidelines

1. Font stroke width to be 1/5 of the character height. This is optimum for readibility. Don't use serif fonts and avoid True Type fonts.
2. Check the PCB manufacturer has the capability to print your character sizes. Typically, 0.8mm height is good for high density boards and 0.6mm is the bottom limit. Try to keep designator character heights the same.
3. To help eliminate ambiguity, orient the designator any way, including sideways and upside down if needed. Rotate the designators so the left most bottom of the first character is nearest the actual component.
4. Avoid having multiple designator strings in the same orientation so close they concatenate. If necessary, offset them slightly; else align them for aesthetic reasons providing they don't concatenate.
5. Strive for 100% of components having designators.
6. Consider back annotating until the first round of documentation is created, such as just prior to a BOM or detailed design document that mentions parts by their designator is released. Then, no more reannotating or back annotating! The initial back annotating helps find components quickly on a busy board.
7. If you are striving for 100% testpoint coverage for ICT on a high density board, avoid using for example, "TP23". Just use "23". The "TP" is redundant. Just using a number for a test point is not ambiguous and I have used this on many designs without any issues. This violated the IEEE standard of using "TP" but it is often much better for high density boards. In fact, it can be the difference between impossibility and possibility.
8. Use IEEE standard reference designators. Don't use multiple types for the same type of part. https://en.wikipedia.org/wiki/Reference_designator
9. Label key test points with their abbreviated function. eg: GND.
10. Consider using callouts. A callouts is a block of designators arranged in an orientation and spacing pattern of a highly dense R, L and C components nearby. Typically a box is placed around the callout and an arrow is used to point to the cluster of components the callout is referring to.
11. Avoid putting designators over vias, even if they are tented. If they are over a via, make sure the round part of a character has the via in its centre. eg: R10. where the via is in the centre of the '0'.
12. Don't have designators hiding under component bodies or over exposed non-masked areas (ENIG, HASL etc), and don't trust the manufacturer to remove these for you.

[free_electron]

1. Font stroke width to be 1/5 of the character height. This is optimum for readibility. Don't use serif fonts and avoid True Type fonts.

with a width no smaller than 0.15mm. anything less will peel or become problematic in printing even when jetted.

3. To help eliminate ambiguity, orient the designator any way, including sideways and upside down if needed. Rotate the designators so the left most bottom of the first character is nearest the actual component.

Only 0 and 90 degrees counterclockwise are allowed. no upside downs or 180 degree rotations
There is nothing worse than to have to spin the board around to find bloody designators.

5. Strive for 100% of components having designators.

Technology advances ,and while this was true 20 years ago , it is simply no longer possible. Parts are so small and density is so high there is no room on the board to fit designators. I can fit 4 0201 resistors in the space occupied by a small designator.
The same goes for silkscreen of any kind. Silkscreen should be used to identify the board , denote big terminals and key components. Anything resistor/capacitor : no silkscreen, no designator. Silkscreen is problematic in multiple ways
- alignment is less precise than soldermask. There is a risk the silkscreen offset covers pads or flows into soldermask openings.
- silkscreen eats into placement space
- silkscreen , when improperly designed or applied can 'lift' component corners. This is problematic on parts like DFN , QFN and flip-chip
- silkscreen becomes a trap for flux remnants ( for example: a circle under a capacitor can become a headache for cleaning the board. ) Same thing for 'boxes' around small parts.

Silkscreen should only be used for polarity indication ( a 0.2mm diagonal dot denotes the cathode or + on a capacitor ) on BGA / DFN / DFN three corners are marked : the corner where pin 1 is is NOT marked. The AOI systems can hunt for this patterns faster than the actual landpattern. once the location is found the AOI will then hunt for the actual copper to do the actual positioning as the silkscreen itself is still unreliable. The silkscreen acts as a 'coarse target' that is easy to find , fine positioning is done based on copper.

Silkscreen also can be a headache if boards need spray-on conformal coating. The silkscreen 'repels' the coating and can cause discontinuities.

6. Consider back annotating until the first round of documentation is created, such as just prior to a BOM or detailed design document that mentions parts by their designator is released. Then, no more reannotating or back annotating! The initial back annotating helps find components quickly on a busy board.

Annotation should be schematic driven. There is nothing worse than to end up with an opamp u7 surrounded by r209 r3 and c214. Clustered parts in the schematic need to have sequential numbering.

7. If you are striving for 100% testpoint coverage for ICT on a high density board, avoid using for example, "TP23". Just use "23". The "TP" is redundant. Just using a number for a test point is not ambiguous and I have used this on many designs without any issues. This violated the IEEE standard of using "TP" but it is often much better for high density boards. In fact, it can be the difference between impossibility and possibility.

ICT machines don't care about the testpoint being marked. Neither bed of nails nor flying probe has an optical system. It may be useful for bench testing by humans , but if it eats space : delete it.

[2N3055]

those switch connector cables were made by alps. i've seen them in hi end audio stuff as well as test equipment ( advantest , yokogawa , mainly japanese stuff )

Ditto.. Those were used much for audio switching, so you can switch near inputs in the back and have knobs on front.
Later relays were simply used...

[SilverSolder]

1. Font stroke width to be 1/5 of the character height. This is optimum for readibility. Don't use serif fonts and avoid True Type fonts.

with a width no smaller than 0.15mm. anything less will peel or become problematic in printing even when jetted.

3. To help eliminate ambiguity, orient the designator any way, including sideways and upside down if needed. Rotate the designators so the left most bottom of the first character is nearest the actual component.

Only 0 and 90 degrees counterclockwise are allowed. no upside downs or 180 degree rotations
There is nothing worse than to have to spin the board around to find bloody designators.

5. Strive for 100% of components having designators.

Technology advances ,and while this was true 20 years ago , it is simply no longer possible. Parts are so small and density is so high there is no room on the board to fit designators. I can fit 4 0201 resistors in the space occupied by a small designator.
The same goes for silkscreen of any kind. Silkscreen should be used to identify the board , denote big terminals and key components. Anything resistor/capacitor : no silkscreen, no designator. Silkscreen is problematic in multiple ways
- alignment is less precise than soldermask. There is a risk the silkscreen offset covers pads or flows into soldermask openings.
- silkscreen eats into placement space
- silkscreen , when improperly designed or applied can 'lift' component corners. This is problematic on parts like DFN , QFN and flip-chip
- silkscreen becomes a trap for flux remnants ( for example: a circle under a capacitor can become a headache for cleaning the board. ) Same thing for 'boxes' around small parts.

Silkscreen should only be used for polarity indication ( a 0.2mm diagonal dot denotes the cathode or + on a capacitor ) on BGA / DFN / DFN three corners are marked : the corner where pin 1 is is NOT marked. The AOI systems can hunt for this patterns faster than the actual landpattern. once the location is found the AOI will then hunt for the actual copper to do the actual positioning as the silkscreen itself is still unreliable. The silkscreen acts as a 'coarse target' that is easy to find , fine positioning is done based on copper.

Silkscreen also can be a headache if boards need spray-on conformal coating. The silkscreen 'repels' the coating and can cause discontinuities.

6. Consider back annotating until the first round of documentation is created, such as just prior to a BOM or detailed design document that mentions parts by their designator is released. Then, no more reannotating or back annotating! The initial back annotating helps find components quickly on a busy board.

Annotation should be schematic driven. There is nothing worse than to end up with an opamp u7 surrounded by r209 r3 and c214. Clustered parts in the schematic need to have sequential numbering.

7. If you are striving for 100% testpoint coverage for ICT on a high density board, avoid using for example, "TP23". Just use "23". The "TP" is redundant. Just using a number for a test point is not ambiguous and I have used this on many designs without any issues. This violated the IEEE standard of using "TP" but it is often much better for high density boards. In fact, it can be the difference between impossibility and possibility.

ICT machines don't care about the testpoint being marked. Neither bed of nails nor flying probe has an optical system. It may be useful for bench testing by humans , but if it eats space : delete it.

I agree with everything you said there, except the last point...   It is super helpful when servicing the product later in life that a board has test points (and ground, and supply voltages) that are clearly accessible and marked.  If done right, the basic checks should be do-able with a DMM without even looking at a schematic...

[free_electron]

  It is super helpful when servicing the product later in life that a board has test points (and ground, and supply voltages) that are clearly accessible and marked.  If done right, the basic checks should be do-able with a DMM without even looking at a schematic...

ok for certain key things like power rails etc. but you cannot label every ict testpoint. And , like it or not, many assemblies today are unserviceable due to construction and part usage.

[SilverSolder]

  It is super helpful when servicing the product later in life that a board has test points (and ground, and supply voltages) that are clearly accessible and marked.  If done right, the basic checks should be do-able with a DMM without even looking at a schematic...

ok for certain key things like power rails etc. but you cannot label every ict testpoint. And , like it or not, many assemblies today are unserviceable due to construction and part usage.

Agree.  Even the "good old" designs didn't label every IC testpoint,  instead each board might have 2 - 5 judiciously chosen points that have to give a defined, expected result.  Super simple, and super helpful when tracing a fault (even if you end up replacing the whole board, you want to be sure you are replacing the right one, for the right reasons!).