symmetry

tirsdag den 7. september 2021 kl. 21.14.54 UTC+2 skrev John Larkin:

same for components, make it obvious when they are not placed correct

When designing custom transformers, I will have one pin of the bobbin removed, so the footprint isn't symmetrical, and it's impossible to install the transformer backwards.

And are you sure mechanical engineers really love mounting hole symmetry?? Too bad they aren't here to defend themselves.

The asymmetry of a typical DIMM socket sometimes is just a diameter-of-pegs thing; I actually saw a motherboard with one DIMM socket installed backward once, it dropped power immediately if there was a module in the wrong socket.

Yeah, asymmetry is a good plan. Plans don't always work.

Yes for this electronics engineer.

I use 3 sets of symmetric plugs to wire up 8 sets of batteries in a 40 pins ZIF socket. I intensionally plug it in backward to balance the charges. Namely, B and ~B pair up different set. D and ~D discharge different set. C and ~C charge up same.

12V charging (completely full at 4V) and 16V discharging (completely empty at 3V). Using 10A at 20% duty with double pins deeply inserted 3M ZIF socket.

'~' plugs in upside down.'

___***_B_***_________***_C_***_________***_D_***

_1__0V_____40______1__0V___0V_40_____1__0V___0V_40 _2__4V__0V_39______2__4V___4V_39_____2__4V___4V_39 _3__8V__4V_38______3__8V___8V_38_____3__8V___8V_38 _4_12V__8V_37______4_12V__12V_37_____4_12V__12V_37

_6_12V_____35______6__0V___0V_35_____6_12V______35 _7_16V__8V_34______7__4V___4V_34_____7_16V__12V_34 _8_____12V_33______8__8V___8V_33_____8______16V_33 _9_____16V_32______9_12V__12V_32_____9__________32

12_____16V_30_____12_12V__12V_30____12__________30 13_____12V_29_____13__8V___8V_29____13______16V_29 14_16V__8V_28_____14__4V___4V_28____14_16V__12V_28 15_12V_____27_____15__0V___0V_27____15_12V______27 17_12V__8V_24_____17_12V__12V_24____17_12V__12V_24 18__8V__4V_23_____18__8V___8V_23____18__8V___8V_23 19__4V__0V_22_____19__4V___4V_22____19__4V___4V_22 20__0V_____21_____20__0V___0V_21____20__0V___0V_21

It's true. In a radar job many moons ago, the array elements were exactly at the center of curvature of the radome, ensuring that focused reflection from the inside of the radome would blow nearby array elements out. Messed calibration up as well. Oops. Radome was moved far closer to the array face.

I had a lot of fun teasing the mechanical lead about his artistic engineers.

But the longer-term solution is to add a requirement that it be impossible to mis-assemble the unit, so long as hydraulic presses are excluded. It's useful if it's obvious to the eye that it can go together only one way, to reduce futzing.

Joe Gwinn

Yes. Everywhere.

I got my machined EOM oven and I'm doing thermal time constant tests. That gives me lots of time to eat, post here, check the news, generally fool around.

Looks like ballpark 1 hour cooldown tau in free air. This will be an interesting control loop.

Lens elements that are nearly-but-not-quite symmetrical are a famous trap for beginning optical designers.

Cheers

Phil Hobbs

Didn't we mess up Hubble like that?

Once such a lens is ground, how do you tell which side is which?

IIRC it was a misplaced spacer, but the gist is the same.

Depends how close. I have a spherometer in my drawer, which is a dial gauge with a small anvil on either side so that it measures the surface curvature. (The dial is calibrated in dioptres (normalized to n=1.5).

So you can do it, given sufficient care, but it's almost always possible to make an equivalently-good design using a combination of really-symmetrical and obviously-asymmetrical elements.

Cheers

Phil Hobbs

>

if this is correct there wasn't supposed to be a spacer

"When the measuring rod was installed in the null corrector, the bracket to secure the lens in place came up short, indicating something wasn’t right. But under pressure to get the job over with, technicians brought the bracket and lens together with three flat washers, the same kind sold in hardware stores for a quarter. The lens setting was 1.3 millimeters askew, 10 times beyond specified boundaries. "

I have a buddy who toasted a (pre SATA) disk drive by plugging in the power cable backwards. I exclaimed, "I didn't think it was physically possible to

*do* that?!" His reply? "Yeah, it was really difficult going in..." <exclaim> And you didn't think, perhaps, to double check what you were doing??

I've often had to put connectors for multiple, competing actuators in a design. I always arrange for the connector shells to be wired differently (so actuator A doesn't work if plugged as actuator B). Where possible, cable lengths and dressing tries to reinforce the "correct" wiring.

But, regardless, shit happens after the sale. E.g., electrician looks at connector A while he's trying to install new actuator on connector B and, thus, exactly mimics it's wiring/keying pattern. Much easier to consult a sample-in-hand (even if it is WRONG) than to dig through the service documents!

So, write the software to act as a third-party to *watch* how the mechanism operates: "I'm telling it to move right and it's moving left... Or *down*! Hmm, something isn't right, here!"

USB B, the big printer connector, is impossible to put in wrong. And USB C doesn't care. All the others are awful.

Choke your chicken?

John Larkin wrote: ================

** In the world of live sound systems, standard practice is to use identical pairs of sockets ( XLR or Speakon ) on the back of speaker cabinets. This allows chaining of such boxes via short cables with identical plugs at each end.

It also allows the connection of *both* channels of a high power stereo amp to feed the same speaker box.

Whenever I see such amps with both power channels badly damaged at the same time, I know what has happened. Major Murphy has struck again.

..... Phil

Making ICs symmetric, 2 or 4-way symmetric, was a dumb idea. Tubes were better than that.

I hear that some (perhaps non-compliant?) USB C cables work better one way around than the other, so then you still need to know, but there is no mechanical feature to help you. I have yet to acquire much that uses USB-C so I'm putting off learning about it. People who have had to read the USB specs tell me they are really awful, like dealing with them is to be avoided at all costs.

Sometimes it could save time, e.g. if the factory will notice that it could be assembled two different ways, and not assemble your product until they get clarification. If you really can't afford a few extra days slip in your schedule, it may help to tell them "put it this way round" so that they have no reason to ask. I'm guessing it might also help with automated inspection.

Similarly on metla layers of chips, text fonts and chip art that violates DRC rules can waste a lot of time getting waivers regardless of whether it could cause any actual manufacturing problem.

There are also less tangible reasons to key a connector. E.g., you may want specific conductors to be on one dies of a *bulkhead* connector to make it easier to troubleshoot, in the field.

For example, a switch typically doesn't "need" a polarized connector as it will function identically regardless of how wired.

But, knowing *which* signal is present on the "top" wire in the connector can save time when you're trying to troubleshoot the functionality associated with the switch.

I was taught a long time ago:

If something can be made symmetric (or fit more than one way) make it so. If it can't, make it obvious it isn't.

It reminds me of an optical manufacturer who couldn't get the right performance of a lens system that should have been close to diffraction limit.

After 3 weeks fruitless testing, following disassembly of components it was found one component had been fitted the wrong way round. One of the surfaces was an aspheric, but close enough to the other surface to cause confusion.