Factor PFC Into Your Power-Supply Design...

[server]

John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote in
news:5n9skhtri8m5o6fdmd3gupe19s1t6q8lj7@4ax.com:

The problem with the power supply biz is the lack of interface
standards. Even similar models from the same manufacturer have
different and bizarre interfaces and connectors. There are horror
stories.

Most are custom and follow the direction of the company who
contracted us for the product.

Most HV designs are very custom as all HV load situations are very
individual.

We did one for Cymer Laser that bolted to the side of the machine
and the HV output lead was a spring loaded flap that made the
connection once installed. It was right in the middle of the potted
cavity in the device and it was only about 5 x 3 x 2.5 inches. 6kV
IIRC.

We had one where we created the connection and connector and there
was an HV insulation paste filled tube that a 6\" long HV lead \"poked
through\" to get to the HV output node at the bottom of the tube.
Were it only air filled the six inches between the output node and
the ground connection at the back panel would have arced.
That one was for LANL. Not able to be mass produced and sold
elsewhere. Very application specific. That one was 50kV.

 

[John Larkin]

On Tue, 18 Oct 2022 14:16:07 -0000 (UTC),
DecadentLinuxUserNumeroUno@decadence.org wrote:

John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote in
news:vf9skhph0rqm0g05ehgd9f9u5nrrab1f70@4ax.com:


Rack panels are standard and look strange if they are thicker than
their neighbors. The heavy stuff needs trays or slides of course.

You have obviously never seen a rack populated with devices from
multiple vendors.

Probably not more than 10,000 times. I don\'t think I\'ve ever seen a
panel that wasn\'t 1/8 thick. It would look weird.

The thickness of the front panel is (and should
be) dependent on what the weight of the gear is it supports.

Do NOT cantelever heavy stuff off the front panel!




Because

ALL of the weight must be borne solely by the front panel as if there
are no other devices below it to add support. And without slides.
The device itself is the \"tray\" for everything it carries. Panel
thicknesses vary, and especially those meant for vibratory
environments such as military plane flights.

The \"standard\" is for attachment bolt location and spread, not face
plate thickness.

Your front panel is too thin for the weight of the gear you include
in your application. You cannot count on devices being below it to
support it. For a ground based application you tray might work, but
if it is meant to go aboard an aircraft I would be amazed if they do
not mention the fact that it looks flimsy.

No complaints so far. My customers know how to mount heavy stuff.

Also IF it is going aboard a plane, you have to reconsider your
cooling considerations for the thinner air of a 70,000 ft flight
altitude, which was the standard we had to follow.

Our gear isn\'t designed to fly. It could fly inside an instrumented
plane but that would be pressurized for working engineers, 7000 feet
maybe. Jet engines are often tested on a flying test bed.

No complaints so far.

And at one time, we made a ruggedized case for a 25\" CRT meant for
rack mounting. After the Tempest level shielding it required, we had
to boost the airflow to get the same cooling level.

You are strange looking and obviously thicker skulled than your
neighbors.

Here is a 2U design I made years ago on Solid Edge (screw solid
works, Siemens product is far better) (they do PCB layout now as
well)(miles ahead of Solid Twerks)

https://www.mediafire.com/file/3utpm7kmg4ofd6k/2U_Rugged_Rack_Tray.pd
f/file

It is a 3D PDf so you may need actual Adobe products to view it.

Can\'t see it with Foxit.

 

[John Larkin]

On Tue, 18 Oct 2022 02:56:58 -0700 (PDT), John Walliker
<jrwalliker@gmail.com> wrote:

On Monday, 17 October 2022 at 03:19:01 UTC+1, DecadentLinux...@decadence.org wrote:
John Larkin <jla...@highlandSNIPMEtechnology.com> wrote in
news:6t9okhhrlggajurb1...@4ax.com:
On Sun, 16 Oct 2022 09:50:23 GMT, Jan Panteltje
pNaonSt...@yahoo.com> wrote:

Factor PFC Into Your Power-Supply Design
pdf for download, some new chips:ADP1047 and ADP1048:
https://www.electronicdesign.com/power-management/power-supply/whi
tepaper/21233724/electronic-design-factor-pfc-into-your-powersuppl
y-design-pdf-download

It\'s impossible to make your own PFC power supplies at the cost of
buying them.
That depends on how many are to be manufactured and sold by \"you\".

Incorporating a supply into a design is one thing and in a COTS
realm choices can be made... however...

Being a power supply maker, part of our realm was incorporating PFC
where it was needed or required. Being primarily a DC-to-DC HV
supply maker, the issue did not arise much.
We buy a kilowatt Mean Well 48 volt supply for $156. It has emi
filters, a nice enclosure, an internal fan, and all sorts of
compliance stickers.
And... infant mortality... It\'s OK... They meant well, from their
accountant\'s Point Of View.

I don\'t know why you are so anti Meanwell. A product I have been involved with
which uses a Meanwell encapsulated power supply module has shipped around
1500 units over a 5 year period and there have been zero reported failures.
The Meanwells were chosen after evaluating a wide variety of different brands,
including some well known US ones. The Meanwells had the lowest conducted
emissions of any of the candidates (based on my own direct comparisons rather
than the data sheets) and there was a detailed test report available
which disclosed a lot of detail about which internal components were most
highly stressed and what temperatures they would reach at maximum rated ambient
temperature. They were also relatively cheap.

John

They also make small PCB-mount stuff.

 

[server]

John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote in
news:smdtkhhofnb1ckdj29jofegoo2f6mce7k2@4ax.com:

On Tue, 18 Oct 2022 14:16:07 -0000 (UTC),
DecadentLinuxUserNumeroUno@decadence.org wrote:

John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote in
news:vf9skhph0rqm0g05ehgd9f9u5nrrab1f70@4ax.com:


Rack panels are standard and look strange if they are thicker
than their neighbors. The heavy stuff needs trays or slides of
course.

You have obviously never seen a rack populated with devices from
multiple vendors.

Probably not more than 10,000 times.

All cheap chassis shit like the one you are going to use?

I don\'t think I\'ve ever seen
a panel that wasn\'t 1/8 thick. It would look weird.

I think that until this thread, you were oblivious to it. Until I
was critical of your choice. Especially ina a 4U design.

The thickness of the front panel is (and should
be) dependent on what the weight of the gear is it supports.

Do NOT cantelever heavy stuff off the front panel!

Bullshit. The spec is that the device MUST be able to be supported
by said front panel and by that alone. The designer cannot know if
there will be another device or stack of devices below it.

Because
ALL of the weight must be borne solely by the front panel as if
there are no other devices below it to add support. And without
slides. The device itself is the \"tray\" for everything it
carries. Panel thicknesses vary, and especially those meant for
vibratory environments such as military plane flights.

The \"standard\" is for attachment bolt location and spread, not
face
plate thickness.

Your front panel is too thin for the weight of the gear you
include
in your application. You cannot count on devices being below it
to support it. For a ground based application you tray might
work, but if it is meant to go aboard an aircraft I would be
amazed if they do not mention the fact that it looks flimsy.

No complaints so far. My customers know how to mount heavy stuff.

However, you obviously do not.

Also IF it is going aboard a plane, you have to reconsider your
cooling considerations for the thinner air of a 70,000 ft flight
altitude, which was the standard we had to follow.

Our gear isn\'t designed to fly. It could fly inside an
instrumented plane but that would be pressurized for working
engineers, 7000 feet maybe. Jet engines are often tested on a
flying test bed.

No complaints so far.



And at one time, we made a ruggedized case for a 25\" CRT meant
for
rack mounting. After the Tempest level shielding it required, we
had to boost the airflow to get the same cooling level.

You are strange looking and obviously thicker skulled than your
neighbors.

Here is a 2U design I made years ago on Solid Edge (screw solid
works, Siemens product is far better) (they do PCB layout now as
well)(miles ahead of Solid Twerks)

https://www.mediafire.com/file/3utpm7kmg4ofd6k/2U_Rugged_Rack_Tray.
pd f/file

It is a 3D PDf so you may need actual Adobe products to view it.



Can\'t see it with Foxit.

 

[John Larkin]

On Tue, 18 Oct 2022 14:55:00 -0000 (UTC),
DecadentLinuxUserNumeroUno@decadence.org wrote:

John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote in
news:smdtkhhofnb1ckdj29jofegoo2f6mce7k2@4ax.com:

On Tue, 18 Oct 2022 14:16:07 -0000 (UTC),
DecadentLinuxUserNumeroUno@decadence.org wrote:

John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote in
news:vf9skhph0rqm0g05ehgd9f9u5nrrab1f70@4ax.com:


Rack panels are standard and look strange if they are thicker
than their neighbors. The heavy stuff needs trays or slides of
course.

You have obviously never seen a rack populated with devices from
multiple vendors.

Probably not more than 10,000 times.

All cheap chassis shit like the one you are going to use?

I don\'t think I\'ve ever seen
a panel that wasn\'t 1/8 thick. It would look weird.

I think that until this thread, you were oblivious to it. Until I
was critical of your choice. Especially ina a 4U design.

The thickness of the front panel is (and should
be) dependent on what the weight of the gear is it supports.

Do NOT cantelever heavy stuff off the front panel!

Bullshit. The spec is that the device MUST be able to be supported
by said front panel and by that alone. The designer cannot know if
there will be another device or stack of devices below it.

What spec? Who enforces it?

If a heavy unit is cantelevered off the front panel, it will be a
hazard to install and remove. As you install it, it will scrape on the
unit below and scratch it up. When someone later loosens the front
panel screws, all its weight will flop onto the unit below, which
already had its own weight to support. The whole stack of units could
collapse.

Users can\'t know, years later, if this will happen or if the unit
below can bear the weight of both.

Don\'t do it. Nobody competant does.

 

[John Larkin]

On Tue, 18 Oct 2022 14:27:26 -0000 (UTC),
DecadentLinuxUserNumeroUno@decadence.org wrote:

John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote in
news:5n9skhtri8m5o6fdmd3gupe19s1t6q8lj7@4ax.com:

The problem with the power supply biz is the lack of interface
standards. Even similar models from the same manufacturer have
different and bizarre interfaces and connectors. There are horror
stories.

Most are custom and follow the direction of the company who
contracted us for the product.

Most HV designs are very custom as all HV load situations are very
individual.

We did one for Cymer Laser that bolted to the side of the machine
and the HV output lead was a spring loaded flap that made the
connection once installed. It was right in the middle of the potted
cavity in the device and it was only about 5 x 3 x 2.5 inches. 6kV
IIRC.

We built the timing controllers for their MOPA DUV lasers and then for
the tin droplet EUV stuff.

The MOPAs are classic capacitive discharge gadgets. Maybe you did the
HV for them. It\'s a lot of avearge power.

The EUV tin droplets are hit by an enormous laser, but that\'s RF
pumped.

We had one where we created the connection and connector and there
was an HV insulation paste filled tube that a 6\" long HV lead \"poked
through\" to get to the HV output node at the bottom of the tube.
Were it only air filled the six inches between the output node and
the ground connection at the back panel would have arced.
That one was for LANL. Not able to be mass produced and sold
elsewhere. Very application specific. That one was 50kV.

The flashtubes at NIF are about 6 feet long. That must take a bunch of
voltage.

 

[server]

John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote in
news:i5gtkhhjtkr7lhajfo4mg16ag294ped50q@4ax.com:

On Tue, 18 Oct 2022 14:55:00 -0000 (UTC),
DecadentLinuxUserNumeroUno@decadence.org wrote:

John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote in
news:smdtkhhofnb1ckdj29jofegoo2f6mce7k2@4ax.com:

On Tue, 18 Oct 2022 14:16:07 -0000 (UTC),
DecadentLinuxUserNumeroUno@decadence.org wrote:

John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote in
news:vf9skhph0rqm0g05ehgd9f9u5nrrab1f70@4ax.com:


Rack panels are standard and look strange if they are thicker
than their neighbors. The heavy stuff needs trays or slides of
course.

You have obviously never seen a rack populated with devices
from
multiple vendors.

Probably not more than 10,000 times.

All cheap chassis shit like the one you are going to use?

I don\'t think I\'ve ever seen
a panel that wasn\'t 1/8 thick. It would look weird.

I think that until this thread, you were oblivious to it. Until
I
was critical of your choice. Especially ina a 4U design.

The thickness of the front panel is (and should
be) dependent on what the weight of the gear is it supports.

Do NOT cantelever heavy stuff off the front panel!

Bullshit. The spec is that the device MUST be able to be
supported
by said front panel and by that alone. The designer cannot know
if there will be another device or stack of devices below it.

What spec? Who enforces it?

Use some common sense. AND ask the dopes you are buying that POS
thin faced turd from. No matter what you think, the front panel
carries all the weight of the unit, just as the specs state, and the
ONLY exception is when \"drawer slides\" have been mounted.

> If a heavy unit is cantelevered off the front panel,

For one thing, that is not even the correct term. Any rack device
will be stable if only held in place by the bottom two screws,
because the entire faceplate is held fast against the rack by the
weight of the device itself hanging behind the faceplate.

it will be a
hazard to install and remove.

You have obviously not spent much time populating racks, much less
examining rack equipment, and obviously particularly not for a
military contractor.

As you install it, it will scrape on
the unit below and scratch it up.

More bullshit. The spec also includes NOT having ANY protrusions
on the top and bottom face of your product. Like screw heads on the
bottom OR the top of your unit. So \"sliding one out\" after the front
panel has been detached ALWAYS has its weight on the unit below it.
If there is no unit there, one must hold the weight as the screws are
removed to keep it horizontally oriented. They are specifically made
to be able to be slid into place on top of an existing unit, and ALL
properly designed rack mount devices have ZERO protrusions on the top
or the bottom. One must use countersunk, flush or below flush
fasteners in one\'s design. There are also not supposed to be any
venting on the top or bottom of a unit.

You said you \"saw\" thousands of units. I INSTALLED (and
uninstallled) thousands of units over several years time and those
racks are in place in over 50 cities around the world. They were ALL
100% fully disassembled after the build and the cabling and rack
units and the racks were shiped to the site and then fully reloaded
and rewired up. Every last one... literally thousands of units
Some 3U and up units do get slides installed. Some do not. Again,
the specification is that the unit must be able to reside in the rack
only held by the front panel with no rear attachment (slide mount
points). Granted many if not most folks incorporate rear attached
slides that support the weight as well, but that is a prefernce, not
a requisite. The spec remains. The only exception are server
chassis, which are typically many more U than a 3 or 4. They ALWAYS
get slides installed.

A good example is rack mounted UPS units. They have very thick
front faces. Because they follow the spec. Not every one gets
installed in the bottom of the rack or gets slides added, and those
that do... oh looky... nothing below to support them. There are
many other heavy devices with thick front panels. Why? Because that
is the spec.

When someone later loosens the
front panel screws, all its weight will flop onto the unit below,

No shit. And no, it does not \"flop\", like your argument here has.
You are sporting 100% flawed logic. It sits on what is below it and
if there is no unit below it, it should be sitting on the tech\'s hand
as the front panel fasteners get removed. And it has been that way
since the onset of the use of this type chassis which we all know of
as racks.

> which already had its own weight to support.

Racks typically, but not always get populated from the bottom up.
Been that way forever. The stack of still installed units below the
one being serviced are supposed to be able to support such weight.
Been that way forever. The bottom unit has no support below it.
Some get slides and some do not.

The whole stack of
units could collapse.

More flawed, non-factual horseshit.

Users can\'t know, years later, if this will happen or if the unit
below can bear the weight of both.

A tech pulling a unit from a rack will normally have access to the
rear of the rack to remove cabilng AND to examine the circumstance
under which they are performing the servicing. More flawed, non-
factual logic on your part... for years running.

> Don\'t do it. Nobody competant does.

All the folks in industry that use racks who know how to spell the
word competent do.

 

[John Larkin]

On Tue, 18 Oct 2022 19:08:44 -0000 (UTC),
DecadentLinuxUserNumeroUno@decadence.org wrote:

When someone later loosens the
front panel screws, all its weight will flop onto the unit below,

No shit. And no, it does not \"flop\", like your argument here has.
You are sporting 100% flawed logic. It sits on what is below it and
if there is no unit below it, it should be sitting on the tech\'s hand
as the front panel fasteners get removed. And it has been that way
since the onset of the use of this type chassis which we all know of
as racks.

So one tech uses his hand, from the back of the rack, to support the
box while another tech pushes the heavy box into the rack and fastens
the front panel screws? Hire really strong techs. Or one with really
long arms.

Try that with this gradient driver amp:

https://www.dropbox.com/s/qbpws16c29asaxc/PP5.JPG?raw=1

 

[server]

John Larkin <jlarkin@highland_atwork_technology.com> wrote in
news:if2ukh1fdd78e23c5sc7aj80gnqqg93qg9@4ax.com:

On Tue, 18 Oct 2022 19:08:44 -0000 (UTC),
DecadentLinuxUserNumeroUno@decadence.org wrote:


When someone later loosens the
front panel screws, all its weight will flop onto the unit
below,

No shit. And no, it does not \"flop\", like your argument here
has.
You are sporting 100% flawed logic. It sits on what is below it
and if there is no unit below it, it should be sitting on the
tech\'s hand as the front panel fasteners get removed. And it has
been that way since the onset of the use of this type chassis
which we all know of as racks.


So one tech uses his hand, from the back of the rack, to support
the box while another tech pushes the heavy box into the rack and
fastens the front panel screws? Hire really strong techs. Or one
with really long arms.

Try that with this gradient driver amp:

https://www.dropbox.com/s/qbpws16c29asaxc/PP5.JPG?raw=1

Before I built all those satellite baseband gateways for all those
US and Aussie cities, I built the simulator for the F-35. Several
racks full of GHz level RF modules... And then another... and then
a third. They were several racks of 9 units. 8 4U 16 channel units
with one 5 or 6U center modules, which were the combiners for those
with a back panel loaded with 64 SMA ports that resolved down to 16.
I built those modules. But none of them had a behemoth like that in
them. They all were Chem etched Aluminum though like that chassis. I
hand built every single module and *then* installed them all into the
racks.

So after months of building units and tying them all together using
same length cables, etc. They all went into a classified lab that
after I placed and wired together all the racks, including a monster
inch diameter welding cable that grounded all of them together that
would be placed in the field in the drop floor when it got to the
Hughes Anechoic facility in Texas. After I got them all wired
together, they program all of the racks for timing so that all 1024
channels were gathering signals that all then got the same timestamp,
and calibrated the offsets down to microseconds (or even faster).
Once we got all that done, I was no longer allowed in that lab, wired
it with white noise generators over the door to the lab and was
literally no longer allowed in the lab because I did not have the
clearance for the final programming segment. A lab you were not even
allowed to take a piece of paper into or any kind of writing
instrument. It became a secret level classified lab. There was a
small wooden box on the wall outside the lab for cell phones. The
door had an electronic lock on it that looked like the dial on a bank
vault. And a sign in / out sheet.

I literally built every module from flat panels and side panels
with angled edges for rigidity. They were pretty thick Aluminum
chassis. Each of the 4U modules had 16 variable phase, variable gain
modules from General Microwave at first. Those were how they all got
calibrated together for matched arrival times on signals. They had
several tens of units (at like $800 each) that would not perform the
370 odd degree span of the electroninc phase control.
Apparently the vendor\'s in house calibrator looked at the two ends
of the spectrum and called them good to go, but in actual use they
puked on wheeling the phase around the full 360 degrees much less
370. So they ALL got rejected and a new vendor used. The real PITA
was having to take the rack modules all apart to replace the 16 units
in each one with the new vendor\'s modules. They worked nice and were
a lot lighter weight than the hobbed out chassis of the originals,
but I had them alter the chassis design to make that rear of the rack
bank of mixers and phase shift modules \"serviceable\". Because that
replacement chore was a real pain. They agreed. So now, they could
be serviced without having to disassemble practically the entire
module. Two central trays instead of just one. I have probably seen
and used more semi-rigid SMA cables than you have, so many that we
got really good prices from the makers and then there were a huge
number of sheathed SMA cables for all the back of the rack links to
that the center splitter / combiner in each rack had. But inside
there were a hundred small diameter \"solder wetted\" of the unsheathed
semi-rigid variety. Various same length links. My shtuff looked
like that Quantum computer they make that gets lowered down into a
liquid Nitrogen tank. And the splitter combiner was a mad house of
links inside. 16 sets of 8 way splitter combiners inside to make the
giant 64 channel center modules. So they were timed together very
good, and the phase shift / attenuator modules had less of a chore to
do to get them all synched up. Making any detected event carry
proper microsecond timed stamps on them for the events that were
simulated in the anechoic chamber in order to calibrate the sensors
on the F-35 they had hanging there in the chamber with all of my
racks full of channels. Fake a missile launch radio event in the
chamber and get all the sensors in the bird calibrated out correctly.
I wish I had been able to go down to Hughes to seee it in place, but
did not have the clearance yet.

But none of the rack devices I built that were 6+U like that had a
hunk of laminated mass in them like that though! The 4 inch wide
slides, right? Dang, dude.

They all used linear supplies because swtchers introduce glitches
ticks and noise into the high frequency events. So bigger than a
switching supply but still small potatoes compared to that monster.
Your gonna need a bigger overhead lift. That looks dangerous for
that amount of weight. How many did you have to make of those? And
you WANT to be a power supply maker... so funny. You move from
small Hi Freq devices to big PS stuff. I question your logic again.
Hehehe...

The biggest we produced when I was at the power supply company
weight wise on this side of the pond was one that had two 2kVA
transformers sitting in the bottom of it, and it looked like a big
server on casters. Like a Weeble, one could not knock it over. All
the weight was down there at the bottom.

But back at the telecom place (F-35 Stimulator) every unit used
slides and they were ALL suspended units that did not touch the unit
above or below, even though the racks were all fully populated from
top to bottom. Any unit could be pulled out and serviced at the
bench easily. They were fairly light.

But when I did the satellite baseband gateway series of racks some
were packed in like sardines, and all were packed with millions worth
of our gear and network switching gear from Cisco and Juniper.
Quadruple redundant. High speed, properly terminated Ethernet links,
and high speed optical links all throughout. All tagged by a whole
crew of girls that came in after I built out the racks. Then ran
power to them and then the software engineers would come in and
install many many sessions of boot / and run VDMs on all the items
that had operating systems or firmware that operated them...
quadruple redundant... oh, did I say that already? Then, the girls
break all the wiring down and put it in a giant box on a pallet.
Each end tagged of course. And I would the place every item back in
the boxes they came in and the only thing that remained in the racks
were the power strips and the big server card chassis. And any blank
slot covers.

Once on site, there were two 8 meter dishes being constructed to
point at the bird in the sky with. And they reset all of the racks,
without their soft shipping pads, and add a couple Ka band RF racks,
which we did not need in the network side series of racks I built,
and put all the cables back in and fired them up for a series of
startup testing... and voila! Satellite baseband gateway. 21 US
cities and 17 cities in Australia, and a couple colleges in the
middle east. Theirs was the most scrutinized, believe that.
One rack was filled top to bottom with high speed hard drive arrays.
None of ours or Australia\'s had that.

 

[Joe Gwinn]

On Tue, 18 Oct 2022 08:15:28 -0700, John Larkin
<jlarkin@highlandSNIPMEtechnology.com> wrote:

On Tue, 18 Oct 2022 14:55:00 -0000 (UTC),
DecadentLinuxUserNumeroUno@decadence.org wrote:

John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote in
news:smdtkhhofnb1ckdj29jofegoo2f6mce7k2@4ax.com:

On Tue, 18 Oct 2022 14:16:07 -0000 (UTC),
DecadentLinuxUserNumeroUno@decadence.org wrote:

John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote in
news:vf9skhph0rqm0g05ehgd9f9u5nrrab1f70@4ax.com:


Rack panels are standard and look strange if they are thicker
than their neighbors. The heavy stuff needs trays or slides of
course.

You have obviously never seen a rack populated with devices from
multiple vendors.

Probably not more than 10,000 times.

All cheap chassis shit like the one you are going to use?

I don\'t think I\'ve ever seen
a panel that wasn\'t 1/8 thick. It would look weird.

I think that until this thread, you were oblivious to it. Until I
was critical of your choice. Especially ina a 4U design.

The thickness of the front panel is (and should
be) dependent on what the weight of the gear is it supports.

Do NOT cantelever heavy stuff off the front panel!

Bullshit. The spec is that the device MUST be able to be supported
by said front panel and by that alone. The designer cannot know if
there will be another device or stack of devices below it.

What spec? Who enforces it?

If a heavy unit is cantelevered off the front panel, it will be a
hazard to install and remove. As you install it, it will scrape on the
unit below and scratch it up. When someone later loosens the front
panel screws, all its weight will flop onto the unit below, which
already had its own weight to support. The whole stack of units could
collapse.

Users can\'t know, years later, if this will happen or if the unit
below can bear the weight of both.

Don\'t do it. Nobody competant does.

There are customers that do care, and test for it by full-scale
vibration tests on whole cabinets to MIL-S-901. A front-cantilevered
heavy components would break free and fly across the room.

Small equipment rates a 500-pound hammer, and medium rates 2000
pounds.

..<https://en.wikipedia.org/wiki/MIL-S-901>

The barge test is a sight to behold.

Joe Gwinn

 

[John Larkin]

On Thu, 20 Oct 2022 14:18:57 -0400, Joe Gwinn <joegwinn@comcast.net>
wrote:

On Tue, 18 Oct 2022 08:15:28 -0700, John Larkin
jlarkin@highlandSNIPMEtechnology.com> wrote:

On Tue, 18 Oct 2022 14:55:00 -0000 (UTC),
DecadentLinuxUserNumeroUno@decadence.org wrote:

John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote in
news:smdtkhhofnb1ckdj29jofegoo2f6mce7k2@4ax.com:

On Tue, 18 Oct 2022 14:16:07 -0000 (UTC),
DecadentLinuxUserNumeroUno@decadence.org wrote:

John Larkin <jlarkin@highlandSNIPMEtechnology.com> wrote in
news:vf9skhph0rqm0g05ehgd9f9u5nrrab1f70@4ax.com:


Rack panels are standard and look strange if they are thicker
than their neighbors. The heavy stuff needs trays or slides of
course.

You have obviously never seen a rack populated with devices from
multiple vendors.

Probably not more than 10,000 times.

All cheap chassis shit like the one you are going to use?

I don\'t think I\'ve ever seen
a panel that wasn\'t 1/8 thick. It would look weird.

I think that until this thread, you were oblivious to it. Until I
was critical of your choice. Especially ina a 4U design.

The thickness of the front panel is (and should
be) dependent on what the weight of the gear is it supports.

Do NOT cantelever heavy stuff off the front panel!

Bullshit. The spec is that the device MUST be able to be supported
by said front panel and by that alone. The designer cannot know if
there will be another device or stack of devices below it.

What spec? Who enforces it?

If a heavy unit is cantelevered off the front panel, it will be a
hazard to install and remove. As you install it, it will scrape on the
unit below and scratch it up. When someone later loosens the front
panel screws, all its weight will flop onto the unit below, which
already had its own weight to support. The whole stack of units could
collapse.

Users can\'t know, years later, if this will happen or if the unit
below can bear the weight of both.

Don\'t do it. Nobody competant does.

There are customers that do care, and test for it by full-scale
vibration tests on whole cabinets to MIL-S-901. A front-cantilevered
heavy components would break free and fly across the room.

Small equipment rates a 500-pound hammer, and medium rates 2000
pounds.

.<https://en.wikipedia.org/wiki/MIL-S-901

The barge test is a sight to behold.

Joe Gwinn

Yes. We tested some marine control consoles. They were installed in a
barge in a pond and explosives were detonated in the water. All sorts
of buttons and knobs flew off, imaged by a high speed movie camera.

They used some sort of magnesium lights for the camera, and that set
some stuff on fire, but that didn\'t count against us.