PCB socket footprint for NL-5440A?
koolatron
I'm in the process of building up a little four-digit NL-5440A clock,
and I had the good fortune to find a small bag of SK-185 sockets that
mate with this particular tube. If you're familiar with the NL-5440A,
you'll know that the pin spacing on the bottom of the tube envelope is
quite tight, making the design of a faux-socket PCB (using Mouser
pins, the sort that you see commonly with IN-18s and the like) very
difficult.
However, it turns out that the SK-185 sockets that fit NL-5440As are
also rather oddball. Here's a link to some detailed photographs and
datasheet diagrams of the devils:
http://www.tube-tester.com/sites/nixie/data/soc/SK-182-burr-cha/SK-185-burr-ch.htm
I've tried several times to generate a sane EAGLE footprint that fits
this socket, and have met with only failure. The socket has 16 pins,
which would at first blush correspond to a 17-pin ring of evenly-
spaced contacts, with one missing. That, however, is not not the case
-- each 90-degree quadrant has four contacts, and they appear to be
spaced unevenly around the circle.
Before I go out and buy a set of calipers to measure this socket, has
anyone managed to generate a working PCB footprint for it?
Sean
JohnK
CAD/Drawing program. Then 'draw' over it on a different layer.
And, if you need measurements, do them on enlarged diagram with CAD or
[shudder] ruler on flat screen?
John K.
Nick
> I've tried several times to generate a sane EAGLE footprint that fits
> this socket, and have met with only failure. The socket has 16 pins,
> which would at first blush correspond to a 17-pin ring of evenly-
> spaced contacts, with one missing. That, however, is not not the case
> -- each 90-degree quadrant has four contacts, and they appear to be
> spaced unevenly around the circle.
>
> Before I go out and buy a set of calipers to measure this socket, has
> anyone managed to generate a working PCB footprint for it?
Harper's "soculator" generates scripts (http://www.desmith.net/NMdS/
Electronics/soculator.html) and then take a look at Eagle's "polar"
coordinate system - that should easily generate an arc of 16 pins @
21.5 degrees from each other, followed by a gap of 37.5 degrees.
I've been meaning to change the soculator to use polar coordinates,
but its a fairly recent Eagle addition and I didn't want to prejudice
users of older versions...
Cheers
Quixotic Nixotic
> However, it turns out that the SK-185 sockets that fit NL-5440As are
> also rather oddball. Here's a link to some detailed photographs and
> datasheet diagrams of the devils:
>
> http://www.tube-tester.com/sites/nixie/data/soc/SK-182-burr-cha/
> SK-185-burr-ch.htm
>
> I've tried several times to generate a sane EAGLE footprint that fits
> this socket, and have met with only failure. The socket has 16 pins,
> which would at first blush correspond to a 17-pin ring of evenly-
> spaced contacts, with one missing. That, however, is not not the case
> -- each 90-degree quadrant has four contacts, and they appear to be
> spaced unevenly around the circle.
A brief observation shows me that it is the engineering drawing that
is throwing your eye out and this does appear to show the quadrant
issues that you see, which would actually mean the pins were bunched
in two of the quadrants, to allow the gap as shown. The actual socket
is not quite the same, the centre line is not where shown on the
diagram, it's offset and the pins are thus equally spaced. A quick
exercise overlaying radial lines at 21.176º over the actual socket
leads me to believe the pins are equally spaced. Thus Sockulator
should be your friend here and make you a part for Eagle.
I did wonder if they did something sneaky and made the angle an
integral 21º with a wider gap, but again a quick exercise shows that
this fits worse than 21.176º.
John S
Quixotic Nixotic
On 18 Mar 2011, at 07:35, Nick wrote:
> On Mar 18, 3:00 am, koolatron <koolat...@gmail.com> wrote:
>
>> I've tried several times to generate a sane EAGLE footprint that fits
> "soculator" generates scripts (http://www.desmith.net/NMdS/
> Electronics/soculator.html) and then take a look at Eagle's "polar"
> coordinate system - that should easily generate an arc of 16 pins @
> 21.5 degrees from each other, followed by a gap of 37.5 degrees.
Are those angles definite Nicko? Sneaky.
> I've been meaning to change the soculator to use polar coordinates,
> but its a fairly recent Eagle addition and I didn't want to prejudice
> users of older versions...
I hacked a version of Sockulator (sic) to allow me to enter diameter
and number of pins. I find this far more versatile than the more
complicated pin measuring system, especially if I do actually know
the diameter. But the pin measuring system is also useful when it
isn't known. It's also easy to hack the javascript to allow more than
the default 30 pins. I needed to lay out clock faces, 60 pins to a
circle. So I'd add the ability to add more than 30 pins in future
releases. I work with the Osmond PCB version, so people, please don't
ask me if I can send you a hacked Eagle version, as I don't have one.
John S
David Forbes
> Hi all,
>
> Before I go out and buy a set of calipers to measure this socket, has
> anyone managed to generate a working PCB footprint for it?
>
> Sean
>
Buy the calipers. They are incredibly useful if you are laying out PC
boards or doing metalwork. They and a stereo zoom microscope are two
big-ticket items that will make your electronics hobby much more pleasant.
--
David Forbes, Tucson AZ
will
calipers that are sensitive enough to measure the width of a human
hair (i.e. .01mm or .0005in precision). They have been indispensable
to me a number of times, namely putting things inside of CAD programs.
Also, a tip if you want to accurately measure the pinout of
something:If calipers can't quite do it (they work fine for obscure
SMT pitches, but circular stuff not so well), then simply put the pins
of whatever you're measuring on a scanner. Scanners preserve the unit
of measurement, and there is software you can use to overlay angles
and measure distances and whatnot on the image. I haven't done this in
a while, but I remember that it worked for some funky shaped
oscillators and things that I took out of an old motion detector a few
years back.
koolatron
(basic) set of calipers and spent a couple of hours with EAGLE putting
together a library for the NL-5440A plus SK-185 socket. This utilizes
21.5 degree pin spacing and resulting 37.5 degree gap, which fits my
observations of the socket itself quite well. I even printed a copy
of the footprint on the office printer to check things out -- looks
basically sane. Thanks all for the help!
Here's a link to my dropbox, in case you have a need for such a thing:
http://dl.dropbox.com/u/709108/EAGLE/NL-5440.lbr
Share and enjoy. This socket actually fits quite a few nixies, but I
only have NL-5440As on hand and didn't bother adapting symbols for
variations on the scheme.
Cheers,
Sean
Nick
On Mar 21, 4:26 am, koolatron <koolat...@gmail.com> wrote:
> Here's a link to my dropbox, in case you have a need for such a thing:http://dl.dropbox.com/u/709108/EAGLE/NL-5440.lbr
tag on tValues (this is important). You haven't connected the left &
right decimal points (pins 6 & 11 respectively).
Some of your tPlace data overlays the pads and will thus be printed on
the copper if you use a silk screen when the boards are made - data
that is for documentation but overlays copper and which is thus not
wanted on the manufactured board should be on tDocu.
You might want to add descriptive text to the package & device and
name the SK-185 package variant something meaningful (rather then
blank), such as "SK-185" so that you can add other packages such as
PCB pins with a different variant name.
Pin allocations and other stuff looks good.
Nick
A.J. Franzman
wrote:
> is throwing your eye out and this does appear to show the quadrant
> issues that you see, which would actually mean the pins were bunched
> in two of the quadrants, to allow the gap as shown. The actual socket
> is not quite the same, the centre line is not where shown on the
> diagram, it's offset and the pins are thus equally spaced.
datasheet drawings. Dimensions and angles stated are usually correct,
however often the drawing itself often doesn't accurately reflect
those dimensions and angles. For example, datasheet drawings of the
NL-840 family pinout often show the four central pin positions to lie
exactly on two parallel lines connecting four of the outer pins, which
is not the case -- the central square is actually slightly rotated in
respect to those outer pins. In reality, one of the diagonals of the
square formed by the four central pins lies exactly on a line between
a different pair of the outer pins.
A.J.