19 Mar
2015
19 Mar
'15
5:09 a.m.
This is a classic case of "Great minds run in the same ruts" (sometimes
rephrased as gutters...)
I've almost designed a 4 MB card twice now. Because I'm a cheapskate, I
didn't want to use FRAM. And there are very nice SRAMs available: for
example, the Cypress CY62167 is a 2Mx8 1Mx16 RAM in 48 pin TSOP. So you'd
need two, that's ~$30 total. (n.b. there is a 4Mx8 but it's only in BGA, so
fuggitaboutit).
So, you'd only need the bus interface (no, I'm not going there) and the
control logic.
These SRAM parts are so speedy (45 ns) that cache isn't needed (obviously).
Maybe I'll almost design it for the third time!
19 Mar
19 Mar
7:29 a.m.
On Mar 19, 2015, at 8:09 AM, Mark Kahrs <mark.kahrs
at gmail.com> wrote:
This is a classic case of "Great minds run in the same ruts" (sometimes
rephrased as gutters...)
I've almost designed a 4 MB card twice now. Because I'm a cheapskate, I
didn't want to use FRAM. And there are very nice SRAMs available: for
example, the Cypress CY62167 is a 2Mx8 1Mx16 RAM in 48 pin TSOP. So you'd
need two, that's ~$30 total. (n.b. there is a 4Mx8 but it's only in BGA, so
fuggitaboutit).
I used to think of BGA as not possible for home construction. Then I read a nice article
? https://www.beta-estore.com/download/rk/RK-10001_76.pdf ? about reflow soldering using
what amounts to a toaster-oven with a temperature control added, which apparently does the
job just fine. There are some details to get right, which the article explains.
paul
9:30 a.m.
On 03/19/2015 09:29 AM, Paul Koning wrote:
I used to think of BGA as not possible for home
construction. Then I read a nice article ?
https://www.beta-estore.com/download/rk/RK-10001_76.pdf ?
about reflow soldering using what amounts to a
toaster-oven with a temperature control added, which
apparently does the job just fine. There are some details
to get right, which the article explains.
I reflow boards in a modified toaster
oven, not too
different from the one in the picture. But, it did NOT work
well when I first tried it. I just hung the thermocouple in
the air, and the boards were burnt to a crisp! The IR from
the heating elements is absorbed VERY well by the boards,
and they heat much faster than the air. So, my trick is to
poke a
very tiny thermocouple into a plated through-hole in the
board, and then it reads the actual temperature of the
board. I still have some problems with hotter and cooler
areas of the oven, but you can usually find a good compromise.
Jon
9:52 a.m.
On Mar 19, 2015, at 12:30 PM, Jon Elson <elson at
pico-systems.com> wrote:
On 03/19/2015 09:29 AM, Paul Koning wrote:
Nice solution. Yes, that?s the main detail the article mentions: the sensor has to be in
a spot that has similar IR response as the actual board. The solution used in the article
is to use a spare (blank) board for a reference. A different board with similar ratio of
copper to etch should also work. Your solution is elegant if there?s a suitable spot you
can arrange.
paul
I used to think of BGA as not possible for home
construction. Then I read a nice article ?
https://www.beta-estore.com/download/rk/RK-10001_76.pdf ? about reflow soldering using
what amounts to a toaster-oven with a temperature control added, which apparently does the
job just fine. There are some details to get right, which the article explains.
I
reflow boards in a modified toaster oven, not too different from the one in the picture.
But, it did NOT work well when I first tried it. I just hung the thermocouple in the air,
and the boards were burnt to a crisp! The IR from the heating elements is absorbed VERY
well by the boards, and they heat much faster than the air. So, my trick is to poke a
very tiny thermocouple into a plated through-hole in the board, and then it reads the
actual temperature of the board. I still have some problems with hotter and cooler areas
of the oven, but you can usually find a good compromise. 
1:59 p.m.
On Thu, Mar 19, 2015 at 11:30:48AM -0500, Jon Elson wrote:
I reflow boards in a modified toaster oven, not too
different from the one in
the picture. But, it did NOT work well when I first tried it. I just hung
the thermocouple in the air, and the boards were burnt to a crisp! The IR
from the heating elements is absorbed VERY well by the boards, and they heat
much faster than the air. So, my trick is to poke a
very tiny thermocouple into a plated through-hole in the board, and then it
reads the actual temperature of the board. I still have some problems with
hotter and cooler areas of the oven, but you can usually find a good
compromise.
Same experience here. Many fried boards until I tried touching the
(bare -- not sheathed) thermocouple to the board (doesn't have to be in a
PTH, although that might stop it from occasionally twanging and sweeping
components off). I've only minimally tried BGAs (just liquid flux w/no
solder paste at all seemed to be the answer there) but I've had fantastic
luck with QFPs, SOICs, TSSOPs, and and all kinds of SMT passive components.
www.oshstencils.com makes it just too easy.
John Wilson
D Bit
6:33 p.m.
On 03/19/2015 03:59 PM, John Wilson wrote:
Same experience here. Many fried boards until I tried
touching the (bare -- not sheathed) thermocouple to the
board (doesn't have to be in a PTH, although that might
stop it from occasionally twanging and sweeping components
off). I've only minimally tried BGAs (just liquid flux
w/no solder paste at all seemed to be the answer there)
but I've had fantastic luck with QFPs, SOICs, TSSOPs, and
and all kinds of SMT passive components.
www.oshstencils.com makes it just too easy. John Wilson D Bit
Right, I've
never tried the BGAs either. I have done many
thousands of SOIC and TSSOP parts, and a few thousand QFPs
up to 128 pin 0.5mm pitch parts. I've probably done over
100K passives, generally 0805, but I have one board that is
0603 that now works pretty well. I do have a pick and place
machine.
I make my own stencils with gear I used to use to make PC
boards. I get brass .003" shim stock and laminate it on
both sides with dry film etch photoresist, just as if it was
a PC board. I have a laser photoplotter I made, and make
the artwork for the solder stencil with that. I make a
normal and a mirror image, and glue them together on a light
table, with a little spacer between to compensate for the
shim stock thickness. I just glue one edge and let the glue
dry. Slip the shim stock with resist between the artwork,
expose both sides and develop and etch. It took a while to
get the whole process dialed in just right, but now it works
quite well. Part of the trick is to reduce the apertures to
be smaller than the solder pads on the board.
Jon
3516
days inactive
3517
days old
5 comments
4 participants
participants (4)
-
elson＠pico-systems.com -
mark.kahrs＠gmail.com -
paulkoning＠comcast.net -
wilson＠dbit.com