"xpyttl" wrote in message
...
"Carl R. Stevenson" wrote in message
...

The packaging is not "impossible" - it just takes some different
techniques
to do the
soldering - read the QEX article on a DDS/PLL synthesiser in the latest
issue of QEX

I'm pretty comfortable with SOP's and can get along with a TSSOP's,
although
I can't say I like it. In that article, he used TQFP's which, although I
haven't yet had the need, I suspect I can deal with. I use the same
technique for TSSOP's that he describes in the article.

But when I tried to order samples of 9854's the other day, the only choice
was BGA's. I have no clue how to deal with them. Yeah, solder paste and
an
oven MAYBE, but without machine placement seems like a crap shoot.

With BGAs you HAVE to reflow solder ... however if the prep is right, they
should
"self-center" (as do most SMDs) on the pads if you're close enough - it's
interesting to
watch manually placed SMD resistors, caps, etc. actually move an center
themselves
up on the pads due to surface tension effects when the solder reflows ... IF
the pads
are well-designed and balanced.

73,
Carl - wk3c

On Sun, 29 Feb 2004 15:06:12 -0800, John Miles
wrote:

In article ,
says...
"Carl R. Stevenson" wrote in message
...

The packaging is not "impossible" - it just takes some different
techniques
to do the
soldering - read the QEX article on a DDS/PLL synthesiser in the latest
issue of QEX

I'm pretty comfortable with SOP's and can get along with a TSSOP's, although
I can't say I like it. In that article, he used TQFP's which, although I
haven't yet had the need, I suspect I can deal with. I use the same
technique for TSSOP's that he describes in the article.

Actually, the 16-pin PLL chips are TSSOPs, while the AD9852/AD9854
packages I used are considered "LQFP" packages. These are a little
thicker than TQFP but are otherwise pretty similar. The DDSes have the
same spacing between pins as the TSSOPs (0.5 mm) but more of 'em (80).


But when I tried to order samples of 9854's the other day, the only choice
was BGA's. I have no clue how to deal with them. Yeah, solder paste and an
oven MAYBE, but without machine placement seems like a crap shoot.

That's not good news. BGAs really *are* the end of the world for
homebrewing as far as I'm aware. I have not heard of any techniques for
soldering them by hand.


I've done a fair number by hand @ work. They self center nicely, and
don't really require any solder other than the balls. Just some fresh
liquid flux and a heat gun, and a little practice. They do self center
nicely. You can also reflow in a toaster oven if you really want to.
(Not the one in the kitchen, please!).



Dave

On Sun, 29 Feb 2004 15:06:12 -0800, John Miles
wrote:

In article ,
says...
"Carl R. Stevenson" wrote in message
...

The packaging is not "impossible" - it just takes some different
techniques
to do the
soldering - read the QEX article on a DDS/PLL synthesiser in the latest
issue of QEX

I'm pretty comfortable with SOP's and can get along with a TSSOP's, although
I can't say I like it. In that article, he used TQFP's which, although I
haven't yet had the need, I suspect I can deal with. I use the same
technique for TSSOP's that he describes in the article.

Actually, the 16-pin PLL chips are TSSOPs, while the AD9852/AD9854
packages I used are considered "LQFP" packages. These are a little
thicker than TQFP but are otherwise pretty similar. The DDSes have the
same spacing between pins as the TSSOPs (0.5 mm) but more of 'em (80).


But when I tried to order samples of 9854's the other day, the only choice
was BGA's. I have no clue how to deal with them. Yeah, solder paste and an
oven MAYBE, but without machine placement seems like a crap shoot.

That's not good news. BGAs really *are* the end of the world for
homebrewing as far as I'm aware. I have not heard of any techniques for
soldering them by hand.


I've done a fair number by hand @ work. They self center nicely, and
don't really require any solder other than the balls. Just some fresh
liquid flux and a heat gun, and a little practice. They do self center
nicely. You can also reflow in a toaster oven if you really want to.
(Not the one in the kitchen, please!).



Dave

On Sun, 29 Feb 2004 01:12:41 +0000, SWbeginner wrote:

I like to experiment with the Analog Devices DDS for a receiver.
Anyone have a favorite one with good features and least hassle?

Well, it all depends on what sort of design you are doing.

If you're working on a very low power portable design, you want to look at
the AD9834. This baby consumes a mere 8ma at 2.5 to 5 volts. The reference
clock is limited to 50 MHz, which means that 20 MHz output is about its
limit with reasonable spectral purity, though you might get up to the 15
meter band with careful filtering.

The AD9850 and AD9851 are popular for a lot of ham designs. The AD9850 has
a maximum reference clock of 125 MHz, so it's good through the 6 meter
band; the AD9851 can handle a reference clock of up to 180 MHz. These
chips consume around 100ma at 5 volts (somewhat less if run at 3.3 volts,
but then the reference clock speed limit is lower); not ideal for a
battery powered design, but still reasonable. The New Jersey QRP Club
offers a small daughterboard designed for the AD9850, and there is lots of
code available online to control them.

The AD9950 series is the new kid on the block. These run at 400 MHz,
making operation up to the 2 meter band feasible. They have a 14-bit DAC,
rather than the 10-bit one used in the older devices, so spectral purity
should be significantly improved. They consume about 85ma; reasonable for
this level of performance. There are some drawbacks... they come in a tiny
package that's hard to solder by hand. They run at 1.8v, and the inputs
only accept voltages up to 3.3v, so you'll want to design with lower
voltage control logic. Still, if you want to push the state of the DDS
art, these are your chips.

I haven't heard of any ham experimentation with the AD9858. This one can
take a reference clock up to 1 GHz, which means that 450 MHz output might
be possible with careful design. It's power-hungry, though; around 750mA.

You will also see some designs using older chips. The AD9830, the first
readily available DDS chip that could handle a 50 MHz reference clock, and
its serial input cousin the AD9835, were popular for a while, but the much
lower power AD9834 is more appealing for new designs unless you need very
fast frequency changes (the serial input lowers the possible rate of
change). The AD9852 and AD9854 have seen some experimentation, but these
chips are power hogs (they can consume more than an amp of power!), so
they are definitely out of the question for portable designs, and it's
hard to see a reason for using one of these in most designs rather than
the AD9950 series. The AD9854 does have the unique feature of having TWO
synthesizers on the same chip, which could be useful for people
experimenting with high-speed digital data modes.

On Sun, 29 Feb 2004 01:12:41 +0000, SWbeginner wrote:

I like to experiment with the Analog Devices DDS for a receiver.
Anyone have a favorite one with good features and least hassle?

Well, it all depends on what sort of design you are doing.

If you're working on a very low power portable design, you want to look at
the AD9834. This baby consumes a mere 8ma at 2.5 to 5 volts. The reference
clock is limited to 50 MHz, which means that 20 MHz output is about its
limit with reasonable spectral purity, though you might get up to the 15
meter band with careful filtering.

The AD9850 and AD9851 are popular for a lot of ham designs. The AD9850 has
a maximum reference clock of 125 MHz, so it's good through the 6 meter
band; the AD9851 can handle a reference clock of up to 180 MHz. These
chips consume around 100ma at 5 volts (somewhat less if run at 3.3 volts,
but then the reference clock speed limit is lower); not ideal for a
battery powered design, but still reasonable. The New Jersey QRP Club
offers a small daughterboard designed for the AD9850, and there is lots of
code available online to control them.

The AD9950 series is the new kid on the block. These run at 400 MHz,
making operation up to the 2 meter band feasible. They have a 14-bit DAC,
rather than the 10-bit one used in the older devices, so spectral purity
should be significantly improved. They consume about 85ma; reasonable for
this level of performance. There are some drawbacks... they come in a tiny
package that's hard to solder by hand. They run at 1.8v, and the inputs
only accept voltages up to 3.3v, so you'll want to design with lower
voltage control logic. Still, if you want to push the state of the DDS
art, these are your chips.

I haven't heard of any ham experimentation with the AD9858. This one can
take a reference clock up to 1 GHz, which means that 450 MHz output might
be possible with careful design. It's power-hungry, though; around 750mA.

You will also see some designs using older chips. The AD9830, the first
readily available DDS chip that could handle a 50 MHz reference clock, and
its serial input cousin the AD9835, were popular for a while, but the much
lower power AD9834 is more appealing for new designs unless you need very
fast frequency changes (the serial input lowers the possible rate of
change). The AD9852 and AD9854 have seen some experimentation, but these
chips are power hogs (they can consume more than an amp of power!), so
they are definitely out of the question for portable designs, and it's
hard to see a reason for using one of these in most designs rather than
the AD9950 series. The AD9854 does have the unique feature of having TWO
synthesizers on the same chip, which could be useful for people
experimenting with high-speed digital data modes.