Is mounting SMD DDS chip on homemade proto board a good idea?

[Jumbaliah]

Hey there,

I'm thinking of making a DDS function generator using one of the AD9x
chips. All of Analog's DDS chips are in SMD packages. I've emailed
others who have made these in the past based on the 50 and 125MHz
10bit chips and they recommend going with one of the better chips for
a cleaner output. So I'm possibly leaning towards the 400MHz 14bit
chips.

I wanted opinions on mounting one of these on a homemade pcb ( similar
to this: http://protoboards.theshoppe.com/qfp_0_8_mm.html ) then using
short jumper wires to connect it to the rest of the circuit. This way
I would only have to solder the chip once (by hand). The board and
chip could easily be removed for use in other circuits or revisions.

My worries are it may cause too much noise or interference in the
output. I don't have much experience in these frequencies so I wanted
to ask what others thought.

Thanks,
Burnboy

 

[Tim Shoppa]

Hey there,

I'm thinking of making a DDS function generator using one of the AD9x
chips. All of Analog's DDS chips are in SMD packages. I've emailed
others who have made these in the past based on the 50 and 125MHz
10bit chips and they recommend going with one of the better chips for
a cleaner output. So I'm possibly leaning towards the 400MHz 14bit
chips.

I wanted opinions on mounting one of these on a homemade pcb ( similar
to this: http://protoboards.theshoppe.com/qfp_0_8_mm.html ) then using
short jumper wires to connect it to the rest of the circuit. This way
I would only have to solder the chip once (by hand). The board and
chip could easily be removed for use in other circuits or revisions.

My worries are it may cause too much noise or interference in the
output. I don't have much experience in these frequencies so I wanted
to ask what others thought.

Using jumper wires to hook up your bypass caps will completely destroy
all the advantage you get by going to a 400MHz chip, and that's
best case. Worst case is that the chip doesn't work at all without
proper bypassing.

If you laid out a PCB with proper ground and power planes and
bypassing, you'd stand a chance.

I would heavily recommend that you buy one of the Analog Devices
DDS evaluation boards if you think you need multi-hundred MHz clocks
and you don't want do do a proper PC board layout. Yes, the eval
boards cost a few hundred bucks each, but you just hook up power
and the interface to a microprocessor and plug in a coax cable to get
the RF and it all just works.

Tim.

 

[Tim Shoppa]

I'm thinking of making a DDS function generator using one of the AD9x
chips. All of Analog's DDS chips are in SMD packages. I've emailed
others who have made these in the past based on the 50 and 125MHz
10bit chips and they recommend going with one of the better chips for
a cleaner output. So I'm possibly leaning towards the 400MHz 14bit
chips.

Oh yeah, I almost forgot about thermal issues after going on
about RF and bypassing issues: above 200 MHz you need the thermal slug
heatsinking via the PC board to use the AD DDS parts.

Again, shelling out for the eval board (even though it's not cheap) will
save you a huge amount of heartache.

Or step back and figure out how you can use a much slower and cheaper
DDS to get what you want. Also determine the difference between what
you want and what you need! Maybe you think you need 100 MHz triangle
waves or something?

Tim.

 

[Mac]

Hey there,

I'm thinking of making a DDS function generator using one of the AD9x
chips. All of Analog's DDS chips are in SMD packages. I've emailed
others who have made these in the past based on the 50 and 125MHz
10bit chips and they recommend going with one of the better chips for
a cleaner output. So I'm possibly leaning towards the 400MHz 14bit
chips.

I wanted opinions on mounting one of these on a homemade pcb ( similar
to this: http://protoboards.theshoppe.com/qfp_0_8_mm.html ) then using
short jumper wires to connect it to the rest of the circuit. This way
I would only have to solder the chip once (by hand). The board and
chip could easily be removed for use in other circuits or revisions.

My worries are it may cause too much noise or interference in the
output. I don't have much experience in these frequencies so I wanted
to ask what others thought.

Thanks,
Burnboy

I wouldn't do it.

But if you DO do it, make sure you get bypass caps and analog power supply
filtering on the same PCB as the DDS.

Also, those DDS's run pretty hot, and have thermal pads on their
underside, I believe. You may have problems with overheating if you hand
solder and don't connect the thermal lug.

Have you thought about building your prototypes in a toaster oven? I
haven't tried this yet, but I would like to try it soon. This would save
you from hand-soldering.

--Mac

 

[Jumbaliah]

Hey there,

I'm thinking of making a DDS function generator using one of the AD9x
chips. All of Analog's DDS chips are in SMD packages. I've emailed
others who have made these in the past based on the 50 and 125MHz
10bit chips and they recommend going with one of the better chips for
a cleaner output. So I'm possibly leaning towards the 400MHz 14bit
chips.

I wanted opinions on mounting one of these on a homemade pcb ( similar
to this: http://protoboards.theshoppe.com/qfp_0_8_mm.html ) then using
short jumper wires to connect it to the rest of the circuit. This way
I would only have to solder the chip once (by hand). The board and
chip could easily be removed for use in other circuits or revisions.

My worries are it may cause too much noise or interference in the
output. I don't have much experience in these frequencies so I wanted
to ask what others thought.

Thanks,
Burnboy

Thanks for the comments guys.

Right now I'm looking for a fuction generator to give me some sort of
signal I can test other circuits with. I'd say 0-20Meg would be
great. I just saw these DDS chips and with the 400Meg ones I may get
something like 0-50Meg out of it (never know what bandwidth you'd need
down the road). With the price of the eval boards I may be better off
just getting a MAX038 (? forgot the part number) based kit.

I've used those nice Agilent function generators in a lab and thought
this DDS would be a great way to get the similar features.

How about the 180MEG AD9851? I'd feel more confident soldering that
SSOP to the main board over the QFP parts. At least there would only
be 2 sides to work on if I needed to remove the chip. Any thermal
lugs on these packages?

What's would be a realistic estimate on max frequency clean output on
these DDS chips? eg. I've read websites where the 9835 50Meg chip was
used and the author was hoping for 0-20Meg output. (
http://hem.passagen.se/communication/dds.html ) I've been
guestimating 1:5. I guess it comes down to the output smoothing
filter's design?

Thanks for the help,
Burnboy

 

[John Miles]

Thanks for the comments guys.

Right now I'm looking for a fuction generator to give me some sort of
signal I can test other circuits with. I'd say 0-20Meg would be
great. I just saw these DDS chips and with the 400Meg ones I may get
something like 0-50Meg out of it (never know what bandwidth you'd need
down the road). With the price of the eval boards I may be better off
just getting a MAX038 (? forgot the part number) based kit.

I've used those nice Agilent function generators in a lab and thought
this DDS would be a great way to get the similar features.

How about the 180MEG AD9851? I'd feel more confident soldering that
SSOP to the main board over the QFP parts. At least there would only
be 2 sides to work on if I needed to remove the chip. Any thermal
lugs on these packages?

What's would be a realistic estimate on max frequency clean output on
these DDS chips? eg. I've read websites where the 9835 50Meg chip was
used and the author was hoping for 0-20Meg output. (
http://hem.passagen.se/communication/dds.html ) I've been
guestimating 1:5. I guess it comes down to the output smoothing
filter's design?

Thanks for the help,
Burnboy

You can hand-wire DDS chips of any speed, as long as you mount them
above a solid copper ground plane. The only constraint is your
eyesight! "Proto-board" of any variety is out of the question.

Chips like the AD9852ASQ will need to be underclocked somewhat, because
of the lack of thermal bonding to their heat sink tabs.

Use chip caps as bypasses directly from the DDS pins to the ground plane
-- no jumpers.

Study the data sheet for the part you're using. The usual rule of thumb
is that the maximum clean output frequency is around 40% of the clock
rate, but higher clock rates are still preferable. An AD9835 or AD9851
will be fine from DC-20 MHz.

-- jm

 

[john jardine]

[email protected] (Jumbaliah) wrote in message
[clip]

What's would be a realistic estimate on max frequency clean output on
these DDS chips? eg. I've read websites where the 9835 50Meg chip was
used and the author was hoping for 0-20Meg output. (
http://hem.passagen.se/communication/dds.html ) I've been
guestimating 1:5. I guess it comes down to the output smoothing
filter's design?

Thanks for the help,
Burnboy

1:5 range is about right, maybe upto 3.5:1 at a push. It seems often
overlooked that beyond this point, no amount of filtering can get the
harmonic distortion down to reasonable levels (say 2%), as the 'distortion'
frequencies or products are moving *downwards* below the output frequency as
the output frequency is increased. The effect is similar to 3rd order mixer
IMD products. Low pass to ones hearts content but those 'sub' frequencies
just can't be stripped out and will insist on messing up the waveshape.
regards
john

 

[Rene Tschaggelar]

Jumbaliah wrote:

How about the 180MEG AD9851? I'd feel more confident soldering that
SSOP to the main board over the QFP parts. At least there would only
be 2 sides to work on if I needed to remove the chip. Any thermal
lugs on these packages?

What's would be a realistic estimate on max frequency clean output on
these DDS chips? eg. I've read websites where the 9835 50Meg chip was
used and the author was hoping for 0-20Meg output. (
http://hem.passagen.se/communication/dds.html ) I've been
guestimating 1:5. I guess it comes down to the output smoothing
filter's design?

Without extensive experience in the RF field, you should leave the
faster DDS alone. Get one with a 50MHz clock. It makes 20MHz
on the output.

Rene

 

[Jumbaliah]

Without extensive experience in the RF field, you should leave the

faster DDS alone. Get one with a 50MHz clock. It makes 20MHz
on the output.

Rene

Thanks for the tips. I've ordered the AD9851 180MHz chip. I'm just
bummed a bit that it doesn't have a 12 or 14bit DAC. But I'll have to
wait and see, the 20MHz output may be pretty good with the 10bit DAC.

It would be a big jump to try the 3-400MHz circuit on my first
attempt. Other than the toaster oven and solder paste methond I don't
know how someone could ( at home mind you with no access to assembly
processes) easily connect the thermal pads on the QFP packages. I
also looked closer at the datasheet and they suggest a grid of Vias
placed under the chip for venting. If I want one of those chips down
the road then I'd probably spring for the evaluation board.

Other than the Analog website is there any good DDS related sites out
there?

Thanks,
Burnboy