AD9833 DDS question

O

Oppie

Guest
Has anybody worked with the Analog Devices AD9833?

Neat and inexpensive little direct digital synthesis chip but it seems that
there is a quirk I haven't been able to resolve yet. Every time that you
load a new frequency to the device, there is a short burst of high frequency
in the output. this happens for the duration of the second word written to
the frequency register.
I had wanted to use this chip to clock a step motor state machine. Can't use
it if the output has any frequency disturbances in it.

Thanks for any input,
Oppie
 
Has anybody worked with the Analog Devices AD9833?

Neat and inexpensive little direct digital synthesis chip but it seems
that
there is a quirk I haven't been able to resolve yet. Every time that you
load a new frequency to the device, there is a short burst of high
frequency
in the output. this happens for the duration of the second word written to
the frequency register.
I had wanted to use this chip to clock a step motor state machine. Can't
use
it if the output has any frequency disturbances in it.
Click to expand...
Hey

Sounds sad, because i had planned to do a frequency sweeper with that device
to test filters etc )):

Kasper
 
I read in sci.electronics.design that Oppie <boppie@-nospam-ludl.com>
wrote (in <e5c5bbe92451e651941a23f3499662b7@news.teranews.com>) about
'AD9833 DDS question', on Fri, 19 Mar 2004:
Every time
that you load a new frequency to the device, there is a short burst of
high frequency in the output.
Click to expand...
High enough to eliminate with a low-pass filter?
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk
 
"John Woodgate" <jmw@jmwa.demon.contraspam.yuk> wrote in message
news:dTkiyWKf+zWAFw5y@jmwa.demon.co.uk...
I read in sci.electronics.design that Oppie <boppie@-nospam-ludl.com
wrote (in <e5c5bbe92451e651941a23f3499662b7@news.teranews.com>) about
'AD9833 DDS question', on Fri, 19 Mar 2004:
Every time
that you load a new frequency to the device, there is a short burst of
high frequency in the output.

High enough to eliminate with a low-pass filter?
Click to expand...
hard to see the frequency but it appears to be near the 25 mhz master clock
frequency. I need a few hundred khz. You are right that it should be able to
be filtered out with a LPF but a tech note published in EDN shows no such
thing.

See pages 4+5
http://www.reed-electronics.com/ednmag/contents/images/10804di.pdf


Oppie
 
"Kasper" <RepzakRemoveme@hotmail.com> wrote in message news:405b3883$0$209>
Sounds sad, because i had planned to do a frequency sweeper with that
device
to test filters etc )):

Kasper
Click to expand...
I bought the Eval board for the chip. who knows, I might have gotten a dud.
Got several calls into Analog Devices for clarification. Nothing in the
datasheets shows that this should happen. The group that handles this is in
Ireland so it may be a while till the St. Paddy's celebrations wind down <my
bad>.
Will post if I get any info. Would make my life easier if the chip would
work.

Oppie
 
On Fri, 19 Mar 2004 19:00:42 GMT, "Oppie" <boppie@-nospam-ludl.com>
wrote:

"John Woodgate" <jmw@jmwa.demon.contraspam.yuk> wrote in message
news:dTkiyWKf+zWAFw5y@jmwa.demon.co.uk...
I read in sci.electronics.design that Oppie <boppie@-nospam-ludl.com
wrote (in <e5c5bbe92451e651941a23f3499662b7@news.teranews.com>) about
'AD9833 DDS question', on Fri, 19 Mar 2004:
Every time
that you load a new frequency to the device, there is a short burst of
high frequency in the output.

High enough to eliminate with a low-pass filter?

hard to see the frequency but it appears to be near the 25 mhz master clock
frequency. I need a few hundred khz. You are right that it should be able to
be filtered out with a LPF but a tech note published in EDN shows no such
thing.

See pages 4+5
http://www.reed-electronics.com/ednmag/contents/images/10804di.pdf


Oppie
Click to expand...
DDS outputs *need* a decent lowpass filter, followed by a Schmitt if
you want a square wave or a clock. But I've found the AD parts to be
clean, if you follow the loading rules exactly.

John
 
I read in sci.electronics.design that Oppie <boppie@-nospam-ludl.com>
wrote (in <b6ff9ae229c3cef05a63ebc12c4b5fff@news.teranews.com>) about
'AD9833 DDS question', on Fri, 19 Mar 2004:
hard to see the frequency but it appears to be near the 25 mhz master
clock frequency. I need a few hundred khz. You are right that it should
be able to be filtered out with a LPF but a tech note published in EDN
shows no such thing.
Click to expand...
Maybe:

- EDN didn't have a fast enough scope to see it;

- the manufacturer 'improved' the design of the chip.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk
 
On Fri, 19 Mar 2004 19:24:50 +0000, the renowned John Woodgate
<jmw@jmwa.demon.contraspam.yuk> wrote:

I read in sci.electronics.design that Oppie <boppie@-nospam-ludl.com
wrote (in <b6ff9ae229c3cef05a63ebc12c4b5fff@news.teranews.com>) about
'AD9833 DDS question', on Fri, 19 Mar 2004:
hard to see the frequency but it appears to be near the 25 mhz master
clock frequency. I need a few hundred khz. You are right that it should
be able to be filtered out with a LPF but a tech note published in EDN
shows no such thing.

Maybe:

- EDN didn't have a fast enough scope to see it;

- the manufacturer 'improved' the design of the chip.
Click to expand...
EDN doesn't test those circuits, they just publish them if the editor
thinks that it might work.

Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
speff@interlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com
 
"Oppie" <boppie@-nospam-ludl.com> schreef in bericht
news:e5c5bbe92451e651941a23f3499662b7@news.teranews.com...
Has anybody worked with the Analog Devices AD9833?

Neat and inexpensive little direct digital synthesis chip but it seems
that
there is a quirk I haven't been able to resolve yet. Every time that you
load a new frequency to the device, there is a short burst of high
frequency
in the output. this happens for the duration of the second word written to
the frequency register.
I had wanted to use this chip to clock a step motor state machine. Can't
use
it if the output has any frequency disturbances in it.
Click to expand...
Perhaps you are messing up the NCO when loading the data. There
are a few option in the control bits, to disable the NCO internal
clock. Perhaps you should try that, disable first, load new values,
and enable. Or load the new frequency in the other register first,
and then switch over, and keep doing so for each new frequency.


--
Thanks, Frank.
(remove 'x' and 'invalid' when replying by email)
 
In alt.binaries.schematics.electronic,sci.electronics.design, "Oppie"
<boppie@-nospam-ludl.com> wrote:

Has anybody worked with the Analog Devices AD9833?

Neat and inexpensive little direct digital synthesis chip but it seems that
there is a quirk I haven't been able to resolve yet. Every time that you
load a new frequency to the device, there is a short burst of high frequency
in the output. this happens for the duration of the second word written to
the frequency register.
Click to expand...
I'm looking at the block diagram on the first page of the
datasheet, and it shows two frequency registers "FREQ0 REG" and "FREQ1
REG" that you can presumably switch between with a control bit. You
should be able to make a glitchless change in frequency by writing the
new value to the frequency register the chip is NOT using a the
moment, THEN set the device to use that frequency register.
The data sheet ought to (I haven't read it all, but I sure will if
I use this thing someday) describe switching between registers this
way to change frequency.

I had wanted to use this chip to clock a step motor state machine. Can't use
it if the output has any frequency disturbances in it.

Thanks for any input,
Click to expand...
Your're welcome, HTH, HAND, send me a check, etc.

Oppie
Click to expand...
-----
http://mindspring.com/~benbradley
 
"Oppie" <boppie@-nospam-ludl.com> wrote in message
news:b6ff9ae229c3cef05a63ebc12c4b5fff@news.teranews.com...
"John Woodgate" <jmw@jmwa.demon.contraspam.yuk> wrote in message
..

High enough to eliminate with a low-pass filter?
Click to expand...
No, the LPF filter doesn't work here. The output of the 9833 is voltage. It
has an internal 200 ohm load resistor. The Eval board comes with a location
for an optional output filter capacitor. I tried values up to 1000pF with
sweep frequencies up to 200 Khz. No Good.
The burst that comes out during the load of the second tuning word (most
significant 14 bits) varies in frequency: Possibly with the value of the
first word (14 least significant bits).

The burst also seems to be counted in the accumulator so the phase is
screwed up. It is almost like the frequency register is not being buffered
properly as the new tuning word is written.

I checked whether there is any errata information on the ADI website but
none was present.
The chip has only the markings 'D' next to the logo and on the next line
'JB'

If you don't have to change the frequency, the device looks very good. If
the frequency has to be swept, there may be a problem. <see below>

The evaluation software came with the eval board and is also downloadable
from the 9833 product page. Works through the parallel port.
Word of caution though, the install ate the MSVCRT.dll file on my win98
machine and I had to reinstall the OS! If during install of the eval
software, you get a 'locked file' message, hit IGNORE. the Reboot option
eats the file.
=================


I just tried as Ben had recommended; preloading the freq0 and freq1
registers then switching from one to the other in ping-pong fashion. The
transition appears to be nice and clean now. This was my original intention
when we got around to writing our own program. Why in blue blazes does the
demo software present the part in the worst possible circumstances.

Thanks to all
Oppie
 
On Fri, 19 Mar 2004 19:16:45 +0100, Kasper wrote:

Has anybody worked with the Analog Devices AD9833?

Neat and inexpensive little direct digital synthesis chip but it seems
that
there is a quirk I haven't been able to resolve yet. Every time that you
load a new frequency to the device, there is a short burst of high
frequency
in the output. this happens for the duration of the second word written to
the frequency register.
I had wanted to use this chip to clock a step motor state machine. Can't
use
it if the output has any frequency disturbances in it.

Hey

Sounds sad, because i had planned to do a frequency sweeper with that device
to test filters etc )):

Kasper
Click to expand...
Don't rush to judgement. I am familiar with two different AD DDS's, and
neither of them exhibits the bad behavior in question. There is probably
some explanation for all this, and I hope that Oppie will keep us advised,
now that he raised the issue in such a public forum.


--Mac
 
"Oppie" <oppie1@cloud9.net> wrote in message
news:105ra3hkq2pdi52@corp.supernews.com...
"Bob Groschen" <rpgroschen@adelphia.net> wrote in message
news:taidnTuemfLUc8HdRVn-uQ@adelphia.com...
|
| I've used the 9830 and 9850. Both worked slick. I used canned LPFs
from
| CoilCraft because they were already spec'ed for 50 ohms.
|
| I think I might still have the original demo board from that era.
|

I seem to have found the problem with help from the group.
We were using the Eval software that came with the demo board. Seems that
when you write to only one frequency register, the second of the two
tuning
words causes a major discontinuity. The correct way, it seems, is to load
one frequency register, write a control word to use that register, load
the
second control register and then write a control word to use that one. Why
the eval <evil?> software uses only one register for performing a sweep
function, I can't for the life of me figure out.Analog Devices still
hasn't
answered my questions. They have been very responsive in the past.

The AD9833 is a really nice device and at $4, quite a bargain. It is able
to
output a sine, triangle or square wave (only one waveform at a time) with
only a master clock input and a 3 wire SPI link (Serial data, Serial clock
and a chip select)

Regards,
Oppie
Click to expand...
 
"Oppie" <oppie1@cloud9.net> wrote in message
news:105ra3hkq2pdi52@corp.supernews.com...
"Bob Groschen" <rpgroschen@adelphia.net> wrote in message
news:taidnTuemfLUc8HdRVn-uQ@adelphia.com...
|
| I've used the 9830 and 9850. Both worked slick. I used canned LPFs
from
| CoilCraft because they were already spec'ed for 50 ohms.
|
| I think I might still have the original demo board from that era.
|

I seem to have found the problem with help from the group.
We were using the Eval software that came with the demo board. Seems that
when you write to only one frequency register, the second of the two
tuning
words causes a major discontinuity. The correct way, it seems, is to load
one frequency register, write a control word to use that register, load
the
second control register and then write a control word to use that one. Why
the eval <evil?> software uses only one register for performing a sweep
function, I can't for the life of me figure out.
Click to expand...
Well, they probably are in short supply of decent programmers so they let a
trainee do that simple job thinking there was no way he (she) could F*** up.
Another example of that assumptions are the mother of all/some f***ups

Cheers

Klaus
 
I read in sci.electronics.design that Oppie <oppie1@cloud9.net> wrote
(in <105ra3hkq2pdi52@corp.supernews.com>) about 'AD9833 DDS question',
on Sun, 21 Mar 2004:
The correct way, it seems, is to load one
frequency register, write a control word to use that register, load the
second control register and then write a control word to use that one.
Why the eval <evil?> software uses only one register for performing a
sweep function, I can't for the life of me figure out. Analog Devices
still hasn't answered my question
Click to expand...
Maybe the guy who set it up that way left 2 years ago, and no-one else
wants to try to explain it away.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk
 
Oppie wrote:

Has anybody worked with the Analog Devices AD9833?

Neat and inexpensive little direct digital synthesis chip but it seems that
there is a quirk I haven't been able to resolve yet. Every time that you
load a new frequency to the device, there is a short burst of high frequency
in the output. this happens for the duration of the second word written to
the frequency register.
I had wanted to use this chip to clock a step motor state machine. Can't use
it if the output has any frequency disturbances in it.
Click to expand...
OK, in your application, you only need a square wave output.

You can 're-clock' the VFO output through a flip-flop which is itself clocked
with a frequency just slightly higher than the one you need to use. Nothing
faster will get through.

Graham
 
"Pooh Bear" <rabbitsfriendsandrelations@hotmail.com> schreef in bericht
news:405DED63.B43F715E@hotmail.com...
Oppie wrote:

Has anybody worked with the Analog Devices AD9833?

Neat and inexpensive little direct digital synthesis chip but it seems
that
there is a quirk I haven't been able to resolve yet. Every time that you
load a new frequency to the device, there is a short burst of high
frequency
in the output. this happens for the duration of the second word written
to
the frequency register.
I had wanted to use this chip to clock a step motor state machine. Can't
use
it if the output has any frequency disturbances in it.

OK, in your application, you only need a square wave output.

You can 're-clock' the VFO output through a flip-flop which is itself
clocked
with a frequency just slightly higher than the one you need to use.
Nothing
faster will get through.
Click to expand...
Better read the entire thread first, before posting this silly
advice.

--
Thanks, Frank.
(remove 'x' and 'invalid' when replying by email)
 
On Fri, 19 Mar 2004 17:50:07 GMT, the renowned "Oppie"
<boppie@-nospam-ludl.com> wrote:

Has anybody worked with the Analog Devices AD9833?

Neat and inexpensive little direct digital synthesis chip but it seems that
there is a quirk I haven't been able to resolve yet. Every time that you
load a new frequency to the device, there is a short burst of high frequency
in the output. this happens for the duration of the second word written to
the frequency register.
I had wanted to use this chip to clock a step motor state machine. Can't use
it if the output has any frequency disturbances in it.
Click to expand...
What is the relevance of "Rset", which appears twice in the datasheet
(once in the test setup- Rset=6.8K and once in the formula for output
voltage:

VOUT = Vref * 18 * Rload/Rset * (1 + (SIN(2*pi*(FREQREG * fMCLK *
t/2^28 + PHASEREG/2^12))))

I don't see it on the schematics. Is it a left-over from other DDS
chips or would it be connected to, say, the COMP pin?

Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
speff@interlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com
 
"Spehro Pefhany" <speffSNIP@interlogDOTyou.knowwhat> wrote in message
news:03hu50daqkmgg0nfe4dab4u0cjnjfefi9q@4ax.com...
What is the relevance of "Rset", which appears twice in the datasheet
(once in the test setup- Rset=6.8K and once in the formula for output
voltage:

VOUT = Vref * 18 * Rload/Rset * (1 + (SIN(2*pi*(FREQREG * fMCLK *
t/2^28 + PHASEREG/2^12))))

I don't see it on the schematics. Is it a left-over from other DDS
chips or would it be connected to, say, the COMP pin?

Best regards,
Spehro Pefhany
Click to expand...
Good eye there Sphero. I didn't even notice it and finally did a search of
the pdf file for rset.
http://www.analog.com/UploadedFiles/Data_Sheets/37150134474160AD9833_a.pdf

Apparently, this is a vestige of another DDS datasheet they used as a
template. If customer support ever answers me, I'll bring it to their
attention.

The AD9830 has a Rset pin that affects voltage scaling.
http://www.analog.com/UploadedFiles/Data_Sheets/3692632ad9830.pdf

Prompts one to ask who is doing the proof reading and verification.

Regards - Oppie
 
Just had a good discussion with a member of the ADI DDS group.

They are sending me another software version to verify if it works any
better on the eval board.

Mentioned that Sphero found an error in the data sheet. Eva had written the
data sheets on both the 9834 (which has the Rset pin) and 9834. Oops and
thank you's.

Got an agreement that a FAQ section and an 'idiot's guide to DDS' is needed
<smirk>

So, hopefully it will be easier for the next person to use the part. I feel
better now that there is a good start.

Oppie
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top