2MHz sine waveform using AD9833 and PIC18F4580

[syamin0712]

Firstly, does it mean the signal generally is correct by using MODE 2? I am right?

[asmboy]

my work with the the part( which i use in several different products )-- led me to believe the 200 ohm value you cite is actually a source impedance. and at frequencies below 100khz made no special difference other than cutting the 200 mv RMS to 100 mv.

however at frequencies above a megahertz i found spectral purity was
very much reduced without it.
https://upload.wikimedia.org/wikipedia/commons/e/e7/1hzto1Mhz_sinewave.JPG
Above is a module with performance from 1 hz to 768khz in 1 hz steps ,
1.00 V RMS with better than .3db flatness, and DC output guaranteed to be less than 1 mv ( due to DC servo) at better than -56db total RMS spurious content. no terminating resistor needed.
the residue is primarily 3rd harmonic energy.

R5, C6 , R8 are added when operation above 1 mhz is required.

my biz partner has used several thousand of these with great happiness.
ANYWAY.....
Try using the external terminator
AND
also try grounding your scope to the PCB common ground and not using the tip -ground clip.


i'd feel a lot more certain of your problem if you showed a PHOTO with the details of your actual hardware under test.

showing how you connect your scope lead too ideally.

[syamin0712]

[Ttelmah]

My comment about the scope probe, is this.
Every oscilloscope probe has to be compensated to match the scope it is being used on. There is a setup procedure on the scope, and a test source built into the scope. You attach the probe to a pin on the scope producing a nice square wave (it is normally a pin on the front of the scope). Set the time-base and amplitude to specific levels for the calibration (scope manual), and adjust the trimmer capacitor on the side of the probe to get a nice waveform without overshoot or rounding.
This is a standard part of using a scope, and unless it is done, the scope will generate smoothing or overshoot on the waveform.

You need to be grounding the scope to the chip under test, especially with your wiring.

[syamin0712]

Hi Ttelmah,

[Ttelmah]

I said the ground of the chip under test, not the circuit.

Point is that there is likely to be noise between the chip, and your master ground 'point' when working at high frequency.

[syamin0712]

[Ttelmah]

No, you are missing the point slightly.

Just because you have two pins connected together, does not make them the same signal!...

When dealing with things switching at speed, there _will_ be noticeable signals, between points that are connected together. You need to actually make a connection directly adjacent to the pin of the chip you are testing.
This is why on HF boards, there are often multiple test point grounds round the board. One close to each major section of the circuit.
Try the experiment of turning the scope gain up a little, and touching the probe to different points on the same 'ground' signal round your board. Be prepared for just how much signal there may be at places.

[syamin0712]

[asmboy]

[syamin0712]

[Ttelmah]

No. The point is that high frequencies and breadboards, are a pretty sure formula for problems. You need a proper board with ground plane, and then very careful decoupling, and careful _layout_. Look carefully at the chips. Work out where the high frequency pins are, then position the chips on the board, so that things like high speed clocks are positioned away from the analog stuff. Even with a ground plane, you need also to ensure that paths from digital stuff back to the supply, do not cross the analog sections.

This is the classic sort of problem that just needs some actual practice, experiment and experience.

There is a lot of guidance online:
<www.ti.com/lit/ml/slyp173/slyp173.pdf>
<http://electronics.stackexchange.com/questions/5223/what-to-think-about-when-designing-hf-pcbs>
<http://www.maximintegrated.com/en/app-notes/index.mvp/id/5100>

The last in particular links to a really superb lot of guidance.

[asmboy]

summary:

the waveform in
http://obrazki.elektroda.pl/7927611700_1440496782.png

shows roughly the right fundamental frequency and amplitude from the Ad9833 , once you ignore the awful analog waveform.

All your remaining issues center on not understanding
the importance of correct layout procedure and how to deal with the ANALOG and power supply aspects of the design you copied for the AD9833.

This is a PIC forum - not an mixed signals design forum.

You are driving way ahead of your ability in the analog field
and you need to connect with somebody who can do a proper layout for a circuit board for you. all the talk in the world won't get you ahead i fear.

BTW - connecting the digital controls for the AD9833 (RE2 RC3 RC5 ) through 47 to 56 ohm resistors is a good idea to prevent glitching if the part is more than a few inches from the PIC .

[syamin0712]

Ok. Thank you very much for the advice and the help.
Really appreciate it.
I will look detail on it