Author Topic: AD9833 waveform generator  (Read 8601 times)

0 Members and 1 Guest are viewing this topic.

Offline DanKalcTopic starter

  • Newbie
  • Posts: 9
AD9833 waveform generator
« on: December 05, 2013, 01:27:43 pm »
Hi,
Few days ago i got my hands on AD9833 so i decided to build my own waveform generator. My first problem was can i use simple quartz resonator or should i use quartz generator.
I also want to adjust the output voltage so i bought AD8051 (110 MHz rail-to-rail 12V amplifier). Gain will be changed using a Digital POT MCP4011 (10kOhm max), maximum gain will be about 2 (AD9833 outs 5V and i want amplitude from 0-10V).

Can anyone check my schematics for errors, its my first "high" frequency project. Will everything work at 12.5MHz?



Datasheets:
AD9833 (http://www.analog.com/static/imported-files/data_sheets/AD9833.pdf)
AD8051 (http://www.analog.com/static/imported-files/data_sheets/AD8051_8052_8054.pdf)
MCP4011 (http://ww1.microchip.com/downloads/en/devicedoc/21978a.pdf)
 

Offline tszaboo

  • Super Contributor
  • ***
  • Posts: 8503
  • Country: nl
  • Current job: ATEX product design
Re: AD9833 waveform generator
« Reply #1 on: December 06, 2013, 02:37:14 pm »
The input capacitance of your opamp is 1.5pF. With 10K feedback, that is about 10Mhz -3db bandwidth on the feedback.
 

Offline krivx

  • Frequent Contributor
  • **
  • Posts: 765
  • Country: ie
Re: AD9833 waveform generator
« Reply #2 on: December 06, 2013, 04:35:25 pm »
The MCP4011 lists a BW of -3dB at 4MHz.
 

Offline DanKalcTopic starter

  • Newbie
  • Posts: 9
Re: AD9833 waveform generator
« Reply #3 on: December 06, 2013, 04:48:40 pm »
Thank you,
I need to figure out other way to change my amplitude.
 

Offline KerryW

  • Regular Contributor
  • *
  • Posts: 112
  • Country: us
Re: AD9833 waveform generator
« Reply #4 on: December 06, 2013, 05:06:52 pm »
I think you'll find that the AD9833 puts out 650mV peak to peak, not 5V.  You'll need a lot more gain than you think.

Kerry
One accurate measurement is worth a thousand expert opinions
- Adm. Grace Hopper
 

Offline codeboy2k

  • Super Contributor
  • ***
  • Posts: 1836
  • Country: ca
Re: AD9833 waveform generator
« Reply #5 on: December 08, 2013, 12:14:01 pm »
You might find it easier to first gain up the 650mv output x15 or so, to get near to 10V output from it, as a first stage.
You might need to do this in 2 stages, with 2 opamps, x5 and x3, in order to keep within the limits of the Gain Bandwidth Product (GBW). For example, if your device has a 100MHz GBW, then at gain of 5, it only has a 20Mhz bandwidth. This is the -3dB point, so it actually starts to roll off its gain way earlier than that, like around 100kHz or 200Khz or so.  This is why you often see high gains done in multiple stages.

Once you have the output gained up to 10V p-p, then you can use the digital potentiometer to control an inverting opamp as an attenuator, to attenuate the signal by 20dB or so. This will give you 0-10V output (approx).  You have to be careful here too, because you are attenuating x10, and you still have to contend with the GBW product of this stage. Furthermore, you don't want the signal going through you digital pot, which does not have the bandwidth you need.

You could also forgo that, and just come up with a VCA (voltage controlled amplifer) but you'll need to design carefully to get the performance you need.

You'll also need your opamps to have very high slew rates for these speeds at 10Mhz and 10V output. Here's why:
Consider that your opamp is running fine, not slewing, and getting a nice sine wave output.

  Then .
The time derivative of the output is and the maximum value of occurs when ,
 i.e. when the angular frequency at time t is an integer multiple of pi.

 It is here that and and this cannot exceed the slew rate of the opamp.

so , and

the maximum frequency and voltage that the op amp can output without slewing is:



In your design, Vp = 5V and fmax = 12.5Mhz, so 2*pi*12.5e6*5V = 392699082 V/s slew rate, or 393 V/us (as you'll find it on the datasheets in V/us). So you'll want your final stage opamp to have a slew rate of at least 1.5 to 2 times that, just to be safe, or 600 to 700 V/us

And finally, you'll also need to consider the Full Power Bandwidth (FPBW), which is the maximum frequency the opamp can put out with it's peak voltage equal to the clipping voltage (the Vout(min) and Vout(max) "rails" ). Going beyond this frequency (FPBW),  the opamp will start full slewing and will reduce it's output voltage to a  peak voltage of slewrate/4f. The peak output is inversely proportional to the frequency and the slewing output is not pretty.. You don't want to operate in this region, so you should seek it out on the datasheets :)

There was another recent thread about higher frequency op amps, and many "young players" look at the GBW and don't consider the slew rate, but the slew rate is equally important and needs to be considered along with GBW, so I thought I'd put the simple math out there in this thread for you and anyone else to understand the GBW, slewrate and FPBW better.

So, your first high frequency design is not so simple, as you can see.  But you can learn from it.

Cheers!

 
The following users thanked this post: testian


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf