Designer’s Guide Community

ywguo

Community Fellow

Offline

Posts: 943
Shanghai, PRC

Oscillations at DAC output
Dec 26^th, 2007, 9:58pm

Hi Guys,

I saw oscillations at the DAC output. Please see my attachment, which is a lab test result of a 10-bit current steering DAC. The output is differential current. Then it is converted using a LM7372 differential amplifier. The clock is 75 MHz crystal oscillator. The die is bonded on the PCB directly. The bonding wire is estimated 2 mm long. I am simulating the DAC with 2nH inductance in series of the DAC output.

Yawei

Back to top

TEK00000.PNG

IP Logged

sheldon Community Fellow Offline Posts: 751	Re: Oscillations at DAC output Reply #1 - Feb 8^th, 2008, 3:19am Yawei, Just curiuos, were you ever able to debug the oscillation issue? BTW, you mentioned the modeling the bondwires, did you also model the rest of the board parasitics, pad capacitance, pad capacitance of the load resistors, etc? Best Regards, Sheldon
Back to top	IP Logged

ywguo

Community Fellow

Offline

Posts: 943
Shanghai, PRC

Re: Oscillations at DAC output
Reply #2 - Feb 12^th, 2008, 6:52pm

Hi Sheldon,

Thank you very much for your comment.

I have no idea about the value of the board parasitics and the pad capacitance of the load resistors. Sad

Since the pad adds relative small capacitance only, it is not included in the simulation. The simulation proves that the oscillation caused by the bonding wire attenuates quickly. The oscillation shown in my attachment does not attenuate in fact.

The clock driver is near the DAC output on my evaluation board. I suspect that crosstalk kills the DAC output. However, I have not proved that point.

Or probably that the harmonics of the clock is coupled into the probe. If I have a differential probe, it is easy to eliminate the coupling.

Best regards,
Yawei

Back to top

« Last Edit: Feb 12^th, 2008, 10:37pm by ywguo »

IP Logged

sheldon

Community Fellow

Offline

Posts: 751

Re: Oscillations at DAC output
Reply #3 - Feb 19^th, 2008, 6:16pm

Yawei,

Don't know what you have tried, how about the following:

If you turn off the data, then only the clock would be running and
you could isolate the effect of clock cross-talk from the output
settling and ringing characteristics.

Also what happens if you change the clock frequency, does the
oscillation change?
If the crosstalk is due to the clock crosstalk, then the frequency
of the clock harmonics should also scale with the clock frequency.
If the oscillation stays the same, then the source of oscillation
may not be clock cross-talk.

Best Regards,

Sheldon

Back to top

IP Logged