Actions
Forum Thread Details
  • Replies 2 replies
  • Subscribers 532 subscribers
  • Views 1021 views
  • Users 0 members are here
Related

FPGA-based DAC

neuromodulator

I have this (alternate) idea on how to implement an FPGA-based DAC, but I haven't been able to find any info in the net...

 

With the traditional approach of delta sigma modulating and then lowpass-filtering the signal, the closer that the modulated signal frequency gets to the filter cutoff frequency, the more that it gets attenuated and phase-shifted...

 

So as an alternative I thought the system could be handled like a control system, but since there is no feedback from the analog output I would run an LPF model within the FPGA in a feedforward way. In this way, if I want to push the modulating frequency closer to the cutoff frequency I would be able to compensate for the attenuation and phase shift so that there is none.

 

The reason that I'm thinking of applying this instead of increasing the LPF cutoff frequency is that the latter would have drawbacks like reduced noise filtering. Compensating for the attenuation and phase shift of course would not be "free", but it still is an alternative to just not being able to generate the desired signal amplitude or keep the phase shift to the minimum.

 

Any ideas? Does this make sense?

  • There are some nifty ways to do DAC with an FPGA.  If you only need a few bits, you can use normal CMOS I/Os with external resistors.  For example, the Lattice iCE40H1K-based Go Board has a 9-bit VGA output, with 3 bits per color (RGB).  It uses 2K, 1K, and 549 Ohm resistors to weight the three bits.

     

    If you are generating audio, you can use an I2S DAC such as the NXP UDA1334ATS which supports up to 24 bits stereo.  I2S only needs 3 wires.  Adafruit has a nifty little UDA1334A breakout board for US$7.

     

    Hope this helps!

  • in reply to johnbeetem

    Thanks for the response John. I'm aware of the techniques you mention, but I'm trying to avoid them. When using resistors unless the resistors match perfectly, you lose linearity, plus its harder to implement for high bit resolution and also requires 1 pin per bit. Using a DAC is more expensive and requires more pins. The advantage of using delta sigma is that it will be very inexpensive and I'll get lots of outputs.