DAC Noise Simulation
I have to calculate amount of total noise which DAC will contribute to the TX chain. Would be enough to take NSD from datasheet and add 10log(bandwidth)? It's not have to be really accurate. What do you think?
Thanks in advance
Marcin
Assuming that the digital input signal (signal properly generated and conditioned) does not add more quantization noise than the thermal noise NSD at the output, then yes.
So you mean that NSD (datasheet parameter) describes only thermal noise? I thought that in terms of DAC it says about quantization noise and internal NF of DAC. Correct me if I am wrong.
Thanks
Well-ish, I expressed it perhaps a bit sloppy, what I meant was that as long as the digital signal is properly generated then it makes sense. Given that you generate a sinusoid through e.g. an N-point IFFT, the quantization noise of the DAC is essentially determined by the N tones occuring in the spectrum. If the signal is not properly defined then you might not capture the quantization noise properly.
For example, to go the extreme - assume you have a digital square wave as input - at a third of the sample frequency - the only tones occuring are the ones at multiples of the signal frequency, there are not necessarily a lot of other tones in the spectrum.
... something like that...
Thanks for answer, so as far as I understand, DAC noise depends on applied signal, am I right? But datasheet parameters are measured under specified conditions i.e. when it comes to NSD, input signal is sinusoid at specified frequency usually few dB under full scale. But still, its just kinda estimation. One more question is, can I calculate noise factor for transmitter the same as for receiver? I mean that I want to somehow estimate noise floor at the output of receiver.
Thanks in advance
Yes, noise depends on the signal, but still can be overlayered by a substantial thermal noise. Assume for example that the DAC is a current-steering one where you terminate the outputs on resistive loads. The resistors will give you noise, but so will also the current sources of the DAC. From a differential signalling point of view, this noise level is independent of the applied DAC code (assuming you use a differential DAC). Which one is dominating depends on the success of the design... (price of product).
I guess you can define NF. There is no definition of gain per se in the DAC as you just represent the digital signal with an analog signal. But I guess you would assume gain to be unity and then set the signal to peak amplitude and stay within a band where the signal power is not attenuated too much (i.e., low frequencies).
Thank you for your help. One more question is: do we usually take into account DAC/ADC when calculationg NF? Most examples do not consider them.