You know that's not how ADCs actually work right. I mean ladder ADCs/DACs exist, but things have moved on since the 1970s.
If your 3bit example. The bottom D1 is comparitor'ing on 0.5V. Thus 1 bit = 0.5V resolution.
What is it comparator'ing on if we make VREF 40mV or 10mV?
If your 3bit example. The bottom D1 is comparitor'ing on 0.5V. Thus 1 bit = 0.5V resolution.
What is it comparator'ing on if we make VREF 40mV or 10mV?
If you are a software engineer and you don't understand what bits are, the difference between magnitude and precision, you are in big trouble or a python developer/script kiddie with no formal education in the topic.
Look reading back through this, the reason it derailed is because you all jumped on the "characterising the DAC".
I'm not. I'm measure the noise on the circuit in test. Just the noise. No signal. The DAC is not the only thing in the circuit. In the image capture posted there is quite a lot of other stuff in that circuit.
Subjectively it sounded quiet to me. But I thought I would see what I can see on the scope. -100dm with a few spikes up to -59 won't win prizes, but it's quiet by my standards, very quiet.
Given I'm not measuring the DAC's signal to noise ratio or it's dynamic range and... 100% confirmed it is playing a stream of 0s, the dynamic range of the DAC is completely irrelevant as it's currently operating at 0 bit depth.
Also, what you would be missing with you cheap little sound card is the real issue / elephant in the room, that would cause this circuit to at very least fail it's RF emissions test. It may even cause it to fail in service or the components within it fail early.
Clock noise with over/undershoot, inductive ringing and a full voltage swing of well over VDD in the circuit.
Your sound card would not be able to dial in onto that noise (in the 3 to 60Mhz range) and see that. In fact the total Vpp of the digital signals is only slightly above where I would like it and that is because the other component you all forgot was there, the Optical SPDIF/ interfaceboard generating the digital I2S stream.... has actually bothered to put a bus transceiver on it's outputs with suitible termination resistors to stop the ringing. The DAC (and of course the breadboard) have no such protection, so it's likely a lot of the clock noise is radiant from the breadboard and the jumpers as would be expected.
Give all of that, I am in fact using the correct tool for the job and seeing far, far more relevant to the circuit information that a PC sound chip will.
I'm not. I'm measure the noise on the circuit in test. Just the noise. No signal. The DAC is not the only thing in the circuit. In the image capture posted there is quite a lot of other stuff in that circuit.
Subjectively it sounded quiet to me. But I thought I would see what I can see on the scope. -100dm with a few spikes up to -59 won't win prizes, but it's quiet by my standards, very quiet.
Given I'm not measuring the DAC's signal to noise ratio or it's dynamic range and... 100% confirmed it is playing a stream of 0s, the dynamic range of the DAC is completely irrelevant as it's currently operating at 0 bit depth.
Also, what you would be missing with you cheap little sound card is the real issue / elephant in the room, that would cause this circuit to at very least fail it's RF emissions test. It may even cause it to fail in service or the components within it fail early.
Clock noise with over/undershoot, inductive ringing and a full voltage swing of well over VDD in the circuit.
Your sound card would not be able to dial in onto that noise (in the 3 to 60Mhz range) and see that. In fact the total Vpp of the digital signals is only slightly above where I would like it and that is because the other component you all forgot was there, the Optical SPDIF/ interfaceboard generating the digital I2S stream.... has actually bothered to put a bus transceiver on it's outputs with suitible termination resistors to stop the ringing. The DAC (and of course the breadboard) have no such protection, so it's likely a lot of the clock noise is radiant from the breadboard and the jumpers as would be expected.
Give all of that, I am in fact using the correct tool for the job and seeing far, far more relevant to the circuit information that a PC sound chip will.
Point is, you can’t switch ADC resolution above hardware determined using just software.
You are also ignoring elephant in the room (noise voltage levels and oscilloscope input sensitivity).
Smarter than me members determined that we are going around in circles here.
Have a nice day. 😉
You are also ignoring elephant in the room (noise voltage levels and oscilloscope input sensitivity).
Smarter than me members determined that we are going around in circles here.
Have a nice day. 😉
Ah... you have never used a scope before have you.
The scope has hardware the £3 sound card does not. Funny that.
So i have a 1104x-e, used the FFT and have a 24bit (32bit) ADC. And i have a software engineering degree.
The 1104 will give you an 8bit dynamic range and i assume you’re using the front end scaling to place that 8bit range & scale to the noise floor.
You may detect noise but quantifying it (for many reasons) is likely to be difficult.
The 1104 may be a scope but there’s room for improvement - everything from internal noise to timing noise to sample rate to linearity at that front end gain across the spectrum.
Once you get into low voltage, the noise of everything starts messing with the result..
Last edited:
I was taught that, because there is a sign bit, 16-bit give 90dB dynamic range, not 96. Can anyone correct me? (It was a long time ago!)