Here2Learn
Active Member
- Joined
- Jan 7, 2020
- Messages
- 112
- Likes
- 113
Hi all, I'm not an EE, I love audio and try to be as objective about it as possible. But not being an EE means one can fall for claims that ultimately one can find no evidence for and has just wasted time.
Many years ago, I intended learning some audio because you know, there's a website called DIY Audio, and one would think it's helpful. The problem is, it's full of as much subjectivism as the general hifi press. Sure, some know some electronic engineering principles, but whether these are mixed with 'beliefs' based on subjective sound preferences that are not congruent with neutral and transparent sound, is another matter.
So if trying to do targeted learning on a subject via DIY Audio, one can find themself more confused than at the outset. Which claims should be investigated and learnt about, and which are just 'beliefs' without supporting evidence? Well, in regards to I/V design and using opamps in their design, that's what I'd like to know.
Where am I? That intention of building a DAC many years ago transpires to mean I have 2 PCM1794 chips laying around, a Crystal SRC (8416 I belive) and probably need something like XMOS XU316 and some input multiplexing circuit, but not really interested in that aspect. I believe Twisted Pear Audio can help out as ar as input switching.
I know how to generally calculate gain with opamps. I don't know how to calculate the resulting bandwidth at a particular gain. Similarly, I know that in an inverting use-case, I'd usually need a capacitor in parallel with the feedback resistor to achieve stability, but I don't know how to calculate this. Additionally, I read that many, if not most opamps, require capacitance on the supply rails and that this is also calculated somehow.
Here's an example thread I read on DIY Audio that makes we question things: https://www.diyaudio.com/community/threads/ideal-i-v-converter-stage.203695/
Towards the end of that thread, some are saying the AD811 with its high slew rate of 2500V/uS makes it ideal for I/V tasks. But, as I understand it, we need any I/V opamp to perform linearly across the audio range (don't know if that means to 20kHz since we're dealing with analog, or if there is some reason for this to at least be the digital sample frequency) and linear performance requires that the available opamp bandwidth for the gain designed for, exceeds this. So for a PCM1794 to output via I/V stage to 4Vrms XLR requires a gain of 20, for which it appears the AD811 would be seriously short of suitable bandwidth. NOTE: I didn't calculate the resulting bandwidth for the AD811 at a gain of 20, but see from published specs, that its high bandwidth drops seriously sharply just at a gain of 2, let alone 20.
Also in the DIY Audio thread link above, this is mentioned: "The snag is that some opamps allegedly can't cope too well with the pulse transitions which accompany the audio" - what does this even mean?
The same post also says: "The ideal solution is probably an inverting opamp, but preceded by a diplexer to send the pulse transitions to a low value resistor. Alternatively a well-designed valve stage (grounded grid?) could cope with the pulses but may add a little second-order distortion." - what does such a diplexer look like? How is one designed for an I/V stage, or is this and "pulse transitions" just BS?
There are a few other longer-running threads on DIY Audio that are like this. I cannot help but perceive some of the claims may be rooted in objectivity and many of them will not be. Furthermore, as all commentary seems based on subjective preference of resulting sound from design types, I am not sure I can trust any of it for simply achieving the best measured performance.
In addition, I am wondering if a DIYer can ever get close to a state of the art I/V stage, assuming a 4Vrms output means ~5.66Vpeak, which to resolve the 22nd bit of DAC performance, must be able to accurately resolve 5.66/(2^22) = 0.00000135V. This seems a ridiculously tough ask, but SOTA measuring DACs (even the top Chi-Fi ones) are achieving this with ease it seems.
To me, requiring an I/V stage to be accurate to within 0.00000135V to resolve to 22 bits seems to require more than just an interest in opamp noise performance, but also its accuracy its peripheral circuit design to output the correct voltage, its stability to maintain it (when required), and how that also probably ties in with exceptionally ultra low noise on the opamp supply rails, which I assume must be equally as stable.
Obviously the last few paragraphs have reached in to my questioning musings, rather than being anything I actually know But in all the areas I've touched upon, what are the facts, what is the fiction, and how does one go about designing for 22-bit performance in an I/V stage? I initially thought DIYers make stuff comparable to SOTA units for less money, but given my illustration above, I am not so sure - it might just be an exercise in folly if one is not a seriously competent and trained/educated EE already.
Many years ago, I intended learning some audio because you know, there's a website called DIY Audio, and one would think it's helpful. The problem is, it's full of as much subjectivism as the general hifi press. Sure, some know some electronic engineering principles, but whether these are mixed with 'beliefs' based on subjective sound preferences that are not congruent with neutral and transparent sound, is another matter.
So if trying to do targeted learning on a subject via DIY Audio, one can find themself more confused than at the outset. Which claims should be investigated and learnt about, and which are just 'beliefs' without supporting evidence? Well, in regards to I/V design and using opamps in their design, that's what I'd like to know.
Where am I? That intention of building a DAC many years ago transpires to mean I have 2 PCM1794 chips laying around, a Crystal SRC (8416 I belive) and probably need something like XMOS XU316 and some input multiplexing circuit, but not really interested in that aspect. I believe Twisted Pear Audio can help out as ar as input switching.
I know how to generally calculate gain with opamps. I don't know how to calculate the resulting bandwidth at a particular gain. Similarly, I know that in an inverting use-case, I'd usually need a capacitor in parallel with the feedback resistor to achieve stability, but I don't know how to calculate this. Additionally, I read that many, if not most opamps, require capacitance on the supply rails and that this is also calculated somehow.
Here's an example thread I read on DIY Audio that makes we question things: https://www.diyaudio.com/community/threads/ideal-i-v-converter-stage.203695/
Towards the end of that thread, some are saying the AD811 with its high slew rate of 2500V/uS makes it ideal for I/V tasks. But, as I understand it, we need any I/V opamp to perform linearly across the audio range (don't know if that means to 20kHz since we're dealing with analog, or if there is some reason for this to at least be the digital sample frequency) and linear performance requires that the available opamp bandwidth for the gain designed for, exceeds this. So for a PCM1794 to output via I/V stage to 4Vrms XLR requires a gain of 20, for which it appears the AD811 would be seriously short of suitable bandwidth. NOTE: I didn't calculate the resulting bandwidth for the AD811 at a gain of 20, but see from published specs, that its high bandwidth drops seriously sharply just at a gain of 2, let alone 20.
Also in the DIY Audio thread link above, this is mentioned: "The snag is that some opamps allegedly can't cope too well with the pulse transitions which accompany the audio" - what does this even mean?
The same post also says: "The ideal solution is probably an inverting opamp, but preceded by a diplexer to send the pulse transitions to a low value resistor. Alternatively a well-designed valve stage (grounded grid?) could cope with the pulses but may add a little second-order distortion." - what does such a diplexer look like? How is one designed for an I/V stage, or is this and "pulse transitions" just BS?
There are a few other longer-running threads on DIY Audio that are like this. I cannot help but perceive some of the claims may be rooted in objectivity and many of them will not be. Furthermore, as all commentary seems based on subjective preference of resulting sound from design types, I am not sure I can trust any of it for simply achieving the best measured performance.
In addition, I am wondering if a DIYer can ever get close to a state of the art I/V stage, assuming a 4Vrms output means ~5.66Vpeak, which to resolve the 22nd bit of DAC performance, must be able to accurately resolve 5.66/(2^22) = 0.00000135V. This seems a ridiculously tough ask, but SOTA measuring DACs (even the top Chi-Fi ones) are achieving this with ease it seems.
To me, requiring an I/V stage to be accurate to within 0.00000135V to resolve to 22 bits seems to require more than just an interest in opamp noise performance, but also its accuracy its peripheral circuit design to output the correct voltage, its stability to maintain it (when required), and how that also probably ties in with exceptionally ultra low noise on the opamp supply rails, which I assume must be equally as stable.
Obviously the last few paragraphs have reached in to my questioning musings, rather than being anything I actually know But in all the areas I've touched upon, what are the facts, what is the fiction, and how does one go about designing for 22-bit performance in an I/V stage? I initially thought DIYers make stuff comparable to SOTA units for less money, but given my illustration above, I am not so sure - it might just be an exercise in folly if one is not a seriously competent and trained/educated EE already.
