If you want to make this even remotely interesting you should post actual recordings of the same device before the part change and after the part change. And accompanied by measurements before and after the part change.
Without looking at the proposed layout by ESS, It seems that all you need in front of the chip is a buffer and a single-ended to differential stage (?).
With optional gain if needed, although you can always digitally manipulate gain in Audacity or other editors, after recording.
Main critical locations would be the opamps of course, AC-coupling capacitors (i.e. input-coupling) and decoupling capacitors around ICs.
And dare I say resistor types, in the analog section.
If you provision the use of through-hole parts in those positions I'm interested!
Power supply is also critical (especially for the analog part) but you already know that I'm sure.
Last edited:
For now, you are free to listen to the ones I posted.
And draw your own conclusions.
And if I see there is genuine interest in what you asked for, I might post that as well.
Provided that you can read a circuit schematic.
Last edited:
Coming from somebody who thinks diy is about changing parts these types of ad hominem attacks are hilarious
I asked genuinely and politely.
You are on the attack and i see no reason mate.
You have your convictions, that is fine.
I am not the one trying to push or prove something.
I know what I prefer, using my own set of ears. That is all it matters to me.
I will never push hard for others to agree with me.
Everyone has different expectations and sense things differently.
So keep that in mind too.
You ask for data, I do not have it.
If you are interested feel free to listen to what i offer, if not move along.
It's that simple.
You are on the attack and i see no reason mate.
You have your convictions, that is fine.
I am not the one trying to push or prove something.
I know what I prefer, using my own set of ears. That is all it matters to me.
I will never push hard for others to agree with me.
Everyone has different expectations and sense things differently.
So keep that in mind too.
You ask for data, I do not have it.
If you are interested feel free to listen to what i offer, if not move along.
It's that simple.
Last edited:
So asking for measurements is now considered an attack?
You have been pushing your view in this thread from page 1. For many of us tweaking existing devices and changing parts is not what diy is about.
You have been pushing your view in this thread from page 1. For many of us tweaking existing devices and changing parts is not what diy is about.
As I said politely in my previous post:
I posted actual recordings for folks that are curious and may be interested in how two devices sound.
Not why, but how different they sound.
And they decide on what they prefer or not.
So i am not pushing anything mate.
If anything, I'm stating my humble opinion.
And you are free to disagree.
I posted actual recordings for folks that are curious and may be interested in how two devices sound.
Not why, but how different they sound.
And they decide on what they prefer or not.
So i am not pushing anything mate.
If anything, I'm stating my humble opinion.
And you are free to disagree.
Last edited:
I think this comes across as an insult, which might not be what you intended, but resulted in insults coming back. It would be nice if you could both decide to stop the argument.
@evonimos - You are exactly right - it just has a buffer & SE to diff converter. Would it be sufficient to just have a header so you can build your own input stage on a breadboard and feed it directly to the chip?
Regarding the power supply, yes I know it is very important. ESS recommend their own special regulator chip and IIRC it has >90dB PSRR for the entire audio band if you input 5V. I was going to feed that from the RPi's power.
I do plan on measuring the actual performance myself, although if I manage to replicate the datasheet performance (unlikely, tbh) then that might surpass the capability of my equipment (which is also DIY).
Reaching datasheet performance with ES9822PRO is not impossible but there is some chip-to-chip variance.
The real THD+N is reachable down to 6-7db lower than the datasheet promises. ESS used usual AP analyzer which is a well-known noisy thing. If you'll test well-designed and THD compensated ADC ES9822 with the low-noise&distortions sine source(DAC+LPF is a good choice), some chips will be able to show THD+N@1kHz -124-125db in the MONO mode. The average ES9822 will be about -123db.
PS: if someone wanna make the same high-performance ADC with TH parts instead of SMD, I bet it will be 100.00% fail.
No. Why can't you seem to resist including ad hominem bits mixed in with your more respectable engineering comments?
When people use 'we' or 'all' in place of 'I' or 'myself' in their arguments, I wonder why they do it.
Every ADC and or DAC needs a time-reference and a voltage-reference (presumably could be a current-reference instead). So don't forget the clocking design including their power if you want good sound. Since typical audio FFTs and human operators don't usually look much for clock close-in phase noise, Vref noise and distortion (including linear distortion), signal-correlated noise modulation, RF pollution of analog signals and their side effects, non-time invariance, etc., its almost as though such things can't exist. Its almost always just steady-state noise and distortion as seen on an AP, or equivalent, as though that's all there is. Kahneman described that type of cognitive bias as WYSIATI. https://facilethings.com/blog/en/what-you-see-is-all-there-is
Last edited:
All these things affect the THD+N directly with no voodoo involved 😉
Possibly, and or maybe to some limited extent. There is no guarantee any useful insight will be developed into the actual time-domain behavior of a circuit, its audibility, and or as to causation and correction if needed. If we measure to satisfy an AP machine, we end up with a device that is good for satisfying an AP machine. That's the basic result. If the goal is to design a substitute AP type measurement device, then maybe all is good. If the goal is to design something that satisfies a desire for convincing music reproduction then maybe things are not so good. Again, two AP spectra can look identical and sound obviously different to anyone. Its not enough in the way of measurement to guarantee human satisfaction with reproduced SQ. https://purifi-audio.com/2019/12/07/amfm/
There is much more that could be said on this subject if anyone is interested. If so, I would be happy to go into more detail. Otherwise don't want to bore people who have already read what I have written about on some of the issues.
Last edited:
if voodoo is involved, you don't need any tech reasons like Vref noise and distortion (including linear distortion), signal-correlated noise modulation, RF pollution of analog signals. If you are coming to the tech side you need to measure all these tech things and fix them. The only two approaches are there voodoo and tech, no mixed arts.
Voodoo refers to magic. There is no true magic in this world. Listening, particularly skilled-listening, is a type of measurement. In some ways it can be more useful than steady-state FFT measurements, because skilled listening can be more sensitive to short-term dynamic distortions and signal-correlated noise in a way that FFTs are not especially useful for. That's what ESS says about things like state variable settling, you saying their engineers are telling voodoo stories? Steady state measurements tend to make all distortions and noise look steady-state, even if that is far from reality. How about call that illusory effect 'voodoo' if you like.
Regarding finding problems and fixing them, nothing is perfectly fixed. There is no perfect minus infinity distortion and noise. If you want to post numbers for THD+N then post all the measurements so people can understand what is or isn't fixed enough for them. We have known for many decades that THD+N is a pretty useless number as a proxy for perceived SQ.
Regarding finding problems and fixing them, nothing is perfectly fixed. There is no perfect minus infinity distortion and noise. If you want to post numbers for THD+N then post all the measurements so people can understand what is or isn't fixed enough for them. We have known for many decades that THD+N is a pretty useless number as a proxy for perceived SQ.
Last edited:
I agree that THD+N is only a narrow technical view. There are some types of inaccuracies that simply comparing output of a fixed signal to what was input will not detect. It's sort of like using a multimeter to measure an AC signal. But I suspect @IVX already knows this.
The problem is that we don't have a standard method to characterise the dynamic components of the response, so that they can be measured and fixed. Some people just use their ears - it's unscientific but it's better than nothing.
Listening to music is a personal experience, that involves a plethora of emotional responses.
So, I'm assuming that it's near impossible to frame it (the experience), in an acceptable quantitative manner by measurements (using instruments).
I mean, you can measure the performance of a device, but how can you measure its effect on an emotional experience?
So, I'm assuming that it's near impossible to frame it (the experience), in an acceptable quantitative manner by measurements (using instruments).
I mean, you can measure the performance of a device, but how can you measure its effect on an emotional experience?
Actually, more than that, it's an experience.
And I dare to assume that what makes it valuable for the individual, is the fact that listening to music itself is an entirely personal experience in its essence.