New 32-bit ADC IC for high resolution Hi-Fi audio playback | Hacker News Reader

New 32-bit ADC IC for high resolution Hi-Fi audio playback

allaboutcircuits.com

10 pointsteleforce8 comments

32 bits of DA/AD is total BS.

If you really had 32 bits of resolution, the lower bits would be representing Brownian motion (sometimes called "thermal noise").

The specs make it clear it isn't a 32 bit DAC:

"At the same time, class-leading low noise and distortion (130dB SN ratio, -115dB THD+N) provide the high performance demanded by high fidelity audio equipment."

In other words, completely normal converter specifications.

If the 32 bits means anything in this context, it's just a sample packaging format, the same as more or less any other DA/AD circuit in general use.

TFA is also more or less 100% BS. Look at this garbage:

"The BD34301EKV is the newest Hi-Fi 32-bit DAC converter from the ROHM MUS-IC product family. This converter has been selected by the Luxman Corporation, a high-end audio equipment manufacturer because it is said to have a “sound that is natural, and easy to listen to."

Why is this even on HN?

This chip has ~22 bits of dynamic range. It's somewhat common to have 32-bit word sizes on ADCs/DACs like this that start to approach 24 bits, mostly as a convenience, and to make sure nobody accidentally ends up clipping in the middle of a filter MAC stage somewhere, but inevitably marketing decides to plaster "32 bits!" all over the datasheet.

I work with an ADC designed for seismic applications that claims similar specs and also has a 32-bit word size, and even those specs (~23 bit DR) are pretty difficult to achieve in practice.

I'm somewhat amused to see this in an audio DAC though, since the power amplifier and speakers are going to be way less linear than anything you can push through this thing. But I guess audiophiles gonna audiophile...

Don't forgot room dynamics - you push a 5.1 (and all the weird speaker factors) and then your corners are muddy yadda yadda.

Yeah it reads like marketing piece

So they're celebrating almost catching AKM.

https://www.akm.com/eu/en/products/audio/audio-dac/ak4493eq/

The power tree issues the article cites are either:

1) a little overblown, or

2) indicative of some issues with the ROHM DAC.

A noise optimized LDO as a reference for the DAC Vref point is typically sufficient. You have more to worry about in terms of other external aggressors than direct coupling into the power rail through some capacitor impedance or LDO impedance across temperature mumbo jumbo. (Source: me, who designed a consumer product with the AK4490, the 4493's precursor part.)

Wonder if they're pin compatible to the AK4490. That would be worth something to Rohm. The AKM fab that caught fire last month was the one that made all of their hifi audio parts. The only one, as a matter of fact.

Boutique audio shops the world over are likely freaking the fuck out about that AKM fire.

It's DAC, not ADC. Signal flows in opposite direction

One issue in hi-fi audio is the surprising lack of good source material given the passage of time.

For example - 20 years ago it feels like you could by a SACD with some classical music on it - and the quality was pretty darn impressive. I remember multi-channel audio in these.

Despite 20 years of tech - when I try to stream or even buy some multi-channel audio for my HiFi rig it's basically super annoying as far as I can tell. No clear disclosure of actual channels / format etc.

Am I missing something? SACD had 5,000+ titles out, you could just grab one and it worked.

What's the current approach here for high quality multi-channel audio?

My home setup is Klipsch 5.1 if that matters, but I'm happy if things play 3.1 if needed.

They claim "3.16 μV of noise on a 1 Vrms signal". So then that's lg2(1/3.16e-6) = 18.3 bits beyond which the rest is noise.

That is not even near top of the line DACs you can get elsewhere, which, for serious $$$, can approach 23 bits

Reply on news.ycombinator.com