That is not US9871530.Interesting patent. Northrop Grumman.
https://patentimages.storage.googleapis.com/e6/ef/7b/aa66c2a13634c3/US9871530.pdf
From the patent it looks like they are converting PCM to PDM (aka DSD of some type, maybe DSD-Wide or some variation of it that can be used for multi-path purposes?).
In any case the overall idea seems more plausible to me if using PDM rather than PCM.
Using PDM kind of also tends to blur the idea of mononicity, at least as it may be conceptualized at DC frequency (i'm thinking about, PDM is not exactly rendered as in PCM staircase models, rather it is an averaged value of an oversampled 1-bit step size).
In any case the overall idea seems more plausible to me if using PDM rather than PCM.
Using PDM kind of also tends to blur the idea of mononicity, at least as it may be conceptualized at DC frequency (i'm thinking about, PDM is not exactly rendered as in PCM staircase models, rather it is an averaged value of an oversampled 1-bit step size).
Last edited:
Unless the resistive divider has the accuracy of the lower DAC LSB, you get a glitch.
Sorry but I will bow out now, this is going nowhere.
Jan
Sorry but I will bow out now, this is going nowhere.
Jan
From the patent it looks like they are converting PCM to PDM (aka DSD of some type, maybe DSD-Wide?).
Is it really that hard to get?
You take your audio signal.
You take all data and send it to DAC#0
You take the same data and right shift it by n Bit (cutting off the MSB's) and you send this to DAC#1
Now DAC#1 will overflow with a full scale signal, DAC#0 will show poor low level resolution at levels where DAC#1 still has full resolution.
So data which does not overflow DAC#1 is send to DAC#1 and data that overflows DAC#1 goes to DAC#0.
All we now need to do (if the DAC's are current out into a virtual ground for simplicity) is to scale the output current of DAC#1 by a suitable factor. Easiest, the way Denon did it back in the day. Very low value precision resistor (say 1R 0.01%) as I/U conversion and then either a suitable series resistor (chain) or a T-attenuator that results in the correct current into the I/U converter node. Even with 0.1% resistors we can probably do better than +/-0.5LSB by design.
Then the entire process is purely digital.
You send silence to whichever DAC is idle and you send send the signal to whichever DAC needs to process it. Extremely trivial and simple. Any FPGA/CPLD in the market can handle this.
Analogue is a simple mixer with one signal attenuated according to requirements.
If there is actually REAL WORLD value to this IN AUDIO compared to (say) using a better DAC instead is another question and if going past 18 Bit ENOB is a valid design goal yet another.
But if anyone wants to replicate this, trivial. Really, really, really trivial.
Thor
Last edited:
They filed a number of patents and applications. WO2016118674A1 is also worth looking at. Actual DAC conversion technology is not relevant for the idea, I'd say.From the patent it looks like they are converting PCM to PDM (aka DSD of some type, maybe DSD-Wide or some variation of it that can be used for multi-path purposes?).
The Audio DAC presented here is most certainly based on standard DAC chips (PCM). It won't take too long until someone posts internal pics...
Unless the resistive divider has the accuracy of the lower DAC LSB, you get a glitch.
Sorry but I will bow out now, this is going nowhere.
Actually, it is the UPPER DAC used LSB. But yes, precision is needed. Is that really such a big deal in 2025 with 0.01% tolerance resistors commodity items?
In the days of carbon film 5% resistors of course, this is like "g*d's domain".
Thor
The patent bohrok2610 linked to goes into some detail about using PDM. In fact, PDM is part of the title. Its for A/D conversion, but it talks about processing PDM streams for muti-path use. Thus maybe there is some connection with how they might process for D/A conversion? (After all, ADCs are always studied first, as they contain an internal DAC anyway.)Actual DAC conversion technology is not relevant for the idea, I'd say.
https://patentimages.storage.googleapis.com/e6/ef/7b/aa66c2a13634c3/US9871530.pdf
Last edited:
Please don't. Take your time to read, for example, WO2016118674A1 to see how it works. Then you will understand that absolute ratio is not important, only stability of the resistive mixer and because of the monitoring and calibration things do actually work glitch-free.Unless the resistive divider has the accuracy of the lower DAC LSB, you get a glitch.
Sorry but I will bow out now, this is going nowhere.
The DAC is a real product and the Mr. L Grou is a serious designer. This thing works as advertised, I'm 100% sure. This is pro stuff, not audiophile BS.
I've been studying this DAC two years ago when it became public. I know exactly how it works.
'Dynamic calibration' is just another word for 'generating glitches'. If you don't see that, well, what can I say?
Jan
'Dynamic calibration' is just another word for 'generating glitches'. If you don't see that, well, what can I say?
Jan
Sure, but that does not mean the actual DAC we're discussion here is NOT based on standard DAC chips.The patent bohrok2610 linked to goes into some detail about using PDM. In fact, PDM is part of the title.
https://patentimages.storage.googleapis.com/e6/ef/7b/aa66c2a13634c3/US9871530.pdf
Thus maybe there is some connection with how they might process for D/A conversion?
Well, PDM has limited resolution. So it needs most help. But for the basic principle you do not even need to do this for digital audio.
It's kinda of loosely related to the old analogue compander system we messed in the 80's, dbx, UC, DOLBY A etc.
That's essentially it.
Thor
Thought I saw in that patent that they produce a high resolution PDM stream?
Anyway, most modern ADCs actually produce RAW output which is oversampled PCM-Narrow or DSD-Wide (depending in part on how you think about it).
Same type of thing occurs in a dac chip. PCM or DSD is converted to 5-bits or so and oversampled, although AKM seems to do it somewhat differently from ESS.
Question I am wondering about it that it seems like calibration of the two dacs and or other DSP processing might be in some way advantageous in PDM format? At least, it looks like something they did a lot of thinking about for the ADC problem.
Anyway, most modern ADCs actually produce RAW output which is oversampled PCM-Narrow or DSD-Wide (depending in part on how you think about it).
Same type of thing occurs in a dac chip. PCM or DSD is converted to 5-bits or so and oversampled, although AKM seems to do it somewhat differently from ESS.
Question I am wondering about it that it seems like calibration of the two dacs and or other DSP processing might be in some way advantageous in PDM format? At least, it looks like something they did a lot of thinking about for the ADC problem.
Last edited:
If there are really issues with audible glitches, they won't succeed. Time will tell.I've been studying this DAC two years ago when it became public. I know exactly how it works.
' Dynamic calibration' is just another word for 'generating glitches'. If you don't see that, well, what can I say?
Thought I saw in that patent that they produce a high resolution PDM stream?
This patent does not directly apply to the Benchmark DAC. There is, as best as I can find no patent or patent application for this DAC (it would fail on "obviousness" and abundant prior art anyway).
The patent illustrates the principle.
Anyway, most modern ADCs actually produce RAW output which is oversampled PCM-Narrow or DSD-Wide (depending in part on how you think about it).
Even better, some output PCM-WIDE-FAST (20 Bit at 1 Megasample per second). But nobody wants such an ADC. I wonder why?
Same type of thing occurs in a dac chip. PCM or DSD is converted to 5-bits or so and oversampled, although AKM seems to do it somewhat differently from ESS.
But as said, the process is fundamentally in the analogue domain. As said, the analogue domain relations are 1980's companders for magnetic tape and LP recordings.
I've been studying this DAC two years ago when it became public. I know exactly how it works.
'Dynamic calibration' is just another word for 'generating glitches'. If you don't see that, well, what can I say?
Maybe. Seeing the attenuation is only 28dB and the DAC's used are some form of bitstream I don't see the great big problem. I suspect that the "dynamic calibration" simply compensates for component tolerances in lieu of a trimmer adjusted manually by an employee.
It adds to level of the Magical Mystery Tour for the audiophile who understand Bahnhof.
It gives the "journalists" something they can copy & pasta to fill out their "reviews" as we can read.
And makes it simpler in production and apparently harder to copy by the proverbial Schenese Kopy Kat:
Thor
Huh? How did the Benchmark DAC get into this?This patent does not directly apply to the Benchmark DAC.
Its an ADC patent by La Grou
https://patentimages.storage.googleapis.com/e6/ef/7b/aa66c2a13634c3/US9871530.pdf
Huh? How did the Benchmark DAC get into this?
My bad, different Pro/Fi crossover company, Millennia Media. My mind ain't what it used to be... Never was, coming to think of it.
Appy polly logies for the brainfade.
And I was referring to the Northrop Grumman Patent. I'll have to read that other patent very carefully, as clearly the 20 Years old expired Northrop Grumman Patent @KSTR linked discloses the same principles.
As does this:
https://patents.google.com/patent/US20070120721A1/en
At some point in the future, the professional audio industry will realize an entire 170dB recording ecosystem, from microphone to power amplifier, and everything in-between
I still cannot stop laughing.
A 170dB dynamic range microphone, presumably without Brownian motion noise from the air on the diaphragm because we record in a vacuum and an electronic noise level of -20dB SPL as we have the whole thing cooled to near absolute zero and with an SPL handling of 150dB SPL (poor musicians to play these instruments... Instantly deaf).
Seriously.
Thor
Last edited:
Maybe 140dB at best.A 170dB dynamic range microphone...
https://www.grasacoustics.com/microphone-guide/dynamic-range
Maybe 140dB at best.
And that only if we get robots to play the instruments. The better LDC's (large diaphragm condenser) will have around 10dBA (SPL) self noise.
This is a mix of pink noise, falling slope with frequency, caused by the interaction of circuit resistances (commonly 1GOhm+) and the capsule capacitance (~ 60pF) and Violet noise, rising slope caused caused by the brownian motion of air molecules. Intersection between the two often happens in region where the ear is most sensitive, so around 2kHz. At higher and lower frequencies noise rises, but for measured (and audible) noise A-weighting/hearing sensitivity change with frequency reduces this.
A lot of instruments get pretty loud close-up, some enough for hearing damage for musicians. But unless you stick the microphone into a kickdrum or inches from the bell of trumpet seeing more than 120dB peaks is rare. Realistically, 110dB is rare. Many recoding Microphones can take 135dB and that is generally my design target for my own microphones, but I never managed to run into these limits, not even with an LDC directly in the hole of rear skin opposite of the beater at a kick drum. Mind you, I never worked with "Animal".
So presuming we have a listening room with ~ 10dB background noise and a system that can play back at 110dB Peak SPL we can do any acoustical recording of civilised music totally uncompressed full justice with a usable/used dynamic range of 100dB. But that's shaving it close. Still, 16 Bit get's us most of the way there.
So gimme 18 good bit's. Real multibit ones and recordings sampled at 64kHz or more. I'll pass on the 270 Bit 10GHz DAC.
Thor
You could not fit that preposterous amount of dynamic range into any living room. The bottom end would be below audibility, and the upper end would shatter the windows and the listener's eardrums. It vastly more than needed for any and all home sound reproduction.
- Home
- Member Areas
- The Lounge
- 27bit DAC -> 162 dB dynamics...