Watts Up...?

[kennyb123]

First thing I noticed was that PGGB sounds really soft and damped, a bit like DSD.

You are doing something very wrong if this is what you hear. No way you did a fair comparison.

I think the most impressive thing I hear from PGGB is how well it salvages transients from DSD. That softness vanishes in most cases to where it’s impossible to tell that the source was DSD. Even the DSD256 material I purchase from High Definition Tape Transfer gains better transients.

A 16fs pggb file played through a Chord Poly/Mojo2 sounds fantastic for on the go.

Yes it sure does.

 

[thePhones]

You are doing something very wrong if this is what you hear. No way you did a fair comparison.

That's what I heard, as I said I used the provided settings (as shown) and JRiver with bit perfect settings. Hugo 2 was showing the same colors when using M Scaler. The original file was also played using JRiver with the same settings (then showing red for 44.1) . Expectation bias was the only thing imo that could have happened, except I was expecting hard sound like HQ Player not soft damped sound. But I am happy to listen to your descriptions as well.

I think the most impressive thing I hear from PGGB is how well it salvages transients from DSD

I can't speak to that as I have not heard a DSD converted file but I think I remember Rob saying somewhere that this is technically impossible. *edit; he said: "The benefit of DSD is that you don't need a sync interpolation filter because it's just the original data." Maybe your listening experience is due to the DSD noise no longer being properly filtered as the Chord dac does no longer treat the pcm data stream with the heavy DSD filtering and therefor you have harder/brighter sound? (just a wild guess)

 

[alphaman]

John Atkinson, way back in 2002, noted the DAC64's FIFO RAM buffer. How has this buffer changed or improved in the past 21 years?

 

[ecwl]

John Atkinson, way back in 2002, noted the DAC64's FIFO RAM buffer. How has this buffer changed or improved in the past 21 years?

https://www.head-fi.org/threads/chord-electronics-hugo-2-the-official-thread.831345/post-16844537

I believe this older post would answer your question.

 

[kennyb123]

Maybe your listening experience is due to the DSD noise no longer being properly filtered as the Chord dac does no longer treat the pcm data stream with the heavy DSD filtering and therefor you have harder/brighter sound? (just a wild guess)

Your assessment was arrived at in a rushed fashion with only a few tracks processed. You then rushed right out to report your negative findings without asking any questions about why you might have arrived with outlier findings.

I’ve been using PGGB since the beginning and was one of the beta listeners. I’ve likely done hundreds or maybe thousands of comparisons across all types of music with three different DACs (and three high-end music servers), including one DAC that’s not manufactured by Chord. I have processed hundreds of albums some multiple times. My system and my room have been optimized around time domain performance as I’m a drummer and the realistic reproduction of drums and percussion is of utmost importance to me. Most of my listening tests had me laser focused on how well ADSR (attack, decay, sustain, release) were handled. Much of the feedback I provided about PGGB was presented in terms of my impressions of how well percussive instruments (including piano) were made to sound more like the real thing in terms of the reproduction of ADSR. Often other beta listeners heard things the same way I did, and I think most of them have better systems than me. Friends who dropped by to listen would report hearing the same kinds of improvements I heard completely unsolicited by me so I could test how well I was keeping my biases in check. As I improved my system, every positive improvement I heard from PGGB became even more pronounced and even easier to hear. I’ve processed over 450 of my own albums but still often listen to music from Qobuz. The drop in sound quality with steaming is significant despite me using HQP for PCM scaling so rarely a week goes by without me hearing the difference PGGB makes. My subscription with Qobuz is the top tier as it allows me to buy music at a discount. If I like the music, I will purchase it and then process it. I have hundreds of albums that were ripped from SACD but struggled with enjoying much of it because of the softness. That my system has been optimized around time domain performance has the downside of making the harm that DSD does even more apparent. PGGB is a miracle worker on this material. DSD takes much longer to process than PCM, but this processing is a must. The improvements PGGB brings to DSD go beyond just removing the softness. PGGB makes the music sound more natural and more lifelike. It does all the things my MScaler used to do but the improvement is much more profound and no fixer boxes are required.

Long winded way me say that I have had thousands of of hours of experience with PGGB and my impressions have been repeatedly validated by other listeners, some of whom have excellent listening skills and remarkable systems. My findings aren’t an outlier. Those with DACs that don’t stomp on what PGGB can do and who have given PGGB a truly fair evaluation have reported exactly the benefits I’ve reported. And, like me, they have also spent their hard earned cash on a PGGB license and they’ve dedicated countless hours to processing their music and countless more being blow away by how much PGGB has improved their musical enjoyment. I think most of the beta listeners including me also owned MScalers at one time and it was our favorable impression of it that got us interested in PGGB. I think we each owned Chord DACs too and some still do because PGGB makes them really shine.

I think Occam’s razor might tell us that the cause of why you and I have completely different impressions isn’t likely to be found with me or my system or my approach.

P.S. The other beta testers and the PGGB developer are some the most honorable and fair men I’ve ever gotten to known. They are high character individuals who have each garnered respect online for their informed view, keen insights and eagerness to help others gain more enjoyment from their audio systems.

 

[audionewbi]

I can't wait for the release of Hugo 3.

 

[alphaman]

https://www.head-fi.org/threads/chord-electronics-hugo-2-the-official-thread.831345/post-16844537

I believe this older post would answer your question.

Well, yes and no.
Not sure why RW reported : "If you wanted low latency, then you had to use an analogue PLL. These devices create huge problems, which hugely affect SQ and the measurements. So I decided to replace the PLL with my DPLL - this took 6 years to perfect - but I ended up with the DPLL sounding identical to the RAM buffer. "
Reason being that canned DPLL solutions for doing DPLL FIFO existed back then. Eg. DIR1701, etc.
https://www.alldatasheet.com/datasheet-pdf/pdf/92668/BURR-BROWN/DIR1701.html

More important is JA's 2002 comment about the switchable RAM FIFO feature of the DAC64.
https://www.stereophile.com/content/chord-electronics-dac64-da-processor-page-2
Notably:

There was no doubt in my mind that, without the RAM buffer, the DAC64 with CD sources was fairly ordinary. Tonally it sounded a bit dry, soundstage-wise a bit shallow. Switching the buffer in-circuit both sweetened the tonal balance and made the perceived stage deeper. (All my following comments refer to the sound with the RAM buffer engaged.)

Nota bene: This is the John Atkinson, editor of Stereophile, who has heard and measured the "best stuff" avail. Even in 2002. Hint, hint. Wink, wink

 

[Jawed]

Those with DACs that don’t stomp on what PGGB can do

Which DACs stomp on what it can do?

 

[Reactcore]

I have processed hundreds of albums some multiple times.

I think it is a clever way of upscaling and salut its maker and users. I was tempted by its lower cost and maybe possible advances over the Mscaler.

But to me its lower cost in money is vaporised by the cost of my already so few time and hassle of big file sizes, not to mention the times i would want to redo the processing after i found a new setting or getting a software update with a new better algorithm.

The possibilly of using Mscaler live with just a second delay is invaluable to me.

Maybe if or when PGGB utilizes hundreds or thousands of parallel CPU cores of say a Nvidia videocard and become able to stream live it will appear on my radar again.

But as i understood from reading in to things going live brings all kinds of challenges like constant power consumption and cost, swithing noise that must be filtered from entering the DAC and offcourse getting all those cores to work as one without too much jitter and delay.

For now im a happy camper using my time listening to my music waiting for what is to come in WTA scalers and software solutions.

 

[ecwl]

Long winded way me say that I have had thousands of of hours of experience with PGGB and my impressions have been repeatedly validated by other listeners, some of whom have excellent listening skills and remarkable systems.

P.S. The other beta testers and the PGGB developer are some the most honorable and fair men I’ve ever gotten to known. They are high character individuals who have each garnered respect online for their informed view, keen insights and eagerness to help others gain more enjoyment from their audio systems.

I think this is a great summary of PGGB vs M-Scaler

PGGB is developed by a programmer who had a lot of loyal testers/customers who helped tune the sound for the optimal filter that are enjoyed by the customers.
M-Scaler is developed by Rob Watts who also tuned the sound himself based on repeated listening tests.

I can say I only tried the original version PGGB after it was tuned by the beta testers and I prefer M-Scaler and solo DAVE over PGGB. Even back then, some (or many?) people preferred PGGB over M-Scaler. Since I mostly stream, I’ve decided for now not to retest the latest version of PGGB as I have always preferred Rob Watts’ engineering design AND his digital filter sound. But I respect others who prefer otherwise.

I really don’t think Rob Watts has to justify or explain why he does what he does or detail the rationale behind his choices in detail. Most audio companies don’t. And to what end? As I have said here before, Rob Watts mentioned he noise shaped the M-Scaler 24/768 output here and then 1-2 months later, HQPlayer adds noise shaping (LNS15) to their program. I don‘t think customers (and non-customers) are entitled to the design details to the products they’re using.

As for the people who say, well, Rob Watts says this so therefore, we are entitled to challenge him. Hmmm. Technically, I guess? But there are so many products with many marketing and engineering claims, to devote one’s focus just to challenge Rob Watts claims seem a bit hypocritical, when one can ask similar probing questions to one’s favorite product designers.

At the end of the day, I think people should just use the audio gear that they enjoy the sound of. And listen to more music.

 

[thePhones]

Your assessment was arrived at in a rushed fashion with only a few tracks processed. You then rushed right out to report your negative findings without asking any questions about why you might have arrived with outlier findings.

It was just really easy for me to hear those differences. More tracks would have helped if I was unsure about what I am hearing, but I wasn't. I used two tracks I am absolutely in love with, listened to it back and forth and it was easy for me to hear what sounds different. I also don't care about being an outlier with my findings, I also was when MQA was riding the hype train. Tidal was out of the game for me on the first day I heard the damaging effects of MQA. Sorry if I am like an elephant in a china store, I am just voicing my opinions in a "North of Germany"-way
It is totally fine with me if you enjoy a different kind of sound or even feel differently about what one should hear.
Maybe you took my initial post a bit too personal as I was making fun of the program. I like to have a bit of fun teasing, so your reaction was well earned. It's all fun and games for me, no (real) harm intended. Please, we take this hobby already way too seriously.
Enjoy your music!

 

[flyte3333]

except I was expecting hard sound like HQ Player

Got a screenshot of your HQP settings that sounded "hard" ?

I have it and have a Hugo2

 

[Jawed]

Maybe if or when PGGB utilizes hundreds or thousands of parallel CPU cores of say a Nvidia videocard and become able to stream live it will appear on my radar again.

I think there was a time when it used GPU acceleration.

To gain access to 128-bit or 256-bit precision, PGGB is using software libraries produced by other companies, which can only run on a CPU.

The beauty of an FPGA is that it's way faster than a GPU for a fixed algorithm at vastly lower power consumption and cost.

I hope, one day, Rob will take seriously the idea of using the FFT algorithm for convolution. It can be done in real time. For a million taps it should be possible in real time using the FPGA in Mojo 2, since FFT will be at least 1000x less computational work than the direct convolution that Rob does.

 

[KMann]

I hope, one day, Rob will take seriously the idea of using the FFT algorithm for convolution. It can be done in real time. For a million taps it should be possible in real time using the FPGA in Mojo 2, since FFT will be at least 1000x less computational work than the direct convolution that Rob does.

But you assume the issue is one of speed and not space. FFTs will help speed for sure, but will not require any less space, if anything it would require more space. Using FFTs also come with it's own set of challenges, fixed point FFT algorithms need to be implemented and OA or OS algorithm would be required (which will add to the memory requirements).

Another issue that will arise with more taps that FFTs cannot help solve even if they are faster is the latency. The latency that will increase with longer filters to a point where it may become unacceptable. I don't know where one would draw the line for latency, a 16M tap Scaler would likely have a latency of 11 seconds, which may be acceptable for music listening if you are a album listener like me and do not keep changing tracks.

While latency issue can be solved in software using partitioned convolution or zero latency convolution, these methods are even more complex and computationally intensive and challenging for RT streaming.

 

[DarginMahkum]

I think there was a time when it used GPU acceleration.

To gain access to 128-bit or 256-bit precision, PGGB is using software libraries produced by other companies, which can only run on a CPU.

The beauty of an FPGA is that it's way faster than a GPU for a fixed algorithm at vastly lower power consumption and cost.

I hope, one day, Rob will take seriously the idea of using the FFT algorithm for convolution. It can be done in real time. For a million taps it should be possible in real time using the FPGA in Mojo 2, since FFT will be at least 1000x less computational work than the direct convolution that Rob does.

I wanted to check the PGGB and it started installing the Matlab runtime. What the hell!?! Selling software for a grand and didn't manage to implement your own functions in C/C++/CUDA or whatever the base implementation is. A big thumbs down. I hit cancel and stopped the installation.