It is definately not silly to consider feedback vs NFB and opamps vs. discrete technology outside a DAC chip.
Feedback in the digital domain is completely different from analog feedback, as it is not used to linearise circuits, but instead used as timing control, noise shaping, DEM etc. which are necessary items all of them.
In an analog stage it is very different, as you throw away around 100 dB signal level just to get your op-amp linear over an acceptable bandwith.
Op-amps mostly are invented for use as operational amplifiers and not audio amps. Still a lot of non audio op-amps are used for audio, because most of them have very good specs, of which different designers emphasizes the parameter they find most important. That´s how slew rate became important.
Also pretty much of the advanced technology used in op-amps, strives towards very high PSRR, which is not necessary for audio at all, you can just design a good PSU yourself. But nowadays it is common to use slow and noisy IC regs.
So if you want components that hardly can be destroyed, runs almost on AC, and have Slew rate figures of more than 1.000Vµsec, THD+N at 130-140dB, and can be linearised simply by adding feedback, op-amps are the item.
If you instead think of the real task of your output stage, which is to drive cables and amplifiers without reacting upon the load, you will not consider them anymore.
As a BUF32S user I'd agree with Kurt. COTS BUF with opamps sounds not involving at all due to damaged middrange as the result of NFB. However, the potential of Sabre could be retrieved by cutting off the whole analog stage and deploying passive stage on trafos (low end LL1527 in my case) or active one on tubes (eg developed by my friend Lukasz Fikus - I heard it many times and liked very much) or even passive SE on caps - all of them sound more natural and real comparing to opamps.
That's really where I do not agree. When I look at the PCB; i do not see anything that seems to have taken years to R&D. I see a very fast implementation of a new DAC chip. Around the chip is a few ceramic capacitors for decoupling. Most likely pure app.note, which is very OK. But it takes only a few hours to draw.
All digital connections and power lines are routed to a screw-terminal.
The analog stage is based on op-amps, which is also a very fast way of R&D. If I am right in my observation of the use of ceramic capacitors in the analog stage for filtering, I know that there has not been spend a hole lot of time tweaking for the right components. Ceramics capacitors has nothing to do in the analog stage.
Very often the time audio manufacturers calim to have used in the R&D, is pure storytelling.
I know a company using another 32 bit DAC chip, claiming to have spend years and years of tweaking their product. Funny thing is, that they launched the product about 6 months after the first samples of the DAC chp was released
. That's High End storytelling. As some of the danish users in here will know, we started a thread on a danish forum more than 4 years ago, to follow the R&D all the way.BTW... When was the new ESS chip released?? And when was the new Buffalo DAC based on this chip released?? The difference between these dates, gives you a figure of how long time has been spend on R&D on the Bufalo.
Remenber, that during this time, some weeks must have been used waiting for samples, PCB production and so on. I wonder how much time have been left for real R&D, listening and teaking before finishing the design?? Guess we will never know....
Last edited:
I must admit to have been mistaken.... I did not know of the SPDIF receiver. Well... That's a plus.
Anyway... It just tells me, that time spend on impementing is even lower
And at that point, I really do not understand, why they did not spend some of the free time on making a better analog stage. Op-amp, op-amp and op-amp..... Just like anything else...
Get your point
but pls note Lukasz system sounds much better than you could imagine. My opinion on various BUF analog stages I based on my own system: shigaraki clone transport -> DACs (BUF or SRC2496 or JVC K2/PCM56) -> AudioSynthesis passive -> Audio Note Conquest -> Triangle Luna -> 50 sqms -> a pair of middle-age ears; and this is not only my opinion, it was confirmed by many callers
It seems that the new ESS chip was shown around Dec 2008. The manual for the Buffalo in version 1.0 was done in May 2009. That gives approx 5 months, if Mr White was around the top priorites in the samples program, when ESS started sending out these. Considering the potential annual use, I would guess not.
But anyway.. 5 month from receiving the first chip, to being able to launch a design that is trimmed in any way, is just not possible. In this time you must:
1) Draw schematic.
2) Route the PCB layout.
3) Order and wait for PCB production (Here in Denmark this normally will take at least a week and more if you don't want to pay for quick service).
3) Mounting of components.
4) Initial listening.
5) Burn in. Especially electrolytic capacitors may take a long time (At least a week), before the sound will no longer change.
6) Tweak using alternative components.
7) Listening test when burned in.
8) Tweak by circuit changes.... And that takes you back to 1) and then the loop is started.
9) Jump out of the loop when satisfied.
We spend 4½ years in this loop, in the search of perfection. Others claim to have used even longer, even though the central components used have only been in production for 6 months
But anyway.. 5 month from receiving the first chip, to being able to launch a design that is trimmed in any way, is just not possible. In this time you must:
1) Draw schematic.
2) Route the PCB layout.
3) Order and wait for PCB production (Here in Denmark this normally will take at least a week and more if you don't want to pay for quick service).
3) Mounting of components.
4) Initial listening.
5) Burn in. Especially electrolytic capacitors may take a long time (At least a week), before the sound will no longer change.
6) Tweak using alternative components.
7) Listening test when burned in.
8) Tweak by circuit changes.... And that takes you back to 1) and then the loop is started.
9) Jump out of the loop when satisfied.
We spend 4½ years in this loop, in the search of perfection. Others claim to have used even longer, even though the central components used have only been in production for 6 months
...Hurtig,
don't you have something better to do ?
Read Buffalo development thread...the chip designer was in on the design of Buffalo, i'm hoping you really have a good dac...but this used to be a nice place before your lack of humility came in....you are in very good company here....don't spoil the place, respect people.....and yes i have read all your posts here and i noticed some of your fellow danes have a very negative opinion of your products ....from your native forum!.
don't you have something better to do ?
Read Buffalo development thread...the chip designer was in on the design of Buffalo, i'm hoping you really have a good dac...but this used to be a nice place before your lack of humility came in....you are in very good company here....don't spoil the place, respect people.....and yes i have read all your posts here and i noticed some of your fellow danes have a very negative opinion of your products ....from your native forum!.
Last edited:
Absolutley Luke,
As far as the Buffalo24 goes i built one and as i also have one of Pedja Rojic's AYA II, i had a pretty wide ( in terms of time ) comparison scope.
The Buffalo, is very nice in the fact that it has detail, very good sound stage, what i noticed with the IVY stage was that it did not have that sweet sound the AYAII has, although upper range was clearer, but then again the TDA1541A is not known for it's upper range.
I like it a lot but maybe will work on a diferent IV stage when i have the time...or just put in a "counterpoint".
By the way, my system at the time of testing: Sony CDP 337ESD ( as transport ) , Pass UGS, ORION+
Last edited:
I'm sorry that you feel this way. But I still believe, that a DAC based on op-amps in somewhat standard applications and ceramic capacitors in the analog stage, is not the new 7th wonder. DAC's like this is real mainstream. Buy an reference board from any semi-company, ad you will get the same, at a lower price and without the storytelling. "Hey look.... we spend 50 years of developement, and ended up with an op-amp in a standard application"
.
Maybe we can start diyaudioxxx.com. Single driver is fine, but multi-way is so much more more exciting.
Last edited:
Please try to make your point with either scientific arguments, or maybe even rock solid empirical experience which can be repeated when called upon.
IMHO NFB can never be used for linearisation purposes, as it also cannot be used to drive a load not included in the calculations of the feedback loop.
This goes for global NFB, and has been prooved by i.e. Matti Otala.
please proove him wrong, it might even bring you a Nobel price
We can hardly wait to be enlightened
Please try to make your point with either scientific arguments, or maybe even rock solid empirical experience which can be repeated when called upon.
IMHO NFB can never be used for linearisation purposes, as it also cannot be used to drive a load not included in the calculations of the feedback loop.
This goes for global NFB, and has been prooved by i.e. Matti Otala.
please proove him wrong, it might even bring you a Nobel price
We can hardly wait to be enlightened
The feeling is mutual, you have presented nothing but the usual "opamps just have this sound". Anecdotal nonsense.
Scientific arguments, please.
I waited patiently until Buy it Now button re-appeared for Buf32. Dealing with Brian was a pleasure. It builds easily in a couple of nights if you rtfm. For $600 I can't imagine a better DAC. The incredibl promise of Sabre32 is true and fulfilled here in Buffalo. It is clear, rich, harmonically dense and detailed, textures and tones are perfectly realistic, massed strings, woodwinds, brass al perfect. Low bass control and dynamics are breathtaking. It completely eliminated the extreme jitter coming from my Sonos ZP-80 net player, that the Altmann Attraction could not fix. Smooth as silk, with iron fisted control. Mine has dual LCBPS and one LCDPS. I used Brian's trafos and switchcraft RCAs, cut up PC power cord.
In comparison to Altmann Attraction DAC (1543 based) with UPCI, no JISCO, 16bit/44.1 only, optima batt. ($1200 new) SEPP 300B amps, Usher Be20, JPS wires, Cary SLP98 preamp. There is hardly any comparison to be made. The Buffalo trounces the Altmann. Significantly improved clarity, coherence, presence, more extended range, superior dynamics, relaxed precise presentation. Much ballsier. The key advantage of the Attraction over of oversampling DACs was the relaxed more realistic, undistorted high treble. The buffalo is much more extended and dynamic in the treble, but I do not hear any of the sizzle or fake sibilance as with most oversamplers. I need a more powerful amp now to keep up with the dynamic demands made by Buffalo.
Last night I compared it to my friend's Oracle DAC1000 ($6500 new), LFD NCSE, JPS, Harbeth 40.1. The Oracle is the local reference standard for DACs. We have all compared our to it and turned green. The buffalo had similar presentation in terms of clarity, intensity, tonal density, dynamics. It was superior in timbre because the Oracle ttends to excite things too much, sounds aggressive, unnaturally intense. The Buffalo is even tempered and mater of fact, confident to just be there. I was glad to hear it hold it's own in the music making categories. The Oracle did reign supreme in the end though, because of the greatly superior spatial presentation. The Buffalo sounded closed in and flat in comparison to the oracle. Also the oracle had an even greater degree of textural detail. Both aspects are typically cut at the knees with high feedback amplifiers.
In other projects where 4562 was tried I have heard the same kind of "precise but not spacious" sound. So another friend is working on a different output stage that might be more appropiate for the incredible performance of this chip. I don't know what he will come up with, but that's what DIY is all about.
Congratulations to Russ White, this DAC is awesome. My battery is running low, gotta stop. Add one thing The Volumite control works excellent. It does not truncate redbook words until -96dB attenuation. I heard perfect cymbal tone with my ear to the tweeter turned down as low as I could hear with preamp at full volume.
Looking forward to trying more Twisted Pear products in the pipeline. I will report on output stage ideas that we try.
Rich
In comparison to Altmann Attraction DAC (1543 based) with UPCI, no JISCO, 16bit/44.1 only, optima batt. ($1200 new) SEPP 300B amps, Usher Be20, JPS wires, Cary SLP98 preamp. There is hardly any comparison to be made. The Buffalo trounces the Altmann. Significantly improved clarity, coherence, presence, more extended range, superior dynamics, relaxed precise presentation. Much ballsier. The key advantage of the Attraction over of oversampling DACs was the relaxed more realistic, undistorted high treble. The buffalo is much more extended and dynamic in the treble, but I do not hear any of the sizzle or fake sibilance as with most oversamplers. I need a more powerful amp now to keep up with the dynamic demands made by Buffalo.
Last night I compared it to my friend's Oracle DAC1000 ($6500 new), LFD NCSE, JPS, Harbeth 40.1. The Oracle is the local reference standard for DACs. We have all compared our to it and turned green. The buffalo had similar presentation in terms of clarity, intensity, tonal density, dynamics. It was superior in timbre because the Oracle ttends to excite things too much, sounds aggressive, unnaturally intense. The Buffalo is even tempered and mater of fact, confident to just be there. I was glad to hear it hold it's own in the music making categories. The Oracle did reign supreme in the end though, because of the greatly superior spatial presentation. The Buffalo sounded closed in and flat in comparison to the oracle. Also the oracle had an even greater degree of textural detail. Both aspects are typically cut at the knees with high feedback amplifiers.
In other projects where 4562 was tried I have heard the same kind of "precise but not spacious" sound. So another friend is working on a different output stage that might be more appropiate for the incredible performance of this chip. I don't know what he will come up with, but that's what DIY is all about.
Congratulations to Russ White, this DAC is awesome. My battery is running low, gotta stop. Add one thing The Volumite control works excellent. It does not truncate redbook words until -96dB attenuation. I heard perfect cymbal tone with my ear to the tweeter turned down as low as I could hear with preamp at full volume.
Looking forward to trying more Twisted Pear products in the pipeline. I will report on output stage ideas that we try.
Rich
Last edited:
The op-amp in the Buffalo isn't used for linearization.
OK. Here's your enlightenment. Use a short cable from an op-amp IV to a good preamp. Do it so the load doesn't adversely affect the op-amp feedback.
Use a decent cable so the reactance isn't a problem. The load can be stable.
Is that such a relevation?
Your DAC is not accessable to the lot of the members reading this. Why don't you make it so?
I have both Buffalo and Buffalo32s and also modded/worked dacs based around CS4397/4398
No way on gods earth would any CS439* based dac no matter what output stage and regulation used would better the Buffalo's.
I've tried dozens of dacs, the Sabres are unlike anything I've used before
AYA II is an excellent sounding dac, its one of my favourite commercial offerings
BTW the caps used in the Buffalo32s are not ceramic, their PPS
No way on gods earth would any CS439* based dac no matter what output stage and regulation used would better the Buffalo's.
I've tried dozens of dacs, the Sabres are unlike anything I've used before
AYA II is an excellent sounding dac, its one of my favourite commercial offerings
BTW the caps used in the Buffalo32s are not ceramic, their PPS
- Status
- Not open for further replies.