Tight bass in a DAC is achieved by a combination of low jitter of the interface or input signal to the DAC as well as power delivery that enables this in the DAC. Power delivery includes power supply, voltage regulators, power storage caps and power decoupling caps. These all must be good designs and high-quality parts of optimum values. The caps in particular must have high "Q". Bass dynamic response depends on it. The output caps affect on bass response will be minimal as long as the load is high-impedance, like 25K ohms. Even 1uFd is sufficient to get good bass. If you plug 100 ohm headphones into the DAC output, it will probably not have any bass, but any preamp or amp will be 25K ohms or higher. Not an issue. I make a very detailed DAC, the Overdrive SX, but it also has superb bass response. Ultra-clear imaging and liquid vocals. If you don't care about DSD, the Overdrive SX is the one to beat. It does not decode DSD. One thing that will make it difficult to compare DACs is if you are not using the same interface, for instance comparing one DAC driven from a CD transport to another using USB from a computer for instance. The first thing is to achieve a really low jitter source. Then, feed all of the DACs with this same source. Your CD transport can become this, but it will need some help. A Synchro-Mesh reclocker can reduce the jitter to around 20psec. Typical Transports have about 800psec of jitter. See these jitter plots: http://www.audiocircle.com/index.php?topic=154408.0 Steve N. Empirical Audio |
lalitk wrote: Speaking of your Synchro-Mesh reclocker, what about DAC’s internal jitter control? Is it not good enough to reduce jitter on a incoming digital signal? I am just trying to understand the need for external reclocker.
My DAC is equipped with a high resolution, low jitter clock. The measurements show Intrinsic Jitter <10 pS. I have yet to hear a DAC that is not sensitive to incoming jitter, even those with reclocking inside. The easy way to tell is to use a cheap S/PDIF cable from a good low-jitter source and then a really good cable. If you hear ANY difference, then it is sensitive to jitter. I routinely ask my customers to do this test and 100% of the time they hear a difference. Jitter almost always matters. Intrinsic jitter is not a good indicator either. This usually means that the internal clock jitter has this jitter specification. This has little bearing on the actual jitter you hear from your DAC. The associated circuitry and power delivery to that circuitry makes all the difference and usually increases that number 10-100 times. I recently did jitter measurements on a device that advertised "intrinsic jitter" of less than a picosecond. The measurements at the S/PDIF termination into 75 ohms showed jitter of 60-80 psec. Lots of manufacturers say they eliminate all jitter. Not possible. Here is some really low jitter at the 75 ohm termination inside the DAC: http://www.audiocircle.com/index.php?topic=154310.0 Steve N. Empirical Audio |
Shadorne wrote: Bass and power supply are linked in every way. This is because they share the same frequencies. The latest DACs are now using switched mode power supplies to great advantage at ultra high frequencies in order to get power supply noise far far away from the audible range. Well, sort-of. The real story is that the regulation on SMPS responds much faster than most LPS. It has little to do with the 50 or 60Hz, more to do with the devices used, the "Q" of the storage capacitors and the technology used for regulation. LPS can also respond very fast, if the circuit is designed for it and the output device is really fast. I provide such a fast LPS for my converter and reclocker products. I also use SMPS in my Overdrive DAC. They both can work for digital. Steve N. Empirical Audio
|
Thank you for your feedback and sharing the link. I guess that’s one way to promote external reclocker by taking a subpar DAC/Streamer. I do not know about anyone else but IMO, SONOS connect box is a piece of crap. It’s also one of the most antiquated streamer/DAC in their lineup.
The final jitter will be exactly the same, no matter what the source is. you can literally compete with a $20K transport using a $300 transport and a $1400 Synchro-Mesh and Dynamo power supply. The input jitter to the Synchro-Mesh is totally unimportant. The Sonos is very popular and many folks like the interface well enough. With the Synchro-Mesh reclocking, the Sonos actually sounds as good as any of my $3K digital interfaces, including USB and Ethernet.
Maybe you can post some measurements with Bluesound Node 2 or Aurender N100 and Synchro-Mesh reclocker. Also, what degree of improvement justifies the cost of external reclocker and digital cables?
I would be happy to if I had one of these on hand. So far, everything I have compared with the Synchro-Mesh has much higher jitter. Having said that, I do recognize the importance of a quality digital source and DAC that employs best designs attributes to minimize or eliminate jitter within digital domain.
I would be open to try an external reclocker between my two sources Aurender N10 / Bluesound Vault 2 and ARC DAC9 in near future. At this time, I am very content and happy with fabulous sound output of my digital setup.
When you are ready, I have 30-day money-back on most of my products, less shipping. BTW, I also now offer a truly world-class S/PDIF BNC/RCA cable for $499. My whole career in audio has been devoted to achieving the lowest jitter. Steve N. Empirical Audio
|
“You can literally compete with a $20K transport using a $300 transport and a $1400 Synchro-Mesh and Dynamo power supply”
- That’s a pretty bold statement, what have you tried and compared in that range to arrive at this conclusion?
I have heard plenty of megabux transports at tradeshows for 15 years. I don't like to name names.
- Wyrd 4 Sound offers an somewhat similar internal upgrade for lot less. Once you add the cost of Connect, Mesh w/Dynamo PS, extra digital cable, you are already at $2250. The Synchro-Mesh plus Dynamo LPS plus BNC reference cable is $1897. Less than the cost of most high-end S/PDIF cables alone. Doing mods inside a transport will not get you there. I used to do mods on transports and DACs myself, for 10 years. Steve N. Empirical Audio |
Stuartk - I think you should address the CD transport before you get a new DAC. The source is at least as important as the D/A, maybe more. Before you drop $2500 on a new transport, consider a reclocker that will actually get you much lower jitter in the S/PDIF signal than a transport at a lower cost, and also provide for galvanic isolation to break ground-loops. You can still use your existing transport. The Synchro-Mesh with Dynamo power supply and BNC reference cable combination will do this. You could start with the Synchro-Mesh OTL and the reference BNC cable for $1100.00. This will get you jitter in the 25psec range. Here are some jitter plots comparing the SM to a popular transport: http://www.audiocircle.com/index.php?topic=154408.0 Steve N. Empirical Audio |
GeorgeHIFI wrote: As I mentioned in my first post, it has output coupling caps, that's why many owners on the net are changing that cap.
If that cap is too small in uF (microfarads) it will roll off the bass too early giving a shallow subdued (cool) bass, increase this cap 3 x in uF and it should fix the problem and give you a deeper stronger bass. If its 1uFd or more, this should be sufficient for most preamp and amp loads to deliver bass. No bass roll-off will occur. The real issue is the quality and combination of these caps. For all caps I have tried with my own DAC, one must add a smaller cap in parallel to get the full range out of it. Highs are ultimately compromised by using larger caps only. The best names, Duelund, Mundorf and V-Cap all need this. Steve N. Empirical Audio
|
What is your opinion on the stereophile J-test?
It measures analog out of a DAC to a test signal - so it checks for both incoming and intrinsc jitter.
Much better to measure the jitter directly, although this is all you can do if the Master Clock interface is inside the DAC. Mildly interesting and only gives you part of the picture IMO because the stimulus is not a true music waveform.
What do you think of the typical SOTA DACs that have a noise floor at -150dB when tested for jitter?
Those are good numbers, but what stimulus is used? A single frequency? Not good enough.
How can perfect performance be improved upon and what do you look for when testing an asynchronous DAC of this level and how could Synchro-mesh or another cable improve upon perfect? I have yet to see or hear "perfect performance". Every time I make an improvement to my own DAC or converter, I think that it cannot possibly get better, but then it does. If I had to choose the best DAC for a demonstration, cost no object, I would choose mine, even though it's not the most expensive or even DSD capable. Performance can always be improved. If you can drive this DAC with S/PDIF signal and hear a difference with different cables, then it is not jitter immune. A simple test. The Synchro-Mesh and a reference BNC cable would likely improve the SQ from it. If it is an internal Ethernet or USB interface, making changes to the power delivery to the clock circuits and buffers will likely improve it. Sometimes even making changes to the ground-plane or voltage regulators will improve it. Performance in high-end audio is mostly about good power delivery, which includes optimum board design, power supply, regulators and decoupling caps. I modded DACs and other digital gear for 10 years and I was able to improve any product a customer would send me. This is precisely what gave me the bag of tricks that I use on my own products. Given enough time tinkering, I can probably improve them even more too. Steve N. Empirical Audio
|
Shadorne - I think the problem lies in where jitter is measured. Most manufacturers measure it at the oscillator or even use the spec sheet oscillator number. What I'm talking about is measuring the jitter of S/PDIF signal as it arrives at the DAC. I have never seen direct measurements of this, except what I published. Companies that make Transports, converters, reclockers and servers all need to measure this directly, but they don't. They don't need to clip probes onto chips to do this. They need a 75 ohm termination inside the scope or AP equipment. Measuring jitter inside the DAC, say on a I2S bus from a USB or Ethernet converter makes a lot of sense, but it requires measuring both the bit clock and the Master Clock. The D/A can use both of these or either one to affect jitter. I doubt if this is done in this J-test. Steve N. Empirical Audio |
Since you are clearly able to build a device that takes 800 psec jitter from an Oppo and turn that into 20 psec jitter with your Synchro-Mesh, then why do you believe that nobody else has a DAC that can correct incoming jitter to completely inaudible levels - around 20 psec or less? Good question. The answer comes from 10 years of modding equipment, seeing the circuit boards exposed in these DACs and hearing them at shows. I don't see the things implemented in DACs that are required to achieve these low jitter numbers. It's not just selecting a low phase noise oscillator, or providing a dedicated power supply for it. It goes WAY beyond that. It has taken me 10 years of modding and tweaking on my own products to learn these tricks. And I don't hear the SQ that would result in exhibit rooms at shows if they had implemented these optimally. After I retire in a few years, I may write a technical book so that other designers can benefit. I have thought about teaching a course at a college, but I live in the sticks. For now, these are trade secrets that differentiate my products. Steve N. Empirical Audio
|
Why would someone buy a tube component that almost sound identical to Solid State? Because Tubes are usually lacking in dynamics, bass tightness or HF extension. You can get the liquid midrange of tubes and the dynamics, bass and HF extension of the very best SS without having that "tubey" distorted sound. Steve N. Empirical Audio |
In general I found that there is a trade of between mid/high to low dynamics and not only in Dacs, gear that sounds very detailed and ’sweet’ (strong at mid and high) doesn’t peoduce strong punchy bass and in contrary gear that has accurate and tight bass tends to be less detailed and musical at the mid/high section.
In general I prefer to sacrifice a little bit at high/ mid to achieve strong and tight bass .
This is not inherent in components, it is a function of the design. This has everything to do with 2-3 things: 1) values and placement of decoupling caps 2) types of decoupling caps, ie; Teflon, Polystyrene, paper-oil 3) jitter If proper attention is paid to 1 and 2, you can usually have your cake and eat it too. My 10 years of modding has taught me that many designers don’t understand power delivery, so they pick capacitors just like other manufacturers pick, or what it recommends in the data book of the integrated circuits. Many of the guidelines in data books are only to make sure the circuit functions, nothing to make sure it delivers great audio quality. The errors just get repeated over and over. This is a bit of a black art I have to admit. Tube equipment is a case in point. Many audiophiles avoid tube equipment because of a "tubey" sound. This is generally due to 2 things: poor choice of tubes and non-existent decoupling caps, particularly for high-frequencies. Designers somehow think that because of the high voltages and low currents involved that these are unnecessary. If they are optimally chosen and placed along with selecting good tubes, it can sound almost identical to SS. Steve N. Empirical Audio |
The J-test uses a known digital signal and analyzes the spectrum of the analog final output of the DAC for spurious frequencies that should not be there. It is actually a complete approach to testing the entire audio chain from digital signal passed through any cable into the digital input of the DAC and a detailed examination of the final analog output. Any jitter at all anywhere that reaches the D to A chip will result in quite obvious spurious analog signals that do not match the known input. Any source jitter, cable jitter, interface jitter, LIM jitter and intrinsic jitter (in fact the sum of all and any jitter) is tested for. I don't believe the DAC output is a good indicator. It will change with every DAC you use in the test, so testing for instance a reclocker or a USB or Ethernet interface would be insufficient. What if the particular DAC used introduces spurious signals unrelated to the jitter of the incoming signal? It is a result of the DAC, not the USB converter driving the DAC for instance. I also don't like using a fixed signal. I would rather see actual music data. I have seen lots of J-test plots in Stereophile over the years. None of them seem to correlate well with sound quality. I think this direct digital measurement I am making is superior to the analog measurement for correlating to SQ. So far, its helping me make decisions on future products. Steve N. Empirical Audio |
I am sure you’re going to tell us if there is component out there that is capable of all of the above without the “tubey” distorted sound. There are a few. My own customized SET monoblocks fall into this category. Others are: Nagra VPA Any Wavac Prima Luna Arte Forma Due Volte - I have the customized version of this one I prefer SET amps, but the Prima Luna sounds really good at shows, and they use quality parts in them. I always enjoy their room. Steve N. Empirical Audio |
Hi Steve. I took a cursory look at the overdrive info on your website http://www.empiricalaudio.com/products/overdrive-dac and user manual. I couldn't find anything about the type/kind of D/A chip(s) or whether or not your technology is considered delta sigma, multibit r2r or something else. Please advise. I decided to use a Analog Devices delta-sigma, but only because I can select the digital filters for it rather than it selecting the filter for me. I cannot seem to get the HF transient response from older ladder chips that I can from newer delta-sigma chips. It also supports 24/192 PCM, as well as allowing for reference voltage-based volume control. I previously offered a tube DAC design using a ladder chip called the "Spoiler". It was very musical, but ultimately did not deliver on the accurate HF sonics that I'm after. It also required hard-to-get expensive NOS tubes to sound really good. The only other option is to design a ladder DAC from scratch or in an FPGA. I don't have the bandwidth for that unfortunately. BTW, audiostream will be reviewing my new SX DAC soon. Steve N. Empirical Audio |
Here is basically my setup, but this one was used at a show, so its not my amps or speakers. I have my own custom versions of both. http://www.arteformaaudio.com.tw/due-volte.asp The Vapor picture at the bottom. Steve N. Empirical Audio |
That Arte Forma SET amplifier seems like a very high tier no nonsense effort. No doubt that it is exceptionally honest, open, transparent and natural in its sonic character. I suspect it is quite capable of revealing the fine nuances of the various upstream DACs and sources. The nice thing about it is that it is DC-coupled, so no caps in the signal path. Also, the version they sell now contains a circuit-board. The one I have was a point-to-point wired prototype that I further modded. It's too bad they don't offer the wired one, but they could not seem to get consistency in the build. I don't use any of their tubes BTW.
Steve N. Empirical Audio
|
However, if a device passes the J-test with perfect results then one can only conclude that all distortion spuriae including any jitter of any sort are all below the noise floor. This assumes that the DAC has excellent HF response and transient response. If it doesn't, the spuriae will be masked, filtered. Take the same source to another DAC that has better HF and transient response and you might see more spuriae. I think it depends on the DAC, which it should not. DAC analog circuits should not be part of the measurement IMO. Steve N. Empirical Audio |
It’s hard to believe that a DAC designer wouldn’t use effective caps in their design. It is, but I would not have had a lucrative modding business for 10 years if it was not the case. I trust my ears also. I have to because measurements just don't cut it IME. A good audio designer must rely on both measurements and their ears. I cringe when I read about someone starting out in this industry and thinks they can navigate on measurements alone. I recently read this on DIY forum... Steve N. Empirical Audio
|
|
Here is what I determined by designing and building many digital cables, and making measurements on those and others: 1) the 75 ohm impedance of the cable is critical and should be uniform over it's length 2) Termination of the 75 ohm cable to RCA plugs is a non-starter. Must be terminated to a proper 75 ohm connector, and a good one, like RF or BNC. 3) Use of pure silver in the 75 ohm cable is preferred, much better tonality. 4) 100% shield is beneficial 5) Conductor gauge is critical. Should be 26 Gauge or smaller to avoid skin-effect Here are a few measurements: http://www.audiocircle.com/index.php?topic=154425.0 Steve N. Empirical Audio |
I believe it to be difficult for hobbyists to determine exactly what they want or need in their stereos, especially in an age where most consumers will make purchases via the internet rather than making comparisons in the dealers show room. This is precisely why I’m doing these jitter measurements and trying to correlate them to sound quality. If a measurement can be proven to be useful, then no amount of snake-oil claims can compete with it. After all, marketing BS is what is mostly written about cables. Most cable manufacturers don’t put ANY measurements on their sites. I’m trying to improve that situation. gdhal - If I could significantly improve my system dynamics, imaging and clarity for $275, I would jump at this. Chump-change, besides it’s 30-day money-back. I recently got a IFI SPDIF iPurifier for my Home Theater. Another no-brainer. I lowers jitter for Dolby Digital and DTS for $150. Improves movies significantly. I can finally hear all of the quiet dialogue in the starting scene of "Gravity". The movie makes sense now. I prefer my Synchro-Mesh for 2-channel. Steve N. Empirical Audio |
FWIW, my next improvement attempt (which will happen in the coming days, not weeks or months) will be to use shorting caps on the inputs to lower my already very low noise floor. This is a mod that will have little effect IME. Inputs that are not selected are not involved. Better off to figure out how to eliminate ground-loops. Now this will actually reduce the noise floor. I have zero ground-loops in my system. Jet black background. Steve N. Empirical Audio |
Why don’t you use an APX555 from Audio Precision?
I cant really tell much from an oscilloscope plot and triggering can be tricky.
Triggering is automatic and programmed. It's a programmable scope with special jitter program. It does have some limitations of scaling the axis etc., but useful so far. The newer scopes don't have as many limitations, but they are way more expensive, like buying a house. This particular scope when new was $130K with the jitter software. I purchased it used. The active probes alone are $5K new each. The AP stuff is great for analog, but this is digital. Even the APX555 only has 1MHz bandwidth. Way too low for digital. Even the clock fundamental can be 49MHz. Edge-rates have GHz components. The Tek scope I'm using has 7GHz bandwidth. This is 7000 times higher than the AP. This is what is needed for accurate digital measurements. The newest version of what I have is the DPO70000, which lists for $315K and that is not including the jitter analysis software which will probably be $25K more. Steve N. Empirical Audio |
I have all of Jonis albums myself, and many in 24/192 downloaded from HDtracks. Steve N. Empirical Audio |
Great news, although it proves that maybe the reclocker is either not 100% immune to incoming jitter or poor signal integrity. You would have to look with a scope to see which one is causing this. I wonder if this would happen if it was powered from a Dynamo rather than the wall-wart or if the reclocker was totally isolated (no OTL mod)? Steve N. Empirical Audio |
