Computer Audio Asylum

Music servers and other computer based digital audio technologies.

Return to Computer Audio Asylum


Message Sort: Post Order or Asylum Reverse Threaded

Musiland Monitor 01 US - more mods

219.132.219.63

Posted on December 27, 2009 at 22:47:29
Thorsten
Manufacturer

Posts: 4209
Location: Somewhere nice on planet dirt
Joined: September 25, 1999
Folks,

A few days back I commented on the Musiland Monitor 01 Devices and some mods for the 01 US specifically:

http://www.audioasylum.com/forums/pcaudio/messages/6/66891.html

Here a pic of the modded 01, stage 1:

Image and video hosting by TinyPic

Since then I got a little time to do some more work. As I had a request for a unit that remained easily portable and USB powered with a good quality line out I did not try separate supplies, super-clocks and all that.

Instead I focused on the two main weaknesses in the DAC section that are easly addressed, namely the filter capacitor for the reference voltage pin and the Op-Amp.

After changing the reference capacitor to 220uF/16V Elna Silmic (feel free to try other too, I just like that particular one) with an SMD film bypass I also changed the Op-Amp to AD8620 (at +/-5V it sounds pretty good).

The result of these two fairly minimal added mods (after the big round of power supply cleanups) is to turn the Musiland 01 US into a fairly capable solid state DAC, running directly from the PC with 192KHz 24Bit (and anything below) capability.

I don't much play around in the world of Solid State DAC's. I am listening routinely to an non oversampling TDA1541 Double Crown with tube analogue stage (fed directly from a Motherboard SPDIF output which is quite a bit modified) and next to this the modded Musiland sounds a little white, not as harmonically rich and a little constrained and not as spacious. The GF observed the same, as it happens, from the side, curled up on the sofa, her first remark was "This does not sound quite so impressive.".

That said, I would say when comparing it to much of the mainstream (non usb) all solid state DAC's out there, this modded Musiland sounds suprisingly resolved, detailed, spacious and smooth. It is definitly very listenable.

And it will do 192/24 for the latest High Rez Downloads.

Now.

So if you can mod your own stuff the Monitor 01 US (or the Monitor 01 Mini as remarked elsewhere) is not a bad starting point, even using the on-board DAC, just sort out the basic problems (powersupplies, nondescript op-amp in the Analog Stage) and it will sing quite well.

But it does make me wonder what a fully "pimped out" version with a linear supply for the digital section, a really good clock and a tube analogue stage (of course disabled switched supplies) would sound like.

Could be a killer tubed 192/24 USB DAC for very little money. If I use what I can readily get from diyhifisupply and know well, it looks like around 750 US in bits and bobs, including the musiland, a case and all, some assembly required, but all the hard work already done.

With more DIY (assuming one is capable to do it) the cost could be quite a bit lower than that.

Ciao T
Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

 

Hide full thread outline!
    ...
RE: Musiland Monitor 01 US - more mods-$750????, posted on December 28, 2009 at 06:32:29
fmak
Audiophile

Posts: 13158
Location: Kent
Joined: June 1, 2002
As I said to audioengr, what is the logic of spending $750 on a $70 processor?

For that amount, the $70 unit with external supply will provide a good signal for an existing external dac. Many have $1000 dacs, quite a few have excellent $5000 plus units.

 

So Fred spending $1500 on a Big Ben is silly also?, posted on December 28, 2009 at 06:47:50
Gordon Rankin
Manufacturer

Posts: 2928
Joined: June 9, 2000
Fred,

Since you feel that using this with the Big Ben is essential. Then why would Thorsten's mode be over spending?

Heck the way I see it you would save $750 in your configuration.

Thorsten knows what he is doing and only pointing out obvious modifications.

Thanks
Gordon
J. Gordon Rankin

 

Can you Read, Gordon?, posted on December 28, 2009 at 07:16:10
fmak
Audiophile

Posts: 13158
Location: Kent
Joined: June 1, 2002
As a 'respected' manufacturer, you do not seem to understand the proper use of words. I could say that your $15000 usb dac limited to 96k is silly, but I have not said so.

I said I did not see the rationale of spending $750 on a $70 dollar device.

Posts like yours are not going to improve your poor image here or help your sales.

If I want to spend $10000 on my dac, that is my perogative. It is stupid for you to comment on any post in such a way. By the way, a Big Ben does not cost $1500 and is is a format converter, clock, relocker and works on video if required.

I do not expect pros like you to post in packs and twist words irresponsibly.


 

RE: Musiland Monitor 01 US - more mods-$750????, posted on December 28, 2009 at 07:37:18
Thorsten
Manufacturer

Posts: 4209
Location: Somewhere nice on planet dirt
Joined: September 25, 1999
Hi,

> As I said to audioengr, what is the logic of spending
> $750 on a $70 processor?

This argument can be made for many modifications.

The counter argument is that many of the resultant devices perform on a far higher of performance than equivalent money spend on commercial devices.

> For that amount, the $70 unit with external supply will
> provide a good signal for an existing external dac.

Of course, that assumes that one has an existing DAC capable of 192/24 and perhaps one with a good tube output stage (assuming one would want one - I do) and with good SPDIF jitter suppression.

> Many have $1000 dacs, quite a few have excellent $5000 plus units.

It stands to reason that such individuals with good 1,000 USD 192KHz capable Tube Dac's with good jitter suppression on the SPDIF inputs will take your advise, as will no doubt those with the 5,000 USD DAC's.

For the tiny minority left over my post may provide some entertainment value and maybe an idea to do something.

FWIW, I know of no nice tubed output (and that does NOT include any "Cathode Follower after a ton of Op-Amp's" ones) 192/24 capable DAC with good jitter suppression for USD 1,000.

Would you care to suggest a few options for those not into DIY and who would like to buy such a device?

Heck, I might get one myself to save me the building!

Ciao T
Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

 

RE: Musiland Monitor 01 US - more mods-$750????, posted on December 28, 2009 at 10:25:23
Like I said before, it takes too much to fix USB.

Something is wrong when it takes so much to make it right.

With this hobby its easy to let the obsessive compulsive behavior lead you astray and wind up spending hundreds or thousands of dollars trying to make or tweak some small details.

More and more it is becoming obvious that some people really don't enjoy music at all, but are just in this hobby as an outlet for compulsive behavior. When you buy something and take it apart, tweaking it while listening for a certain sound over and over, month after month, year after year how much real music are you listening too? Reminds me of a friend I knew who spent years working on his race car making it perfect but have yet to run more than a few races. Meanwhile people who love racing were out there every month running with what they had...

 

RE: Musiland Monitor 01 US - more mods-$750????, posted on December 28, 2009 at 10:42:11
Thorsten
Manufacturer

Posts: 4209
Location: Somewhere nice on planet dirt
Joined: September 25, 1999
Hi,

> Like I said before, it takes too much to fix USB.

!!??

I did not notice it was broken, outside the pages of a certain magazine anyway...

> Something is wrong when it takes so much to make it right.

I would agree.

Modern "value engineering" tends to omit essential and simple/inexpensive circuit features which by their absence seriously impact on the resulting sound quality.

> When you buy something and take it apart, tweaking it while
> listening for a certain sound over and over, month after month,
> year after year how much real music are you listening too?

True. This is why I always have one or even several pretty steady systems with minimal changes that I use to listen to music. But as I happen to mess around with a lot of audio electronic stuff for a living, anything is grist for the mill.

> Reminds me of a friend I knew who spent years working on his race
> car making it perfect but have yet to run more than a few races.
> Meanwhile people who love racing were out there every month running
> with what they had...

And even others where running every month what they had (gets you ton's of experience) while working on something they figured could be better or at least equal while doing more.

In fact, little real progress would be made without such continuous tinkering. The Dynaudio Speakers you seem to like are in principle the exact same thing as the original Dynamic Driver from Rice and Kellog (no rice crispies jokes please). But Dynaudio seems to have gone through an aweful lot of hassle to get this same basic recipe "just so" over decades.

If you want to see the kind of progress that is being attained within corporate research and development programs of big corporations Microsoft can serve as shining beacon of achievement, as no-one there tinkers that way.

Which is why Windows 7 is hailed as such an extraordinary achievement... It actually almost works!

Ciao T
Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

 

RE: Musiland Monitor 01 US - more mods-$750????, posted on December 28, 2009 at 11:23:41
audioengr
Manufacturer

Posts: 6017
Location: Oregon
Joined: April 12, 2001
Takes the same amount of money and effort to fix Firewire or ethernet devices IME. I dont get your point.

 

RE: Musiland Monitor 01 US - more mods-$750????, posted on December 28, 2009 at 11:57:09
Ryelands
Audiophile

Posts: 1867
Location: Scotland
Joined: January 9, 2009
Like I said before, it takes too much to fix USB.

USB is, as the name implies, a flexible technology. You may prefer to stick with external SCSI for scanners and backup drives, LPT1/Centronix for printers, serial interfaces and who-knows-what for anything remotely non-standard, etc etc. I don't - USB is an ingenious technology.

OTOH, the transfer of audio data is a real-time process and sensitive to timing errors (No, No, not the "j" word, please . . .).

To get the best results from audio on the USB means sacrificing its flexibility to achieve timing accuracy: the interface needs to be properly configured. If you know what you are doing, that's not that hard to do. If the chosen device has drivers that address the issue, even better. All this has been discussed ad nauseam on the list.

Can I just close by saying that I heartily welcome Thorsten's recent contributions to this list. IME, he has a record of effective innovation, design competence and a refreshing willingness to share his ideas. Why this is seen by some as a licence for dog's abuse, I've no idea. (Well, to be honest, I've got a pretty good idea but I'd rather not go into it right now.)

Surrogate Steve

 

RE: Musiland Monitor 01 US - more mods-$750????, posted on December 28, 2009 at 12:51:04
Old Listener
Audiophile

Posts: 2090
Location: SF Bay area
Joined: February 6, 2005
> Can I just close by saying that I heartily welcome Thorsten's recent
> contributions to this list.

I do too.

> To get the best results from audio on the USB means sacrificing its
> flexibility to achieve timing accuracy: the interface needs to be
> properly configured.

It is early days for USB audio. There is no reason that cheap, high volume USB receiver chips could not be designed with support for async mode, 24/96 data rates and general concern for noise and jitter issues.

> If you know what you are doing, that's not that hard to do.

Products like these require new skills and expertise to get a really high quality result at present. I don't think that most US high-end companies have all the required expertise. Some of the manufacturers who post on this and other PC audio forums seem to be somewhat ahead of the others.

However, I note that much of the interesting development activity is now coming from Asian and to a lesser extent, European companies. Until recently, I had not considered that I should learn Chinese so that I could read the best audio forums.

Maybe Pro-audio companies can fill the void but their main market has different needs so we'll be poor relatives asking for the left-overs.

Bill


my blog: http://carsmusicandnature.blogspot.com/

 

RE: Musiland Monitor 01 US - more mods-$750????, posted on December 28, 2009 at 13:19:28
Mercman
Audiophile

Posts: 6580
Location: So. CA
Joined: October 20, 2002
I sometimes think we place too much emphasis on the computer interface used. I have been playing with some different Firewire DACs and believe me, the differences I hear in these DACs is not due to the particular interface they are using. USB 2 Audio will probably support 24/384 with no problem. I'm sure Gordon will release his 32/192 as soon as Microsoft supports USB Audio 2 with native drivers.

The real art as I see it, is in the design of the power supplies and associated circuitry. As Throsten intimated, these are not secrets.

 

RE: Musiland Monitor 01 US - more mods-$750????, posted on December 28, 2009 at 13:24:00
Ryelands
Audiophile

Posts: 1867
Location: Scotland
Joined: January 9, 2009
Products like these require new skills and expertise to get a really high quality result at present.

What I said "the interface needs to be properly configured," I was not talking of designing products, only of configuring a computer to get the best out of a USB port in the audio context (much as you'd expect to adjust VTA on a turntable arm).

It's not hard for anyone with minimal PC and, to take it a bit further, DIY skills. I was questioning the notion that USB (esp for audio) was inherently and irrevocably flawed. It isn't. True, it's at its best when properly implemented on the motherboard but you can scarcely blame Intel for that.

I can't say whether Firewire is better for domestic use, likely to survive in the marketplace, can be daisy-chained or externally clocked bla bla bla - all I'm saying is that the idea that USB doesn't work properly for audio is misleading.

Surrogate Simon

 

RE: Musiland Monitor 01 US - more mods-$750????, posted on December 28, 2009 at 14:12:11
Old Listener
Audiophile

Posts: 2090
Location: SF Bay area
Joined: February 6, 2005
I wasn't in any way disputing what you said.

> all I'm saying is that the idea that USB doesn't work properly for audio
> is misleading.

I don't think that USB audio is inherently flawed either.

Bill

my blog: http://carsmusicandnature.blogspot.com/

 

RE: Can you Read, Gordon?, posted on December 28, 2009 at 18:47:50
Bob_C
Audiophile

Posts: 2667
Location: NY
Joined: July 31, 2000
Yes Gordon can read and so can everyone else here! That is why most people here try to ignore you, and are thankful for posts by people like Gordon and Mr. T...

 

"Posts like yours are not going to improve your poor image" LOL nt, posted on December 28, 2009 at 21:42:10
Beetlemania
Audiophile

Posts: 1217
Location: Utah
Joined: November 1, 2003
>>>Posts like yours are not going to improve your poor image

 

RE: Musiland Monitor 01 US - more mods-$750????, posted on December 28, 2009 at 22:03:02
fmak
Audiophile

Posts: 13158
Location: Kent
Joined: June 1, 2002
I agree and have been saying so for sometime.

 

RE: Can you Read, Gordon?, posted on December 28, 2009 at 22:06:01
fmak
Audiophile

Posts: 13158
Location: Kent
Joined: June 1, 2002
If you choose to ignore my posts, why are you replying?

As I said, a few vendors' associates post in packs when they choose not to reply directly.

 

RE: Musiland Monitor 01 US - more mods, posted on December 29, 2009 at 03:27:54
soundchekk
Audiophile

Posts: 2424
Joined: July 11, 2007
"Folks"

I hope you didn't miss the note where to potentially buy the needed parts for tweaking! ( As Fred said we're talking ten times the amount of the actual discussed USB-device)

I doubt that Thorsten (as a manufacturer) has/had any relationship with the company he is bringing up here.

If he would have (had) any relationship with this particular company, you might consider the whole discussion as a hidden marketing campaign!


Beside that the whole discussion shows again what it means to have it "measured". You ask ten guys and you get 10 different measurements. Obviously only the manufacturers know what to measure and how to measure it.

Folks - trust your ears and not the marketing messages! ;)


Cheers

-----------------------------------------------------------------

blog latest >> The Audio Streaming Series - tuning kit pCP

 

RE: Musiland Monitor 01 US -Have learnt, posted on December 29, 2009 at 03:45:11
fmak
Audiophile

Posts: 13158
Location: Kent
Joined: June 1, 2002
Musiland apparently offers the chipset to manufacturers and I'd be surprised that a company like them (seems big from their website)will not ultimately make a tubed dac out of it. It will probably cost less than $820. The current Musiland products have support issues in relation to drivers and firmware as well.

The reason why I made the post is that, over the years, I have learnt not to invest big sums on low cost items to 'improve' them. Start with something well made and there is a much better chance of longer term reliability and satisfaction.

 

RE: Musiland Monitor 01 US - more listening, posted on December 29, 2009 at 08:24:30
Thorsten
Manufacturer

Posts: 4209
Location: Somewhere nice on planet dirt
Joined: September 25, 1999
Folks,

Had a chance to listen to modded Monitor 01 US in a different system.

This one normally runs a (now modded - see other posts) Monitor 01 USD as USB2SPD into another non-oversampling, valve analog stage, TDA1541 DAC.

The good news, just plug the 01 US in, driver recognises it and everything plays straightaway, truly "plug & play" (using ASIO and an ASIO only Music Player).

The bad news, someone just brought in and opened a big jar of angry bees.

Yup, we have a ground-loop and a bad one. The Monitor 01 USD isolates the Computers chassis, earth and ton's of noise from that front from the HiFi system. The 01 US does not.

Nothing a cheater plug cannot fix (the angry bees decided to depart for other realms quite promptly), but it makes me wonder how often "bad PC Audio Sound" is not even bad setup or software, but a plain old earth loop?

We spend quite a bit of time listening and generally confirmed yesterdays listening at my place. Compared to the TDA1541/Valves DAC the very basically modified Monitor 01 US lacks the warmth and body, also the immediacy of non-oversampling is missing, yet the overall results are surprisingly good and very listenable, especially in light of the overall maybe 250 Euro total (including counting my time for the actual modification).

This BTW is something I have been observing with a few DAC's recently that use chip that are quite unfashionable because of perceived technical shortcomings just as is the case with PCM1793 from Burr Brown (another used an Analog Devices Chip, the third an obscure Japanese chip).

Something all these had in common are comparably "primitive" digital filters and hybrid (bitstrean and multibit mixed) DAC sections. And yes, to my ears these sound subjectively better than the latest super duper DAC's where the on paper technical performance of the filter and DAC is much better, but the sound quality seems lacking, while the theoretically worse chips delivered much more subjective sound quality. Go Figure.

Anyway, I remain surprised at just how good this very basically modified and small little box turned out to sound. Really unexpected.

For the record, the modifications did involve around 30 Euro worth of parts, no need to spend 750 US, nor do I think i will, at least until I get much more 24/175.4 and 192 material than the few test tracks I have now (my current PC/DAC handles up to 96KHz perfectly well, if truncated to 16 Bit [shrug]).

For anyone who can solder SMD components [or maybe at least see them well enough to solder them ;-) ] these modifications should take maybe one or two hours. So anyone can decide if they CAN and WANT to try it.

From previous experience in modding many a DAC and CDP I still feel that upgrading the clock and analog stage will be well worth it. Clocks are now as many as sand by the sea, only when it comes to easy replacements of analog stages is there some trouble, I do not know if the Zap Filter is still available, I quite liked the results it provided, though "No tooobsz!" and op-amp swapping only gets one so far.

I guess really DIY minded folk can follow the schematics in my decade old article on how Valves and DAC's can go together (Thermionic Valve Analogue Stages for Digital Audio - A short overview of the Subject) or peruse Lukasz "Lampizator" Fikus's Website (LAMPIZATOR web) and make a Kwack Clock.

Ciao T
Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

 

RE: Musiland Monitor 01 US - more listening, posted on December 29, 2009 at 09:10:39
jkeny
Manufacturer

Posts: 502
Joined: May 4, 2001
Hi Thortsen,
Good to hear your positive reports on these devices - there certainly is a new wave arriving of inexpensive/good sounding devices such as the Musiland & M2Tech HiFace. I fully concur with your approach & agree that modding something like this makes much more sense than buying a $1,000 or $5,000 device:)

I found the same issue with ground loops on the 01US & used also a cheater plug (if this means one without a ground connected?). I also found that running my laptop on batteries or linear supply also improved the sound. I forgot to say that I used transformers as my output stage & this had a huge improvement over the op-amp output (as well as killing any ground loops)

I have a suggestion for you for a different op-amp in the portable configuration - a OPA365 - a single 5V supply, 50MHz, low noise, low distortion high CMRR, op-amp. This would possibly allow the decommissioning of the +/-5v Dc switched supplies & their noise issues?

I also query whether a low jitter clock would improve matters much. The existing 24MHz crystal feeds the Cypress chip & is then passed to the Xilinx chip where audio clock(s) are synthesised. This synthesis is most likely the cause of the majority of jitter found on the MCLK at the DAC.

I tried a low jitter Crystek clock fed by battery replacing the crystal & initially noticed an improvement but this was done with a number of other mods & when I changed back to the crystal I didn't notice any deterioration in sound. Others may have had a different experience.

 

RE: Musiland Monitor 01 US - more listening, posted on December 29, 2009 at 10:02:43
glt
Audiophile

Posts: 39
Joined: May 20, 2001
Thortsen, enjoy reading your adventures with these devices very much. Keep them coming.

I do have the same question as jkenny regarding replacing a crystal for a clock. The cypress chip uses a crystal to feed the clock that is inside the chip. If one uses a clock instead of a crystal, the cypress chip will still use that frequency to synthesize the clock inside the chip. In addition, "conventional wisdom" seems to indicate that an external clock with all the additional wiring will negate any improvement even if there is one to start with.

Looking at crystal specifications, If one selects a quartz type, the specifications between crystals do not vary much, and most are of the 50ppm type.

I have read in some specs that the crystals are sensitive to external perturbations, thus adding some damping material to the metal can/case may be a cheap tweak...

Thus in order of investment, adding damping material may result in some improvement. Replacing the crystal with one with better spec, may result in additional improvement. Replacing the local 3.3v reg with one of lower noise may also result in clock improvement as the clock is inside the chip.



-----------------------------

H I F I D U I N O

 

RE: Musiland Monitor 01 US - more listening, posted on December 29, 2009 at 11:41:53
jkeny
Manufacturer

Posts: 502
Joined: May 4, 2001
One other point - the PCM1793 is a Vout DAC with a bias voltage of 1.65 on each differential output leg - so the op-amp is only performing DC blocking function & also upping the output signal voltage to 2V (RMS?).

I used a 1:1 transformer and a cap as output devices - both of which sounded better than the existing op-amp but I preferred the natural sound of the transformer.

 

RE: Musiland Monitor 01 US - more listening, posted on December 30, 2009 at 09:50:06
Thorsten
Manufacturer

Posts: 4209
Location: Somewhere nice on planet dirt
Joined: September 25, 1999
Hi,

> I do have the same question as jkenny regarding replacing a
> crystal for a clock. The cypress chip uses a crystal to feed
> the clock that is inside the chip. If one uses a clock instead
> of a crystal, the cypress chip will still use that frequency to
> synthesize the clock inside the chip.

Hmmm, the crystal plus the circuit inside the cypress chip form a so-called Pierce Oscillator. This is quite sensitive to powersupply noise. Inside a given chip there will be always a bit extra supply noise (quite a bit extra actually, plus a goodly bit of ground bounce) from the inductance of the chip's lead-frame. This is the ultimate limit of the "goodness of the oscillator", which is not particularly good to start with.

An external oscillator will not be subject to that kind of noise, if it is fed from a clean supply. So there is a possibility of improvement.

Inside the Cypress Chip the clock is used for the USB Functions. It is also passed to the Xilinx chip, where a so-called fractional-n PLL is used to synthesize the audio clocks from this base clock.

The best PLL circuits manage to create such clocks from a "jitter-less" base clock (the 24MHz in the Musiland) with under 50pS (not ppm) Jitter, which is as good as the real-world performance of many a "super-clock" once integrated in a given product (remember the ground-bounce).

The PLL in the Xilinx is not as good, real numbers are a bit hard to come by, around 100pS seem possible though. If our "reference clock" has more jitter than that (which the on-chip clock may very well have) it will become the limiting factor.

And of course the Jitter of the PLL is mostly noise (if the powersupply is sufficiently quiet), while the clock of the Cypress Chip may very well have serious deterministic jitter from the USB processing (presumption, not tested).

So it would still seem that improving the clock may still produce significant overall improvements, of course, cleaning up the supply noise would need to come before that.

> In addition, "conventional wisdom" seems to indicate that an external
> clock with all the additional wiring will negate any improvement even
> if there is one to start with.

I do not follow this at all? How does the wiring come into this, if it is designed to give the correct transmission?

> Looking at crystal specifications, If one selects a quartz type,
> the specifications between crystals do not vary much, and most
> are of the 50ppm type.

The ppm specification has relevance only to the accuracy of the resonance frequency (ppm = Parts Per Million - or Hz deviation per MHz of clock for 1ppm). It has zero bearing on jitter performance.

> I have read in some specs that the crystals are sensitive to
> external perturbations, thus adding some damping material to
> the metal can/case may be a cheap tweak...

Marantz used to do this, I felt that pulling off the damper improved the subjective sound (but using a physically much larger crystal that was rigidly attached to the PCB which was rigidly attached to a chassis that could be made much more rigid with a wooden base - did much better).

Somehow it seem that applying "damping" to something resonating makes the problem a little worse.

> Thus in order of investment, adding damping material may result
> in some improvement.

Or dis-improvement, subjectively. One would have to try it, please do. Though the small size of the whole Musiland Gizmo makes resonances much less of a problem than larger structures.

> Replacing the crystal with one with better spec, may result
> in additional improvement.

This is unlikely. One may be able to improve the oscillators performance with crystals having a higher Q for their resonance (this is usually not specified, you would have to measure it) but with a pierce oscillator this is not going all that much.

> Replacing the local 3.3v reg with one of lower noise may also
> result in clock improvement as the clock is inside the chip.

It may be possible, the chip that is there is not quite that bad at low frequencies. Better can be done, but not really in the standard format, so some kind of "super-regulator" would be needed, maybe a teddyreg?

Moreover, the regulator chip has zero effect at high(er) frequencies where a lot of our problems stem from. The whole issues of lead frame inductance etc. is also affected by the regulator.

So, I think a case can still be made for a separate clock. The best choice might be one that avoids referencing the clock signal to the ground of the Cypress Chip, where it would messed up by the ground bounce from the lead frame of the IC (which was likely one of the reasons jkenny's experiment with a separate clock was not very successful. Other factors may have also contributed.

Ciao T
Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

 

RE: Musiland Monitor 01 US - more listening, posted on December 30, 2009 at 10:01:15
Thorsten
Manufacturer

Posts: 4209
Location: Somewhere nice on planet dirt
Joined: September 25, 1999
Hi,

> I found the same issue with ground loops on the 01US & used also
> a cheater plug (if this means one without a ground connected?).

Yes, a Cheater Plug is one which "cheats" the earth connection.

> I forgot to say that I used transformers as my output stage &
> this had a huge improvement over the op-amp output (as well
> as killing any ground loops)

Yes, I have suggested this approach for a long time for Voltage output DAC's. Tube stages are often still more to my liking.

> I have a suggestion for you for a different op-amp in the
> portable configuration - a OPA365 - a single 5V supply, 50MHz,
> low noise, low distortion high CMRR, op-amp.

I guess you mean OPA2365? Looks pretty good, actually.

> This would possibly allow the decommissioning of the +/-5v Dc
> switched supplies & their noise issues?

Probably. But you loose output swing, so you need to redesign the analog stage to output less than 5.6V PP (which it does now).

> I also query whether a low jitter clock would improve matters much.

I suspect it depends on application. The conventional way of injecting a clock signal into the Xtal-In pin is subject to the chip's internal ground-bounce which is material, so your nice low jitter clock gets mangled.

> The existing 24MHz crystal feeds the Cypress chip & is then passed
> to the Xilinx chip where audio clock(s) are synthesised. This
> synthesis is most likely the cause of the majority of jitter found
> on the MCLK at the DAC.

I know dedicated fractional-n clock synthsisers that manage under 50pS additional jitter over the base clock, given a clean supply to the PLL and good ground routing to the PLL ground...

Ciao T
Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

 

RE: Musiland Monitor 01 US - more listening, posted on December 30, 2009 at 10:44:02
jkeny
Manufacturer

Posts: 502
Joined: May 4, 2001
>I guess you mean OPA2365? Looks pretty good, actually.

Yes that is the dual version

>> This would possibly allow the decommissioning of the +/-5v Dc
>> switched supplies & their noise issues?
>Probably. But you loose output swing, so you need to redesign the analog
>stage to output less than 5.6V PP (which it does now).

The output can swing to within 10mV of the Rails.


>> I also query whether a low jitter clock would improve matters much.
> I suspect it depends on application. The conventional way of injecting a > clock signal into the Xtal-In pin is subject to the chip's internal
> ground-bounce which is material, so your nice low jitter clock gets
> mangled.

There is an interesting post on the Squeezebox forum where a member has reduced the voltage of the PS to his Tent clock which is directly supplying a Xilinx CPLD & he reprted that the sound improved significantly when clock voltage was dropped from 3.3V to 1.6V. Thread here: http://forums.slimdevices.com/showpost.php?p=498889&postcount=133

>> The existing 24MHz crystal feeds the Cypress chip & is then passed
>> to the Xilinx chip where audio clock(s) are synthesised. This
>> synthesis is most likely the cause of the majority of jitter found
>> on the MCLK at the DAC.

>I know dedicated fractional-n clock synthsisers that manage under 50pS >additional jitter over the base clock, given a clean supply to the PLL >and good ground routing to the PLL ground...
As you said in the reply to glt - my clock experiment may have been a less than optimal configuration - I would be open to repeating it but how would I isolate the clock ground from the Cypress ground which you suggest may well be dirty?

 

RE: Musiland Monitor 01 US - more listening, posted on December 30, 2009 at 16:56:18
glt
Audiophile

Posts: 39
Joined: May 20, 2001
Thorsten,

Thanks for the reply.

>An external oscillator will not be subject to that kind of noise,
>if it is fed from a clean supply
>So there is a possibility of improvement.

I suppose you imply that the external oscillator (even though it is based on a similar crystal) feeding the pierce oscillator in the cypress chip will be better than a crystal feeding the pierce oscillator. Wouldn't the resultant clock be subject to the same PS noise because the internal pierce oscillator cannot be bypassed?

-----------------------------

H I F I D U I N O

 

RE: Musiland Monitor 01 US - more listening, posted on January 1, 2010 at 20:06:16
Thorsten
Manufacturer

Posts: 4209
Location: Somewhere nice on planet dirt
Joined: September 25, 1999
Hi,

> I suppose you imply that the external oscillator (even though it
> is based on a similar crystal)

This is a basic fallacy. The crystal is just a basic resonance circuit, in a can. Depending on the actual oscillator circuit used you get different levels of power supply noise immunity and self noise.

The Single Gate Pierce circuit is among the many "less good" oscillators. So the same cyrstal in a different oscillator circuit may perform very different. And a high quality "canned" oscillator or a so-called suplerclock may use a very different kind of actual crystal, as different ways to shape (cut) the crystal result in a different Q and hence different non deterministic (noise-like) jitter.

> feeding the pierce oscillator in the cypress chip will be better
> than a crystal feeding the pierce oscillator.

The crystal is not FEEDING the Pierce oscillator, it is part of it, together with two capacitors and a single cmos logic gate. You can find details at wikipedia (among others).

A Pierce type oscillator is very sensitive to powersupply noise. While researching VCXO's (I decided thater that VCXO's where so 1980's and do not use them) I found a patent using a Pierce Oscillator with it's power supply voltage varied as "low cost VCXO".

> Wouldn't the resultant clock be subject to the same PS noise
> because the internal pierce oscillator cannot be bypassed?

Well, in "internal Pierce Oscillator" is not there if it is not used as oscillator. If not used as oscillator it is a simple gate.

It is still subject to switching point modulation due to the power supply variation. However this can be addressed by not feeding the clock signal referenced to ground and if using a ground referenced clock is used by minimising the rise time.

So "bypassing" the Pierce oscillator is not the issue nor relevant.

The on-chip ground-bounce is one of the things that cannot be overcome and forms the final barrier to low jitter. From observations of the results of others and of my own I'd estimate this "barrier" to be in the region of 50pS to 100pS jitter for 44.1/48KHz sample rates. Higher sample rates tend to be subject to more jitter.

By using a set of flip-flops (preferably a single flip flop case per signal) as single bit flip-flop we could get lower jitter than that, however this requires a directly available clock with the desired frequency that also synchronises the the data source (something sadly not possible with the Musiland).

AND it would require a DAC not subject to significant internal ground bounce, which essentially eliminates anything running at high oversampling factors, with integrated digital filters and with delta sigma modulators or in other words which eliminates each and every single DAC Chip in current production.

Ciao T
Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

 

RE: Musiland Monitor 01 US - more listening, posted on January 2, 2010 at 06:24:42
Thorsten
Manufacturer

Posts: 4209
Location: Somewhere nice on planet dirt
Joined: September 25, 1999
Hi,

> The output can swing to within 10mV of the Rails.

So it can swing a little less than 5V PP and will clip at the last 2dB or so. And as some USB outputs produce less than 5V DC it may clip even earlier.

I think one would have to redesign the output stage to produce around 4.5V PP to be on the safe side.

> There is an interesting post on the Squeezebox forum where a member
> has reduced the voltage of the PS to his Tent clock which is directly
> supplying a Xilinx CPLD & he reprted that the sound improved
> significantly when clock voltage was dropped from 3.3V to 1.6V.

Interesting. I used to use Tent Clocks quite a bit. I remember guido always recommending feeding the clock with 5V and to drop the signal by a divider following the clock. I always found this better than dropping the supply in my mods, though ultimatly using a seperate buffer (including seperate supply) before the divider isolating the Clock from the load was an even better choice.

Ciao T
Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

 

RE: Musiland Monitor 01 US - more listening, posted on January 2, 2010 at 09:14:37
glt
Audiophile

Posts: 39
Joined: May 20, 2001
Thorsten, again thanks for your contribution and Happy New Year!

>Well, in "internal Pierce Oscillator" is not there if it is not used as >oscillator. If not used as oscillator it is a simple gate.

> ...

>So "bypassing" the Pierce oscillator is not the issue nor relevant.

This part makes a lot of sense.

I agree that different cut crystal will result in different jitter, thus my thought of replacing the crystal and the 3.3V regulator. But those "special crystals" are impossible to source, and crystal manufactures such as Crystek seem to be using the simple pierce oscillator in most of their products (except perhaps their top end models that come in a large can). If you look at their technical literature it is all about pierce-gate oscillators.

-----------------------------

H I F I D U I N O

 

RE: Musiland Monitor 01 US - more listening, posted on January 2, 2010 at 09:52:48
Thorsten
Manufacturer

Posts: 4209
Location: Somewhere nice on planet dirt
Joined: September 25, 1999
Hi,

> But those "special crystals" are impossible to source,

Not exactly impossible, but a little difficult, plus you get pretty substantial MOQ.

> and crystal manufactures such as Crystek seem to be using the
> simple pierce oscillator in most of their products

Yes, but even then at least you have separate ground and supply pins and you can supply it from fully independent supply, very low noise too (BTW, batteries generally are not VERY low noise), something that is non-trivial if at all possible with on chip oscillators.

Also, who said you had to buy crystek.

There are quite a few mil spec suppliers that are used to unreasonable demands (you should see the kind of specs the military geeks turn out - making a nice audio grade oscillator is actually almost a holiday to those guys) and are very happy to make you a very low jitter all analogue sinewave oscillator with discrete circuitry for low jitter "clipping" circuitry to give a good fast slope square output, for a (high) price and around 100 pcs MOQ.

Of course, even at group-buy levels you are unlikely to get enough subscribers for the MOQ.

Ciao T
Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

 

RE: Musiland Monitor 01 US - more listening, posted on January 2, 2010 at 16:38:19
Tony Lauck
Audiophile

Posts: 13629
Location: Vermont
Joined: November 12, 2007
"AND it would require a DAC not subject to significant internal ground bounce, which essentially eliminates anything running at high oversampling factors, with integrated digital filters and with delta sigma modulators or in other words which eliminates each and every single DAC Chip in current production."

Not necessarily. It would depend on how carefully the chip layout was done and the various pinouts.

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

 

RE: Musiland Monitor 01 US - more listening, posted on January 2, 2010 at 20:52:14
Thorsten
Manufacturer

Posts: 4209
Location: Somewhere nice on planet dirt
Joined: September 25, 1999
Hi Tony,

> Not necessarily.

Actually, necessarily.

> It would depend on how carefully the chip
> layout was done and the various pinouts.

If this is done well it will help some, but as long as you keep the DAC on the same chip as various DSP processing and have a DAC running at high switching speeds that "some" is not a lot.

Any DS type DAC that has leadouts is still subject to ground-bounce, even if it has completely seperate leads. Anything switching causes on chip ground bounce (and supply voltage modulation) due to inductance of the leadframe and bondwire's resistance.

A traditional full multibit DAC is much less subject to the problem, as it only switches at the sample rate (normally as low as 384KHz). In typical current manufactured DS and Hybrid DAC's the underlying switching frequency is around 25MHz.

As for ESS, their DAC is subject to the same limitations, just changing pinning cannot overcome. As I have not seen independent jitter measurements for it I do not know if it gets below the 50pS jitter Barrier.

Based on what is in their datasheets and other literature I doubt it, but I'll wait for the Stereophile Review of Gordon's Async USB DAC with the Sabre Reference. I was looking putting it into a product, but the pricing ESS has is just ridiculous and their way of dealing with potential customers is so off-putting, forget it.

I can buy a full stereo PCM1704 solution Chipset for less and that gives a single ended current out and can be run non-oversampling and so gives me exactly what I want. So I just buried the ESS papers in a dark recess.

Ciao T
Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

 

Superclock4, posted on January 3, 2010 at 10:32:12
audioengr
Manufacturer

Posts: 6017
Location: Oregon
Joined: April 12, 2001
Thorsten wrote:
"The best PLL circuits manage to create such clocks from a "jitter-less" base clock (the 24MHz in the Musiland) with under 50pS (not ppm) Jitter, which is as good as the real-world performance of many a "super-clock" once integrated in a given product (remember the ground-bounce)."

The new SC4 is specified at 2 psec RMS jitter. It is actually better than several other custom monolithic digital clocks rated at 2psec, which I also use.

Steve N.

 

High-preformance parts, posted on January 3, 2010 at 10:39:04
audioengr
Manufacturer

Posts: 6017
Location: Oregon
Joined: April 12, 2001
Almost all really high-performance IC's are only sold in large quantities. This is because they are usually low volume and expensive to make or "tune". It's the run-of-the mill parts that are cheap and available in qty of one.

This is one of the reasons why DIYers can only go so far with their mods.

It takes a large investment to make really high-performance designs. One company can make this investment and then spread the cost over many hundreds or thousands of products.

Steve N.

 

RE: Superclock4, posted on January 3, 2010 at 15:28:06
Thorsten
Manufacturer

Posts: 4209
Location: Somewhere nice on planet dirt
Joined: September 25, 1999
Hi Steve,

The specified jitter (even measured) of a "Superclock" (any such) is one thing.

The resultant "in system" jitter when the the clock is applied as after market mod into a given piece of equipment or even designed in is another story. It was this "real world system performance" I was referencing in the above post.

Ciao T
Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

 

RE: Musiland Monitor 01 US - more listening, posted on January 3, 2010 at 19:47:02
Tony Lauck
Audiophile

Posts: 13629
Location: Vermont
Joined: November 12, 2007
I don't think the switching frequency is what matters. It's the audio frequency that matters. At least that will be the case if the output of the DAC is glitch free. For an illustration, if a DAC is reproducing a DAC signal then jitter won't have any effect on the analog signal output by the DAC.

In any event, I agree that the only thing that matters is the effect of jitter on the actual analog signal output. All of the clock signals could be absolutely perfect to no avail if the clocks have become noisy at the only place where they count: the actual analog switches.

I suspect that ESS has addressed these issues in their expensive chips, including various ways of balancing out errors using dynamic element matching. (Assuming that their 132+ DB S/N numbers aren't just marketing smoke.)

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

 

RE: Musiland Monitor 01 US - more listening, posted on January 4, 2010 at 08:11:03
Thorsten
Manufacturer

Posts: 4209
Location: Somewhere nice on planet dirt
Joined: September 25, 1999
Tony,

> I don't think the switching frequency is what matters.

Sure it does.

Look at it that way. We got what is called LIM by Ed Meiter.

If the DAC is fed by a suitable digital signal and switches at a suitably high frequency we will observe sifts in the switching point voltage.

Now the signals even on a single chip do not have infinitely steep slopes. So, as we modulate the switching point we modulate the timing - so we have jitter.

This jitter will be lower if the switching frequency is lower.

I suspect the lowest systemic incidence of this type of jitter would be happening with a DAC operating at 1Fs Data without any Delta/Sigma modulation and re-clocked by using individual dual flip-flops separate for all the signals of the I2S bus.

The higher the Delta/Sigma switching frequency the more ground bounce and supply modulation will be caused by the inductance of the actual internal lead-outs and other traces outside the actual IC and will in turn cause LIM (type) Jitter.

Now all this assumes the best clocks and practices everywhere.

From the reviews of other peoples designs I find that not much out there seems to better 100pS Jitter and very little even gets down to 50pS, even if the Clock is a near unobtainable 1ppm/1ps type and all else is maximally optimised.

ESS poignantly omit relevant jitter measurements even in the kind of literature that you have to signa endless NDA's to get it, so I cannot comment how much better (if any) ESS is.

> I suspect that ESS has addressed these issues in their expensive
> chips, including various ways of balancing out errors using dynamic
> element matching. (Assuming that their 132+ DB S/N numbers aren't
> just marketing smoke.)

One can take measurements many ways. I found some very cute ways of abusing an AP System Two to measure stuff it was never intended to and to use a lot of averaging to tune out all sorts of noise, which I use to get a clearer picture of deterministic stuff normally burried under the noise.

Without a full and complete disclosure how a given measurement is done it is hard to to be sure measurement results are truly comparable.

So, I'll wait for the first Stereophile review with full measurements of an ESS toting product (then I will know how it compares to other stuff in measured terms, including my own), while I get on with other stuff.

Ciao T
Thor

At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?

 

RE: Musiland Monitor 01 US - more listening, posted on January 4, 2010 at 09:11:34
Tony Lauck
Audiophile

Posts: 13629
Location: Vermont
Joined: November 12, 2007
LIM is just noise caused by digital logic that couples into the clock circuitry. Nothing magic, except that it has been trademarked by Meitner. (Snake oil flag goes up...)

There are various ways to minimize these effects. As you say, time will tell how well various approaches work out. However, in the case of the SABRE chips there are already good reports as to their sound, at least in the Twisted Pear Buffalo DAC kit.

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

 

Page processed in 0.041 seconds.