Audio Asylum Thread Printer
Get a view of an entire thread on one page
|For Sale Ads|
I have been aware of these Musiland units for a while (and had some samples around). While doing some work for diyhifisupply we also included one of them in a heavily modded version as USB2SPDIF, this project may or may not go into full commercial mode, if interested anyone can read about at the diyhifisupply section on thishere site.
Meanwhile, having these little Musilands lying about (together with loads of other eval and demo boards, experiments and so on) encourages their use.
One of the reasons is the really good and solid software support, easy to get to work and rock steady (more than I can say about some other stuff even from big names), work bitperfect, isolated SPDIF outputs etc.
They generally get used as impromptu signal generators (with some software) and the Monitor 01 US I occasionally attempt to use as portable headamp to go with my Laptop (not really seriously, my windows smartphone always sounded better!!!).
Okay, so much for background.
After finding major problems to have a very nice SPDIF receiver lock reliably onto the Musiland Monitor 01 USD SPDIF signals I decided to look seriously inside either of the "Monitors" to see what is going on.
Looking closely I was VERY disappointed to find many of what I would consider serious and very unnecessary design flaws.
Fred has already touched on the Monitor 01 USD output level. On the unit in my hands I measured a little over 200mV peak-peak of actual signal and extra 100mV spikes of overswing at the transitions. That is the picture on the left.
(please note the "grass" to the two sides of the cetral slope is how my digital 'scope reacts to SPDIF data, it cannot really trigger that well on this signal or I'm too daft to make it trigger well, it is not actually in the signal)
Measurement was using 1.5m Belden 75R Video Cable (the good plenum stuff, forgot the #) with either decent RCA's that measure not too far off 75R (cheap stuff made for video) or 75R BNC's with the far end terminated with a socket to which two 0805 SMD resistors of 150 ohm where soldered with minimum lead length as load, measured at the output of the Musiland with a calibrated 10:1 probe, sample rate 96KHz.
Well, it seams that with the cable length so short the transmission line effects are not very pronounced, but the output transformers leakage inductance together with the cable capacitance ring like mad.
In order to get the trace on right side I removed the voltage divider Musiland included, which is wrong on all levels. The output transformer is driven from 90 Ohm (not 75) and the level is correct if the load is NOT 75 Ohm terminated but less than halve of the IEC Standard minimum if it is.
The combination of these two can give several recent designs of SPDIF receivers major problems to lock, only "old faithful" CS8412/14 did fine, but the output was so jittery, you could see the jitter on the 'scope with a naked eye!
When I had tested the Musiland in my system at home it had done sonically much worse than the modified motherboard SPDIF output from dedicated PC Player, sounding gritty, forward and spatially challenged.
I first tried to simply drive the transformer directly from the buffer with just a cap and find a resistive divider on the secondary that would match the transformer impedance and give about correct levels, no luck, probably way too much parasitic inductance to squash using this methode.
So I replaced them with parts (sorry, you can't buy these anyway, so don't ask) that I know work under this kind of application and fitted the PCB (I did not want to waste more time) and applied a suitable divider after the transformer that presented the correct load to the transformer AND the correct 75 Ohm output impedance (well, close, I did not much about with the snubber to offset the rather low leakage inductance). This ended up with around 800mV peak-peak output, pretty much smack bang in the middle of the range specified in IEC standard.
This gave something close to the right trace and my SPDIF circuit finally got a solid lock. But the traces where very furry. The same "fur" appeared on the power supplies. The power supply filtering in the Musiland is vestigial at best and noisy USB Power rides straight across the input filtering and through the regulators to appear on the rails and eventually on the output. I measured nearly 150mV peak-peak on the various chips, all stuff above around 60KHz or so and really wideband. I need to get a faster scope (mine is only 150MHz) to measure just high it went. As a result I would expect different USB cables and USB ports and the like to make a disproportionately large difference in sound quality.
Anyway, Musiland has a place for a filter choke but elected not to fit it and the capacitors they use in the power supply do not well at high frequencies to filter out incoming crud.
Okay, I fitted a simple filter using an SMD choke, some ceramic SMD caps and the biggest Os-Con I had around. This finally cleaned the fur up pretty well, noise on supplies all across the board (meaning the circuit board) is now in the region of a few mV, showing that Musiland really did an excellent job with board layout and decoupling the chip supply lines.
Given all this technical excellence in the difficult part of the hardware and the software the oversight of such extremely basic stuff as I had to fix and the way in which it cripples a potentially excellent and affordable product is really disappointing, especially as fixing these issues at the design stage costs pennies (well, okay, except the Os-Con(s).
If you use a Musiland Monitor 01 USD and you have problems with your DC not locking or major jitter issues, try to find someone who knows what they are doing to fix these issues, sadly after such a modification job the Musiland may not be so good value anymore.
The Powersupply noise issues can be fixed using an external supply (easily attached actually) but the problems with SPDIF outs are much more serious and not easily fixed.
PS, a quick and dirty fix for the output problems of the Musiland is to remove the transformer and the (SMD) resistor divider and to then short the series resistor (tiny SMD soldering required, sorry) and to use a 220 Ohm Series resistor plus a 110 Ohm resistor to ground soldered in instead of the transformers (can use 1/8W types and you can actually see the resistors when soldering).
This gives around 0.55V peak-peak and 75 Ohm impedance, but no isolated output. Depending on the computer used this lack of isolation may be a deal breaker. But the traces look quite clean and in practice I found this working well, much better than the original output anyway, not that this is saying much.
Edits: 12/17/09Follow Ups:
I located the datasheet for the Output Transformers:
It is specified to match the ancient Crystal receivers/transmitters and is specified for 7Mbps (that is basically 48KHz sample rate).
This does not seem very good to transmit the four times faster rate at 192KHz.
Indeed, the waveforms I saw seem not too great for 44.1KHz, though still better than some of the show of horrors shown on Lampizators website:
(All these are SPDIF output traces from CD-Players/Transports long discontinued at 44.1KHz, so it is possible to do much worse than Musiland)
I measured the rise/falls times at 176.4 k. They are almost the same at about 7 nS.
It is not a bad thing sometimes to 'slow' the signal. By adding a little output inductance, I have virtually no overshoot, very little ground bounce, and very clean rise and falls. The sound has also improved.
I am going to use the musiland monitor 01 to add a USB input to my dac. Could yuo please suggest a way to automatically switch the proper input when a USB signal is fed to the musiland?
I looked again at the digital output of the externally powered Musiland 01 usd thru a 75 R BNC connector which replaced the phono socket on the device. There was no sign of the issues you displayed. The signal is very clean, has the classic spdif shape with low time axis spread, and a little ground bounce with the first cycle at about 110 MHz, decaying to about 80 MHz in the 2nd cycle. Te transition edge was clean and the only issue is a degree of noise on the top hat which was cleaned up when I applied a 20 MHz filter. The 50R bnc output that comes as standard was not much different overall, but does sound slightly 'mellower'
Maybe further bypassing of the internal regulators can improve things but this is another story. I don't mess with battery supplies 'cause they are long term nuisances (from experience).
The 'slow' pulse transformer did not appear to be an issue at 176.4k which was what I fed into the device. The waveshape was as good as any other PC device I have seen and is nearly on par with the HiFace. However, the pk-pk volatge was indeed lower than standard.
The set up was:
Vaio usb - 10 cm shielded usb cable with Vbus cut and power injected from 5V alw regulator with 7 uV noise re 1MHz bandwidth - 3 m bnc silversonic 75R cable - 75R termination - Tetronix 2465B scope. The same trace shape was there with a different PC and a 1.5m Illuminati XV2 cable.
The one serious issue I have, is that, having updated the firmware and updated the driver to 188.8.131.52, subsequent reinstall of any software at 184.108.40.206 or above fails as device is then not recognised. The Musiland website is so 'secretive' that I cannot register onto the forum to post the issue. They really need to sort this out if they want world sales growth.
If anyone has resolved the issue, please post. For myself, I have tried the usual tricks and I think it is soemthing that is written onto, or missing from the registry. I do use a reg cleaner and may be this has taken its toll. The reg key from Musiland after re-registration cannot be added just by clicking on it as all I get is 'device not found'.
There is also the issue of transfer among PCs where nothing higher than 220.127.116.11 works, and the posted install instructions do now work either..
I don't have the same issue that you have identified; the spdif output being decent and there is no furring as shown in your picture. What I have found is that the output frequency deviation is a constant 100 kHz.
It is my prectice to relock spdif outputs or use a dac with v good clock cleaning and regeneration capabilities.
I agree that the 18.104.22.168 software is solid. However, the 1.0.5 drivers has been hacked such that the digital levels depend on the PC and instaed of 100, shows 78 or 156! To my ears the sound is more mellow and worse.
The 1.0.7 driver does not work on any of my PCs at all (firmware not recognised).
When using the 22.214.171.124 driver, the best sound is obtained by backing off digital levels to about 96.
> I don't have the same issue that you have identified; the spdif
> output being decent and there is no furring as shown in your picture.
Interesting, what 'scope do you use? What is the frequency limit?
Also note that I measured with a 1.5m SPDIF cable attached and terminated at the far end, not directly into a 75 Ohm terminator at the unit, as I am interesting in the practical performance, not theory...
Both the impedance mismatch and the excessive leakage inductance of the output transformer are shown up well when measuring like this (and yes, the far end looks even worse).
> It is my prectice to relock spdif outputs or use a dac with v
> good clock cleaning and regeneration capabilities.
That certainly helps.
> I agree that the 126.96.36.199 software is solid.
Yes. And I like the sample rate auto adjustment.
> However, the 1.0.5 drivers has been hacked such that the
> digital levels depend on the PC and instaed of 100, shows
> 78 or 156! To my ears the sound is more mellow and worse.
> The 1.0.7 driver does not work on any of my PCs at all
> (firmware not recognised).
Then don't use them... :-)
Analog Tetronix > 400 MHz scope, with 75R termination at the scope.
The stock transformer doesn't look bad in terms of waveshape and there is less ground bounce and ringing than the M2Tech HiFace. Trace brodening is marginally worse thyan the HiFace but better than many pci sound cards.
Which software driver are you using and have you tried KS on the WDM panel?
> Analog Tetronix > 400 MHz scope, with 75R termination at the scope.
Hmm, here digital 150MHz HP, with 1:10 probe to minimise capacitive loading. And as said, a real cable attached.
> The stock transformer doesn't look bad in terms of waveshape and
> there is less ground bounce and ringing than the M2Tech HiFace.
> Trace brodening is marginally worse thyan the HiFace but better
> than many pci sound cards.
Okay, the results are not bad for PC audio, but by high-end HiFi standards the results are poor. After my mod's the musiland outputs match my high end designs and sound much bettet.
> Which software driver are you using
The one that came in the Box, 1.0.5
> and have you tried KS on the WDM panel?
My playback software (not CPlay) is strictly ASIO.
I have an HP 150MHz with FFT module. It is not really much good for looking at fast digital waveforms; there are more dots than I want to see! My scope observations do not involve using probes but through a high bandwidth 75R 1.5m cable terminated into 75R at the scope.
I don't believe that one can mod the Musiland to high end audio standards. For a start, the F sythesis does not output a tight sampling rate (100k out). The only way that it starts to approach high end sound is by relocking the output or by playing thru a dac with v low jitter clock cleaning/internal clock. I used a Universal Audio 2192 and a Buffalo dac for listening. An external power supply such as a Jung type regulator is also needed. The usb V bus must be cut as well.
Does the 1.0.5 driver on your system give 100% scaling on the asio or wdm boxes?
> I have an HP 150MHz with FFT module. It is not really much
> good for looking at fast digital waveforms; there are more
> dots than I want to see!
Maybe, I find it very useful and nowadays very rarely get the Tek analog 'scope out.
> My scope observations do not involve using probes but through a high
> bandwidth 75R 1.5m cable terminated into 75R at the scope.
So you are looking at the end with the 'scope in 1:1 mode, while I am looking at the beginning with a probe that minimises capacitance. So rather different methodes. Also, mine was powered of the USB bus.
> I don't believe that one can mod the Musiland to high end audio
> standards. For a start, the F sythesis does not output a tight
> sampling rate (100k out).
Sorry, I do not follow? How can the sample rate be 100KHz out? I am getting the expected frequencies everywhere. Do you mean 100Hz?
Anyway, I was merely referring to the actual measured behaviour of the SPDIF output, not to the overall performance. What I have found however, is that jitter with the mods I did (01 USD) is low enough that my AP can't show it, which is not very, very low, but lower than many CD-Players/Transports and lower than any PC Audio device I had at hand.
> Does the 1.0.5 driver on your system give 100% scaling on the asio
> or wdm boxes?
Sorry, I do not follow? How can the sample rate be 100KHz out? I am getting the expected frequencies everywhere. Do you mean 100Hz?
Sorry, 0.1k not 100k. Still locks on my Big Ben and UA2192.
> Sorry, 0.1k not 100k. Still locks on my Big Ben and UA2192.
Hmmm, need to test again, but I did notice this in the lab...
Is it 100Hz at 44.1KHz and scaling or 100Hz at any rate?
At all rates.
Your 10:1 probe; you are measuring something like 0.03V over 1.5m with no termination? The HP, from memory, is not that sensitive
> Your 10:1 probe; you are measuring something like 0.03V
> over 1.5m with no termination?
> The HP, from memory, is not that sensitive
My one is 1mV/div.
And the second trace was measured the same way, so the 'scope and probe where not the limit.
I have just done a re-comparison of asio v ks, and ks even with 5 only buffers, sounds clearly better; more transparent and clearer with no aggression.
By mistake, I plugged in a 317 power supply instead of the alw and this sounded both aggressive and not nice.
Re: 317 vs alw: is that before the local regulators or replacing the local regulators?
Happy Holidays ~ Rio
I feel of all the Musiland models, the easiest to mod is the MINI. I have some pictures in my blog
The other receiver that this device in stock form works well with is the Wolfson WM8804/05
I couldn't tell any differences with battery power in lieu of USB power or by using large bypass capacitors on the on-board ps regulators, but my own experience (and ears) tells me that the accumulation of small improvements will at some point show up as an improvement, so I keep doing these small improvements hoping for some tangible improvement at some point
The spdif is already "pre wired" in the MINI and I just added a resistor which is just a voltage divider with the 75 ohm in the receiver end
The i2s wires are also easy to tap into, but the only drawback is the the mck for 44.1K material is at 128fs. Perfect for BB DACS but not so perfect for Wolfson, AKM and other DACs
Here the respective Mods on my Musilands:
And a second view:
Modification list in short:
USB power CLC Filtering added, with big os-con
SPDIF Output Circuit fixed with different output transformers
SPDIF Output Circuit fixed with changed attenuation circuit/values
SPDIF connections made with 75 Ohm miniature PTFE Coax
And here the Monitor 01 US:
As can be seen this has been massively os-conned...
Modification list in short:
USB power CLC Filtering added, with big os-con
LC filtering added between main power and switched supplies (big os-con)
All local decoupling Capacitors upgraded to various Os-Cons
After this I find the devices rather good sonically.
I suspect adding a 'super clock' (how I hate that term) and a good linear 5V supply (+/-5V for the 01 US and bypassing the switchers) would take these devices to an even higher level of performance. Well, maybe over the christmass break...
If you are a music lover and not an electrical engineer it sounds pretty good, especially for $70. I usually use mine with toslink, but it sounded as good via coax as the SPDIF produced by my Lynx L22. I appreciate your point that it can be improved for a small investment, but it would be inaccurate to leave people with the impression that it is a piece of junk. It is in fact a great way to play high rez files via a USB port.
The easiest way to improve performance is to cut a usb cable at the PC output (Vbus Red) and connect a decent 5V supply to this and ground.
Also change the bnc/phono sockets to 75R. I just changed the phono to 75R and this sounds better than the 50R output.
Back off the digital output level to prevent 'hot' files from overloading the dac (from 100 to 95) and the unit sounds v good; much better than the 0404 usb. Also try KS using the WDM drive mode instead of ASIO. Disable the other not used.
I bought the Musiland after you praised it in a post a few weeks ago. I have been using mostly ASIO4All v2 with J River which sounds great. It sounds substantially better than the Lynx coax SPDIF using the Lynx ASIO driver, which surprized me. Is that the KS you recommend?
I may try the battery mod. Does it matter whether the + or - goes to ground? Does this make a dramatic improvement? Since I am primarily using toslink, I probably will not change the bnc/phono sockets.
The receiver chip in my DAC is a Wolfson wm8804. Maybe that makes a positive difference.
No, it is kernel streaming mode in windows xp. Foobar supports it but I have never used J River. Have a look under OUTPUT and see if this is available.
It is not possible to find a tight 5V supply using batteries. Li is too expensive and requires voltage regulation. Li-ion likewise for voltage. You may be able to use akaline at 1.65V in series but I don't suppose an AA cell setup will last that long. Still it may be worth trying. Bypass with a good 470 to 1000u cap. and DO NOT connect polarity wrongly. I use a ALW Jung type regulator.
You could always go directly to 3.3V supplied from a battery, lift the 3.3V leg output pin of the 3.3 V regulator & feed the battery 3.3V to this output pad! This 3.3V then feeds to a 1.2V regulator.
The local regulators are spec'ed at 1.3V maximum dropout voltage. So the minimum input voltage is 3.3+1.3=4.6V. I'll say 4 nimh cells should work well. (4.8-5V). I think the device consumes 200-300 mA, mostly the USB controller, so a 2500mah cell should last many hours.
Okay, it may seem that my comments on the Musiland stuff are very negative. That is perhaps because I am generally impressed with the engineering and maturity of the product and I feel that in the light of this the issues on such basic levels as I fond them are by far more infuriating than they would be in a bad product.
If the SPDIF outputs work well for you with your DAC, all the better. I can think of some most excellent recent DAC's (using non Cirrus Logic SPDIF receiver chips) that would/could struggle with this signal though.
Okay, what I like about about Musiland monitors:
Rock solid, easy to use driver software, works well, easy to get the best on that front.
General ease of use, having tested a number of alternative options this one is great.
What I think is bad:
No Mac or Linux Drivers. No option to use external power supplies (this could help tremendously).
What I think is positively UGLY:
Wrong SPDIF output impedances and levels (and poor waveform) in the 01 USD (note, these tend to be more common in PC audio products than you may think - Musiland are not alone there), which with many will kill the potential of High Rez sound files and in my experience even CD.
Badly handled and completely unnecessary switch mode supplies in the 01 US that make a mess out of the sonic potential.
That said, compared to some other stuff out there that costs more (EMU0404 USB anyone) the Musiland devices are quite an achievement...
Thanks Thorsten, some excellent analysis backed by measurements on the Musiland units. I have stayed away from the 01USD as I was interested in I2S to a local DAC (in my case an ESS9022 - a 2 channel Sabre Vout DAC), both the Musiland, DAC & clock fed from individual 3.3V battery supplies. Sounds very good.
I've since got a M2Tech Hiface to compare to the Musiland. Unfortunately, I can't listen to it stock as I don't have a SPDIF receiver so I'll be opening it up to take out I2S to the same local DAC. In theory, this should be a lower jitter configuration compared to the Musiland.
Have you heard or measured a HiFace?
PS. I meant to also say a BIG welcome back to the forums & hope your high quality contributions can continue - I look forward to them & have miised your posts over the last 3 years or so!
After the above experience with the digtal output only Musiland USB Gizmo I decided to take a second hard look at the one with the integrated DAC (the PCM 1793 is not the last word but not THAT bad and the OP275 analog stage ain't exceptional but should do okay) to see why I could not stand listening more than 10 to 15 minutes to the thing and then only at quiet volumes.
Well, first, there are two switched mode supplies on board, they create around +/-5V Supplies for the Op-Amp output stage and probably also for the DAC's analog supplies.
These switched supplies where not a total disaster, but left around 50mV peak-peak noise at the 100KHz+ switching frequency, as the reservoir capacitors had trouble handling this frequency well. A pair of big (220uF/10uF) Os-Cons fitted instead plus one each 10uF SMD ceramic cap squashed the switching noise to less than 1mV peak-peak though one switcher started to have maybe 2mV p-p of around 2KHz ripple, no idea where that came from, but one can live with that I think.
What was worse was that these two switchers threw over 200mV peak-peak noise at their switching frequency into the USB Power line which of course fed the rest of the device so this same noise was on each and every power supply node minimally attenuated!
Next stop was a SMD choke and big Os-Con separately for each switcher to kill this noise. After that what was left was the same USB Port supply noise I had noticed with the 01 USD, so I attacked it the same way. As I had already dropped five (!) expensive Os-Cons into the gizmo, I finished off replacing the remaining original (6pcs) local decoupling caps with Os-Cons as well.
Now poking the 'scope around the PCB reveals very little noise anywhere, again, surprisingly quiet for such a complex and high speed logic design.
Sound quality is much improved I would say, at least i can listen on 'cans quite loud and by now several hours without fatigue. I suspect the modification of the 01 US the way I did will beyond the abilities of most of the people here.
If you happen to buy a 01 US to pick off the I2S signal from the chip (in principle smart move) make sure to remobe the DAC IC, Op-Amp AND to turn off (cut the trace to each switcher) the switching supplies, plus power the whole PCB from a clean 5V supply instead of the USB bus and you might have something quite passable in return.
The Musiland platform should be capable of around 100pS or thereabouts jitter if fed a clean supply, which not a lot of DAC's and CD-Players equal. Just add your favourite DAC chip (make mine a TDA1541 please), analog stage (make mine a Tube, maybe the Universal Tube Stage diyhifisupply sells) and you have a nice DIY asynchronous USB DAC on a modest budget.
The result may not have Gordon or Charles quaking in their boots, but could be nice for "cheapo fun".
Maybe someone will pick up the OEM solution Musiland offer and incorporate it into an affordable kit or something? That way one might end up with an excellent solution that does mess up SPDIF signals and power supplies and has a better clock solution.
I hope this helps some of those of you who have been asking privately for info and where I lacked time to answer.
This is why I posted an opinion that the 01USD is better as a general converter.
The only compelling reason to get the 01US is because of access to the I2S signals which aren't available on the 01USD. But as glt has shown, a Mini 01US is an better (& cheaper) option, has I2S as well as SPDIF signals but is perhaps a bit small for modding!
Fred, interesting observation on the drivers!
Post a Followup:
Post a Message!
This post is made possible by the generous support of people like you and our sponsors: