PDA

View Full Version : Squeezebox Duet



leo
29-06-2009, 22:53
First we'll start with the external psu, supplied with the Duet comes a cheap plug in SMPS , I decided to try a well filtered linear based PSU using LM1086 to see if it offered any performance boost in this particular application, I found it slightly improved SPDIF and analogue output, because I felt the move from SMPS to linear improved performance I decided to go a little further.
Apart from Paul Hynes I don't know of any other alternative upgraded psu's in the UK for the SB's so I tried one of his PR3 modules, this required a small 12v transformer which I already had , other bits needed like the case, cable etc came from Maplin

Heres the PR3 installed in the Maplin case

http://i273.photobucket.com/albums/jj220/kingbusoms/008-3.jpg

If you don't want to go the diy route Paul sells complete Squeezebox and Duet Receiver upgraded psu's called SR1 which can be found here http://www.paulhynesdesign.com/page7.html

So how does the PR3 compare against my diy linear? In all honesty it was noticeably better! What was very interesting is that the Receivers analogue out didn't sound bad at all now! Its quite surprising what influence psu's have on a units performance , I'd say the analogue out was on par with most decent mid priced cdps I've heard where as before with the SMPS it was more like a cheap DVD player

The next stage was some internal mods, the first round is going to be basic stuff easy for most to try if they wish, nothing new here:)
Heres the standard Duet pcb

http://i273.photobucket.com/albums/jj220/kingbusoms/041.jpg

First thing was to replace the filter cap on the input DC socket, the standard one is a general purpose SMD type which is probably next to useless especially if using the external SMPS, I replaced it with a Oscon , any low ESR should be ok here, stability using low ESR in this position was fine

http://i273.photobucket.com/albums/jj220/kingbusoms/cap.jpg

The 74HCU04D is part of the SPDIF and XO circuit , its supply comes from the onboard 3v3 switcher used to supply other various bits and bobs, ideally you want a clean supply here so separating the line and giving it its own regulator is worthwhile ,for now I fitted a basic VBE supplying a 3v3 reg, this will be upgraded to something better later (I'll post results) .
I may also try using separate Hex inverters instead of the one shared for XO and SPDIF but the reg mod alone is certainly worthwhile for both analogue and mostly SPDIF out


Instead of butchering the track I lifted the supply leg and ran the new regulator straight to it
Warning, this is fiddly, its very easy to snap off the leg

http://i273.photobucket.com/albums/jj220/kingbusoms/74hcu04.jpg

With reg installed

http://i273.photobucket.com/albums/jj220/kingbusoms/reg.jpg

The Receiver also uses SMD ferrites on both the SPDIF out and analogue out, probably there to help filter possible noise because of the SMPS, A guy called ART posted some interesting measurements on Audiocircle about these ferrites used in the SB3 (only used in the SPDIF output for the SB3) , the SPDIF measured better without it so I also decided to bypass both the SPDIF one and also the pair on the analogue out

http://i273.photobucket.com/albums/jj220/kingbusoms/duetbits.jpg

What I intend to do later is make use of the internal I2S lines in the Receiver but for now I wanted to see what improvements could be done to both the SPDIF out and the internal dac

The dac is a cheap WM8501 , not exactly state of the art , these don't even require an additional active output stage
Datasheet can be found here http://www.wolfsonmicro.com/uploads/documents/en/WM8501.pdf


I next changed the output coupling caps which was a pair of general purpose 10uf SMD's, you can use what you like here as long as theres room in the case, I just used what I had to hand, a pair of Panasonic for audio (metallic blue sleeve) with a Wima film cap strapped across them, I'll faff about comparing alternatives later;)

http://i273.photobucket.com/albums/jj220/kingbusoms/capsreceiver.jpg

Theres a small 5v SMD reg for the WM8501 dac, this would certainly be worth upgrading too if your using the Duets analogue outs, I'll change this later along with decoupling caps and post results.
Theres more to try yet so will update as soon as I can, this will mainly be internal regulation.
I'm currently running in the Receiver but initial impressions are very promising

Covenant
30-06-2009, 06:03
Hi Leo,
As usual your ability and confidence is to be admired.
I am glad you noticed a difference with a linear supply-I certainly did with my SB3 classic yet the designer says it does nothing-strange.
Do your mods improve both the analogue and digital output?

leo
30-06-2009, 16:37
Hi Leo,
As usual your ability and confidence is to be admired.
I am glad you noticed a difference with a linear supply-I certainly did with my SB3 classic yet the designer says it does nothing-strange.
Do your mods improve both the analogue and digital output?

Thanks:)
It improved both SPDIF and analogue out to my ears, more so with analogue out imo, also the range in noticeable improvement with spdif varies depending on what external dac was being used so I guess mileage will vary.
The upgraded linear supply on its own is a good start, the other bits will be mainly tune ups

Covenant
01-07-2009, 13:04
Thought you might like to see my homemade (not by me) SB3 power supply. Its in a wood box that used to house knives and forks! The Virtual Dynamics power cable had to be disconnected because being a garden hose it doesnt bend very well.
I thought the linear supply was a good upgrade for 100 but its full potential is probably hampered by the SB3's internal regulators.

leo
01-07-2009, 17:58
Thought you might like to see my homemade (not by me) SB3 power supply. Its in a wood box that used to house knives and forks! The Virtual Dynamics power cable had to be disconnected because being a garden hose it doesnt bend very well.
I thought the linear supply was a good upgrade for 100 but its full potential is probably hampered by the SB3's internal regulators.


Looks good!
The internal regs in the Squeezebox's are obviously cheap but I think its that external smps which lets the performance down most.
One of the problems with SB3 is that noisy display polluting the things

leo
01-07-2009, 18:02
Just an update, I removed the wireless card out of the Receiver , the sound doing this has definitely improved some more (slightly but noticeable)
The only snag is the front led flashes pink continously , it thinks it has a fault:)

jkeny
06-07-2009, 01:39
well done Leo,
You certainly are a circuit Junkie - what external DAC are you running this into - Sabre?

leo
06-07-2009, 20:40
well done Leo,
You certainly are a circuit Junkie - what external DAC are you running this into - Sabre?

Ta:)

I use it with several different dacs, obviously these things are limited especially with higher res formats but their pretty convenient in use and don't sound bad once modded

jkeny
06-07-2009, 21:11
How does it sound with the Sabre? I'm interested in how the Sabre's ASRC handles the jitter that this SB puts out? Any better than sounding than into other DACs?

I'm attracted to some network solution like this & I was thinking along the lines of a Network Media Tank (NMT) device like the Popcorn Hour. Now these are primarily for video playback so their clock arrangements are not the best so I wondered just how good the Sabre is at handling jitter & would it make one of these into a decent playback device?

leo
11-07-2009, 00:06
How does it sound with the Sabre? I'm interested in how the Sabre's ASRC handles the jitter that this SB puts out? Any better than sounding than into other DACs?

I'm attracted to some network solution like this & I was thinking along the lines of a Network Media Tank (NMT) device like the Popcorn Hour. Now these are primarily for video playback so their clock arrangements are not the best so I wondered just how good the Sabre is at handling jitter & would it make one of these into a decent playback device?

It sounds pretty good , actually jitter from Duet isn't bad compared to a few CDP's
I've never tried one of those Network Media tanks so can't comment on those, even though Sabre is reported to use some fancy patent Time Domain Jitter Eliminator I can still tell the difference with the sources used, I actually tried feeding it from a cheap DVD player and found it sounded noticeably worse than from the Duet

I tried a separate clock using Tent XO with the Duet, tbh I didn't think results was worth the effort, I'll look into it later for better options

Biggest most noticeable gains so far for me was from the supplies

jkeny
11-07-2009, 00:32
Ah yes PS has more significance than we give it credit for - we just don't know yet all the various PS distortions/perturbations & their effects at various parts of a circuit digital & analog.

Hence my attempt to investigate PS in my Sabre tweakers board with individual battery PS's - I believe a clean PS is more important than the sub-optimal routing of a protoboard!

I had heard that the Sabre sounded different with different sources so there's either more at play here than jitter or the "jitter eliminator" does not live up to the marketing departments claims. Still, does it do a better job at minimising the sonic effect of jitter than other SRCs?

leo
11-07-2009, 00:48
I'd say it does a better job than the external ASRC chips I've tried which did more harm than good, even the ones which are claimed to have the feature of pass through

The Sabres are very impressive technically , just like other dacs I've tried though, they won't suit every system

jkeny
11-07-2009, 00:51
Thanks Leo,
Very good information!

Covenant
25-07-2009, 16:07
Hi Leo,
Just wondered if you have got any further? I am probably going to ask Tirna (who do the 4032=capacitor upgrade for the 7520) to quote for doing some of the mods you suggest to my SB3 but only in the digital domain and nothing that involves loosing wireless reception.

leo
25-07-2009, 19:28
Hi Jerry,

I've been holding off until I can get some better regulation like Paul Hynes's to replace the one for the Hex inverter (replace the little one I knocked up) in the spdif circuit and maybe for the internal dac, I'd like to see what I can get out of that as well as the digital output


Theres a few other things I want to try too, soon as I get time I promise to post some details:)

You could add a few of the simple things to start you off which is cheap
The cap on the DC input socket is certainly worth replacing to a low ESR (especially if using the standard external SMPS), an Oscon or Panasonic FM works ok
Also if you don't want to risk a seperate reg on the hex inverter in the spdif adding a parallel Oscon across the ceramic bypass on the supply pin would be a cheap and easy tweak to help filter noise

EDIT
Bugger, just noticed you have a SB3 which is different, these mods will also be able to be transferred to a SB3 but the person doing the mods will need to know the unit internally because the layout etc is different

Covenant
28-07-2009, 20:44
Hi Leo
What is your view on the mods done here:
http://myweb.tiscali.co.uk/nuukspot/decdun/sb3.html
Seems OTT to me

leo
29-07-2009, 17:50
Hi Jerry,

DD is a great little site, believe it or not its what first got me into diy several years ago after just browsing the net, :lolsign:
First thing I ever tried diy was a Gainclone

If your using an external dac just a few things helps, you don't have to go mad, things to help control the noise and crud floating around on the circuit brings the biggest gains:)

jkeny
11-12-2009, 03:13
Hi Leo,
I know that this has been dormant for a while but I'm interested in how this eventually turned out?

Did you ever go the I2S route to an external DAC?
I'm looking to buy one of these & this would be the way I'd want to go along with batteries for 3.3V & 1.2V (I don't think I need anything else if using it on a wired network & I2S out?)
I just got an Itouch & don't want to wait for a SBTouch!

leo
12-12-2009, 21:22
Hi Leo,
I know that this has been dormant for a while but I'm interested in how this eventually turned out?

Did you ever go the I2S route to an external DAC?
I'm looking to buy one of these & this would be the way I'd want to go along with batteries for 3.3V & 1.2V (I don't think I need anything else if using it on a wired network & I2S out?)
I just got an Itouch & don't want to wait for a SBTouch!

Hi John,

The Duet is not bad after some modding, getting rid of that awful switcher helps IMHO, I've not tried batteries yet although it should also be a good way of powering the thing.
The main limitation is that it can't output the high res stuff so your stuffed if you want to try 96,176.4 and 192.
I have tried feeding the Sabre I2S direct which worked fine, main reason I went back to s/pdif is because of the distance I have the Duet away from the dac, also as you know I'm toying about with the Musiland USB to spdif converter so using spdif is the quickest way to compare things

I doubt I'll be bothering with anymore of the SB based products, not until it offers higher res output anyway

leo
25-01-2010, 10:03
Hi Leo,
I know that this has been dormant for a while but I'm interested in how this eventually turned out?

Did you ever go the I2S route to an external DAC?
I'm looking to buy one of these & this would be the way I'd want to go along with batteries for 3.3V & 1.2V (I don't think I need anything else if using it on a wired network & I2S out?)
I just got an Itouch & don't want to wait for a SBTouch!

Hi John,

I noticed on diyaudio you was having a few problems with the Receivers dac, I've just had a quick look at the datasheet, Isn't the voltage on Vmid pin controlled by the Enable pin?

Have a look at page 12 under Hardware control modes, its seems they do this so it acts as a slow start up so the digital section is given chance to start up first

Cheers,
Leo

leo
25-01-2010, 10:16
may even be possible to add switches so you can compare against the standard voltage or against an external one, I've not looked into it all properly yet tbh, I just did a few easy simple mods, this thread never really took off so I just left it as is:)

jkeny
25-01-2010, 10:41
Brilliant Leo, I thought I did a search on all VMID refs within the datasheet, doh! I knew it wouldn't work as it stands! I guess RTFM applies here :)

leo
25-01-2010, 10:53
Well mate I've not tried it yet so can't be 100%
Problem with a lot of dacs is that they don't give enough details, the internal diagram for the wm8501 is only basic, it doesn't show us all the details

jkeny
25-01-2010, 11:12
BTW, I'm well into modding this unit now. Here's what I've done so far:
- removed 3.3V switching reg & replaced with external supply
- removed 5V reg & replaced with external supply
- tapped I2S to a socket that I can plug my I2S ESS DAC into (ES9022 at the moment - ES9018 eventually)
- although I wasn't particularly interested in the WM8051 on-board DAC I decided to experiment with it as it will drive headphone loads without problem.
- Removed output coupling caps & wired directly - sounds good through external caps
- played around with VMID & tried a transformer output between VMID & Vout (cancelling the 2.5V bias) - sounded distorted
- supplied external 2.5V on one leg of trafo primary instead of VMID - sounds wonderful

So with your suggestion I'll now try to disable VMID & supply it with external 2.5V

Things left to do:
- remove 1.2V switching reg & supply from external
- remove wifi board (I'm only interested in wired connection) - this can be done with the 1.2V & 3.3V external supply in place & unit doesn't report a hardware error
- remove 11.289MHz clock & provide external clock to Xilinx pin 42
- disable SPDIF on 74HC chip
- further de-couple all sensitive chips.
- would love to provide a clean 3.3V directly to Xilinx chip as I think it would benefit from a clean supply but the PS is provided through a power plane & no traces showing that can be cut. This has to share the 3.3V PS with Realtek & 74HC & WM8501. I could remove the 74HC & WM off the 3.3V bus but not the Realtek.

Anyway, the device has good potential & I've order a low jitter clock to complete my mods. Sound from the WM DAC via transformers is very good.

leo
25-01-2010, 16:18
You've tried a lot more than me with the Receiver, anyway It will be interesting to see how you find the modded receiver compares as a transport against the modded Hi-face and Musiland:)

UKPD
26-01-2010, 19:46
BTW, I'm well into modding this unit now. Here's what I've done so far:
- removed 3.3V switching reg & replaced with external supply
- removed 5V reg & replaced with external supply
- tapped I2S to a socket that I can plug my I2S ESS DAC into (ES9022 at the moment - ES9018 eventually)
- although I wasn't particularly interested in the WM8051 on-board DAC I decided to experiment with it as it will drive headphone loads without problem.
- Removed output coupling caps & wired directly - sounds good through external caps
- played around with VMID & tried a transformer output between VMID & Vout (cancelling the 2.5V bias) - sounded distorted
- supplied external 2.5V on one leg of trafo primary instead of VMID - sounds wonderful

So with your suggestion I'll now try to disable VMID & supply it with external 2.5V

Things left to do:
- remove 1.2V switching reg & supply from external
- remove wifi board (I'm only interested in wired connection) - this can be done with the 1.2V & 3.3V external supply in place & unit doesn't report a hardware error
- remove 11.289MHz clock & provide external clock to Xilinx pin 42
- disable SPDIF on 74HC chip
- further de-couple all sensitive chips.
- would love to provide a clean 3.3V directly to Xilinx chip as I think it would benefit from a clean supply but the PS is provided through a power plane & no traces showing that can be cut. This has to share the 3.3V PS with Realtek & 74HC & WM8501. I could remove the 74HC & WM off the 3.3V bus but not the Realtek.

Anyway, the device has good potential & I've order a low jitter clock to complete my mods. Sound from the WM DAC via transformers is very good.

Hi,

My first post here. Found just what I was looking for. Wow thats an impressive list you have there.:stalks: ATM I am at the PSU upgrade stage and decoupling of the critical stages, not swapped the regulators yet.
Care to share any pictures ?

Peter.

jkeny
27-01-2010, 00:39
Hi,

My first post here. Found just what I was looking for. Wow thats an impressive list you have there.:stalks: ATM I am at the PSU upgrade stage and decoupling of the critical stages, not swapped the regulators yet.
Care to share any pictures ?

Peter.

Hi Peter,
I'm not getting notifications from this forum so apologies.
I have to report the sad news of the demise of my SBR - a minutes silence, please :). My latest mod - removed the 1.2V reg & connected a wire - while connecting my 3.3V battery the gnd wire of the 1.2V supply touched the + end of 3.3V battery. It did n't die immediately, seemed to be OK for a while & then died a short time later :( Now the front led just slowly flashes through different colours & the ethernet port does not seem to be working or the led on the optical out. Any ideas? Kapput? I was hoping it might have been a diode blown but I fear the worst - these batteries can put out 120Amp instantaneous current!! Beware of loose wires & late nights - I usually have all my connectors in place before connecting anything. just when it was getting interesting too!

Anyway, sob, sob, I don't have any pics of my set-up yet but here's a site with pics & info -
http://www.lampizator.eu/LAMPIZATOR/Squeezebox/squeezebox.html

And another that I have posted on http://forums.slimdevices.com/showpost.php?p=498889&postcount=133

BTW, Leo, I contacted Wolfson about ENABLE/VMID so I wouldn't kill the WM8501 (how ironic is that). The support engineer has been just quoting the datasheet for 2 days now - can't get a straight answer from him. I might escalate it?

Anybody know of a cheap WORKING SBR - I have a good case so damaged case would be fine?

UKPD
27-01-2010, 16:53
Sorry for the late reply(back from work). I feel your pain,:(
ARe you shure it's all dead? Maybe you can use Wi-Fi node instead of LAN.
I do have a spare squeezebox but I'ts V-1 (if thats any good for you).

Peter

jkeny
27-01-2010, 17:22
Sorry for the late reply(back from work). I feel your pain,:(
ARe you shure it's all dead? Maybe you can use Wi-Fi node instead of LAN.
I do have a spare squeezebox but I'ts V-1 (if thats any good for you).

PeterYes, I'll look into the wifi side but I suspect the worst

Thanks for your generous offer but I wanted to stick with the SB receiver as I believe that it has the best potential for sound & I've already put some research into it (as you can see)

Anyway, I'll help here, if I can, while I wait for a cheap SBR to become available.

jkeny
29-01-2010, 12:43
What do you think of this reply from Wolfson about how to supply an external 2.5V - I think it's piffle
I appreciate the datasheet is a little confusing but I do not believe it to be incorrect.

Internally the VMID is generated using a simple resistor divider circuit and therefore you should be able to change it using an external source without turning the internal supply off. You should also be able to AC couple a signal onto the internally generated VMID supply.

There is no way of turning the internal VMID supply off without disabling the whole device through the use of the ENABLE pin.

I hope this helps.

leo
02-02-2010, 10:29
Behind again:doh:

How you going with the Receiver John? did you manage to get it working?

I got itchy fingers last night, after searching about in my stash I found one of those Burson clocks, I removed the onboard cheap oscillator and replaced it with a Guido Tent, supply was tapped into the main Paul Hynes PR3 , its a 5v XO so a 100R resistor was added from the out of the clocks inverter to ground because of the Receivers Xilnix is 3.3v.
The Receivers stock onboard XO was removed and the new clocks output ran straight to the Xilnix

Seriously this made one hell of a difference! everything just locks into place and that bass :eek: it trounces the modded Musiland for 44.1 material
Obviously this lash up is not ideal but was mainly as a test and gives me an idea how a better clock affects this unit, i wasn't actually expecting this much of an improvement with the Receiver , this was clearly heard with all three dacs

Amazes me those who claim all spdif sources are the same, I read some total crap on another forum from one of the ex members on here , that guy must have huge amounts of fur growing in his ears:mental:

leo
02-02-2010, 10:40
What do you think of this reply from Wolfson about how to supply an external 2.5V - I think it's piffle

Hmm, it does seem like piffle doesn't it? so if the Enable pin is disabled it turns the whole device off not just the voltage bias :scratch: doesn't seem to say that in the datasheet unless I'm missing something

jkeny
02-02-2010, 11:55
Behind again:doh:

How you going with the Receiver John? did you manage to get it working?

I got itchy fingers last night, after searching about in my stash I found one of those Burson clocks, I removed the onboard cheap oscillator and replaced it with a Guido Tent, supply was tapped into the main Paul Hynes PR3 , its a 5v XO so a 100R resistor was added from the out of the clocks inverter to ground because of the Receivers Xilnix is 3.3v.
The Receivers stock onboard XO was removed and the new clocks output ran straight to the Xilnix

Seriously this made one hell of a difference! everything just locks into place and that bass :eek: it trounces the modded Musiland for 44.1 material
Obviously this lash up is not ideal but was mainly as a test and gives me an idea how a better clock affects this unit, i wasn't actually expecting this much of an improvement with the Receiver , this was clearly heard with all three dacs

Amazes me those who claim all spdif sources are the same, I read some total crap on another forum from one of the ex members on here , that guy must have huge amounts of fur growing in his ears:mental:

Great Leo, I have found exactly the same sound with my modded hiFace (clean PS to on-board low-jitter clocks) - the bass is deep, tight, musical & has texture as well as the rest of the frequency range just being so much clearer - the sound of low jitter = very good analogue :)

jkeny
02-02-2010, 11:57
Hmm, it does seem like piffle doesn't it? so if the Enable pin is disabled it turns the whole device off not just the voltage bias :scratch: doesn't seem to say that in the datasheet unless I'm missing something

Yes, that's what I thought! The communication with Wolfson has gone dark :( - no reply to my last email which questioned what he said above & asked him to explain exactly how an external 2.5V could be provided to the VMID pin - hopefully he's off checking with colleagues :)

leo
02-02-2010, 12:29
Yes, that's what I thought! The communication with Wolfson has gone dark :( - no reply to my last email which questioned what he said above & asked him to explain exactly how an external 2.5V could be provided to the VMID pin - hopefully he's off checking with colleagues :)


I'd have thought it would be in their interest to help you mainly because people like us diyers post results and opinions on the net which works out in their favour especially if your happy with the performance.

Anyway keep at em John :eyebrows: nobody is going to know more than their own engineers, if they don't want folks like us hassling, the online datasheets should be more detailed:lolsign:

jkeny
02-02-2010, 15:45
Here's a application note on the use of external supply to VMID (sent to me by Wolfson)

Edit: Sorry, i see that the PDF didn't upload - zipped it

So here's what it says:
The WM8501 is designed to generate an internal VMID reference voltage however due to the internal architecture of the device any voltage applied to the VMID pin will be accepted as the internal reference of the device. Figure 3a and 3b show the connections to the external supplies required for DC and AC voltage supplies to the VMID pin of the WM8501.

OK, so it seems to say that the DAC chip will turn off the internally generated 2.5V when an external supply is connected - fine! Why the mention of an AC supply connection to VMID?

Dan Giovanni
03-02-2010, 23:51
The internal reference won't be so much turned off as over-ridden. It used to be pretty common to provide external references to some DAC chips to improve performance.

I'm scratching my head why an AC signal would be used though - obviously it will modulate the output of the DAC. Perhaps there's some non-hifi application for it.

jkeny
04-02-2010, 00:37
The internal reference won't be so much turned off as over-ridden. It used to be pretty common to provide external references to some DAC chips to improve performance. Improving performance is exactly the object here! Surely though the internal generation of 2.5V on VMID pin has to be turned off in order to feed an external 2.5V supply into this pin? if there still is 2.5V being generated internally then external supply is meaningless, no?

I suppose one way of testing this would be to provide a 2V external supply & measure the voltage on VMID. If still 2.5V then internal supply is on & not over-ridden.


I'm scratching my head why an AC signal would be used though - obviously it will modulate the output of the DAC. Perhaps there's some non-hifi application for it.I agree, the more info I get the more I'm confused :scratch:

Dan Giovanni
04-02-2010, 20:20
Surely though the internal generation of 2.5V on VMID pin has to be turned off in order to feed an external 2.5V supply into this pin? if there still is 2.5V being generated internally then external supply is meaningless, no?

The internal reference is formed using a potential divider - two resistors between another voltage supply. Hence it is a high impedance point. Supplying a low impedance Vref will "take charge" of this node.

In other news .... found this kit for USB isolation:

http://www.circuitsathome.com/products-page/usb-interfaces

Thought you might be interested - I am! Someone has already used it with a USB DAC and claimed it improves sound quality .... :)

jkeny
04-02-2010, 20:32
The internal reference is formed using a potential divider - two resistors between another voltage supply. Hence it is a high impedance point. Supplying a low impedance Vref will "take charge" of this node.Cool, thanks - I just needed an explanation that makes sense (I don't have the E'ee background to work it out myself :) Now any idea why they would also show the possibility of an AC supply feeding VMID - what possible use could it serve?


In other news .... found this kit for USB isolation:

http://www.circuitsathome.com/products-page/usb-interfaces

Thought you might be interested - I am! Someone has already used it with a USB DAC and claimed it improves sound quality .... :)Sorry, but it won't work on hi-speed USB2 - 480Mbps which is what the HiFace (& Musiland use)!

Dan Giovanni
04-02-2010, 21:23
The only thing I can think of Re. AC supply for VMID is some kind of frequency synthesis i.e not hifi use. I wouldn't worry about it. If I were looking to mod one of these, I'd use a filtered Vref and an op-amp buffer.

Re. the isolator, it is claimed to support "full speed" USB 2.0. Is there a difference between "full speed" USB and "hi-speed"?

jkeny
04-02-2010, 21:38
The only thing I can think of Re. AC supply for VMID is some kind of frequency synthesis i.e not hifi use. I wouldn't worry about it. If I were looking to mod one of these, I'd use a filtered Vref and an op-amp buffer.Yep, sounds good idea


Re. the isolator, it is claimed to support "full speed" USB 2.0. Is there a difference between "full speed" USB and "hi-speed"?Yep, about 468Mbit/s difference :) 12 for full speed & 480 for hi-speed

MyNaim
10-02-2010, 03:16
Behind again:doh:

How you going with the Receiver John? did you manage to get it working?

I got itchy fingers last night, after searching about in my stash I found one of those Burson clocks, I removed the onboard cheap oscillator and replaced it with a Guido Tent, supply was tapped into the main Paul Hynes PR3 , its a 5v XO so a 100R resistor was added from the out of the clocks inverter to ground because of the Receivers Xilnix is 3.3v.
The Receivers stock onboard XO was removed and the new clocks output ran straight to the Xilnix

Seriously this made one hell of a difference! everything just locks into place and that bass :eek: it trounces the modded Musiland for 44.1 material
Obviously this lash up is not ideal but was mainly as a test and gives me an idea how a better clock affects this unit, i wasn't actually expecting this much of an improvement with the Receiver , this was clearly heard with all three dacs

Amazes me those who claim all spdif sources are the same, I read some total crap on another forum from one of the ex members on here , that guy must have huge amounts of fur growing in his ears:mental:

Thanks Leo for the very good job you are doing. I do follow this thread very closely since your very first post ;)

Could you possibly post some pics of the new job done ? It will definitely help to understand/follow ; I'm very new in the DIY sport but very much interested in following your mods)

Cheers!
Olivier

leo
10-02-2010, 14:49
Hi Olivier,

Thank you, thats much appreciated

It would be really hard to see whats happening with just a pic as is because of the other mods so I tried to edit an earlier pic before the inverter had its separate reg etc

The PR3 now runs Duet Receiver and also acts a pre reg for the clock (which has its own onboard regulation to the XO)

The clock I lashed up isn't optimal but was good enough to give me an idea what sort of results to expect, I hope to compare others at some later stage.
For now the Burson module using the Tent XO is giving very good results

The resistor R104 on the output of the inverter may be better off removed altogether if running a new clock direct to the Xilinx pin42

http://i273.photobucket.com/albums/jj220/kingbusoms/clockmod.jpg

janos_904
11-02-2010, 09:01
All this work is seems to be a great and amazing microsurgery! What about the sound :-)?

leo
12-02-2010, 02:31
All this work is seems to be a great and amazing microsurgery! What about the sound :-)?

I also have a stock Receiver so i its easy to compare, lets just say I'm quite pleased with the results so far;)

MyNaim
12-02-2010, 04:51
Hi Leo

Thanks for the picture and the drawing, it has been really helpful.

I'm still not 100% XO-clear :) about where to connect the output of clock, it has to run to both output of 74HCU04 (R104 to be removed) and pin 42 of XILINX, is that correct?

Cheers
Olivier

leo
12-02-2010, 10:48
I didn't remove resistor R104 yet although I'd imagine it may be better to remove it

Standard unit is

Stock XO (clock) - 74HCU04 inverter - output resistor R104 - pin42 Xilnix

My unit , upgraded clock (tagged on output of R104) - pin42 Xilnix

Depending on the the new clock your fitting you can run it to the input of 74HCU04 or straight to pin42 Xilnix (main clock input).
The spdif output (digital out) and stock XO's all share the same 74HCU04 which is a compromise, because the clock I used was suited to run the Xilinx direct I ran it to the output of 74HCU04 (tagged onto the output side of the 74HCU04 output resistor R104) rather than 74HCU04 input if that makes sense

Sorry, I'm crap at explaining things by text :)

MyNaim
16-02-2010, 04:16
...Depending on the the new clock your fitting you can run it to the input of 74HCU04 or straight to pin42 Xilnix (main clock input).
The spdif output (digital out) and stock XO's all share the same 74HCU04 which is a compromise, because the clock I used was suited to run the Xilinx direct I ran it to the output of 74HCU04 (tagged onto the output side of the 74HCU04 output resistor R104) rather than 74HCU04 input if that makes sense

Sorry, I'm crap at explaining things by text :)


Hi Leo,

Thanks again for your replies and you're not crap at explaining; got another explanation I'm crap at understanding :lol:

When you said depending on the new clock, how do you know where it's supposed to go better ? For instance, the new tentslabs clock or the one from audio-gd ?

Cheers
Olivier

ar-t
16-02-2010, 20:53
I assume the clock you are using has a build-out resistor, on its output.

Pat

leo
17-02-2010, 21:00
Hi Leo,

Thanks again for your replies and you're not crap at explaining; got another explanation I'm crap at understanding :lol:

When you said depending on the new clock, how do you know where it's supposed to go better ? For instance, the new tentslabs clock or the one from audio-gd ?

Cheers
Olivier

Hi Olivier,

Better to ask again than just fitting something if not 100% sure:)

I'd ask the place you buy the clock from just to be sure if it can run the Xilnix direct, with mine I basically didn't want it sharing the same inverter as the one used for spdif

Leo

leo
17-02-2010, 21:28
I assume the clock you are using has a build-out resistor, on its output.

Pat

Hi Pat,

Theres an output resistor on the clocks onboard inverter, It was all just a temp quick lash up tbh, I need to look into something better suited as a full time replacement
Was thinking of using a PFM Flea and simply adjusting its onboard reg to 3.3v to the XO

Leo

MyNaim
21-02-2010, 18:00
Hi Olivier,

Better to ask again than just fitting something if not 100% sure:)

I'd ask the place you buy the clock from just to be sure if it can run the Xilnix direct, with mine I basically didn't want it sharing the same inverter as the one used for spdif

Leo

Thanks Leo for the explanation; new clock is ordered, I'm really impatient to plug it.

Cheers mate
O

MyNaim
06-03-2010, 20:23
Tried the trick with a chip clock to see how a new clock can improve things. Went for the one from audio-gd and honestly I don't know whether the ones from burson or tentlabs are better but what I can say it's WAHOO !

And the bonus is that Kingwa (conceptor, seller) is happy to help and advise, very nice person indeed.

One advice, go for it. One of the best mod i have made so far (with a good linear PSU).

And of course, thank you to Leo for his wonderful work.

an happy man !

leo
06-03-2010, 20:32
Thats excellent Olivier, I'm pleased to hear it went well for you :smoking:

I'm hoping to try a couple of other clocks at some stage, should prove interesting

steveinspain
06-03-2010, 22:30
Having just read through this entire thread I can tell you I understand nothing at all !
I can join 15mm copper pipe for central heating, but thats as far as my skills go.
What do you guys do when you finish a project, or do you never finish - is there always a next stage to try ?
I started reading as I am idly wondering about getting a duet - would you do this sort of work for a fee for me if I got a duet ? Are there people who do mods for a living to stuff like duets/sonos etc ?
Respect to all of you who know how to and then do these mods just to see what happens !

leo
06-03-2010, 23:13
Once you start to understand how things work and start to see how certain compromises can be improved, modding can become pretty addictive :eyebrows: when you start hearing the improvements it also becomes quite satisfying, obviously theres a lot more to it than just swapping bits about.

I sometimes do mods for friends when I get time, at one stage it got a little crazy so I had to limit them, doing your own stuff means you can do it whenever its convenient.
If you can't diy there is places like Audiocom who provide a modding service, they give you options etc to suit your budget

ar-t
07-03-2010, 17:31
The answer is yes. How much? Well, we can measure the difference. Whether one can hear that or not, is another story.

Without going into too much extraneous gibberish, we have the use of something called a Wavecrest DTS-2077. It allows us to measure the amount of jitter on a signal. Its strong point is that it can separate jitter into its two main components: random and deterministic.

Random jitter is just what it is called. It is Gaussian in nature, and exists in everything.

Deterministic jitter is a bit more involved. To simplify things a bit, we can say it is predominantly correlated jitter. That is, jitter that bears a strong resemblance to the data stream itself.

In the Squeezebox products, (which are very good, especially for the price), the designers have to make certain compromises, to meet the price/performance goal.. (All products do, but ones at the low price end have to do more of it.) The main compromise is using a hex inverter, to function as 2 clocks, and the SPDIF driver stage.

So, having said all of that, one would expect to see deterministic jitter, and yes, it does exist. It is a tad higher than the random jitter.

Ok, the question you all want to know:

Does changing the external supply affect any of this?

Yes, it lowers them. Not a great deal, but we can measure a difference. Can you hear the difference? Probably. Maybe. I dunno. It depends. All very valid answers.

The other question you want to know:

If one separates the clock and SPDIF functions, totally, what happens to the SPDIF signal?

If you do it properly, the deterministic jitter goes completely away. The random jitter can be much lower, depending on the quality of the clock.

Hope this sheds some insight.

Pat

leo
07-03-2010, 17:52
Excellent post Pat thank you
Must be great having all that nice test gear whenever needed :eyebrows:

ar-t
08-03-2010, 03:41
Glad it was beneficial.

It is nicer to do work for outfits that can afford a machine that costs USD 77K, with a USD 6K s/w package.

Ok, they bought it on eBay. Probably cost $12K, when all the costs are added up. I think the most expensive piece of equipment we ever bought cost around $2K. Thanks to the telecom industry imploding (putting guys like me on the streets), there is lots of good stuff to be had. Just takes money.

Pat

MyNaim
19-03-2010, 07:58
...
The resistor R104 on the output of the inverter may be better off removed altogether if running a new clock direct to the Xilinx pin42



not sure it has brought something better but it's definitely not a problem to remove it

janos_904
19-01-2012, 09:47
Almost 2 years went by since the last post. Leo made a remarkable work, but is it any progress right know on Duet? By chance, I found a clean spdif signal originating from Xilinx chip leg 3 (it goes to the Toslink). I found on the lampizator page a description how to get a signal out from Xilinx (10-100 nF cap in series + resistance of 200 ohm, 75 ohm between GND and signal). It works, and this output is much better. The tweak works on SB3 too. Did somebody similar trick, is any advices of choices of the cap and the resistor types?

ar-t
20-01-2012, 00:37
Those resistor values will give the wrong level and impedance.

The shunt resistor has to be larger than the desired output impedance.

janos_904
20-01-2012, 10:08
Thank you ar-t for your fast response. What values (C, R1?, R2?) do you recommend to get a good impedance? This is lampizator's description (maybe I misunderstood it?):

"4. Middle red circle shows the real source of S/PDIF signal.
Just gently desolder R102 attached to pin 3 of the XILINX. Solder a capacitor - one leg to the lower soldering pad left after removal of R102. It will be directly connected to Pin 3 of the Xilinx. The other leg of that cap (any foil type like MKP, MKT, MKS, or even a ceramic, size between 10 nF and 220 nF, ) connects to a 200 Ohm resistor. The other end of this resistor connects to the center pin of the RCA with SPDIF. First - cut off the RCA center pin from the pcb and float it. Last point: solder across the RCA - between center and GND surrounding golden frame - a resistor 75 Ohms. (or 58 Ohms if you want to be close to real RCA and coax impedance). "

ar-t
21-01-2012, 01:06
No, you did not misunderstand. I think someone else is just wrong.

Off the top of my head.......without aid of calculator, or even scratch pad................try 220R and 100R. That will get you closer.

But, you know how to calculate the resistance of 2 resistors in parallel. That will get you close.

In addition, I bet you can figure out a voltage divider. If you want to get really technical, you need to know the output Z of the chip where you get the signal from. That depends on the logic family, etc., and operating voltage. (R goes up, as rail voltage goes down.)

Don't know who originally came up with the values. I bet he neglected all of that. Or, maybe he thought it was not important.

I won't say who, but I bet most here will know who I am talking about, in the following.

A "competitor" copied the values in the TI data sheet. Ooops...........TI also was wrong! The impedance was wrong, and the voltage was twice what it should have been. I believe they corrected it. But still wrong in the data sheet, even today.

So, just because it is in a data sheet........well, they are as good as a lot of the free advice you get on the 'Net!

lenta
24-04-2012, 08:05
I would realy like to get your comments about this direct xilinx spdif output.
Becouse I read somewere that its bad idea, but it wasnt clear why.

If one separates the clock and SPDIF functions, totally, what happens to the SPDIF signal?

If you do it properly, the deterministic jitter goes completely away. The random jitter can be much lower, depending on the quality of the clock.

Is it proper way? At least it seperates spdif from clock of the 74hcu chip.

Regards,
Martin

Marco
24-04-2012, 08:57
Hi Martin,

Welcome to AoS! :)

Before you go any further, however, please read this thread and comply with the request therein: http://theartofsound.net/forum/showthread.php?t=17333

All newbies must pop into the Welcome area and say hello - cheers! :cool:

Marco.

ar-t
25-04-2012, 02:02
I would not try to drive anything, other than another gate, with the output of a FPGA. They are not designed to drive tons of current.

A single picogate D flip-flop can do it. Also reclocks the signal.