Ok, GuidoR, It's over to you - I showed you mine
yes, understood.
But you can argue from my avatar that speed is not my specialty
During this w.e., I promise, I'll work on it
Ciao
Guido
No problem, GuidoR, your friend Vale told me that you are getting a Musiland unit also - I would be interested in your comparison with the M2Tech. I'm particularly interested in how the lack of USB cable on the M2Tech is giving it a sonic advantage. Can you try your HiFace with an extension USB cable?
Modding HiFace: how to
The board is very small, so only limited mods are possible.
Here the most important ones.
Taking out I2S: the points are on the following resistors:
MCLK -> R3
SCLK -> R5
LRCK -> R6
SDATA -> R7
In theory, it could be better to steal the signal upstream of the res, in case of long wires, downstream in case of short ones.
The "up" is the side near the DIT for R5 and R6, the opposite for R3 and R7.
In fact, due to the extremely small dimension of the 47R res (0603) is not advisable to attach wires directly on them, the risk of burning them (as I did) is very high. (you can see from the pics that I replaced the 0603 with 0805, but due to the limited space one has to be mounted on edge).
DO NOT REMOVE THOSE RESISTORS!!!!
My solution, after some disasters, has been oriented to the better "mechanical" solution, in order to avoid any stress to the very fragile pads. Following the tracks from the resistors, you can see four vias that are connected, the other side, to the Xilinx chip (the one with the big green dot).
I ran copper wires (0,35mm - AWG 27), duly bent and tinned, applied some flux, lined them up on the legs of the chip and just a tip of a soldering iron - a magnifier is a must. The attentive reader can note that the connexions are mixed - upstream and downstream: it works just the same.
You can steal the ground connection from the second uninsulated pad on the side of the board (near the square one).
Replacing PS: the HiFace is fed by USB, so we can isolate the V input.
To make the change reversible, I disconnected the Vin pin by only lifting them, and connecting an insulated wire. The wire is fixed with a tie-wrap to improve the mech (and to avoid stress to the pads).
On the pics I have evidenced in red the vias and the contacts.
Enjoy (I did!!)
The board is very small, so only limited mods are possible.
Here the most important ones.
Taking out I2S: the points are on the following resistors:
MCLK -> R3
SCLK -> R5
LRCK -> R6
SDATA -> R7
In theory, it could be better to steal the signal upstream of the res, in case of long wires, downstream in case of short ones.
The "up" is the side near the DIT for R5 and R6, the opposite for R3 and R7.
In fact, due to the extremely small dimension of the 47R res (0603) is not advisable to attach wires directly on them, the risk of burning them (as I did
) is very high. (you can see from the pics that I replaced the 0603 with 0805, but due to the limited space one has to be mounted on edge).DO NOT REMOVE THOSE RESISTORS!!!!
My solution, after some disasters
, has been oriented to the better "mechanical" solution, in order to avoid any stress to the very fragile pads. Following the tracks from the resistors, you can see four vias that are connected, the other side, to the Xilinx chip (the one with the big green dot).I ran copper wires (0,35mm - AWG 27), duly bent and tinned, applied some flux, lined them up on the legs of the chip and just a tip of a soldering iron - a magnifier is a must. The attentive reader can note that the connexions are mixed - upstream and downstream: it works just the same.
You can steal the ground connection from the second uninsulated pad on the side of the board (near the square one).
Replacing PS: the HiFace is fed by USB, so we can isolate the V input.
To make the change reversible, I disconnected the Vin pin by only lifting them, and connecting an insulated wire. The wire is fixed with a tie-wrap to improve the mech (and to avoid stress to the pads).
On the pics I have evidenced in red the vias and the contacts.
Enjoy (I did!!)
Attachments
Hi Jkeny
Unfortunately I, from the very beginning, tried HiFace only with a cable extension (1m long) and I2s out, so no possible comparisons from me on this particular topic. As you can see from my last post (and after Vale's first attempt) I've further modded my HiFace replacing the USB PS with a shunt reg (ZenMod's Shiny or Salas Shunt, I'll decide tomorrow depending on the PCB I have on hand...).
BTW, you also excel in microsurgery, eh?
Guido
Does anybody have a link to the clock/crystal shown in the above HiFace pic? I'm wondering what their specs are & where they might be purchased. I've searched for all the combinations of the letters that I can make out on the pic VEC C1009HC or NEC C10D9HC. Always looking for proven low jitter clocks at reasonable cost
Ok, here's the start of the new clock install (micro-surgery is fun ):
- First pic is crystal removed with the smd Crystek clock beside it
- Second pic is the back of the smd clock with a small board cut from a matrix board - I had put solder on the clock pads as I had intended to solder to the board but I didn't use this method - instead I did this:
- Third pic - there are little grooves from the pads that go up the sides of the smd clock - here is a component lead cut-off soldered in the groove.
- I did this with all 4 pins (although I don't need pin 1 connected) & put the pins through the board from the top so I could solder the pins underneath to the copper side of the matrix off-cut board.
- Pic 4 - This gives a very solid & small structure. I soldered the out pin to the Xtalin (pin 12) where the original crystal had been.
- As a quick & dirty job I soldered the clock VDD to the 3.3V regulator Vout that goes to the Cypress USB chip; the clock gnd pin to the USB gnd at the USB socket.
- Plugged in USB cable but unit is now not recognised by the laptop. Not surprising really as I haven't changed the load cap that was on the crystal (I told you this was quick & dirty to see if I could get it working straight off). The other issue that may be causing it not to properly initialise is that there is a timing condition that an external clock has to meet for the Cypress to operate correctly. I will probably try to power the clock from a 3.3V battery as I originally intended & have this powered up before the Cypress chip.
):- First pic is crystal removed with the smd Crystek clock beside it
- Second pic is the back of the smd clock with a small board cut from a matrix board - I had put solder on the clock pads as I had intended to solder to the board but I didn't use this method - instead I did this:
- Third pic - there are little grooves from the pads that go up the sides of the smd clock - here is a component lead cut-off soldered in the groove.
- I did this with all 4 pins (although I don't need pin 1 connected) & put the pins through the board from the top so I could solder the pins underneath to the copper side of the matrix off-cut board.
- Pic 4 - This gives a very solid & small structure. I soldered the out pin to the Xtalin (pin 12) where the original crystal had been.
- As a quick & dirty job I soldered the clock VDD to the 3.3V regulator Vout that goes to the Cypress USB chip; the clock gnd pin to the USB gnd at the USB socket.
- Plugged in USB cable but unit is now not recognised by the laptop
. Not surprising really as I haven't changed the load cap that was on the crystal (I told you this was quick & dirty to see if I could get it working straight off). The other issue that may be causing it not to properly initialise is that there is a timing condition that an external clock has to meet for the Cypress to operate correctly. I will probably try to power the clock from a 3.3V battery as I originally intended & have this powered up before the Cypress chip.Attachments
Success
The clock needs 1-2mS to initialise so it needed to be powered up before the Cypress chip as I suspected. I re-did the clock - I'll post pics later.
How's the sound, I hear you ask
. Obviously all this is on my own system so whatever the abbreviation is BUT
Bass is much more solid & powerful, mids & highs are more crystalline. What I use for listening is Herbie Hancock's River (The Joni Letters) recorded in WAV to disk from CD. As I said before, I'm using a modded Lepai amp (Tripath TA2020) with no caps on the input (because I'm using a transformer). So no caps from DAC o/p straight through to amp input.
My test is to get the brush strokes on the cymbals to sound like brush strokes on cymbals with texture, etc rather than a haze of sound - this gets me closer to it with some texture showing through, not all the way. I haven't heard a system yet that renders this properly so maybe it's in the recording. As for the rest of the album, the voices sound amazingly lifelike (Tina Turner, Corrine Bailey Rae, leonard Cohen, Joni herself, etc), the piano sounds perfect, Wayne Shorter's sax playing is a wonder on this album too & you can just about hear the spit in the mouthpiece
Anyway, I would recommend this mod to anyone - cheap too ($3 clock, $8 battery). A lot of the sonic improvement may well be providing a clean PS to the clock, something I couldn't do to the crystal to test it.
The clock needs 1-2mS to initialise so it needed to be powered up before the Cypress chip as I suspected. I re-did the clock - I'll post pics later.
How's the sound, I hear you ask
. Obviously all this is on my own system so whatever the abbreviation is BUTBass is much more solid & powerful, mids & highs are more crystalline. What I use for listening is Herbie Hancock's River (The Joni Letters) recorded in WAV to disk from CD. As I said before, I'm using a modded Lepai amp (Tripath TA2020) with no caps on the input (because I'm using a transformer). So no caps from DAC o/p straight through to amp input.
My test is to get the brush strokes on the cymbals to sound like brush strokes on cymbals with texture, etc rather than a haze of sound - this gets me closer to it with some texture showing through, not all the way. I haven't heard a system yet that renders this properly so maybe it's in the recording. As for the rest of the album, the voices sound amazingly lifelike (Tina Turner, Corrine Bailey Rae, leonard Cohen, Joni herself, etc), the piano sounds perfect, Wayne Shorter's sax playing is a wonder on this album too & you can just about hear the spit in the mouthpiece
Anyway, I would recommend this mod to anyone - cheap too ($3 clock, $8 battery). A lot of the sonic improvement may well be providing a clean PS to the clock, something I couldn't do to the crystal to test it.
It's stable as it's legs are component leads & so stiff - one of these is soldered into the XTALIN hole vacated by the crystal, the ground lead is soldered to another fixed spot, the ground connection on the board (you'll see it better in the pics). BUT this is not the optimal arrangement for the clock - I believe it could be made more stable & secure & closer to the Cypress chip - I might even put it on top of the Cypress with a short, fine wire to pin 12 (XTALIN) - this was all pretty much a proof of concept thing.
Here's the pics:
1. Pic of full DAC, trafos & amp. the trafos are Sescom MI-97. The white top is an octal tube socket into which the Sescoms plug.
2. Battery powering the clock & USB cable powering the rest. Noticeable sound improvement - haven't had a chance to test with external power to the DAC, yet
3. The clock is mounted on the board about 0.5" above the board- it's clock out wire lead soldered into hole vacated by crystal. Gnd lead connects to where the green wire is soldered to the board. I de-soldered the original cap that was on XTALIN board track from the crystal to ground. I put a 22pF styrene load cap across the clock out to gnd underneath the clock board - you can just see it sticking out of the far side of the board. I also soldered a 0.1uF cap under the clock board between VDD & gnd.
Here's the pics:
1. Pic of full DAC, trafos & amp. the trafos are Sescom MI-97. The white top is an octal tube socket into which the Sescoms plug.
2. Battery powering the clock & USB cable powering the rest. Noticeable sound improvement - haven't had a chance to test with external power to the DAC, yet
3. The clock is mounted on the board about 0.5" above the board- it's clock out wire lead soldered into hole vacated by crystal. Gnd lead connects to where the green wire is soldered to the board. I de-soldered the original cap that was on XTALIN board track from the crystal to ground. I put a 22pF styrene load cap across the clock out to gnd underneath the clock board - you can just see it sticking out of the far side of the board. I also soldered a 0.1uF cap under the clock board between VDD & gnd.
Attachments
Last edited:
One thing I wanted to check with you guys:
As my test CD, I'm using Herbie Hancock's River (The Joni Letters) recorded from CD as WAV to HDD on my laptop. On the recording there are parts (30 secs or so) in some songs that sound like a hiss & I have always assumed that it was a cymbal being brushed but not resolved properly by my system. In other parts of the recording I can distinctly hear cymbal brushing so I'm unsure what this noise is. It's on the first song & the Tina Turner one that I can remember & probably others. I'm not at home so can't give the exact time refernces for these but anybody who has the CD will know what I'm referring to.
So with my modded system I can hear still hear this hiss even though the rest of the recording has more texture & detail to it. Can anyone confirm or otehrwise what I'm hearing?
As my test CD, I'm using Herbie Hancock's River (The Joni Letters) recorded from CD as WAV to HDD on my laptop. On the recording there are parts (30 secs or so) in some songs that sound like a hiss & I have always assumed that it was a cymbal being brushed but not resolved properly by my system. In other parts of the recording I can distinctly hear cymbal brushing so I'm unsure what this noise is. It's on the first song & the Tina Turner one that I can remember & probably others. I'm not at home so can't give the exact time refernces for these but anybody who has the CD will know what I'm referring to.
So with my modded system I can hear still hear this hiss even though the rest of the recording has more texture & detail to it. Can anyone confirm or otehrwise what I'm hearing?
Mi scuzi,but e-mu 0202 cost under 100 bucks - and can i2s with
any 32-192 sample rate..)))
Oscilators in emu - are not top grade but aren't totaly ****...
Don't understand what are you looking for - pro audio is sonicaly **** ,but ! it will work as digital interface flavecely...
Jkenny pcm1793 is rubish as 1798 1796 1795 and other ****.
Try pcm1792 1794 . Or analog.
Your will be totally satisfied.
any 32-192 sample rate..)))
Oscilators in emu - are not top grade but aren't totaly ****...
Don't understand what are you looking for - pro audio is sonicaly **** ,but ! it will work as digital interface flavecely...
Jkenny pcm1793 is rubish as 1798 1796 1795 and other ****.
Try pcm1792 1794 . Or analog.
Your will be totally satisfied.
Don't worry, I'm not saying the PCM1793 is the best DAC - it's just the one that's inside the Musiland 01US & it's sounding good with output transformers. I'll be using an external Sabre32 with either I2S or SPDIF
I don't know what the rest of your post is about or is in response to
?
Last edited:
Hello jkeny,
the guys from diyhifisupply did also some mods like you on the Musiland 01 USD and afterwards a listening session with several people.
After the mods the 01 USD sounds as good as their reference CD-player:
http://www.diyaudio.com/forums/diy-...erver-computer-transport-hop.html#post1976858
the guys from diyhifisupply did also some mods like you on the Musiland 01 USD and afterwards a listening session with several people.
After the mods the 01 USD sounds as good as their reference CD-player:
http://www.diyaudio.com/forums/diy-...erver-computer-transport-hop.html#post1976858
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.