New HQ DIY DAC

[Deleted]

but if you feel that the TOSLINK works OK you might be ready to listen to the Lightspeed as it uses photons too.

When Hell freezes over !
My preamps and P.A. have very close to 4 zeroes in the distortion figures, and I am not about to degrade it by a factor of more than 100 times by controlling an ANALOGUE signal with a photo electric semiconductor junction. Nelson Pass has already confirmed this graphically in THAT thread.
I prefer coax SPDIF, but The Samsung LED TV has Toslink, and most day to day listening is via the TV.
I realise that conversion to Toslink and back again is not the greatest way to do things, but I make sure that area of the DAC has the cleanest possible DC supply, as any noise in the power there would be able to increase timing Jitter.That's part of the rationale behind the 3.3V PH regulator that I use in that area.
Regards
Alex

[Will]

Hi Guys,

My simplisitic view on spdif cable/receiver mismatch is that it has a similar effect to a transmitter with an incorrectly sized aerial for the frequency it's transmitting. If the aerial is incorrect, then a proportion of the signal is reflected back down the line, into the power stage. With spdif, this reflection increases the jitter present. As Greg says, with the 74HC (which acts as a buffer, and ensures that all spidf signal levels are 'equalised') that Jitter will still be present.

I think that the resistor change is due to matching the spdif input to the 74HC, rather than the dir9001.

Jon Clancy has been investigating spdif input here Which I think is important in trying to reduce jitter levels.

[Deleted]

I think that the resistor change is due to matching the spdif input to the 74HC, rather than the dir9001

Hi Will
The resistor change is to correctly terminate the cable to help avoid reflections, and not the 74HCU04 which has a high input impedance.
Alex
P.S.
I was unable to notice any difference after making the change
when playing 24/96 "Queen-Another One Bites the Dust"
However, 1.5M is widely reported to be an optimum length of cable for minimum reflections.

74HC Supply: 2 to 6V, small fluctuations are tolerated.
74HCT Supply: 5V ±0.5V, a regulated supply is best.
Inputs have very high impedance (resistance), this is good because it means they will not affect the part of the circuit where they are connected. However, it also means that unconnected inputs can easily pick up electrical noise and rapidly change between high and low states in an unpredictable way. This is likely to make the IC behave erratically and it will significantly increase the supply current. To prevent problems all unused inputs MUST be connected to the supply (either +Vs or 0V), this applies even if that part of the IC is not being used in the circuit!

[Deleted]

Dec 2, 2011 6:29:41 GMT @sandyk said:

but if you feel that the TOSLINK works OK you might be ready to listen to the Lightspeed as it uses photons too.

When Hell freezes over !
My preamps and P.A. have very close to 4 zeroes in the distortion figures, and I am not about to degrade it by a factor of more than 100 times by controlling an ANALOGUE signal with a photo electric semiconductor junction.....
Alex

Hi Alex,

You need to open your mind to other possibilities. To laws of physics do not apply to hifi.

Everyone knows photon induced distortion is benign and can not add to electron distortion. It's like mixing water and oil. ;D

I'll bring the Lightspeed along to our next GTG and we can listen to it for a couple of hours.

regards

[Deleted]

I'll bring the Lightspeed along to our next GTG and we can listen to it for a couple of hours.

Hi Greg
That will be YOUR next GTG. You can play it in line with a valve preamp and a couple of GCs to your heart's content.
Compare that against Carlos's DX and marvel about how good GCs sound in comparison! Luckily, I will be too many suburbs away to hear it!
Regards
Alex

[Deleted]

Dec 2, 2011 8:59:18 GMT Will said:

Hi Guys,

My simplistic view on spdif cable/receiver mismatch is that it has a similar effect to a transmitter with an incorrectly sized aerial for the frequency it's transmitting. If the aerial is incorrect, then a proportion of the signal is reflected back down the line, into the power stage. With spdif, this reflection increases the jitter present. As Greg says, with the 74HC (which acts as a buffer, and ensures that all spidf signal levels are 'equalised') that Jitter will still be present.

I think that the resistor change is due to matching the spdif input to the 74HC, rather than the dir9001.

Jon Clancy has been investigating spdif input here Which I think is important in trying to reduce jitter levels.

Hi Will,

I think you are correct. Back in the old days when coax was used for networking, the cable had to be exactly the right length, all connectors were BNC and the ends of the cable had terminations to prevent reflections. A dicky cable, connector or terminator caused the whole leg of the network to fail.

I also believe they are trying to match input impedance which includes the first 74HCU04 gate. Its just a matter of how your understand the function of the first gate. I think it is set up as a linear amplifier. If you were to ignore the circuit to the right of the resistor, the value should be 75 ohm.

Yes I think John could help solve this issue for us as he seems to have a good understanding of what is going on. I know he has a bunch of input transformers to play with.

regards

[Deleted]

Dec 2, 2011 9:09:04 GMT @sandyk said:

I was unable to notice any difference after making the change
when playing 24/96 "Queen-Another One Bites the Dust"

Hi Alex,

It would be interesting to keep playing with this resistor value until you can notice a difference. It lools like you have proven that matching impedance is a non-issue.

regards

[Deleted]

Dec 2, 2011 20:22:43 GMT @sandyk said:

I'll bring the Lightspeed along to our next GTG and we can listen to it for a couple of hours.

Hi Greg
That will be YOUR next GTG. You can play it in line with a valve preamp and a couple of GCs to your heart's content.
Compare that against Carlos's DX and marvel about how good GCs sound in comparison! Luckily, I will be too many suburbs away to hear it!
Regards
Alex

;D

Question: Can you have a GTG with one person?

[Deleted]

Question: Can you have a GTG with one person?

Do you have a split personality ?

[Deleted]

I also believe they are trying to match input impedance which includes the first 74HCU04 gate

The 74HCU04 is a CMOS device, and is high impedance, so it can be neglected as far as impedance matching is concerned.
I had another listen to the same track again last night, and I am not convinced that the change has made any audible improvement,
and it may in fact be slightly worse.It would also seem likely that individual 74HCU04 may differ a little in their hysterisis points, so this may also affect results, as the input level will be a little higher with 300R instead of 75R.
You really need to fine tune these things with the aid of something like a 100MHZ C .R.O.
A question for Marc or Frans . Does a 1.5M length of 75R RG59 etc. even present an impedance of 75 ohms with such a short length ?
Alex

[Deleted]

NEXT MONTH in Silicon Chip (February 2012)

Crystal DAC
For the very best performance from 24bit /96kHz recordings,this new board uses the Crystal CS4398 DAC,and a discrete transistor output stage. It can be dropped into our original Stereo DAC project without any major changes,replacing the Burr-Brown DSD1796 DAC I.C. with the high end Cirrus Crystal CS4398. It's harmonic and intermodulation distortion figures are significantly lower than before and surprise,surprise, the differences can be discerned.Try it and find out for yourself.

[Will]

Jan 2, 2012 8:49:28 GMT @sandyk said:

NEXT MONTH in Silicon Chip (February 2012)

Crystal DAC
For the very best performance from 24bit /96kHz recordings,this new board uses the Crystal CS4398 DAC,and a discrete transistor output stage. It can be dropped into our original Stereo DAC project without any major changes,replacing the Burr-Brown DSD1796 DAC I.C. with the high end Cirrus Crystal CS4398. It's harmonic and intermodulation distortion figures are significantly lower than before and surprise,surprise, the differences can be discerned.Try it and find out for yourself.

That looks interesting Alex. I do wonder what they've based the term very best performance on, most likely specs rather than listening. Very interested in the output stage though, so would be interested in what that brings up.

Looks like the whole DAC/IV board will be replaced though, along with those nice metal jacket lme49710's

[Deleted]

Jan 2, 2012 8:49:28 GMT @sandyk said:

NEXT MONTH in Silicon Chip (February 2012)

Crystal DAC
For the very best performance from 24bit /96kHz recordings,this new board uses the Crystal CS4398 DAC,and a discrete transistor output stage. It can be dropped into our original Stereo DAC project without any major changes,replacing the Burr-Brown DSD1796 DAC I.C. with the high end Cirrus Crystal CS4398. It's harmonic and intermodulation distortion figures are significantly lower than before and surprise,surprise, the differences can be discerned.Try it and find out for yourself.

Hi Alex,

Any news on this?
As the casework for my PK/SCHA is passing the halfway stage I'm thinking it's time the DAC had a permenant home too, preferably with all mod cons

[Deleted]

Apr 17, 2012 22:20:37 GMT @cjarchez said:

Jan 2, 2012 8:49:28 GMT @sandyk said:

NEXT MONTH in Silicon Chip (February 2012)

Hi Alex,

Any news on this?
As the casework for my PK/SCHA is passing the halfway stage I'm thinking it's time the DAC had a permenant home too, preferably with all mod cons

Hi Chris
There wasn't a kit released by the usual suppliers, and the pre-programmed micro did not appear (at least yet) on the Silicon Chip list of available items.Reluctantly, I have decided to pass on this, even though I would have loved to try the front end balancing techniques in it's discrete output stage.
Kind Regards
Alex

[Deleted]

Cheers Alex. Maybe I'll leave a bit of extra space, just in case (a first for me )

[Deleted]

Apr 17, 2012 22:43:37 GMT @cjarchez said:

Cheers Alex. Maybe I'll leave a bit of extra space, just in case (a first for me )

Hi Chris
I have been having another look at this one. Silicon Chip sells the PCB for around $20. The Atmel Microcontoller can be supplied pre-programmed for $15. Element 14 sells the Crystal DAC for around $21, but you would need to mount this SM device yourself. It uses a couple of dual FETs for muting, that aren't really necessary.
It can even run in basic mode without input selection,or variable volume control just by solder jumpers, and without replacing the existing Atmel Micro. The rest of the stuff are run of the mill components with lots of BC549 and BC559 transistors.
Seems that in Blind testing half the SC staff couldn't hear a difference, and another couple thought it sounded brighter.
Distortion at medium to HF is quite a bit lower.
It should be possible to improve front end balance markedly with a couple of simple track cuts and 2 signal diodes, or even by inserting a selected resistor value between the current mirrors and the differential pairs.
Perhaps further down the track I may try it.
Regards
Alex

[rbrook]

OK I've finally got around to getting this HQ DAC kit from Jaycar. Now comes all the reading to find out the best mods to do to it.
Has anyone already put together a list of mods for this project?

[Deleted]

Hi Richard
To get the very best out of this kit you will need several JLH PCBs to start with, but at present there are currently none available.
If you haven't any, you may be able to get some singles via the Group Buy for the Class A. The best person to contact there is Shaun, to see if you can be added to the list.
Do you want to start with the standard SC filter values for the DAC PCB, or go straight to the wider bandwidth of the DSD1792 type filtering similar to it's data sheet ? I would also recommend using the OPA134 for all except the output I.C.s where the LME49710 can be used, and the metal can LME49710HA version is a little better again. You need to decide if you want to go straight to the wider bandwidth filtering, in order to know what value polypropylene capacitors to order.If you take reasonable care and have experience with unsoldering and replacing components, you could of course go there later on.There are lots of small modifications that can be done progressively.On the input PCB, a good place to start would be to replace the 4.7nF and 68nF on pin 22 of the DIR9001 with polypropylene types, which are closer tolerance types.
Regards
Alex

[Deleted]

Richard
Don't forget to check your DAC PCB as per post 450 from Chris.
Some PCBs had a missing track there.
Also, if you have a PCB drill, now would be a good time to fit a .9mm PCB pin as shown. Later on, a heavy duty piece of insulated wire can be connected from there back to the JLH's earth , in parallel with the existing earth to the 3 pin terminal block..
Regards
Alex

Attachments:

[rbrook]

Cheers for the info Alex.
I'm fortunate to have a few extra JLH PCB's in my stockpile.

Now time to do some parts shopping.

[Deleted]

Jun 4, 2012 0:32:07 GMT rbrook said:

Cheers for the info Alex.
I'm fortunate to have a few extra JLH PCB's in my stockpile.

Now time to do some parts shopping.

Hi Richard
Anything further you want to know, just ask.
There are a lot of little changes needed to get it to the state where it can outperform even some of the DACs such as W4S, and the Perfect Wave Transport, and a couple of members are already there or close to it.

Don't let the photo frighten you though, as it can be done in stages. The advantages of the external Xtal Oscillator are debatable though, but I don't want to go backwards to verify that.
(The additional earth lead is not shown in this earlier photo)

Regards
Alex

Attachments:

[Deleted]

SC DAC UPDATE 31-08-2012.
In the attached main photo, the additional earth from the JLH PCB to the main earth of the DAC PCB can be seen. This results in a further improvement in channel separation etc.
Some of you may also be aware of a Silicon Chip magazine article about the high AC voltage present at the earth side of the output RCAs and coax SPDIF with sources using SMPS.
Although it is from a high impedance, and safe, it can give a nasty bite when plugging/unplugging the CD/DVD/STB into an amplifier that is earthed via a 3pin mains plug.
This AC residual also degrades ultimate performance. Contact me if you require further info regarding this aspect.
For those who have built this DAC and are using SMPS CD/DVD players connected via Coax SPDIF, and that do not have SPDIF transformers inside them (most these days) then it may be worth while trying a 100nF 275VAC X2 capacitor between the earth side of the Coax SPDIF Input and chassis.This capacitor can be seen in the attached photo of the Input PCB. To make this easier to install, I drilled a 3mm hole near the coax socket, with a solder lug fitted.
Before doing this, check with a DMM that the Coax input socket and
the Output RCA sockets "earth " sides are not connected to chassis.
I read around 23 ohms between the earth side of the Coax Input socket and the earth side of the RCA output sockets, but no reading to chassis.
My CRO now shows a very worthwhile improvement in residual noise between Coax socket's earth side and chassis when it is connected to my preamp and power amplifier. (My 15W Class A already has 10ohm "earth lift" resistors at it's input)If looking at this residual "noise" with a C.R.O. do not forget to first connect the DAC to the rest of the gear.
Preliminary listening appears to show a noticeable improvement when playing a home made DVD-A from my Oppo DV981HD.
Alex

P.S.
I only tried a 100nF X2 capacitor as that is what I had on hand.
It may be possible to obtain a similar improvement as seen on a CRO with either a 22nF X2 , or 47nF X2 capacitor.It would be best to use the lowest value that gets rid of the obvious sine wave, but you would need a CRO to check that.
I intend trying both those lower values in a few days time, as it does alter the sound slighly.As these capacitors are less than a dollar each usually, you could try using different values.
IIRC Silicon Chip used 22nF in a "tingle eliminator" some time back.

[Deleted]

Pretty pictures always make it so much easier, cheers Alex

[Deleted]

Aug 31, 2012 11:41:08 GMT @cjarchez said:

Pretty pictures always make it so much easier, cheers Alex

Hi Chris
Hang fire on the Coax SPDIF bypassing to IEC earth until I try those lower capacitor values. Although DVD-A from the Oppo definitely sounds quite a bit better, DTV audio via Toslink did not sound quite as good this evening. Best results for coax input may require the use of a quality SPDIF transformer as a Sydney DIYAudio member has done with his, where a SMPS source is used.
Regards
Alex

[Deleted]

Hi Alex,

yes, I did take note of your orginal caution there and will watch it develop.

I tend to DL the photos, mark and text them appropriately as they appear, rather than playing catch-up later and wading through lengthy threads

Cheers.