Looking in a little detail at the datasheet for the CS4328 DAC in my Creek CD42, it has a built in Mosfet output stage which will (when combined with a simple cap/resistor filter) drive a set of interconnects directly, with no need for further amplification or filtration. How good is this stage, using it directly will allow me to bypass the OPA2604 chips, which can then be removed, reducing the load on the PSU (which is another good thing).
On the other hand, I imagine Creek knew what they were doing when they put them in there, so does anyone have any experience of using this chip with no external output stage?
Cheers,
Ed
On the other hand, I imagine Creek knew what they were doing when they put them in there, so does anyone have any experience of using this chip with no external output stage?
Cheers,
Ed
If you read the datasheet (I have it) you see that this chip has a built-in filter and as external filter is only a R+C needed. Download my schematics from my homepage if you want to see how it can be done.
Those of you who wants the datasheet (which isn't availabe from Cirrus), send me a private message.
This mosfet thing is a some sort of "sound enhancer".
You can infact use the signals directly from the dac only via this really simple filter.
Those of you who wants the datasheet (which isn't availabe from Cirrus), send me a private message.
This mosfet thing is a some sort of "sound enhancer".
You can infact use the signals directly from the dac only via this really simple filter.
Thanks for the quick reply! I have read the datasheet and it is because of this that I am puzzled that Creek chose to use an OPA2604 between the DAC and the outputs, as this should not be necessary. Ideally I can remove this and gain some sound quality, but I want to confirm this (if possible) before I get to work with pliers and soldering iron.
Cheers,
Ed
Cheers,
Ed
You can remove everthing except for the single filter .... but beware of this chip is no longer available, be carefull when you connect 
Oh, I forgot, the opamp is probably the deemphasis thing which is to 99.9% unnecessary. The CS4328 has not built-in deemphasis filter.
Oh, I forgot, the opamp is probably the deemphasis thing which is to 99.9% unnecessary. The CS4328 has not built-in deemphasis filter.
I have two output, one straight out and one through an opamp (OPA627) which also contains deemphasis.
I can't hear any difference when I switch between these two outputs. But the problem is that my preamp has a small delay when I switch inputs. During this short time my hearing tends to forget. => not very important difference if you can't clearly hear a difference. OPA627 is very good
I can't hear any difference when I switch between these two outputs. But the problem is that my preamp has a small delay when I switch inputs. During this short time my hearing tends to forget. => not very important difference if you can't clearly hear a difference. OPA627 is very good
quad cdp
The Quad cd67 uses the CS4328 too. It sounds beautiful.
Does anyone have the dac and output stage schematic?
I guess that having the opamps prevent loading of
the mosfet gain stage at the dac chip output. I have used
AD827 loaded with 5k ohm and they do not like that.
The bass lines sound very muffled, like hitting a pillow
instead of a kick drum.
They prefer at least 10k or higher. It seems that even though
the output stages can drive lower impedances, those low
impedances are not neccessarily the optimal impedance
when dynamics are called for.
Yv
The Quad cd67 uses the CS4328 too. It sounds beautiful.
Does anyone have the dac and output stage schematic?
I guess that having the opamps prevent loading of
the mosfet gain stage at the dac chip output. I have used
AD827 loaded with 5k ohm and they do not like that.
The bass lines sound very muffled, like hitting a pillow
instead of a kick drum.
They prefer at least 10k or higher. It seems that even though
the output stages can drive lower impedances, those low
impedances are not neccessarily the optimal impedance
when dynamics are called for.
Yv
CS4328 is kind of special, this is a two chip solution in the same package. This was an expensive way of production but I would say that this chip is one of the best 16-bit DAC's. I'm amazed over the output section with +-5 voltage and complete analog filtering except for a simple LP-filter.
I know though that top of the line DAC's today has equal or better performance at 16 bits but CS4328 was very good at the time and still is.
I know though that top of the line DAC's today has equal or better performance at 16 bits but CS4328 was very good at the time and still is.
Oh yes, the CS4328 was a GREAT DAC chip. I have a soft spot for it as it was the heart of my first DIY DAC the X-DAC 1.0 back in 1991.
Kluge circuit is proof of concept prototype for the ECL master clock circuit. Even in this lashup the improvement a good clock made was astounding.
An important article that introduced us to the wonders of ECL based clocks
This led to the now out of production X-DAC 3.0.
Its optional case.
And ultimately the X-DAC 3.0 Signature
OK, enough time walking down memory lane. Back to the question of using output/buffer stages with CS4328. All that has been written above is correct and I agree. What I want to add is some real world experience gained in my listening room and reports from X-DAC 3.0 users in the Americas, Europe, and Asia.
Like everything in engineering its a trade-off. By-passing the output buffer stage yields maximum transparancy and the finest level of musical detail. The down side is a reduction in bass slam and overall dynamics. A good buffer amp with great power supplies gets you almost all the way there as far as transparancy and resolution of musical details while bringing to the party a fuller measure of bass slam and both micro & macro dynamics.
The results described above are system dependent. If you have an easy to drive device following CS4328 (i.e. high-Z preamp + low cap cables) the negative effects of doing without the buffer amp will be largely elimanated. On the other hand if you want to drive 10 meters of cable into a passive attenuator 'ya better keep that buffer onboard. Put another way CS4328+buffer is a more universal solution while CS4328 naked needs careful system matching to do its best.
FYI my favored buffer for this application is the AD811 curent feedback video op-amp. Tons of bandwidth, low enough noise and as its designed to drive double terminated 75 ohm video lines it barely notices audio loads.
I'll finish up with a couple more hints on tweaking CS4328s. This device will work fine with very simple low cost supporting circuits. However it will respond to every improvment in its surroundings rewarding you with more and more music. As always in hi-fi there is no such thing as too good a power supply system. Get those power supply by-pass caps off pins 2, 5, & 16 replaced with Black-Gates, OS-CONs, or low Z Panasonics ASAP. Careful regulator upgrades will be richly rewarded. Also note pin 28 VREF. The caps off this pin bypass the voltage reference the DAC is using and have a very audible effect. Again low Z types are called for. I liked 0.1uF COG ceramic + 10uF BG. Also heard good resuls using audio quality plastic film types (MITs were favored back then) if they can be made to fit. Try and keep the ground side of the VREF by-pass caps terminated near the CS4328's pin 1 to keep your ground current loops as small as possible. On the X-DAC 3.0 Signature version I soldered the caps leads on the bottom of the PCB directly on the pads for pins 1 & 28.
happy listening
Norman Tracy
Audio Crafters Guild
www.audiocraftersguild.com
An externally hosted image should be here but it was not working when we last tested it.
Kluge circuit is proof of concept prototype for the ECL master clock circuit. Even in this lashup the improvement a good clock made was astounding.
An important article that introduced us to the wonders of ECL based clocks
This led to the now out of production X-DAC 3.0.
An externally hosted image should be here but it was not working when we last tested it.
Its optional case.
An externally hosted image should be here but it was not working when we last tested it.
And ultimately the X-DAC 3.0 Signature
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
OK, enough time walking down memory lane. Back to the question of using output/buffer stages with CS4328. All that has been written above is correct and I agree. What I want to add is some real world experience gained in my listening room and reports from X-DAC 3.0 users in the Americas, Europe, and Asia.
Like everything in engineering its a trade-off. By-passing the output buffer stage yields maximum transparancy and the finest level of musical detail. The down side is a reduction in bass slam and overall dynamics. A good buffer amp with great power supplies gets you almost all the way there as far as transparancy and resolution of musical details while bringing to the party a fuller measure of bass slam and both micro & macro dynamics.
The results described above are system dependent. If you have an easy to drive device following CS4328 (i.e. high-Z preamp + low cap cables) the negative effects of doing without the buffer amp will be largely elimanated. On the other hand if you want to drive 10 meters of cable into a passive attenuator 'ya better keep that buffer onboard. Put another way CS4328+buffer is a more universal solution while CS4328 naked needs careful system matching to do its best.
FYI my favored buffer for this application is the AD811 curent feedback video op-amp. Tons of bandwidth, low enough noise and as its designed to drive double terminated 75 ohm video lines it barely notices audio loads.
I'll finish up with a couple more hints on tweaking CS4328s. This device will work fine with very simple low cost supporting circuits. However it will respond to every improvment in its surroundings rewarding you with more and more music. As always in hi-fi there is no such thing as too good a power supply system. Get those power supply by-pass caps off pins 2, 5, & 16 replaced with Black-Gates, OS-CONs, or low Z Panasonics ASAP. Careful regulator upgrades will be richly rewarded. Also note pin 28 VREF. The caps off this pin bypass the voltage reference the DAC is using and have a very audible effect. Again low Z types are called for. I liked 0.1uF COG ceramic + 10uF BG. Also heard good resuls using audio quality plastic film types (MITs were favored back then) if they can be made to fit. Try and keep the ground side of the VREF by-pass caps terminated near the CS4328's pin 1 to keep your ground current loops as small as possible. On the X-DAC 3.0 Signature version I soldered the caps leads on the bottom of the PCB directly on the pads for pins 1 & 28.
happy listening
Norman Tracy
Audio Crafters Guild
www.audiocraftersguild.com
cs4328 application
Hi Norm
If I use a 'dead-bug' construction method, can a huge
common ground plane be used as an alternative to separating
analog and digital ground planes (and joining them
only at the psu)?
btw, any plans for follow on products
to the xdac 3.0?
thanks for the inputs
Yv
Hi Norm
If I use a 'dead-bug' construction method, can a huge
common ground plane be used as an alternative to separating
analog and digital ground planes (and joining them
only at the psu)?
btw, any plans for follow on products
to the xdac 3.0?
thanks for the inputs
Yv
Oops and 10nf cap recommendations
Well, I have removed the op amps, but due to a momentary lapse in concentration, I melted a good chunk off one of the 10nf film caps between the DAC's audio outputs and ground, giving me a dead channel. They will obiovusly both need to be replaced, any recommendations? (10nf non polarised types)
Cheers,
Ed
Well, I have removed the op amps, but due to a momentary lapse in concentration, I melted a good chunk off one of the 10nf film caps between the DAC's audio outputs and ground, giving me a dead channel. They will obiovusly both need to be replaced, any recommendations? (10nf non polarised types)
Cheers,
Ed
10nf caps
Hi mcai7et2
Use teflon, polystyrene or polypropylenes, in that order
of preference.
Yv
Hi mcai7et2
Use teflon, polystyrene or polypropylenes, in that order
of preference.
Yv
Re: cs4328 application
Ah yes, I've done some "dead bug" hand built protos in my time. Here is one of the last big ones:
As you can see its a solid uncut copper plated blank PCB drilled to accept sockets and parts. Some of the mini coax used is seen on the top side. Below you find more coax along with soldered wire for power and wire-wrap depending on what that node needs.
Yes I would use "a huge common ground plane" and skip "separating analog and digital ground planes (and joining them only at the psu)" in both dead-bug prototypes and PCB layout. The once popular practice of using split ground planes has fallen out of favor with most engineers. The problem is the BIG-O current loops which result from those signals which must go from the digital to analog side. See Supply Decoupling by Guido Tent on the ACG Articles and Essays page for more background.
Yes a replacement for the X-DAC 3.0 kit is in the works. A massively overloaded To-Do list has left the ACG web site home page without a link to the project. I do have some info up for those who follow ACG's R&D efforts. It can be found at these two links:
XD0 Schematics and XD0 circuit background and theory of operation.
enjoy!
yves said:
Ah yes, I've done some "dead bug" hand built protos in my time. Here is one of the last big ones:
An externally hosted image should be here but it was not working when we last tested it.
As you can see its a solid uncut copper plated blank PCB drilled to accept sockets and parts. Some of the mini coax used is seen on the top side. Below you find more coax along with soldered wire for power and wire-wrap depending on what that node needs.
Yes I would use "a huge common ground plane" and skip "separating analog and digital ground planes (and joining them only at the psu)" in both dead-bug prototypes and PCB layout. The once popular practice of using split ground planes has fallen out of favor with most engineers. The problem is the BIG-O current loops which result from those signals which must go from the digital to analog side. See Supply Decoupling by Guido Tent on the ACG Articles and Essays page for more background.
Yes a replacement for the X-DAC 3.0 kit is in the works. A massively overloaded To-Do list has left the ACG web site home page without a link to the project. I do have some info up for those who follow ACG's R&D efforts. It can be found at these two links:
XD0 Schematics and XD0 circuit background and theory of operation.
enjoy!
Done it!
Well, I removed the ouput stage and replaced the 220ohm resistor and 10nf capacitor with 51 Ohm Welwyn RC55Y resistor and 10nf MKP Prolyprop cap (these arethe values on the DAC datasheet and it seemed a good place to start). I also replaced the Gemcon 10uf decoupling caps on the DAC with OsCon SCs.
The net result of these changes is staggering, the CD player is now (subjectively of course) performing like a player twice the price. Thanks for all your help guys, going to tackle the PSU and regulators next - I will be in touch.
Cheers,
Ed
Well, I removed the ouput stage and replaced the 220ohm resistor and 10nf capacitor with 51 Ohm Welwyn RC55Y resistor and 10nf MKP Prolyprop cap (these arethe values on the DAC datasheet and it seemed a good place to start). I also replaced the Gemcon 10uf decoupling caps on the DAC with OsCon SCs.
The net result of these changes is staggering, the CD player is now (subjectively of course) performing like a player twice the price. Thanks for all your help guys, going to tackle the PSU and regulators next - I will be in touch.
Cheers,
Ed
- Status
- Not open for further replies.