A humble and weird request of help

[ginetto61]

Hi !

I am a little embarassed to ask you this
I would like to build a line stage (wait a little before move the 3d ... please ... )
Looking at some measurements on cd/dvd/br players i see often excellent measurements in terms of distortion and noise
To cut short, i would like to take out the PS and analogue stage of a good player, place a pot before the analogue stage and build a line preamp
I ask these because is so much easy to find a used player, maybe with a broken spinner in the market
Maybe I could upgrade some components like caps
What do you think ?
As i said i had this idea looking at the measurements of some players
They look even better than those of a typical line stage
Or another idea would be to buy just a pcb of a dac in kit with a particularly good and strong output stage and build just the analog stage (maybe even the best choice)
I cannot start from scratch ... I am not that good
I madly need a pcb and also easy to work with (i.e. not too crammed with components)
And of coure with smd components I should not have to solder in the circuit
Do you have any suggestion/advice ?
I am interested only in solid state circuits
Thank you very much indeed
Kind regards,
gino

 

[DF96]

Do you have any suggestion/advice ?

Re-think. The amplifier following a DAC is doing a different and somewhat harder job than a line stage, so its performance is likely to be inferior to a line stage designed with the same degree of skill. If you compare 'audiophile' line stages with ordinary commercial digital stuff, you may well see that the measurements are better for the digital stuff. That just tells you something about some 'audiophile' line stages!

 

[ginetto61]

Re-think. The amplifier following a DAC is doing a different and somewhat harder job than a line stage, so its performance is likely to be inferior to a line stage designed with the same degree of skill.
If you compare 'audiophile' line stages with ordinary commercial digital stuff, you may well see that the measurements are better for the digital stuff.
That just tells you something about some 'audiophile' line stages!

Thank you but i do not understand
Actually my idea comes from some tests seen on magazine
A lot of dvd/blue ray layers show spectacular noise and distortion performance
so their output stage must be excellent
Are you saying that they cannot be converted to line stage ?
I would rip apart the platter and the digital circuits to leave only PS and output stages
So also a unit with broken mechanism could do just fine
Then a decent pot before the output stages and it is done
Am I wrong ?
I think i should look at old unit without smd components
Is it not feasible ?
Thanks and regards,
gino

 

[DF96]

Most older DAC chips produce an output current which is fed into a virtual ground at an op-amp input. The opamp then provides gain to bring the signal up to a couple of volts. It may also include some analogue filtering to augment the digital filtering usually done by oversampling. So we have virtual ground, voltage gain and filtering. Quite different from a line stage: high input impedance, little gain and no filtering. Different purpose; different circuit.

Newer DAC chips often produce a differential voltage output. A line stage does not usually need to be differential. As I said, a line stage ought to be simpler and better than a DAC output stage.

You need to do much more reading and research.

 

[ginetto61]

I was clearly wrong
Bad idea indeed
Thanks againg
Kind regards,
gino

 

[twest820]

Whether a DAC is current mode versus voltage mode output is more about what price/performance point it's aimed at and who makes it than whether it's old or new. Also, the I/V resistors for current mode DACs are usually chosen to deliver a couple volts RMS. The main value of gain in the line stage/output buffer/analog antialiasing filter op amps is to provide sufficient voltage swing on the interconnect that common mode noise between grounds and the line receiver's noise don't swamp the DAC's noise floor. The higher end of the line driver/line stage performance range requires instrumentation amplifier type topologies in order to keep the feedback network's impedance low enough that the line driver's noise floor is below that of the DAC's. It's slightly easier to implement the line driver front end with current output than voltage mode output DACs, though that's beside the point if you happen to be working with something like a CS4398 or WM8471.

THAT's design support documents provide a good introduction to line driver and receiver design considerations (Rod Elliot also hosts a couple articles written by the Jensen transfomer folks who collaborated with THAT and may be a bit more accessible). To summarize, the intrinsic noise of line drivers, receivers, preamps, and power amps is a microvolts happen kind of thing. Common mode differences between audio component grounds or even boards in the same chassis are a tens of millivolts happen kind of thing. So if you don't want to blow the DAC's noise floor on the interconnect both gain and 60+dB CMRR tends to be required. It's wise to choose good gain structure so that a higher noise floor is acceptable and tracking supplies offer mitigation in some cases. But usually 40+dB CMRR ends up being desirable. This is easy to achieve with transformer coupled single ended (unbalanced) line drivers and receivers but balanced interconnects are an order of magnitude cheaper.

My experience is there's a curious hole the line driver market with respect to DAC output buffer solutions and even the output buffers on manufacturer's eval boards are often somewhat dubious. Ginetto, some places you might start with are schematics for Cirrus's eval boards, the THAT 1606 datasheet and relevant design notes and papers, and figure 14 of the LME49724 datasheet. This'll give you some sense of the design tradeoffs in antialias filters, common mode level shifting, output impedance matching, and floating versus anchored outputs.

 

[ginetto61]

Thanks a lot for the kind and valuable advice
But I understand that is not the easiest way
I will look at a line stage kit with opamp like ne5534 or similar
In order to have a minimum chance to do something working
Kind regards,
gino

 

[DF96]

Also, the I/V resistors for current mode DACs are usually chosen to deliver a couple volts RMS.

I thought most current output DACs have to operate into a virtual ground. The resistor provides feedback around the opamp. Very few current mode DACs can cope with more than a few 10's of mV at the output pins, as the switched current sources are not designed to cope with more.

 

[qusp]

Gino, are you sure you actually need a line stage? why dont you pick a player that has digital volume and sufficiently low output impedance and use directly with your amp?

 

[twest820]

I thought most current output DACs have to operate into a virtual ground. The resistor provides feedback around the opamp.

Accurate IV translation requires the current be delivered to a reference, but that reference can be a variety of things. In the audio case with a line driver a parallel RC in an op amp's feedback loop is probably the most convenient and best performing place. The op amp's input pin voltage does provide a virtual reference but it's not necessarily ground. Burr-Brown/TI sets up their IV so that it is ground but in a level shifting IV like ESS uses it'll be the DAC's midvoltage reference and in a fully differential IV like Twisted Pear's IVY it's perhaps best thought of as a floating reference between the cross coupled input and output.

Very few current mode DACs can cope with more than a few 10's of mV at the output pins, as the switched current sources are not designed to cope with more.

I just checked a few eval board docs and datasheets I have handy and the pins'll take the usual ground - diode drop and Vcc + diode drop or better (sometimes by a couple volts) and the resistors are sized so the typical ~4mA full scale output current translates to a volt or two (in the passband before the lowpassing cap takes effect).

 

[twest820]

But I understand that is not the easiest way

True that, though the THAT 1606 and 1646 do provide pretty turnkey options if the priority's to get something working.

 

[DF96]

The op amp's input pin voltage does provide a virtual reference but it's not necessarily ground.

I meant AC ground; usually DC elevated by a reference pin.

 

[twest820]

And often seen as level shifted due to power supply mismatches. There's also some fun with bias currents, what the other end of the cable happens to implement if galvanic isolation isn't used, and output as well as input CMRR. Even if noise is held to microvolts the offsets tend to be millivolts to tens of millivolts.