OpAmp update/mod in the Sony CDP-XA7 ES

[Crossy]

Well,
for starters, the Sony CDP-XA7 ES was one of the last Sony "high-end" CD players (available since 1995
up to 2005). It features a Bitstream DAC with internally balanced IO (although not taking fully advantage
of it in its original configuration). It also uses DC coupling in the output stage instead of a coupling capacitor.
(through an opamp based DC-Servo - OPA-27GP from Burr Brown, IC461/561)  8-)

To follow this thread, the reader needs a Service Manual that can be obtained from various sources,
for example from www.manuals-in-pdf.com via direct .pdf download.  :-D

Since the design is already high-quality (in typical Sony-ES series fashion), most of my
hints will involve merely chaning opamps to more current types, as there has been
TREMENDOUS progress in this field over the last decade with regard to noise immunity
(opamp voltage / current noise), output current capability, linearity and speed (slew rate, settling time).   8-)

http://www.thevintageknob.org/SONY/sonyes/CDPXA7ES/CDPXA7ES.html
http://cdp101.hp.infoseek.co.jp/cdpxa7es/cdpxa7es.html
http://translate.google.com/translate?u=http%3A%2F%2Fcdp101.hp.infoseek.co.jp%2Fcdpxa7es%2Fcdpxa7es.html&langpair=ja%7Cen&hl=fr&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools

For most of the post-DAC filtering, the XA-7 ES originally relies on the AD-712. All these opamps are present  on the PCB as SMD types.
Note that the AD-712 is a FET opamp - in a dual opamp package. I found this puzzling as only the first stage after the DAC (labeled "de-emphasis" filter)
seems to require high-impedance (and hence a FET input opamp). Beyond that it appears to me that ordinary BJT opamps would do the job and provide
reduced noise and better sound also. Maybe back then only the AD-712 was available as an SMD chip.
Also present are NJM-2114D types in DIP package - I guess they are used where higher output current is required. Also, these are ordinary BJT opamps
(In the CDP-XA7 as Headphone driver and variable unbalanced output path amplifier). Essentially, they can be considerd "lower noise" NE-5532 replacement,
they calso can handle higher output current than the plain-vanilla NE-5532. The NJM-2114D are typically found in japanese Audio / Video gear and the
prefix "NJM" stands for "New Japan Radio Corp".

My plan is to get "optimal" state of the art opamp types in the audio signal path - for each of the functions the respective stage performs. So my gut feeling tells me
to minimize the parameters noise, "settling time", voltage and and maximize Slew Rate when possible. The problem is - what parameters take precedence in each function ?

IC-504, IC-404 ... OPA-27GP ... DC Servo - doesn't seem to require any change

IC-401,IC-501 (AD-712) post DAC Filter "De-Emphasis" handles balanced audio out from DAC.... appears to require High input Impedance
preference: replace with THS-4032  (dual type, BJT but 2MegOhm Rin, +/- 16V, 100Mhz, 1.6nV noise, 60ns settling time, 100V/µS slew, 2MegOhm Input Impedance, drives 90mA output, -90dB THD 1Mhz)
fallback: replace with AD8620 (dual type, slightly more noise, FET opamp, +/-13V max, 25Mhz, 6nV noise, 5fA noise, 600ns settling time, 50V/µS slew, THD 0,0006% 10kHz)

IC-402, IC-502 (AD-712) labeled "mix amp" (only 1 opamp of the dual AD-712 was used originally !) - makes single ended audio signal from balanced signal
replace with OPA211 (SiGe BJT opamp, +/- 18V, max. 80Mhz, 1.1nV noise, 1pA noise, 400ns settling time, 27V/µS slew rate, THD -136dB 1kHz)
     it appears to me that a BJT opamp can be used here to make the balanced signal single ended. Single type opamp possible since the original opamp used only one half od the dual AD712
     no need for a high-impedance opamp such a FET input opamp and the OPA211 should easily work here (not particulary demanding type wrt. oscilation)

IC-403,IC-503 (AD712) labeled "buffer" one for fixed out (appears to be I/V converter), another for Headphone stage
prefenence: replace with THS-4032  (dual type, BJT but 2MegOhm Rin, +/- 16V, 100Mhz, 1.6nV noise, 60ns settling time, 100V/µS slew, 2MegOhm Input Impedance, drives 90mA output, -90dB THD 1Mhz)
fallback: replace with AD8599 (dual type BJT, +/- 18V, 17Mhz, 1.1nV noise, 2000ns settling time, 17V/µS Slew, -120dB THD 1kHz) ... settling time a tad high, but a typical "audio friendly" opamp

IC-461, IC-561 ... (NJM-2114D) Preamp/Driver-Stage for Headphone Out and variable Out
preference: replace with THS-4031/32 (drives 90A, BJT but 2MegOhm Rin, +/- 16V, 100Mhz, 1.6nV noise, 60ns settling time, 100V/µS slew, 2MegOhm Input Impedance, drives 90mA output, -90dB THD 1Mhz)
fallback: replace with AD8599 (dual type BJT, +/- 18V, 17Mhz, 1.1nV noise, 2000ns settling time, 17V/µS Slew, 50mA output current, -120dB THD 1kHz)
           
IC-460, IC-560 (NJM-2114D) LowPass Filter for Balanced Out
preference : replace 2x AD8599 per channel with one OPA1632 and change the filter configuration accordingly (OPA-1632 is a fully balanced output opamp)
                  OPA1632 (full balanced audio output opamp, 1.3nV nosie, 50V/µS Slew, +/-16V, 150Mhz BW, 75ns settling time, THD 0,0000022%)
fallback: replace with AD8599 (dual type BJT, +/- 18V, 17Mhz, 1.1nV noise, 2000ns settling time, 17V/µS Slew, 50mA output current, -120dB THD 1kHz)

any input is much appreciated !

best rgrds from Vienna/Austria
Crossy

[Crossy]

Okay,
yet another update and refinement

Cognizant of Per-Ander's assertion that replacement opamps should be audio-friendly and not prone to
oscillation, I figured it could be adviseable to use this "fully differential" new opamp OPA-1632 
also in place of the DE-Emphasis (IC501,IC401) and change the circuit design accordingly
(instead of having both hot and cold signals filtered by different chips, the OPA-1632 would handle
both simultaneously)

This is also a BJT opamp (like the THS-4031/4032) yet input impedance (Rin) is even higher,
this time - 34 MegOhm (instead of the 2 MegOhm of the THS-4031/4032). The filter circuit would have to be changed
somewhat since in place of two conventional single ended AD712 opamps, one fully differential OPA-1632 opamp would be present.

The OPA-1632 should be less prone to ringing and oscillation. According to specs
it should handle output currents of up to 150mA, even higher than the 90mA the THS-4031/32 can handle.

http://www.thevintageknob.org/SONY/sonyes/CDPXA7ES/CDPXA7ES.html
http://cdp101.hp.infoseek.co.jp/cdpxa7es/cdpxa7es.html
http://translate.google.com/translate?u=http%3A%2F%2Fcdp101.hp.infoseek.co.jp%2Fcdpxa7es%2Fcdpxa7es.html&langpair=ja%7Cen&hl=fr&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools

IC-504, IC-404 ... OPA-27GP ... DC Servo - doesn't seem to require any change

IC-401,IC-501 (AD-712) post DAC Filter "De-Emphasis" handles balanced audio out from DAC.... appears to require High input Impedance
    preference 1: replace 2x single ended AD-712 with 1x OPA1632 (balanced IO) and change filter layout accordingly (OPA-1632 can handle hot and cold phase simultaneously)
         OPA1632 (full balanced audio output opamp, 1.3nV nosie, 50V/µS Slew, +/-16V, 150Mhz BW, 75ns settling time, THD 0,0000022%, 34MegOhm Input impedance, drives 150mA output)   
        this idea probably makes best useof the balanced output from the DAC and processes all post-DAC filtering with fully differential components as long as possible !
        it would also upgrade the balanced signal path as settling time of the OPA-1632 is ultra-low compared to AD-712 and NJM-2114D and slew-rate is way higher
   preference 2: replace with THS-4032  (dual type, BJT but 2MegOhm Rin, +/- 16V, 100Mhz, 1.6nV noise, 60ns settling time, 100V/µS slew, 2MegOhm Input Impedance, drives 90mA output, -90dB THD 1Mhz)
        fallback: replace with AD8620 (dual type, slightly more noise, FET opamp, +/-13V max, 25Mhz, 6nV noise, 5fA noise, 600ns settling time, 50V/µS slew, THD 0,0006% 10kHz)

IC-402, IC-502 (AD-712) labeled "mix amp" (only 1 opamp of the dual AD-712 was used originally !) - makes single ended audio signal from balanced
       replace with OPA211 (SiGe BJT opamp, +/- 18V, max. 80Mhz, 1.1nV noise, 1pA noise, 400ns settling time, 27V/µS slew rate, THD -136dB 1kHz)
       it appears to me that a BJT opamp can be used here to make the balanced signal single ended. Single type opamp possible since the original used only one half od the dual AD712
       no need for a high-impedance opamp such a FET input opamp and the OPA211 should easily work here (not particulary demanding type wrt. oscilation)

IC-403,IC-503 (AD712) labeled "buffer" one for fixed out (appears to be I/V converter), another for Headphone stage
       prefenence: replace with THS-4032  (dual type, BJT but 2MegOhm Rin, +/- 16V, 100Mhz, 1.6nV noise, 60ns settling time, 100V/µS slew, 2MegOhm Input Impedance, drives 90mA output, -90dB THD 1Mhz)
       fallback: replace with AD8599 (dual type BJT, +/- 18V, 17Mhz, 1.1nV noise, 2000ns settling time, 17V/µS Slew, -120dB THD 1kHz)

IC-461, IC-561 (NJM-2114D) Preamp/Driver-Stage for Headphone Out and variable Out
       preference: replace with THS-4031/32 (drives 90A, BJT but 2MegOhm Rin, +/- 16V, 100Mhz, 1.6nV noise, 60ns settling time, 100V/µS slew, 2MegOhm Input Impedance, drives 90mA output, -90dB THD 1Mhz)
       (am besten wohl mit 2x 4031 mittels SMD/DIP Adapter um Übersprechen nochgeringer zu halten)
       fallback: replace with AD8599 (dual type BJT, +/- 18V, 17Mhz, 1.1nV noise, 2000ns settling time, 17V/µS Slew, 50mA output current, -120dB THD 1kHz)
           
IC-460, IC-560 (NJM-2114D) LowPass Filter for Balanced Out
     preference : replace 2x AD-712 per channel with one OPA1632 and change the filter configuration accordingly (OPA-1632 is a fully balanced output opamp)
                     OPA1632 (full balanced audio output opamp, 1.3nV nosie, 50V/µS Slew, +/-16V, 150Mhz BW, 75ns settling time, THD 0,0000022%)
    fallback: replace with AD8599 (dual type BJT, +/- 18V, 17Mhz, 1.1nV noise, 2000ns settling time, 17V/µS Slew, 50mA output current, -120dB THD 1kHz

The essence of these mods would be to upgrade all audio signal paths materially. (Voltage) Noise is reduced to 1/10 to 1/20 of the original configuration,
THD drops from -110dB to less than -125dB
Slew rate is increased  from 13 to at least 27 (50 in the balanced audio path)
settling time from 3000ns (original total signal path) to 800ns or less (400 in the balanced audio path)

best rgrds,
Crossy

[peranders]

You can edit your post, no time limit. Just a hint.....

[Crossy]

P-A,
thanx for the reminder ! Already altered and slightly corrected both posts !

Another goodie I found today browsing on recent opamp developments.

Bruno Putzeys is one of the key guys behind this odd recent TI/Burr-Brown full-differential audio amp, the OPA1632 - he wrote a lot of TI's application notes on fully differential opamps...
http://recforums.prosoundweb.com/index.php/mv/msg/24704/0/0/0/

He recently linked to this extensive bench-measurement comparison (>370pages !). Most common opamps are being compared for audio purposes there !

http://www.sg-acoustics.ch/analogue_audio/ic_opamps/pdf/opamp_distortion.pdf

best to you and all
CROSSY

[Crossy]

Yet another potentially interesting opamp using
the SiGe process... OPA-827 from TI/BB ... to me
looks like an improved OPA-627 replacement. In
Europe, AVNET appears to be stocked with the type..


OPA-827 : http://focus.ti.com/lit/ds/symlink/opa827.pdf

to be compared with the well-known highend OPA-627:

OPA-627 : http://focus.ti.com/lit/ds/symlink/opa627.pdf

the new OPA-827 seems to offer even lower voltage noise (4nV * sqrt(Hz)) compared
to the 4.5nV of the OPA-627. Settling time is unchanged, however slew rate is reduced
to 28V/µS from 55

CROSSY

[Crossy]

Another issue to consider is how far to go for modifying
the CD/DAC output stage. Just swapping some opamps and/or
discretes or going for an entire "plug and play" board level
I/V converter and/or filter replacement

Unfortunately, lacking the background of an electrical engineer,
I cannot do circuit designs myself, so I guess I will be bound
to "off-the shelf" PCB replacements and small-scale modifications
to them.

I do know for example of a IV and filter replacement board from
LC-Audio called the "Zapfilter" now in its 2nd edition. However
not even its manufacutrer, LC-Audio of Denmark is publishing
THD, slew rate and settlingtime equivalents of their circuit.

I guess what I would like to know or find out is whether op-amp
replacement to the initial circuits with currently available
state-of the art parts (OPA-827, OPA-627, AD-8620, OPA-211, THS-4031, AD797A etc.)
comes close to the ZAPfilter by itself.

this is a fundamental question for me as I got TWO not one
Cd-Player to mod (I have an RCD-971 from Rotel and the CDP-XA7 ES
from Sony). Actually I would love to preserver the "character" of the
respective player but upgrade the noise, THD and the like and
maybe also make the thing more "responsive" (slow filter circuit
don't sound that good). I'm not inclined against an "analytical"
character in any way, if that is the nature of the original machine.

If I do upgrade the DAC for example, it should be with a similar sounding
model with a similar technology. (replace multibit with multibit).
For example the RCD-971 uses one PCM-63 per channel.
There are upgrade replacements available that let you swap the
Burr-Brown PCM-63 with the quieter PCM-1704K
Both DACs are of the same type ("colinear" R2R, Multibit DAC)

I found "Monarchy Audio" on the web where one can purchase the
PCM-1704K on a PCM-63 compatible pinout socket for around $45 per unit
not sure whether I'm going to use them as I will have to look what my
RCD-971 uses originally : if the PCM-63 uses "K" Type chips from the
factory - allegedly some do but not all, the gain in quality should be
somewhat limited. Otherwise, it could be quite a leap.
(see the datasheets)

best rgrds
CROSSY

[Crossy]

Okay,
spent some time reading Walt Jung's articles on the subject and some chapters of the
Opamp Book (2002) available on his website.

For opamp selection, Jung's book emphasizes AC parameters, but states that
tight DC performance of the opamp (low bias current, low offset voltage) is
also an important factor.

Now low bias current would call for an opamp with FET input transistors,
whereas low offset voltage would call for a precision BJT opamp. Worth
noting is however that some DACs, for example the PCM63 in the Rotel RCD-971
however do have an offset adjustment pin on the DAC, which can be used
to improve these. I'm sure this method can lower offset voltage seen from the
IV opamp, not sure however whether the bias current that an imperfect opamp
always shows is lowered (improved) with this compensation pin in the DAC as well.

In an article on opamp audio, he mentions FETs to have a more linear
transconductance behaviour than BJTs - this would also call for an opamp
with FET input transistors.

Aside from this, Slew Rate seems to be important, greater than 20 V/µS
should be enough not to cause slew-rate introduced distortion in the
post-DAC IV and filter stages.

Now the original RCD-971 circuit uses FET opamps for both - IV converter
and Filter stage - namely the OPA-2604. New FET Opamps (AD8620, OPA627,
OPA827) of course are a big time improvement in terms of slew rate,
noise and lower offset voltages and bias currents - DC performance often
better by a factor of 10 !

However it's interesting that new low-noise BJT opamps such as the AD-797B
virtually MATCH the DC performance paramter where FET opamps used to excel
- i.e. low bias currents..

looks like the best opamp candidates here are the OPA827, OPA627, AD8620, AD797B
but still not sure which one to select for the IV stage. It might be and I get
the impression that these modern opamps used in an IV filter could match or even outperform
discrete designs such as the ZAPfilte but I'm not sureyet . Haven't found any measurements
or bench-tests (an "IV-shootout") here so far !

best rgrds
CROSSY