[D555]

Hello,

I have a D-EJ1000 and the consensus is that the sound is not particularly good. I would tend to agree especially after comparing it to the D-555. The D-EJ1000 is a "flagship" model so the hope was that this player would sound great but it doesn't, so why?

Here are some details that might help solve the puzzle:

***CAUTION: informational purposes only. ***

Devices are SMD, the circuit board is multilayer and can be easily damaged.
Any changes attempted as a result of the folowing information are at the modifier's own risk.

Circuit details and impressions:

The CX03029R DSP chip has two sets of outputs: one for line out and one for the headphone amp.

The line out is pretty simple:

A current limiting resistor and a 4.7 ufd tantalum coupling capacitor straight to output. Note: the block diagram for the CX03029R does show that the line out does utilize a Low Pass Filter internal to the chip. Apparently, this filter is not used for the headphone output.

The headphone out is a whole other matter:

The output from the DSP pass through a series of filters utiling 470pf, 3300pf, and 2220pf capacitors, which are then coupled with a 2.2uf tantalum capacitor to the TA2152FN headphone amp. (Not sure what the purpose of the series of filters is. The block diagram for the player does not indicate them. Strip out DSP noise or ultrasonics?) The TA2152FN chip is interesting. Data sheet:

http://www.semicon.toshiba.co.jp/td/..._datasheet.pdf

Have a look a page 5. It can be seen that that the TA2152FN has provision for EQ adjustment in the lower bass region at pins 5 and 8. Depending upon the compensation capacitors there can be a boost or a cut. According to the D-EJ1000 schematic and parts list, Sony has elected to use a metal chip resistor of value of 0 ohms, which, I think, translates into "no compensation". If that is the case, then the D-EJ1000 has a built in bass rolloff beginning at 200hz and becomming -7db at 20hz.

The output is then coupled with some 220uf 4v tantalum capacitors which passes through some 47uh inductors.

Not having a spectrum analyzer, I can't say what the effects of the DSP are. Does it compensate for this rolloff accurately? Do the BASS 1 and BASS 2 compensate for this roll off? Would BASS 1 or BASS 2 produce a spike at around 50hz if the rolloff is not compensated for accuratey?

What would the effects of the 47uh inductors be? I would think it would strip high frequencies and, possibly produce phase problems at the upper frequencies.

Also, from my expereiences with D-555 I'm not a huge fan of large value tantalum output capacitors.

There isn't much room inside the D-EJ1000 and as much as I would like to replace the 220uf output tantalums with some aluminium electrolyic types I can't find capacitors small enough to fit (my estimation 5.2mm long 3.2mm height) and there isn't any room to squeeze in standard leaded electroytic cans.

Modification:

Still, I did remove the 47uh inductors and replaced them with 2.2uh. It seems to have helped with the high notes a bit. Still a bit muffled but better.

More intriguing would be to remove some or all of the small pf capacitors that form the filter between the DSP and headphone amp. This modification has not been done yet. The capacitors are very small, and are located on both sides of the circuit board. It is very cramped there. It might be easier to rig a bypass.

Also intriguing is the headphone amp lower bass rolloff. Might try some .22uf chip capacitors and see what the results are. This area of the board relatively easier to get at.

Anyway, I hope this information has provided some insight into the D-EJ1000 sound. I wonder if the D-EJ2000 is similarly constructed?

Best regards,

Paul

 

[usc goose]

i barely understood half of what you just said.

but thanks.

 

[D555]

Hello usc goose,

Maybe I've increased the mystery!

It's all technical gobbly-gook but to simplify:

1) I suspect capcitors between the DSP chip and the headphone out chip filter out high frequncies.

2) The headpone out chip has a built in bass rolloff that I think Sony is utilizing.

3) Between the headphone out chip and the headphone jack are large value tantalum capacitors. I think these tantalum capacitors contribute to a "muddy sound". In general,
tantalum is probably not a good thing for audio sound. However, the Sony D-555 uses them (except for the output) extensively as I'm sure many/most Discman do -- even the vintage good ones. I'm beginning to think the problem with tantalum is most noticible with *large* values.

4) Between the headphone out chip and the headphone jack, in line with the tantalum capacitors are large value inductors (basically, a coil of wire) that I suspect are rolling off the high frequencies.

I think all of these items contribute to the mediocre sound of the D-EJ1000.

My initial modification was with item #4 and that was to significantly reduce the value of the inductor by replacing the original high value one with a much lower value one. In theory, the modification should improve the high frequency response. However, I think the effects of item #1 are still having a significant impact to the high frequencies.

Best regards,

Paul

 

[D-EJ915]

The most obvious reason sony would put a bass rolloff is that most crap-ass headphones can't handle any real bass at moderate to loud levels, so in order to keep them from being destroyed, they simply removed the possibility of that happening, if they didn't have it I think we'd have a lot less people with streetstyles and other cheap things around

 

[Whitebread]

Hi, please mod IMP-400, kthxbye

 

[puppyslugg]

D555,

Have you tried bypassing the tantalum caps at the line/headphone out with a pp?

 

[D555]

Hello,

Thanks for the suggestion.

No room inside the D-EJ1000 for almost anything. Again, the original tantalum caps are about 5mm x 3mm and are almost right up against each other. Maybe 1mm or 2mm separates them. Leaded polypropylenes and other film types are way too large. There are some SMD polypropylenes that are quite small but I'm not sure that they will fit for even bypassing purposes. They certainly won't fit on top of the tantalums, maybe the sides have enough space.

I've been looking for some SMD aluminum electrolytic chips that are the same size of the tantalums but unfortunately the ones I've seen are *just* a bit too large.

I cant stress enough how tight the space is inside the D-EJ1000 is.

I'm probably going to focus on the filter caps between the DSP and the headphone amp -- that's where I suspect much of the high frequency rolloff is. The mod doesn't require any special parts, just removing caps already present. The issue there is going to be the process in removing them very carefully. It is extremely tight quarters where these caps are located. There might be an issue about introducing DSP noise or hiss once the filters are removed.

The bass mod for the headphone amp shouldn't be too bad, there's enough room for a bit of expermentation with values.

Best regards,

Paul

 

[D555]

After thinking about this some more, I am starting to form a picture of what is going on and more into why the the D-EJ1000 sounds mediocre at the headphone out.

The DSP chip has two outs: a line out and a headphone out. In the DSP is what is labeled in the block diagram as an LPF for the line out. Does the D-EJ1000 utilize oversampling? Is 44.1 or 88.2 being filtered out by the LPF. The headphone out shows nothing in the way of an LPF.

My opinion of what is being done is that the line out of the DSP is using a LPF, while the headphone out has no filter and that downstream a crude "brick wall" filter is being created by the capacitors between the DSP chip and the headphone amp. The 47uh inductor may play a role in this as well.

All told there is about 5970pf of filtering on each channel. To me, that would result in major high frequency loss. Removing the capacitors completely would probably result in harsh sound, because of distortion as the 44.1 (or other frequencies) mix with the audio. There is also a possibility of frying headphones and the headphone coupling inductors with the high frequency energy.

My guess is that "carefully* tweaking the values for the capacitors between the DSP and headphone amp would restore much of the high end "sparkle" and yet still provide protection from high frequency energy and keep distortion low.

A more invasive solution might be to tap off the filtered DSP line out and route to the headphone amp.

Anyway, I think the above explains why the D-EJ1000 doesn't sound particularly good. It might also explain what is going on with all recent Sony Discman players.

Best regards,

Paul

 

[aos]

Quote:

4.7 ufd tantalum coupling capacitor

You don't need to look anywhere else for the cause of bad sound. Are you sure those caps are tantalum? It is hard to believe that anyone would really use them for signal coupling. You really don't need to look anywhere else before you fix that (if that's your intention). Tantalum will give you more capacitance per volume though so that is probably Sony's justification, but the sound is hurt so badly that it's hard to imagine anyone would leave it in after hearing it... responsible Sony engineer must've been loaded with sake at the time.

 

[D555]

Hello AOS,

Yes, the lineout coupling capacitors are specified as 4.7uf tantalum in the D-EJ1000 parts list. The headphone coupling coupling capacitors are specified as 220uf tantalum. Visual inspection would lead me to believe that these capacitors are truly tantalum.

Except for the headphone out coupling and in the power supply the excellent sounding D-555 makes extensive use of tantalum for coupling (even in the line out).

As a result, I'm beginning to think: large value tantalum for coupling = bad.

I would like to replace the D-EJ1000 headphone coupling capacitors with something better -- there just isn't any room. Initially, I think I'm going to focus on the capacitors that are between the DSP and headphone amp. I strongly suspect that most of the "highs" are being lopped off there.

Paul

 

[PeterR]

Could you perhaps post the schematic?

 

[D555]

Hello,

Amateur attempt to explain what is happening:

DAC's for CD can operate at, for example, 44.1kc, 88.2kc, 176.4kc: 2x, 4x, 8x oversampling, respectively. Operating at these frequencies produces measurable output at these frequencies and their harmonic frequencies. For example a CD player DAC opeating at 44.1kc would produce output at 44.1kc (the fundamental) and at 88.2kc, 176.4kc, etc (the harmonics). For a player operating at 44.1kc the output of the fundamental frequency and the harmonics need to be reduced or eliminated by filters. If not then noise, distortion, and other problems will occur. If the player utilizes 44.1kc then the filters must operate close to audible frequencies thus running the risk of damaging the sound. Oversampling techniques utiling 88.1kc (4x) or 176.4kc (8x) as the fundamental would reduce the need for a filter to operate near audible frequencies. That's part of the benefit of oversampling.

What does the D-J1000 use? Don't know but if aggressive filtering needs to be used at or near audio frequencies, I'm tempted to say it may be 44.1kc. If so, then the frequencies of concern would begin with 22kc (upper range of audio for most people) then 44kc, 88kc, 132kc, 176kc, etc. The output at these frequncies would need to be reduced/removed. These days, well crafted digital filters would remove this noise. It appears that Sony has rolled back the clock and has elected to utilize some sort of analog "brick wall"-type for the DAC/DSP output to the headphone amp. I think in the D-EJ1000 the "brick wall" is a poor choice, perhaps implemented poorly resulting in mediocre sound.

However, if the D-EJ1000 utilizes high oversampling then that would be good news. That means less aggressive filtering should possible. (Does oversampling eat up batteries?)

To pick the D-EJ1000 "brick wall" apart:

Stage 1: The 470pf would provide a gentle rolloff.
Stage 2: The 3300pf would provide a rather sharp dropoff.
Stage 3: The 2200pf would provide somewhat less of a dropoff.
There are resistors between each stage so the rolloff is affected (increased) by those components as well (not looking at the schematic at this time so can't state the values). The affect of all of these stages would be cumulative.

The 470pf is probably for super ultrasonic noise. Probably should be left in, relatively negliable effect on audio, I suspect, and would help prevent circuits and headphones from getting fried by
incredibly high frequencies. Would help combat amplifier instability as well.

The 3300pf probably combats the 44kc freq aggressively as well as other frequncies.

The 2200pf probably suppliments the 44kc removal and higher
frequencies.

Adding a filter or attempting to design all-new circuitry would be out of the question in the D-EJ1000 as there is virtually no room and I would say the circuit board essentially un-hackable. What remains is to experiment with the capacitor values.

The experiment would involve:

1) Removing either the 3300pf or the 2200pf.
2) Reducing the values of the capacitors in an attempt to reshape the upper frequencies.

Trade offs would be:

1) Risk amplifier instability
2) Risk frying headphones/inductor/amp
3) Increase noise and distortion

My guess is the 3300pf is the main culprit of the rolled off highs.

I'll plunge in and let all interested read the results.

Peter, I may scan the schematic of the affected part of the audio and post.

Paul

 

[D555]

Hello,

Just about everything you want to know about the D-EJ1000 inner workings.

This is a BIG 201 page .pdf spec sheet of the CXD3029R which is the main DSP:
http://www.sony.co.jp/~semicon/engli...1/a6802964.pdf

Note the provisions for 4 step bass boost and treble boost, compression and other features.

Note: No LPF for the headphone out. The spec sheets shows a circuit similar to what is used for the D-EJ1000.

Paul

 

[D555]

Hello,

C105 and C205, both 3300pf were removed from the D-EJ1000. The process of removal is not for shaky handed. Extremely fine tipped low-temperature soldering iron is needed. The capacitors are about the size of a *small* grain of sand.

The results:

In a word, SPECTACULAR. Removing these two capacitors really opens up the high end of the D-EJ1000. Cymbals and tamborines sound realistic. Lots of low level high frequency detail is now heard. The sound is no longer muddy. "No EQ" and "BASS 1" settings have been tried
("BASS 2" is just too extreme for me. It boosts both bass and treble too much, in my view). No weird noises or additional hiss noted. I am doing a lot of comparison with the D-555 and am listening for any added distortion. Long term usage is going to determine if the modification will cause any failure.

In any event, the initial indication is that the modification works exceedingly well.

To sum up the D-EJ1000 modifications thus far:

L201 47uh ---> 2.2uh
L301 47uh ---> 2.2uh
C105 3300pf ---> removed
C205 3300pf ---> removed

Best regards,

Paul

 

[puppyslugg]

Way to go, Paul!!