Toshiba SD-3990: audio out bypass -- help needed with mods

[PeAK]

Quote:

Originally Posted by headfone Looking fwd to your 'scope-based analysis. I wasn't that impressed with the Auricaps tweak. Mechanical damping was worth the time/effort...

Stock Impressions:

• Sins of omission but fairly listenable (4588 limitations?)
• simplified menu versus 3950
• Operational differences from "paused" state for track selection
• access tracks directly above track 9 from remote do not use 10+ key
• Quieter than 3950 drive
• Play on Load

It is a bit early to really say much as my impression from the first hour of listening differered towards the end. Part of this was due to changes in termination to unused outputs(co-axial and video) and to different footers. Swapping things back, I could not get back the "1st hour" sound. Putting back in the modded 3950 was a not "night and day" but a welcome improvement. The 3990 probably has potential.

At the start, I would say that it sounded like it was voiced more for long term listening (relaxed treble), slight loss of detail, fairly dynamic. At the end, it became more shrill and less relaxed (???). Still early but definitely better than the pocket money players from Wal-mart in terms of picture quality and sound.

Update Nov 25:

1. Changes from last:
2. Terminatons on unused co-axial and composite output
3. Rollerballs changed to pop-can based ones
4. Burn-in

The unit spins down the transport when tracks on done but keeps the output board and display powered on. The unit does bring a level of satifaction that caught me off guard in its stock form. Upgrading the op-amp should be a revelation.

PeAK

 

[mb3k]

I ran out and purchased a 3990, here's a photo of the psu

Click for full size.
Is there any 3990 mods for the psu regime?

 

[Nerull]

I actually was thinking about making a totally new power supply for the thing. It needs two +5V power supplies and a +12/-12V supply. Get two TREADs (or STEPS, if you prefer) and two +5V power supplies (jung regulators or whatnot) and fit them in there.

That should make quite a difference

.

~Tom

 

[PeAK]

Quote:

Originally Posted by headfone ...I'm still a bit in the dark about the resistors and caps *between* the DAC and the opamp in the *stock* Toshiba...I *think* the resistors here may be considered "load resistors" . . .A million q's -- I know! Just want my mod to be right the first time thru

Headfone...I put up some more info up at my 3990 page. The output is DC coupled (from the differntial DAC output) to the RCA output via a differential-to-single ended op-amp gain stage.

cheers,
PeAK

 

[headfone]

Quote:

Originally Posted by Nerull /img/forum/go_quote.gif I actually was thinking about making a totally new power supply for the thing. It needs two +5V power supplies and a +12/-12V supply. Get two TREADs (or STEPS, if you prefer) and two +5V power supplies (jung regulators or whatnot) and fit them in there. That should make quite a difference

Probably. But not as much as true, linear DC

I.e. an all-battery sol'n. One of the modders for the 3950 ultimately ended up building a lead-acid, battery-based supply. Did the same for his DAC. He noted the changes were so dramatic that going back to an AC-based source is now unthinkable.

 

[headfone]

Quote:

Originally Posted by PeAK /img/forum/go_quote.gif [From PeAK's notes...] Upgrading the op-amp (similar to Vinnie's mod) is one possible path. One recomendation would be the LT1361 in the appropriate S8 surface mount package.

I *think* I may have noted this earlier in this thread but ReferenceAuddioMods.com seems to be using the BUF634 op-amp for their Toshiba and other upgrades. So this op-amp may be another possibility. As noted earlier, I've got a couple of samplers (op-amps) on order. As soon as they arrive *and* I free up some time, I'll experiment and report back.

 

[headfone]

Quote:

Originally Posted by PeAK /img/forum/go_quote.gif Unlike the 3950/3960 generation, the output from the Toshiba is differential and not single ended. This means that a Swenson type passive mode is not possible...unless you happen to have Balanced inputs on your pre-amp, in which case you may implement a balanced version of the Swenson mod.

The Swenson mod in and of itself may be out but there is still potential for the 3990. Dual differnetial outputs on the DAC, as is the case with the Zoran in the 3990, may be used to send signals into a coupling transformer, which cancels out the DC internally and, hence, no coupling cap is probably necessary. An op-amp, like the aforemntioned BUF634, is probab. needed after the xformer to add a little gain.

As far as the coupling caps go, I'm told ones made by audio-consulting.ch are ideal. But they are very $. Not sure whether cheaper audio xformers may be used, such as this Xicon avail from Mouser.

Is this appraoch similar to the Sallen and Key/Art DI/O you (PeAk) were referring to?

 

[PeAK]

Quote:

Originally Posted by headfone /img/forum/go_quote.gif ...An op-amp, like the aforemntioned BUF634, is probab. needed after the xformer to add a little gain. Is this appraoch similar to the Sallen and Key/Art DI/O you (PeAk) were referring to?

The DI/O (or SK approach) allows the damping factor to be altered for sharper cutoff. It will also modified the compensation and possibly the voicing of the output stage....more fun...you bet!

cheers,
PeAK

 

[PeAK]

Quote:

Originally Posted by headfone /img/forum/go_quote.gif I did remove Q317 and Q318 today (see image in last msg) which yielded a modest but noticible improvement in sound fidelity (a bit more "focus", perhaps).

One note of caution is that with the muting/shorting transistors removed, some nasty DC levels and signals (DTS) could come out of the RCA jacks. For the DC levels, this can happen at powerup as the rails powerup at different rates for a design dependent upon complimentary (i.e. +/-11V) supplies to realize "zero" output offset.

In the case of an AC coupled output, the series cap limits the maximum level and slew (hf content) of any power related glitches.

With the muting circuit in, the transistors short to ground and limit the level at the RCA L/R when enabled during powerup. This is just one thing to think about/measure before removing the muting transistors. In the past, these transistors could be quite large but the latest SMD should have a very small amount of non-linear capacitance and less effect than in the earlier days (late 80s) when the mod was first suggested by Walt Jung.

PeAK

 

[headfone]

Quote:

Originally Posted by Nerull /img/forum/go_quote.gif I actually was thinking about making a totally new power supply for the thing. It needs two +5V power supplies and a +12/-12V supply. Get two TREADs (or STEPS, if you prefer) and two +5V power supplies (jung regulators or whatnot) and fit them in there. That should make quite a difference .

I like the Tangent power supplies myself. Not sure how one would integrate new supplies with Toshiba's power-control system, though (on/off, auto-off, etc.).

Speaking of power supplies, I did finally replace the first four rectifier diodes w/Schottky's (see mb3k's image of the pwr supply board from earlier in this thread).

Question:

I have a high-quality Nichicon 250V 470uF cap. Can I safely use this to replace the C902 which is 250V 33uF (see mb3k's image again). Reason I ask is that a modder of a similar Toshiba player noted that extra capacitance may mean more "surge current" thru the switching power supply. While the new Schottky's are 4A (the original 1N4007's were 1A), I'm not sure whether they nor any other part of the PS circuit would either benefit or become a safety risk due to the extra capacitance.

BTW: some activity at Headwize on this player.