advertisement


Sony 337ESD mods.

Can you have too much of a good thing?

I have to admit that I'm a bit disappointed with the 337 at the moment. The last few mods have added more and more detail and attack to the point where I'm finding the sound the sound has become a little too 'hard' and bright for my liking. :(

Looking back I was over the moon with how it sounded in the spec I had back in post #16. After that I added a bunch more of the Charcroft CAR Z-foil resistors at the output and I suspect that may be where my problems are. The Charcroft resistors have an amazing 'no loss' sort of sound and whilst having a few in there added a pleasing degree of 'air' and detail I suspect switching to a full compliment of them may have made the sound overly bright.

Of course I have also changed the output coupling caps a few times since then too but I wouldn't expect the high quality Ansar polypropylene caps that I'm currently using to be harsher than the cheap yellow polyesters which I had in before -you'd expect quite the opposite.

I've also been rolling different op-amps recently to see if I can find the balance that I'm looking for. The op amps I've tried are:

JRC5532D
LM4562
OPA2604
Muses02
OPA2107

All are good but my current favourites are still the OPA2604 followed very closely by the OPA2107. The Muses02 have an incredibly clear top end which I like but their slightly more forward presentation exacerbates the over brightness which I'm trying to solve right now. The laid back nature of the OPA2604 is therefore my favourite right now.

Has anyone else here tried z-foil resistors? What did you think and how did they affect the sound?

When I have a spare couple of hours I need to resolve myself to swapping back to the Takman resistors or the original carbon films to see if my hunch is right.

I've also blown some more cash on another expensive upgrade in the hope that throwing more money at it might help! I'll reveal all soon.
 
Good morning folks.

Although it has been a long time since I last updated this thread I've actually been very busy on the 337ESD, I just haven't got around to posting about it until now.

So here we go, I'll bring you up to speed:

First, I'll cover the subject of Op-amps - As you'll read in my earlier posts I've tried quite a few different op-amps in the 337. The standard fitment are JRC5532D which sound surprisingly good despite their lowly reputation. I've also tried OPA2604 (one of my favourites), LM4562, Muses 02 and OPA2107.

Since my last post I also tried the latest V5 version of the Burson Audio discrete op-amp and although the sound was promising I had a problem with very clearly audible distortion in quite passages of music which I couldn't solve despite exchanging emails with the technical department at Burson regarding the problem.



Next I wanted to try OPA627 as I was very happy with these in my old Arcam Alpha 5 project. But....OPA627 are of course singles and I need doubles, so I need 2x OPA627 per channel in adaptors. I chose 2x OPA627AU which are the S08 package rather than DIP8 as they are much smaller meaning that they can be mounted in a very compact way either side of a small converter PCB. OPA627 are very wide bandwidth devices which need high quality tight local decoupling with minimal lead inductance from the supply pins to the ground plane. I therefore believe that this 2x SO8 rather than 2x DIP8 approach is far far better. To keep lead/trace inductance to a minimum I soldered them directly rather than fitting them in sockets as you'll so often see.



For local decoupling around the op-amps I used 10nF Wima MKP from each supply pin to ground taking care to trim the leads as short as practically possible.



I'm very happy with this op-amp choice and they'll be staying for the foreseeable future.
 
Back in post #41 I was having some trouble with a slightly over-bright presentation which was proving tricky to track down and rectify. Back then I had a hunch that the Z-foil resistors in the output where to blame. In the end the Z-foil resistors turned out to be a red herring and I'm delighted to be able to say that any hint of over-brightness has been completely abolished.

Sometimes on these projects you need to backtrack and one of the modifications which (perhaps surprisingly) I had to reverse was to remove the small Oscon caps which I'd fitted between the -5v and -15v supplies of the TDA1541A DAC chips. On my old Arcam Alpha 5 project fitted a 15uF Oscon here was a substantial improvement and so I'd automatically done the same on my 337. To begin with I was using 15uF SC series and then later my friend Phil gave me a couple of 10uF SP to try (see below).



Tying the -15v and -5v supplies together certainly changes the sound and is certainly worth trying but for me on the 337ESD it added a etched/grainy/sterile aspect to the sound which I didn't quite get on with a prefer the player without.
 
Next up I'd like to cover the subject of the power supply decoupling capacitors around the TDA1541. The first thing to say is that these capacitors make an absolutely enormous difference and changing from one variety of capacitor to another can profoundly alter the sound character of the player. I know some sceptics will be rolling their eyes reading this but honestly the differences can be remarkable.

The TDA1541 is a tricky and unusual animal and the +5v and -5v supplies appear are apparently BOTH analogue AND digital which I've a hunch is what makes the best decoupling cap choice so tricky.

For the -5v and +5v supply decoupling caps I've tried a number of different types, values and combinations including:

220uF 25v Elna Cerafine on both +5v and -5v
100uF Elna Durorex on both
330uF Elna Cerafine on both
470uF Cerafine on +5v and 470uF Panasonic FC on -5v
470uF Oscon SEPC on both
56uF Oscon SP on both
180uF Oscon SP on both
Elna Silmic 220uF 16v on both

I've also tried the majority of the above with and without 47nF SMT PPS film bypasses!!

The conclusion that I've come to is that my favourites are definitely the Silmics which to my ears sound the most natural and organic of the lot and will be staying.

The 470uF were very good in the bass and the Oscon SP get an honourable mention for being very 'crisp' and detailed. However whenever I try Oscons in this position, whilst at first I think 'Wow' they always leave me feeling cold after a while as they seem to be slightly grainy in the high frequencies with a rather sterile/artificial quality to the sound. Of course as ever your own personal preference and your system may be different to mine so please feel free to give them a try.

I also found that I prefer the sound without any film bypass caps.

So there's a theme developing here: Oscons are superb capacitors and while they remain a firm favourite of mine for digital duties whenever I try them in analogue circuits they just don't quite sound 'right' to me.
 
OK, so now we're up to speed with my endless capacitor and op-amp rolling I can tell you about something altogether more exciting!

You may remember that earlier I had a dedicated power supply fitted for the CXD1144 digital filter chip and associated IC's but that when I fitted the Tentlabs clock upgrade this was removed in order to make room for the clock PSU.

Well, I've now managed to squeeze BOTH the clock PSU AND the dedicated PSU for the CXD1144 etc inside and all can say is "WOW!".

Here's the new installation:



With both of these power supplies fitted at the same time it's unbelievable good, the soundstage is huge and deliciously detailed without any sense of artifice. Quite simply amazing. It reminds me of when I fitted a dedicated supply to the SAA7220 in my Arcam.

The PSU consists of a 7VA 9v toroid, Schottky 11DQ10 rectifier diodes, 3300uF Rubycon ZLH smoothing cap and an LM317 reg with 100uF bypass cap on the adjust pin with the output set at about 11vdc. This then cascades down to a +5v Spower discrete reg on the main PCB near the CXD1144 which is followed by a 1500uF Panasonic FM and then Oscons and then 47nF ceramic caps right on the pins at each IC (under the PCB).



I've also fitted an improved 'mk2' version of aluminium shelf which I made to house the Tentlabs clock. The new part extends out further brining the clock much closer to the filter chip etc thereby making the wiring from the clock output considerably shorter. I've also been messing about with different wiring here and I've settled on a twisted pair of solid core wires for this (silver plated copper core with Teflon jacket no less).

 
I did some more 'lily gilding' yestrday. Here is the new Fidelity Audio +5v Spower super reg which replaces a 7805 in the servo section. Not quite a drop-in replacements as I had to fit longer pins and then also swap the heatsink to a taller one so that there was enough contact area between the reg and the heatsink.
 
Hi Mike,

What an excellent thread, very informative, good pics and is a credit to the hard work you have done.

I wanted to ask you about the brightness you encountered (and solved). My cdp is different of course (cd3.5) but you may be able to help me.

My problem is excess sibilance on vocals. I thought it was oscillation between anologue stage alwsr and opa627 but that was not the only cause as i tried an avondale tpr4 and some sibilance still exists.

My mods have cumulated in a very detailed sound which may be too much and revealing the limits of the tda1305 dac chip (not highly regarded) or it may be that i am hearing too much of the recording !

Having clamped down the alwsr i feel something i did to my dac is causing the sibilance. I too have oscons decoupling the dac and did this on acoustica's excellent advice. Do you think i should maybe change these out first - prime candidate for sibilance ?

Other things i did was give all 3 dac supply pins their own transformer winding, replaced dac anologue supply lm317 reg with a pfm flea and the other two supply pins got lm317 acoustica treatment and a gyrator proceeding them. Naim have small series resistors between regs and dac pins. I removed the series resistor on the anologue supply when i fitted the flea regulator - maybe i should refit a 1 ohm resistor to see if there is a problem with flea output and dac oscon decouplers.


Thanks, stu
 
Hi Stu, these things can drive you crazy, you have my sympathy!

I've got a few suggestions based on my personal experience:

1) As a general rule I don't like Oscons for anything analogue as they seem to give an overly detailed/bright/forward sound. This would apply to opamp decoupling for example. I don't know much about the TDA1305 but the TDA1541 is a strange beast as the three supplies +5v, -5v and -15v operate in both the analogue and digital domains. I've tried various Oscons for decoupling the TDA1541 power supplies and similarly found that they bring the sound more detail and brightness in a way that was initially impressive but becomes tiresome in the long-term. I'm now using Silmics for decoupling on the opamps and DAC chips and the sound smoother and much less fatiguing. Oscons are of course excellent for digital decoupling.

2) I've also found that small value film bypasses caps can be a source of sibilance. If you've been added film bypasses anywhere then give these some consideration. Try removing them. For example on my 337 I preferred the sound without film bypasses to the DAC decoupling.

3) I've sometimes found that decoupling opamps rail to rail can sound bright/forward/harsh compared to decoupling rail to ground. Sometimes rail to rail seems to work and sometimes I don't like it. However, I pretty much always seem to find rail to ground OK.

4) Try 10nF Wima FKP (rail to ground) on the opa627's instead of 100nF. Be sure to get the lead length (inductance) to a minimum.
 
Thanks Mike for taking the time to write that very helpful reply.

The cd3.5 has single rail power supplies throughout (positive voltage only) so there is only rail to ground decoupling.

Naim have fitted 0.1uf wima film caps on all the important supplies. The opamps have the wima's 5mm from the opamps supply pin and no other decoupling. The dac has 2 anologue supply pins and one digital. Naim fitted the 0.1uf wima's and 10uf tants in parallel here. The acoustica dac decoupling mods involved removing the tants and replacing with oscons. On one of the pins the wima was removed as well.

Just for clarification Mike, you only use a single silmic for decoupling at the opamps and dac supply pins ? If so what capacitance?

I will try removing the oscons from the dac anologue supply pin and try a silmic. Pity they take so long to burn in. I will assess the situation doing one change at a time until i am happy

Many thanks, Stu
 
mega lord said:
Just for clarification Mike, you only use a single silmic for decoupling at the opamps and dac supply pins ? If so what capacitance?

I will try removing the oscons from the dac anologue supply pin and try a silmic. Many thanks, Stu
Click to expand...

Hi Stu,

The opamps in my 337Esd are 2x OPA627AU on browndog adaptors to make them duals. I have 10nF Wima FKP rail to ground right at the opamps. A couple of cm away there are two 100uF Silmics (+15v to ground and -15v to ground) and then a further 10cm away there are 1000uF Silmics after the regs (Fidelity audio Spower). So in summary the decoupling for the opamps is: 1000uF Silmic ----->100uF Silmic----->10nF Wima

The TDA1541A DAC chips require three supply voltages: +5v, -5v and -15v. The -15v is derived from the same -15v reg that feeds the opamps their -15v supply so there's the 1000uF Silmic after the reg (as mentioned before) and then a 100uF Silmic near the DAC. So the -15v is: 1000uF Silmic----->100uF Silmic

The DAC chips each get their own reg for both the +5v and -5v supplies (4 regs in total). These are again Spower regs and there is a just a 220uF Silmic decoupling cap after each reg close to the DACs supply pins.

Trying something other than Oscons on the DAC analogue supply pins seems like a sensible idea and easily reversed if necessary.

I recommend getting yourself a selection of NPO/COG caps to play with too, they're brilliant for local high speed decoupling and have no down sides as far as I can tell.
 
Thanks Mike for the top tips.

I took all the oscons out today (4 of them). I am going to put a couple back but of higher capacitance and replace the anologue supply oscon with a silmic. Then refit a series resistor between flea and dac anologue supply and see where that gets me.

If there is still too much top end i might play with the anologue output stage signal path resistors. Naim have fitted their usual metal film resistors here but i may be able to find something a little warmer sounding. TBH, i don't know which would be suitable and would fit on the board

Stu
 
Mike P said:
HiFi Collective stock lots of different types. The Takman carbon films are very reasonably priced and ought to be a shade warmer than most metal film types.
Click to expand...

What physical property causes the "warmth" of those carbon film resistors?
 
Carbon resistors have a very high temperature coefficient (5,000 ppm / C); at low frequencies this causes measurable low order harmonic distortion. In single rail circuits this will be second harmonic, in symmetric dual rail circuits probably dominantly third.
 
PigletsDad said:
Carbon resistors have a very high temperature coefficient (5,000 ppm / C); at low frequencies this causes measurable low order harmonic distortion. In single rail circuits this will be second harmonic, in symmetric dual rail circuits probably dominantly third.
Click to expand...

Measurable yes. But audible?
 
You must log in or register to reply here.


advertisement


Back
Top