advertisement


It's Taylor DAC v2.0

...according to the articles posted by Martin I'm listening to a noise generator (lead-acid gel battery, mentioned in the article by Werner) powering a heavy distortion device (NOS 1543 DAC, not mentioned in the article).

Uh... er... I still like the sound. :)

But I'll consider a well-regulated mains based PSU in the future if I get fed up with the recharging/listening cycles.
 
Perhaps predictably, I've gone back to the 220uf OSCON SG coupling cap.

Less edge, more middle, shakes the room.
 
Recent investigations that the quadruplet is the best decoupling "cap" by far, and that an inductor after the LM7808 is essential.

I'm using a 56uH Epcos, of the whopping great 1.2A type I ordered by accident. Don't know if this is sonically better/worse than the smaller 200mA I was using, but they're hugely less fragile.

With the quadruplet these give an explosive, forward sound with a pleasingly thick tone. Compare this to a single 4.7uf SMD X7R cap and no inductor, which gives a veilded and processed.

The latter does have the advantage of being softer on the ears though.
 
I now have minimalist antivibrationaries to match the minimalist electronics.

The dac now sits on top of the preamp supported by four inverted plastic PCB pillars, from Farnell. Being light and stiff they should allow vibrations to escape easily. The tip has a "nipple" about 1mm across, making the actual contact area very small. The base is hollow, so a generous quantity of glue can be poured in to make a good contact with the dac board.

I've also swapped the loopfilter resistor for Welwyn RC55 100R, the mathematic match for a 4.7uf cap.

I sinned and implemented both changes at the same time.

It sounded peculiar in the extreme to start with, gritty, with bass reduced to a tuneless subsonic murmur and no imaging to speak of.

Now it sounds rather splendid. Timing is the main improvement, with each instument ploughing a distinct and independent rhythmic furrow, rather than sounding glued together in a blob. Tonality is drier rather than gooier.
 
Now each LM78XX has a 56uH inductor before it.

Sounded a bit leaden and tuneless at first, but a few hours later the sound was a more solid and less distorted. It's subjectively less dynamic at the same volume, but it sounds more convincing at higher volumes.

I think the prior sound was cooking up a mirage of high dynamics at low volumes through aberrations.
 
Nice to see your still working on the TDA1543 Kit. it seems your learning a fair bit;)
I'll be posting some pictures of my new TDA1541 dac soon once its finished, thanks again for those CS8412 chips

Ron
 
On pin 20, I've changed the cap to ground before the 100R resistor from a 10nf polyprop to one of my 4.7uf 1206 ceramics.

More sparkle and low level detail (mouth piece and breathing on saxophone, etc), tape hiss more emphasised.

I've ordered some 1206 resistors which should allow me to make the loopfilter incredibly short, especially if I cut the legs off the CS8412 and just solder to the little "shelf" at the top.

I was planning on trying these for I/V conversion too. As they're very small and 01.25w, are there heat issues I should worry about?

TIA
 
Two new mods to report, 56uH 200mA-odd Meggitt inductors on each CS8412 pin and Multicomp 3k0 1% 0.125w 1206 thick film I/V resistors with 1k5 of the same as Vref.

The former dredged up more sludge, made things warmer overall and the upper mid a bit harder.

The new resistors sound hotter and more explosive, with that quality of being able to turn the volume right up without noticing just how loud things are. Loads of tape hiss, and the instruments sound bigger and nearer the listening chair.

Together the mods make quite an afterburner!
 
For your pleasure:

Loopfilter is 100R Welwyn RC55 & 4.7uf 1206 Y5V
Same cap across pin 20 & 21
Same cap decoupling each CS8412 supply pin
Same cap bypassing input and output of each LM78XX
56uH 1.5A Multicomp inductor before each reg
Same after TDA1543 reg
56uH 200mA-odd Meggitt inductor on each CS8412 pin
Quadruplet bypassing TDA1543
220uf Sanyo SG output coupling cap
75R Welwyn RC55 on CS8412 pins 11, 12 & 26
I/V is 3K0 Multicoomp 1% 0.125w thick film 1206 resistor
Vref is 1k5 of same
Four 1/2" PCB pillars as feet
 
The coupling caps are now 4.7uf 1206 Y5V, soldered right next to the TDA1543 body.

Good sound, smooth, thick and hard, quite cool and a little on the thin side.

They did sound a bit ringy (wonder what causes this?) when I first tried them though, and this became quite a problem. As the sound had already become glassier with a 56uH inductor on the CS8412 analogue supply pin, this was ditched and the problem eased.

Sound is incisive and brilliant with being too brutal, but still needs a bit more tweaking.

The old LM7805 got a bit mangled in desoldering. I twas replaced with the "AT" version of the same reg, which is apparently a superior grade part.

Oh, and the loopfilter is now all-SMD.
 
Still no digital camera :(, but I've some more reports you might be interested in.

At the moment I'm using a 75R 1206/10uf 35v Rubycon ZA as loopfilter. The dac takes a second to lock on to the transport, but it seems to have gained a bit of oomph. Ditto a 4.7uf ZA on pin 20. I've ditched the 75R resistors on pins 1 & 2 of the 1543.

Everything earths to the same ground plane. Separate DIG and ANAL for a cleaner sound and combine for a beefier one.

All LM317 as regs makes the sound airier and easier, with 1206 SMD resistors and ADJ and Vout bypassed with 47uf 50v Panasonic NHG as per Avondale TPR.

5.6uH is before and after each reg.

The 1543 supply is currently LM7808 > 5.6uH > 1543 > 4.7uf X7R 1206/35v 3300uf Panasonic FC to ground.

In a fit of puritan radicalism I decided that supports don't matter. But the sound was too "ringing" in the upper mid and not lush enough. A square of foamed butyl rubber (from zanash - please check your PM) under each bit of kit did the business. The foam looks like the same material used to make camping matts.

Note that as the DAC has no case, it sits on a 30cm square anti static bag to stop it getting zapped.

I bought one of those hot glue guns, which has proved extremly useful for securing leads and components.

Oh, and the coupling caps are good old 220uf 10v Sanyo SG.

Y5V 1206 ceramics? Puh-leeze.
 
Does anyone else fancy trying the SPDIF buffer chip ron mentioned? I was going to ask Oliver if he'd order some on my behalf, so may get all the orderes at once.
 
Here's a little tip I picked up from God knows where.

Use SCSI ribbon cable for P2P wiring and signal earths. It's thin, very flexable and I believe has low resistance.

Does anyone know how much current it can carry?

TIA
 
Kit - it shuld be good for 100ma, more than enough anyway. (SCSI used 5v lines and passive terminators were 50ohms IIRC).

INteresting stuff with teh DAc. How many iterations have you been though now? are you still using the self-same chips as at the start of the thread? I'd have killed a few by now!
 
Martin

I'd say 10 generations at least, given that I've started from scratch a couple of times. The CS8412 is amazingly resilient, the of heat it's had I'd have assumed it'd be toast by now, but no, it keeps on trucking.

I've binned loads of TDA1543 though, they only cost 50p!

Small tweaks keep making a big - and highly subjective - difference. A Pana NHG on the LM7808 output has made the sound much brighter without it sounding grainy or glassy. Is it better? *pffft*
 
The SPDIF input caps are now 10nF NPO 1206, NPO appeantly having good properties for this application. A 1206 resistor solders neatly across for a very neat package.

Does it sound different? Maybe a bit sharper and steelier, but consider that the whole system had been off overnight.

Currently tearing my hair trying to make an AD8561 work as input buffer.
 
Here's one for the "I wish I'd thought of this earlier" file.

It's quite tricky to solder chip legs and especially metal foil with a pointed iron tip. But use a chisel shaped tip, and it's easy as pie.
 
Time for another update, as after lots of small tweaks the actual sound of the dac has changed quite a bit.

Inductance (hope that's the right term) after each reg is a good idea. I use two 100uH/12.5R Epcos mini core chokes in series on the 1543 and 8412, and one on each opamp. Note that the AD8561 comparator is on sick leave, but still gets the Taylor seal of approval.

The effect is like turning up the colour and contrast on a TV. The sound is vivid, solid and visceral. Timbre and texture is more convincing, though this does mean some records sound brighter and grittier.

A potential minus is that the inductors have a subjectively compressive effect. There isn't such a huge gap between quiet and loudest. But whilst there's less vertigo, the overall effect is positive, making for more satisfying lower level listening and reducing the need for extreme volume to get the full effect.

I've had some problems with distortion, er, induced by inductors. I think this is a function of the voltage drop caused by their resistance. So assuming two 12.5R inductors add up to 25R, I've added the voltage drop at typical current to the regs, hence:

CS8412 VD+ / 5.18V / 754R ADJ
CS8412 VA+ / 5.5V / 816R ADJ
TDA1543 / 6.25V / 960R ADJ

Definitely worth trying.
 
You must log in or register to reply here.


advertisement


Back
Top