discrete output for single tda1543

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi PreSapian,

Why not use the easy to do 1543's output design of John - ECdesigns -
Just few component (no dc coupling cap, no opa...) cheaper than benchmark many expensive opa, discretes and shunts to powersupply them. (keep them for an another project).
You have the opportunity with the 1543 to do without all these things if you want.
Search in his famous ultimate thread, there is a shematic.
i'm having a bit of fun with 1543 iterations now a days.

i built an overkill version of cmoy with opa-earth for the buffer/headphone amp of my 1543 and though it does take away the nuisance, the dynamic does benefit a lot for the single chip board.

of which, i'm powering with a nicad battery. i also have a couple of 10 chip boards made- one is 317 regulated, the other teddyreg.
teddyreg does sound more refined but the bass is less.

i did also try john's passive output but scapped it after realizing it took away too much of finer detail- i guess less is more is especially applicable to tda1543 implementations.

i'm now itching to try differential output version and a shunt regulator.
am thinking of piecing together a zener regulator this week just to see if it will have any positive attributes.
Hi presapian,

I had a good result with the TDA1543 with this :

tda1543 with just an OS-CON SP 220 uf (not 1uf in // with mkt, etc) for I/V,

then an old JRC OP that works well here: NJM 4580D(x), the caps are very important here : old nichicon Vx 100 uf that you find in old philips CD players, I don't know why better caps doesn't work better here,

then a passive low pass filter façon Marantz in their old players, then

then 2 OPA GD Sun with 2 Silmic 100 uf by OP for the two rails on it, you can use shunt here if you want,

then DC blockers, i use old nichicon bipolar muse that better than the newer::: the older of the 90s' are very better, you can use too 10 uf KP, PP, lundhal if you have it, but muses are enough with this low budget conf,

to me it dramaticaly sings if you respect the choosen caps... better than a Luxman 105 d with tubes I found in public garbadge 2 weeks ago...

I prefer most of the time a black Nichicon Z muse caps for power than silmic 2.

It's just the 50 cents classical humble empirical opinion of my hears,
thanks Eldam, sounds like a worthy mod to try.

surprising that you liked os-con for decoupling- many have experienced (me including) that they sound harsh in analogue path. perhaps that's not an issue in this location.

have you ever tried mundorf mlytic for that location? 100uf bypassed with a russian pio 0.1uf gave me the most pleasant sound out of all the other config i tried- 220uf bg std, 100uf silmic1 (and 2)+ mundorf supreme 0.1uf, and even a jantzen crosscap 47uf. mlytic is also the biggest electrolytic cap i've seen. i could only acquire the raw version but plain is suppose to be even bigger and better performing.

ultimately i want to try a tube output but i want to fiddle with power supply some more to determine if 1543 can be made good enough. but i suppose a dc coupled 1543 with a minimal trace length to a tube pre serves the same goal.
Hi Presapian,

In fact I just talk about OS Con for the local decoupling cap of the TDA1543.
The same for the others caps I talked about:local decoupling. I don't talk of the main powersupply of these stages which win (of course) to be separate.
Note that I talk about an empirical & particular configuration for the TDA 1543.
So the old blue VX nichicon caps serie for the I/V is just for the JRC NJM4580D(x) (which was used for head speakers...) and the Silmic 2 for the OPA GD Sun (which is ok at 12 V, Burson for example need 15 to 16 V for singing).

if I talk about "most of the time situation" : I'm agree with you for Silmic (1) which softness can save sometimes too. I heard that silmic was a special ask of Marantz to Elna for the bypassing of their CD players outputstages... Don t know if it's true or not.
I don't like ol black gates, even BG NX (in digital) that a experienced engineer like Jean-Paul (for example) likes (and I think he tried a lot because he gives a lot of time...to time for trying many confs). But agree with him when he prefers Nichicon Z muse to silmic 2...most of time.

I don't benchmark new OS-CON technologie with their famous old one (out of production) because both are hard to genuine source. I saw that Thorsten seems to use the old one in his last AMR Cd player (the velvet caps on the motherboard (not the blue ones easier to source...). But I will try the last Nichicon polymere caps whic have the lowest ESR for digital but I surmise that the ESR is not the single fact of the good result of os-con in digital stage (maybe snake oil?).

I will try M-lytic like you already wrote to me for power stages, they are easy to source cause made by our neighbours friends of Germany.

About tubes: I surmise that it's easier to have good result because their use is well known and documented but I think it's possible to live without them... I will be happy to change my mind if I am wrong here, no problem.

I want to play with analog Low pass filters, in fact I ask myself if it's the only way to seriously tune a digital stage to make it sings ????

Happy you give us your results and you play with this little TDA1543 chips.
An externally hosted image should be here but it was not working when we last tested it.
current state of things- single
battery 1543, jantzen crosscap 47uf decoupling, ohmite resistors and i2s from simple lm317 regulated waveio. yesterday i matched it with a b1 buffer
i put together overnight and the sound is sublime. with b1 the lost bass is back along with greater drive, clarity. i was looking into trying out 20x1543 but instead rather would go full battery power on b1. 1543 is fun but b1
is simply amazing!
:) After the B1 you can maybe use the CEN or SEN... known to be good but I never heard it;

Before the B1 which cap did you use with the John's TDA1543 output stage ? Use here may be two smd 1uf Cornel Dublier used by John with the big 470 uf Silmic 2 for the lost details you talked about... tell me please. May be the layout ? I say 2x1uF because I read somewhere that the theory is 5 to 10% of the main value to avoid resonance...but I'm not a technician, check it before but you have to try. I use the empirical method:eek:!

I have sometimes this problem with Silmic2 : awsome bass but lack of details in relation to a nichicon Z muse cap!!!!!;) oups the M-lytic...I forget !

I see you are at the beginning of a great quest with the TDA1543: I don't know if you read that ( you can apply most of the advise of the TDA1541 to the 1543 ):
Last edited:
i used green sanyo 470uf 100v- only thing i had readily available, with mundorf supreme 1.5uf. i'm no technician either but i thought that 470s were basically paralleled reverse polarity along with 1uf for decoupling.

i didn't know 1541 and 43 were so similar- will definitely start tackling that mega thread.

i feel confident enough for a 1541 creation now so multi 1543 vs 1541 will happen soon enough. ;)
just tried out simplest zener shunt regulator for the psu. extremely pleasant and smooth sound, this is the ticket! distortion seems to be more than lm317 but it seems to be h2 dominant since the sound only becomes sweeter as the volume goes up, unlike other regulators i've tried. I have yet to try salas shunt but as sweet as this simply zener sounds, I'll probably keep one as it is.

p.s. snr is also down but it's a 1543 after all- can't expect too much in that department :p
swapped out zener resistor to a caddock thick film and it's a winner.
much more ambiance and texture than any other setup i've tried including teddyreg.
317 cannot even begin to compare. when salas shunt is not an option i think this premium component zener reg would be hard to beat.
it still sounds a tad bit bright but i still haven't matched the zener with optimum resistor value. will try 1ohm caddock within a week.

on a side note, could it be that different production runs of 1543s have different sound? (rohs lead-free i.e.) because this time i built the dac with the chips that say thailand on it without the polarity strip and even though every other components are the same, treble is not as dark as my previous builds.
harsh treble further tamed by using only one mur860 as a half wave rectifier. sounds more intimate/small venue with richer harmonics. i am using a rather oversized ei transformer for the job so maybe it's all just ac induced effect?
regardless, it sounds good and that's what counts with 1543, i think; not accuracy. ;)
swapped out zener resistor to a caddock thick film and it's a winner.
much more ambiance and texture than any other setup i've tried including teddyreg.

Hi Presapian,

thank you for the news.

Teddyreg is known to have one of the flatest impedance curve, maybe a reg with a curve jumping down before high frequency couyld be a deal here ?!

Have you a shematic of your last DAC with the parts used to understand more your progress ?
You feed the TDA via a module usb to I2S ? What are your listing environment (speakers, wires, amplifier...) ? I would try to understand why the acoustic signature of a part seems to sound better than another and what is compensated in the listing chain.
Hi Eldam,

I'm really a caveman with this stuff and the setup is so simple that i should just list it here.

EI trafo 9v tap> single mur860 halfwave> panasonic tsha 10000uf, caddock 4ohm, 1n5??? 6.2v zener reg> 47uh inductor >100uf silmic bypassed with mundorf audiophile 4uf> ohmite little devil bias resistors.

caps after the inductor i need to change to a 47uf film cap (2 would be even better heh)
and 1ohm caddock is coming in the mail.
this really is the best dac i've had, and gives my other 1541 dac run for it's money in the 'fun' factor.
i think of it as an adolescent son to 1541 mother, full of energy and unfortunately not able to ever mature.... ;)
Hi Presapian,

thank you. I like the way you go : learning from a simple circuit, here a regulated powersupply with a polarized zener diode, but with the optimal parts to set up and compensate the parts together to learn from them...

So we need good theory, good circuit (not to reinvent all the times) and cooking with parts to set up. And simply have fun and play with that... a possible hobby for modest diyer who want learn without annoy mixing theory and practice for relaxation. And step by step try to habe a better skill... that the way I go but I'm very slow... I am a perpetual beginner !

Do you know a Korean caps company : G-Luxon. If you can find the blue caps of the 80s' : GR serie (general purpose) you can learn how a very bad cap can be sometimes a very good tool after the bader reg when you have no money for shunt regs. Little tip for little money.

have you try a simple pi filter before the resistor which polarize the zener ?
Something like the MUR diode > Cap (C1) with high ESR (general purpose , 105°), then if you have enough voltage: 100 to 200 homs (for example 600 h x 3 in //) > C2 with 1,5 the value of C1 but with low ESR (Silmic 1 or nichicon KZ or maybe silmic2. Then your resistor with the new value to feed the zener ? Can you hear a difference with that or not ?
First you can try with C1= 7000 uf; C2 = 10000 uf to benchmark with your shematic...then try with divide by 2 or 3 for both C1 & C2.

If you can find Sanyo Os-con 47uf, you can benchmark too the M-Lytic with it...
Your USB to I2S is the Lucian one ? Amanero ?
hi eldam i will certainly keep your suggestions in consideration.. next j want to try russian kbg caps that are said to be netter than k40. also i went little nuts with rfi shielding on my battery dac today. everythings grounded to pin4 of the tda, tho i'd have to count out the copper pad over the tenor chip to consider the whole dac star grounded. sound seem to have become thicker with harmonic and also calmer but i'm not sure- hence the word 'seem'
An externally hosted image should be here but it was not working when we last tested it.

right now im using te8802 board from audiogd since i dont feel the superiority of xmos is audible enough with 1543. infact te8802 has more pronounced bass that the 1 chip battery iteration sorely needs.

te8802 sounds very very good on its own, btw.
Last edited:
i've been fooling around with the zener regulator setup some more. it sounds quite good, much better than lm317 to my tin ears, actually, if you use the right resistor- which is caddock, of course. ;)

right now i'm trying to figure out where the best place for a cap bypass is- i've already tried snubber fillter for the zener and ended up preferring cap-only bypass. also put an inductor in series to the zener but am wondering if the the cap should parallel inductor and the zener or just the zener; i can't tell much difference from the a/b comparison.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.