Building the ultimate NOS DAC using TDA1541A

I have not yet tried low frequency dem clocking - it's on my 'to try' list. It may not work with large value dem caps (100uf)

I use a Li-ion battery pack (+ - 18-20v, with taps at 7-8v for the 5v supplies) so no mains side effects,.

Was not a game changer to me than the low frequency DEM clock and even less the LiFEPo on the clock side (inductance ?) . I listen to more than I measure so it is subjective,, although I think it is better than the opposite. But of course I do know ears are more focused on tonal that anything else, so changes in the device with ears checking only migth be a trouble...
 
Nichicon KL 680uF/25V at around 70mOhm.
It's more of an 270mOhm range for 16v one, 234mOhm for 25v one. Significantly more than 75mOhm, but still quite low for a low leakage cap.

Seems plausible to my primitive brain in digital sector 😀 Wondering if you though of implementing this as a mix with the idea you had, of using 10 tda chips? That one really got my interest going.
 
Thorsten, you said you haven't tried the simultaneous mode yet ( in order to have a point of view about it, I look forward you try it,, did you already?).

I will use it. If I get bored I might try a comparison. I might install jumpers.

I want 384k capabilities, so SIM is necessary.
What about the analog output referenced to 5V ?

What about it? It's just one thing to take into account.

The diferential input Z of the opa637 is not a bit high ?

Input Z is created by feedback. In the Audio band it's low. OPA637 has 80MHz GBWP. Adding 10nF from -in to +in (and GND) keeps the circuit stable and lowers input impedance at higher frequencies.

What about for instance : opa861, AD811 (the old flavor of the moment here in this forum) I haven't tried that one) : AD1633 (dif op amp), AD828 if low curent noise and capacitance input matters ?)

There are many options. Previously in restorations I used OPA637 and it worked well.

You use an op amp + bjt + Mosfet hybrid scheme on your Ifi home DAC best home DAC nowadays, is that easier maybe with the high output and low noise of a PCM1793 ?

Ifi Pro? It actually has a passive output from the DAC(s). RLC filtered.

A fully discrete circuit derived from pro audio gear (non Op-Amp, not even discrete, non usual) is gain/line/headphone drivers.

Op-Amp's are servos.

I focused on your input on some modern chips above in this thread : TDA1541A musicality beated ? (I mean with 16/44 materials)

IMNSHO No.

Edit : I wish I had a good 6072A and 5687, I had to be satisfied with solid state or an E188CC. Will you use the same tubes or confs with your nowadays knowledge VS the CD-77 or maybe your personal one that migth be even better cause it is hard to source NIB todays ?

No, I will use a solid state current conveyor (or shall we call a cascode for Iout?) with ~ 100V rail. A suitable resistor (adjustable, precision wire wound) to produce up to +20dBu output and provides I/U conversion.

It is to be followed by a follower (also solid state) that also implements a sallen key lowpass that will also act as EQ for the SINC rolloff of 1FS digital.

Output coupling, Elna Silmic Bipolar.

Thor
 
Why not using LRCK DEM slaving in spite of Blck ?

It is preferred to have Fdeem at 4 x Sample Rate or higher, due to the way it is designed.

Which is the most critical between both ?

The design of the TDA1541 gives a degree of "build in" noise like jitter, plus, depending on precise design, extra RF noise. Excessively low jitter is unlikely to offer anything useful.

I will reclock all outputs with MCK from the source.

We cannot be 100% sure what latches when and how, we'd need the full schematics. A key factor is to ideally limit the input swing to 200mV P-P centered around 1.3V at 25 degrees. This quiets down things a lot.

MVAL's design still has around 600mV P-P and thus allows the input stage to saturate and to throw noise I to the rails and dgnd, if a lot less than without signal conditioning.

Thor
 
Thanks Thorsten,

Yup I understand if the op amps has very low output impedance it helps a lot vis à vis of the input pin if I understood well the basic (My tech skills are near 0 DB though).

Btw advised to try the ad828 (fet input as well; very low dht <2 K hz) I found it has a very dark analog side on the natural side (not the average op amp sound, imo a rare behavior) but I found the mid to treble to be to thin : I didn't try hard and just with a low 5V VDC that is said to be a limitation maybe (whatever Cf I used between 20 pF and 1 NF COG it sounded thin - not with a TDA1541A but with a an AD1862, but for sure the thin sound in the mid highs is not dac is related.

Have you heard about that job (sorry I digress, your input is not so about I/V but more about DEM managing) : https://www.diyaudio.com/community/...ry-low-distortion.217459/page-40#post-3353436 any thougths ?

Funny enough Pedja uses the two serie Elna S2 (or a O-core output traffo at choice) ; Elna BP are good too, but whatever I try with caps, (film too), DC coupled has some more ... I dunno how it is called, but more good music .Maybe just a myth as it is AC... Perhaps autobias (but for sure the tined SCR or Mundorf works better to my ears or Hifi setup at least : loudspeaker, etc)

"IMNSHO No" .... 😀
 
Last edited:
It's more of an 270mOhm range for 16v one, 234mOhm for 25v one.

I will measure it soon.

Wondering if you though of implementing this as a mix with the idea you had, of using 10 tda chips? That one really got my interest going.

I am struggling getting "A few good TDA1541", never mind large numbers.

4x the wordclock relative to DEM clock frequency is the Grunding scheme, rigth ?

No. It's actually Henk ten Pietrik (I am sure I butchered his name). It is the actual WCK applied to TDA1541, so not Grundig.

Grundig works if using SAA7220 Digital Filter, it sync's DEEM and WCK.

Thor
 
I have not yet tried low frequency dem clocking - it's on my 'to try' list. It may not work with large value dem caps (100uf)

Look up.MV Audiolab, works fine it seems.

I use a Li-ion battery pack (+ - 18-20v, with taps at 7-8v for the 5v supplies) so no mains side effects,.

Everything in your system is battery powered and in a faraday cage?

Thor
 
May you please give us the average output Z of the TDA1541A

Very high.

and it soutput capacitance please ?

A few pF.

I think what you are really asking, is what is the best I/U conversion resistor.

Without offset trim 6.25 Ohm. With offset trim 12.5 Ohm.

You can go higher, with offset trim, 50 Ohm are usable. I would not go higher than that.

It is better to use a current conveyor and decouple the I/U conversion resistor from the TDA1541 output completely. The TDA1541 is really designed for 0V on the output. That means 0 Ohm, or as close as possible.

Have you heard about that job (sorry I digress, your input is not so about I/V but more about DEM managing) : https://www.diyaudio.com/community/...ry-low-distortion.217459/page-40#post-3353436 any thougths ?

Let me look.

Funny enough Pedja uses the two serie Elna S2 (or a O-core output traffo at choice) ; Elna BP are good too, but whatever I try with caps, (film too), DC coupled has some more ... I dunno how it is called, but more good music .Maybe just a myth as it is AC...

DC coupling is preferred, but it comes with other problems. Often enough, "DC coupling" simply shifts a VISIBLE coupling capacitors into a position that is "invisible" (e.g. power supplies, feedback loop of Servo Op-Amp).

They can do as much sonic damage, but people think "it's not in series with the signal so it's unimportant".

Hence we need to consider the entire circuit. Once we do that, often simple AC coupling is the better choice.

The same applies to a lot of items. Listening needs care. It is too easy to fool oneself. In fact, you (or I) are the easiest person to fool.

If I expected something to sound better and know I have implemented it, it will usually sound better, regardless of fact. So care is needed.

I tend to listen "do I enjoy listening and want more", that is emotional reaction and shift from baseline, not so much specific sound aspects (stage, treble purity, treble prominence etc.).

Thor
 
Last edited:
"Look up.MV Audiolab, works fine it seems"

Has to be heard first ! He also like better cathode input and grounded grid, so migth be very subjective (I personaly didn't find it so good the 14x 100 uF, I am even not sure John has ever test it for real on a listening session) but the layout with what I tried is so-so for that (hard to deal with the inductance of long leads when you have just a smd layout, you find the ground where you can, hey !) . Maybe it is good, maybe it is not so, needs more real world tests imho (listenings).

Edit : I tried around your 50 ohms input good enough limit with ECC88 like (mu33 and the help of the follower preamp), but the impedance output TDA1541A ask was more in case to choose active I/V, so not Riv related, more perhaps after first pole stability : peaks after the first pole if of an any importance btw ? ( And thanks for your patience, it is very above my head ). Lower Cf value in mind, flatter bigger bandwidth. Not easy with 3 stages op amps ? Equal to : give up op amps ?

But I digress, please continue the flicker that poped up 🙂

Edit : I readed it was around 600R output?! AD1862 is 2k6 for illustratiion. But I have no idea !
 
Last edited:
He also like better cathode input and grounded grid

It is often better, it just creates another set of problems.

I personaly didn't find it so good the 14x 100 uF

It needs to be seen in context.

In my view the return of these capacitors is -15V, returning to AGND makes everything super critical.

Secondly, these capacitors do not handle the switching noise glitches well, so these ride through.

So we still also need small size, RF capable bypass caps with short current loops. Looking at the scope trace John showed, the switching glitches look 10...20nS wide, or in effect pulses that are half of a 50...100MHz square at 50% duty cycle. That is a lot of potential to cause trouble.

but the layout with what I tried is so-so for that (hard to deal with the inductance of long leads when you have just a smd layout, you find the ground where you can, hey !)

That is trouble.

the impedance output TDA1541A ask was more in case to choose active I/V, so not Riv related, more perhaps after first pole stability :

We are looking at a bunch of BJT Collector outputs. Cascoded and cascoded again for MSB's. That is easily MOhm.

peaks after the first pole if of an any importance btw ?

I think so.

Lower Cf value in mind, flatter bigger bandwidth.

Problems with RF noise.

With CF it rides through past the OPA at high enough frequencies where the Op-Amp is out of steam. No IMD. We can filter with RC on the Op-Amp out.

Have a cap from output to +5V (say 10nF) and we also reduce the noise before it gets to the Op-Amp in and Cf.

Edit : I readed it was around 600R output?!

You just need to look at the various internal schematics.

AD1862 is 2k6 for illustratiion. But I have no idea !

AD1862 is R2R.

TDA1541 is current steering.

Totally different animals.

Thor
 
  • Thank You
Reactions: diyiggy
Look up.MV Audiolab, works fine it seems.



Everything in your system is battery powered and in a faraday cage?

Thor
Thanks.

Yes but battery pack is external (uses two banks of five 22650 8000ma cells)

PS I use Sowter I/V transformers direct from the 1541 outputs. Therefore I don't have any need for any other type of I/V or amplification. IThey remove the need to choose what opamps or tubes to use.

They are very expensive but being bifilar wound, give excellent cancellation of common mode noise.
 
Thank you @ThorstenL !

A genuine and naive question of a newbie 🙂 If DEM should return to -15V and AOR/L to +5V then why Philips engineers did not specify that🙄? Out of the convention that "everything goes do the ground" not to confuse engineers designing CDP's?

I was thinking of a low I/V in the 10R region but the THD profile from MV: https://www.mvaudiolabs.com/digital/tda1541-iv-resistor-selection/
shows domination of H3, H5, and H9 for low R. 50R has more HD but seems better distributed with H2,H3,H4 dominating. Would be interesting to hear your thoughts on that.

Very nice idea on bypassing electrolytic DEM caps. I'm thinking of 100nF C0G 1206 caps soldered between the Elcaps leads on my existing board.
 
............ The analogue stage MUST link it's signal current return to +5V, NOT AGND. It needs low input impedance. If filter capacitors (or impedance compensation capacitors to compensate rising impedance with frequency) are used on the AOL/AOR pins, they must link to +5V not AGND.

This is my analysis so far. I am looking forward to comments, criticisms etc.

Thor

This is what Mvaudiolab told us years ago here : https://www.mvaudiolabs.com/digital/tda1541-iv-stage-options/ , but he did not get any deeper in explanations , ground return work perfect if very well implemented , even better than hes option as far as measurements are telling , sound wise it's too subjective to tell , matter of taste , biais , overal system , etc .....

and I remenber he had temperature issue with the TDA at 62 ° , when normal chip temp even in summer doesn't go over 45 ° without any heatsink , could hes choice be responsable of that ?

what is your personal point to go +5v ref return instead of ground , and why MUST instead of could

and tell us more about the idea of returning the dem cap to where they are from -15 v ? , or maybe I miss something

.
 
Last edited:
A genuine and naive question of a newbie 🙂 If DEM should return to -15V and AOR/L to +5V then why Philips engineers did not specify that🙄? Out of the convention that "everything goes do the ground" not to confuse engineers designing CDP's?

Convention.

We often see circuit nodes returned to ground, which then actually have currents return to a different circuit node from "ground".

shows domination of H3, H5, and H9 for low R. 50R has more HD but seems better distributed with H2,H3,H4 dominating. Would be interesting to hear your thoughts on that.

It's not just which harmonics dominate, but also what the the levels are.

Very nice idea on bypassing electrolytic DEM caps. I'm thinking of 100nF C0G 1206 caps soldered between the Elcaps leads on my existing board.

Very bad idea.

Thor