Many moons ago I built a basic TDA1541A dac using standard OS opamp output. It was pretty good but not stunning.
More recently I've built/modded several dacs using AK4393 & CS4397 dacs (all with passive transformer output stages) and these have all been substantially better than my (poor implementation) TDA1541A dac.
I have a matched pair of TDA1541A S1 chips - is there any point trying to build a dac using these or have the newest generation dac chips surpassed the best quality achievable from the TDA1541A?
I don't want to go to a huge effort with the TDA1541A if there is no potential to beat a decent modern dac.
More recently I've built/modded several dacs using AK4393 & CS4397 dacs (all with passive transformer output stages) and these have all been substantially better than my (poor implementation) TDA1541A dac.
I have a matched pair of TDA1541A S1 chips - is there any point trying to build a dac using these or have the newest generation dac chips surpassed the best quality achievable from the TDA1541A?
I don't want to go to a huge effort with the TDA1541A if there is no potential to beat a decent modern dac.
TDA1541A is out of production, but with NOS I guess you mean "no oversampling" in stead of "new old stock"...?
Oversampling or non oversampling has nothing to do with dac architecture.
A dac is 16, 20, 24 or whatever bit; has current and/or voltage outputs.
Hi, You seem to want some hand holding but this is an imperfect world.
Perhaps you want others to decide for you,but I think you should
be the one that matters as you are going on this audio journey.
If your experience has been that the current 1 bit actually 5 bits dacs
are better than the R2R 16 bit ones like TDA1541A then so be it.Learning
and discovery takes effort and I think you will be shortchanged if you only
let others determine your choices about audio.Yes others can help and guide
but the major part requires your evaluation and consideration.
Hifi is controlled by big business and they determine what use use and hear.
Here is where you take back control and decide what is best for you
instead of marketing hype.Did you learn why the old dac technology(16 bit)
was changed for the "superior" 1 bit dacs? If you did then you would know
that it costs more to make the old 16bit type than the current dacs
because the old type requires more silicon than the new so for a given wafer
you can squeeze more chips than a dac like TDA1541A.
Sound quality is very subjective,different people hear differently,that has
to be recognised and understood.The seemingly better sound from the new
dacs like CS4398 is the different distortion spectrum between it and the old
16 bit type.It is researched that 1 bit dac has a spectrum in the high frequencies
so giving the perception of more resolution but is "weak" in low frequency area and
the opposite is true for 16 bit dac.So it can be seen that man's technology is not perfect
but business is about profit so what do you think will be foremost on their minds?
To be fair to the two dac technologies it must be compared like "apples
with apples" but in reality it is difficult so a compromise has to be made.
Some like the new and others the old.So the only way you are going to
"compare" is to implement each type in the best topology suitable for each.
Have you tried that and do you know what that should be? More learning is required.
In the end it is your ears that matter. Best of luck in your journey.
Last edited:
Hifi is not controlled by anyone - certainly big business has a big influence - magazines seem to have the most influence - but that's changing (reducing) as online forums like this one gain traction.
Yeah its a good idea to ask for what 'superior' might mean if such a claim is being made. 'Superior sound' or 'superior measurements' for example?
Cost in semiconductors is changing all the time. TDA1541A is made with an older process technology than newer DACs so, given the larger feature size, the chip size will be bigger. Bigger chip normally means bigger cost for sure.
Have links for the research? In my own studies I've come to the conclusion that 1bit (bitstream) types suffer from noise modulation - this is largely invisible because nobody's agreed on a common measurement for noise modulation. I take it that 'weak bass' stems from this effect but I have no research, just hand-waving arguments I must admit
Seconded.
abraxalito said
Hi, I don't think I can locate relevant literature for you as it's been ages
when TDA1541A was hot.I do agree about the noise modulation though
and perhaps this is one reason to explain the difference between R2R
vs sigma delta dacs.
I believe even if the lastest wafer technology is used the number
of elements that need to be etched and the number of layers would
make for a costly chip,granted it may be possible to squeeze more
semiconductor elements now on a chip it would still be probable it
would take up a larger area compared to bitstream type dac.Not
only this but the precision triming and QC testing I think.
Hi, I don't think I can locate relevant literature for you as it's been ages
when TDA1541A was hot.I do agree about the noise modulation though
and perhaps this is one reason to explain the difference between R2R
vs sigma delta dacs.
I believe even if the lastest wafer technology is used the number
of elements that need to be etched and the number of layers would
make for a costly chip,granted it may be possible to squeeze more
semiconductor elements now on a chip it would still be probable it
would take up a larger area compared to bitstream type dac.Not
only this but the precision triming and QC testing I think.
Over here I get the impression its still hot - lots of parts for sale on taobao for example. Guys designing and selling boards based on this chip are still making a reasonable living and I don't see that ending any time soon. That's one reason I'm helping out a bit with working on a design for a digital filter to replace the aged SAA7220.
Yeah and multibit dacs have low level linearity problems which restrict the ambience details. So bitstreams do have some advantages.
I believe even if the lastest wafer technology is used the number
of elements that need to be etched and the number of layers would
make for a costly chip,granted it may be possible to squeeze more
semiconductor elements now on a chip it would still be probable it
would take up a larger area compared to bitstream type dac.Not
only this but the precision triming and QC testing I think.
I'm not of the view that smaller feature sizes are necessarily better for audio applications incidentally - I think there are trade-offs in every 'advance' in semiconductor technology.
Hello.
You can consider building a better i/v conversion. A simple one like this http://www.diyaudio.com/forums/digital-line-level/9226-simple-i-v-tda1541.html or that one seen on a french forum (still in progress): Audiyofan.org • Afficher le sujet - Pré-ampli, mu-follower hybride de 6N16B.
You can consider building a better i/v conversion. A simple one like this http://www.diyaudio.com/forums/digital-line-level/9226-simple-i-v-tda1541.html or that one seen on a french forum (still in progress): Audiyofan.org • Afficher le sujet - Pré-ampli, mu-follower hybride de 6N16B.
Posted by abraxalito
Hi, I don't know whether TDA1541A is still hot or not but I think it's the favourable reputation of this dac and mainly it's fans,the bitstream people
simply detest it. I hope those designers succeed but it seems the bitstream
has already gained a foothold what with the likes of Wolfson, AKM,TI/BB and cirus Logic etc.
Mostly the boards that I've come across are only improvements in power
supply (individual ones for the dac) and different implementations of output
opamps,passive IV or tube/valve buffers and not forgetting NOS (non oversampling without the digital filter).
Many people felt or is it heard literally that SAA7220 was not the best and
you say you want to help design a replacement? Is it a fpga or dsp form? Legend has it that the best digital filter that some audiophiles quietly comment was Pacific Microsonic's PMD100 and PMD200 which was aquired by Microsoft that is the software implementation of PMD200.So far microsoft has chosen to put it in cold storage. If there was a white paper if it exist at all would be helpful for you.Wish you success in your project.
Hi, I don't know whether TDA1541A is still hot or not but I think it's the favourable reputation of this dac and mainly it's fans,the bitstream people
simply detest it.
I hope those designers succeed but it seems the bitstreamhas already gained a foothold what with the likes of Wolfson, AKM,TI/BB and cirus Logic etc.
Mostly the boards that I've come across are only improvements in power
supply (individual ones for the dac) and different implementations of output
opamps,passive IV or tube/valve buffers and not forgetting NOS (non oversampling without the digital filter).
Many people felt or is it heard literally that SAA7220 was not the best and
you say you want to help design a replacement? Is it a fpga or dsp form? Legend has it that the best digital filter that some audiophiles quietly comment was Pacific Microsonic's PMD100 and PMD200 which was aquired by Microsoft that is the software implementation of PMD200.So far microsoft has chosen to put it in cold storage.
If there was a white paper if it exist at all would be helpful for you.Wish you success in your project.
Last edited:
Hi, I don't know whether TDA1541A is still hot or not but I think it's the favourable reputation of this dac and mainly it's fans,the bitstream people
simply detest it.
Quite irrational that isn't it? Perhaps the 'bitstream people' can't accept that all new technology isn't necessarily a step forward in respect of the bigger picture. Kinda undermines their belief in 'progress'
I think some, if not all of those guys have realised there are drawbacks of 1bit. I know Wolfson and TI/BB are using multi-level quantisers in their sigma-delta modulators, ADI do too. These are supposed to fix jitter sensitivity and noise modulation, and to a degree they're successful. Just I'm not sure that they totally fix these problems. Hence my renewed interest in the TDA1541.
Strictly speaking its neither of those, its a common-or-garden MCU. I have looked into both FPGA and DSP and they both look like considerable overkill for this application where I estimate about 10MMACs to be sufficient. In addition to being sledgehammers cracking the proverbial nut, they're much less DIY-friendly. An MCU is relatively amenable to people with a smattering of digital experience as well as being more affordable in development terms for when people get into designing their own filters as I hope will be the case
Certainly going by their datasheets they're the best ones I've come across technically. Can't say as I've evaluated them sonically though. The PMD200 is in essence a DSP56300 (24bit DSP originally from Motorola, now Freescale) which makes it a bit power-hungry when compared to my proposed solution. Admittedly they're using quite a lot more grunt but some of that must be dedicated to the HDCD system.
So far microsoft has chosen to put it in cold storage.
Yes if HDCD turns out not to be too difficult to implement and the details are made available, it might undergo a renaissance. Thanks for your kind wishes
The one I'm playing with, the ARM M0 isn't considered to have DSP features (though it does have a single cycle multiplier which attracted me!) - but its bigger brother the M4 has.
I bought an Atmel
Dragon programmer cum emulator some time back but have not tried anything with it.Trying to learn mcu you see.
I suggest you leave the Atmel on the shelf and get in the mainstream - ARM Cortex
Plenty of relatively affordable ways in to the ARM ecosystem nowadays, I shall write a bit about that on my blog shortly.Hi, I think I should learn to walk before I run.
The ARM mcu are not cheapand a basic development programmer and software is not in the budget
of a novice like me.But you are a professional so you have the resources...
I think the ARM mcu would be suitable for your digital filter project with it's
processor speed though.Good luck.
Hi, I think I should learn to walk before I run.
I used to say that - but recently I read something online which pointed out that babies don't do it that way. They do indeed try running before they've mastered walking. It was a very interesting observation
Development of skills isn't a linear process although we often like to think that it is.I really appreciate your feedback, no kidding
It just shows me that my blog posting is getting a little overdue if your perception is held by many. You'll be surprised to hear I just yesterday received 4 dev boards for the LPC1114 - an ARM M0 CPU. The total price was around $35. Perhaps I should start selling on diya? No programmer is needed, its all on-chip programming via the 2wire debug port. These dev boards also amazingly include the USB-debug port interface chip, but I don't yet have the software that interfaces with it. Some tools are also free, I will mention these on my blog.No, I'm an ex-professional returned to hobbying
I think so too - I'm just trying to convince it.....
Hi,
And you probably used 3-Pin regs in the PSU as well? What a perfect way to waste the TDA1541.
Good, be happy.
Good, send them to me, I'll make sure to dispose of them according to the necessaries environmental guidelines.
This has a lot of "depends" to it.
The "old" lower bit multibit DAC's (specifically TDA1541, PCM56, AD1865) have as good or better SNR as most later Multibit Types and they do not require noiseshaping plus massive averaging to achieve this (all modern DS and Hybrid DAC's have very poor SNR, unless you average loads of samples to get the noise-modulated/dithered noisefloor out).
The Linearity of these DAC's is typically also very good and again, this is REAL linearity, not some pretence that relies on averaging techniques.
So in principle there is no reason such a DAC should not do as well or likely better with CD standard digital audio.
HOWEVER, the TDA1541 is probably the most difficult DAC to employ well. It has many unusual and interesting design features that contribute to it's exceptional performance, however it is easy to downgrade this potential with generic power supplies, ill considered interface of the digital signal, ill considered use of digital filters and a number of other factors, it is also extremely sensitive to PCB layout.
For CD standard digital audio at least it is possible to make something crushingly superior.
Ciao T
And you probably used 3-Pin regs in the PSU as well? What a perfect way to waste the TDA1541.
Good, be happy.
Good, send them to me, I'll make sure to dispose of them according to the necessaries environmental guidelines.
This has a lot of "depends" to it.
The "old" lower bit multibit DAC's (specifically TDA1541, PCM56, AD1865) have as good or better SNR as most later Multibit Types and they do not require noiseshaping plus massive averaging to achieve this (all modern DS and Hybrid DAC's have very poor SNR, unless you average loads of samples to get the noise-modulated/dithered noisefloor out).
The Linearity of these DAC's is typically also very good and again, this is REAL linearity, not some pretence that relies on averaging techniques.
So in principle there is no reason such a DAC should not do as well or likely better with CD standard digital audio.
HOWEVER, the TDA1541 is probably the most difficult DAC to employ well. It has many unusual and interesting design features that contribute to it's exceptional performance, however it is easy to downgrade this potential with generic power supplies, ill considered interface of the digital signal, ill considered use of digital filters and a number of other factors, it is also extremely sensitive to PCB layout.
For CD standard digital audio at least it is possible to make something crushingly superior.
Ciao T
Hi, Just curious what do you blog about? 4 boards for $35... are they just
populated with a low end ARM mcu and basic IO with usb ISP? What brand?
You know you could start a small cottage industry by offering mcu controlled
soft start sequencing for tube amps but the challenge would be the high voltage relays coming in after the heaters.Plus monitoring of bias voltage and
current with adjustment which would entail a minimum 2 line LCD backlighted of course.It would look cool and just before the start a personal graphic signature of the amp owner's name or moniker would appear for some seconds plus overload protection or maybe optional just like the one an engineer friend made for his amp.
Thank you for encouragement for poor diyers who still believe in TDA1541A and other R2R dacs.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.