Here's a good article on analog vs digital DAC volume control:
What’s Wrong with Digital Volume Controls?
An important factor is the DAC's noise floor and external volume control noise that you add to the signal. The PGA2311 has 2.5uV extra on whatever comes out of the DAC.
I just measured the level that I normally listen at. What goes into the amplifier, and it was around 40mV rms. The DAC has a 2V rms output so that's 34dB down.
If your DAC's noise is let's say 6uV, irrespective of the digital volume setting you're having that down the signal chain.
If you employ analog volume control after the DAC, and bring that down let's say 34dB in my case, the noise drops from 6uV to 0.12uV. Of-course you add to that the noise of the PGA2311, so around 2.62uV total at the same listening level, without considering any other noise source in the chain. With digital volume control you're loosing dynamic range. Digital volume control is more of a convenience than better performance, the way I see it.
Of-course I may have understood it wrong and maybe someone can correct or add to the picture.
Chasing nV of supply noise but using digital volume control seems wasteful.
edit: the pga2311 is basically a stepped ladder attenuator, with digital control. as in something like this but in chip smd format:
Last edited:
I am not downplaying "this project" at all. Respect to the designer to make existing stuff better. OP stated clearly that the design is for existing LM317 designs. New designs can do better from the start. The thread title is clear: D-Noizator: a magic active noise canceller to retrofit & upgrade any 317-based V.Reg.
So if you wish to think negativity is the core of my posts: I am downplaying YOUR project which is also not the topic of this thread. Using PGA2311 (designed in 2001) while AK4458 has excellent volume control and "stuff you already have" is a perfect example of that.
Exactly this.
You know this is an absurd simplification as it is chockfull of CMOS switches. Also an outdated chip with some peculiarities which need to be addressed already surpassed by way better performing ICs. But it is your party, have fun. I would use LM741 for the output stage. Here an example how things can be done another way: AK4458 / AK4468 DAC - Share Project - PCBWay
Last edited:
The Nonoiser is clearly not for adapting existing lm317 designs, it's been enough said in this thread.
My project is an example use for the Denoiser/Nonoiser, what are you talking about? It's exactly what I could use a low noise power supply for.
You're just snarky for calling you out, that's all.
I just explained the downsides of digital volume control.
Why don't you give me some examples of good volume control chips then if you know better ones? That would be a useful contribution.
I will use LM4562 for output filtering and then LME49724 for driving the signal into my amp. That LM741 is a bit expensive for 8 channels, and is a single package. You're kinda like that car mechanic that you go to replace your breakpads and he always tries to sell you competition level full ceramic brake system when you've got an average car. No doubt they're better, just that it doesn't really makes sense in the grand scheme of things. ]
edit: I already posted in zarandok's thread. I think he made the design your referenced.
My project is an example use for the Denoiser/Nonoiser, what are you talking about? It's exactly what I could use a low noise power supply for.
You're just snarky for calling you out, that's all.
I just explained the downsides of digital volume control.
Why don't you give me some examples of good volume control chips then if you know better ones? That would be a useful contribution.
I will use LM4562 for output filtering and then LME49724 for driving the signal into my amp. That LM741 is a bit expensive for 8 channels, and is a single package. You're kinda like that car mechanic that you go to replace your breakpads and he always tries to sell you competition level full ceramic brake system when you've got an average car. No doubt they're better, just that it doesn't really makes sense in the grand scheme of things. ]
edit: I already posted in zarandok's thread. I think he made the design your referenced.
Your design is a NEW design so make it the best you can! Call me whatever you want. You are not aware of some things. LM4562 has a high chance on very annoying popcorn noise. Too bad as it is a good sounding opamp. Not all have it but quite a few do. Tip: first design the device and then start filling in part numbers depending om required features/specifications.
MUSES 72320
Please start a new thread on your DAC design as the thread is polluted by too much off topic stuff.
MUSES 72320
Please start a new thread on your DAC design as the thread is polluted by too much off topic stuff.
Last edited:
I came here to ask if I can chain two denoiser/nonoiser circuits for my DAC needs and you started pushing me away from this, then started mocking my design decisions.
If you don't want to get offtopic, please stop going offtopic or deferring people from this project. I really don't understand what's into you for doing this. If you can't answer my questions regarding this project, just stop posting here. I really don't like the way you go about it. Question people on their decisions, then mock their decisions then sending them away while accusing THEM of going offtopic after you pushed the discussion that way.
Still didn't get an answer to my initial question.
If you don't want to get offtopic, please stop going offtopic or deferring people from this project. I really don't understand what's into you for doing this. If you can't answer my questions regarding this project, just stop posting here. I really don't like the way you go about it. Question people on their decisions, then mock their decisions then sending them away while accusing THEM of going offtopic after you pushed the discussion that way.
Still didn't get an answer to my initial question.
View attachment Scan20201001113302.pdf
Take a look at the attached schematic. I have a question about diodes D6 and D9 in the power supply. Should they just be left attached or do they need to be routed to something else.
I do know I disconnect and attach the denoiser to the - leg of C34 and the + leg of C36
Take a look at the attached schematic. I have a question about diodes D6 and D9 in the power supply. Should they just be left attached or do they need to be routed to something else.
I do know I disconnect and attach the denoiser to the - leg of C34 and the + leg of C36
I've built 4 of them and they work great ( when removing ceramic cap at the out put.
Why still use them in new designs ?
LM3x7 + denoiser cost 10 to 20 times less than LT30xx .
LM3x7 come in easy to configure 3 lead packages.
LM3x7 are easy to cool in TO220 and DPAK TO252 .
LM3x7 + denoiser perform practically the same as the LT30xx types for what we use them here.
LM3x7 can take 40V , most of the 30xx only 20 V and with transformers giving excess voltages when barely having a load , even a 15 V transformer can go way over the 20 V.
Most of the LT30xx come in super tiny packages , making an PCB and soldering them is near impossible . Cooling 3 mm packages well , is ridiculous for average DIY'er.
You can get LM3x7 everywhere. LT30xx are for some (me) , hard to buy.
So an easy to use, cheap regulator , easy to buy , with the denoiser as good as the new expensive ones , why would we not use it for new designs?
OK , 1 thing that the LM doesn't have is LDO , no low drop voltage .
Prone to oscillate ? Yes with denoiser and low ESR , there is trouble.
Are the LDO LT30xx absolute stable ?? Do they never oscillate ??
using these denoisers without proper measuring equipment does not guarantee expected results
Isn't that for all regulator cicuits ?
What use case is that? Lately we have been discussing power supplies to AK4458. LT304x easily outperforms LM3x7+denoiser for this.
Apparently you have not gone through the whole thread. The improved denoiser (with Sziklai pair) oscillates with higher ESR (above 500mOhm). OTOH regular denoiser (the one in the 1st post) is not happy with LM317 datasheet circuit as it requires more than 1uF output capacitance.
Your initial question was "I need clean 5V for AVCC for a DAC. Needs about 50-70mA. I am planning on doing +-15V with LM317 + denoiser for opamps, and I wondering on the best way to make 5V for AVCC of DAC."
Your question has been answered by many already.
In what objective way apart from the overshoot and time to stabilize? I also have no problem with burning the extra current required by the denoiser/dienoiser if I go with either. Also Nonoiser is on the table, still haven't decided. Currently still thinking about volume control, which by the way, it amazes me that people talk about nV in power supplies yet regularly benefit from the dynamic range of a 16bit(or even less) DAC. Still trying to understand if maybe I got something wrong or people usually tend to not understand what they are actually doing.
Last edited:
-less noise
-requires much smaller space (probably leads to even less noise compared to LM3x7+denoiser)
-proven design
-less tricky with output capacitance
...
As far as I understand you are trying to build an AK4458 DAC that would perform on-par with commercial offerings. Do you really believe that you can achieve this with LM317 PS?
I have built a couple of Dienoizers and they measure great, work like a charm and cost pennies. Perhaps it is time to do some sort of recap listing the best working Denoizer and Dienoizer positive and negative designs and make a sticky post. I think they are here to stay. Kudoos to Elvee.
is it tho? I remember Elvee having lower numbers than lt3042 spec sheet, with the Nonoiser. Also you need really good design to get to the lt3042 spec sheet noise numbers.
With a good implementation of the Nonoiser I can for sure reach the found specs on noise. Especially since Elvee breadboarded it for testing.
Also using a CS3308/CS3318 might allow me to benefit a higher dynamic range than someone with a ES9038pro using its digital volume control.
Would I really need the specs of the Nonoiser or lt3042 for the AK4458/68? Not sure, but since I already have the needed parts, why not?
Modern DACs (including AKM4458) operate with 32 bits so the loss of resolution of digital volume control is small. The article you referenced was written in 2012 (so way before the current generation of DAC chips) and more of an opinion piece without any actual measurements.
I agree. It would also be beneficial to have actual measurements of the designs instead of just simulations.
Everyone would like that, just as some here would like to read of LM317 denoiser success stories. Mind you, not everyone has equipment to measure such low levels of noise, output impedance or high PSRR.
Elvee has made thorough measurements and important fact is that real results were close to the simulations. It is a fair assumption that this is valid for many other designs.
If simulation predicts 140 dB PSRR for the LM317 dienoiser, I don’t care if it is 110 or 120 dB in reality. Anyway, it will be an overkill result.
Some recapitulation at first post would be most helpful to any new thread visitor, who are now lost and angry, unable to find anything useful among thousand posts. However, this can’t be done without OP and administrators’ involvement. Elvee is currently occupied with more important things than to deal with bunch of us - arguing, yammering, nitpicking posters (me included in that bunch
)
Things might have changed indeed. I need to better understand how the AK4468 operates from this point of view. Manufacturers might have implemented some extra tricks in the meantime. I'd be glad to give up the chip and extra design headaches if the loss is not that great with a digital volume control.
It is an easy measurement to do for yourself, provided you have some basic equipment. Take for example an AK4497 and measure two scenario's: 1. FS to DAC, with -40 dB internal volume control and 2. -40dB to DAC.
I did some of these measurements, and the internal volume control presents rather less distortion than attenuation by stripping bits.
My conclusion is twofold: use internal volume control on a modern DAC with confidence, and furthermore, this issue has no place in this thread, so let this be the last on it.