To answer your questions, in order:
1. In the near field, you can hear whatever it truely sounds like.
2. No. Only ever happened to me once that I couldn't stand a CD because of the ugly sound quality. I did like the music though. The best solution was to listen to the mp3, which is why I bought the CD in the first place.
3. They sound like they sound, at least the DAC doesn't sound bad. I probably like it the least if the music is bad, the recording is bad, and the reproduction system is bad.
4. I find that has to do with other things. I dislike listening to satellite radio in my car because the SQ is so very bad. I noticed just how bad when I ripped some CDs to mp3 to play in the car. MP3 and AAC sound much better.
5. People vary a lot in how they hear things and what they find objectionable. I am aware that some people claim modern oversampling DACs sound bad, etc. I have to wonder if they ever heard a good one, not just one they thought was supposed to be good but found they didn't like. I don't like most DACs myself, of any kind. As for you individually, I don't know. I do know that usually the best fix I have found for music recorded in such a way as to leave it with ugly digital artifacts which I do not happen to like hearing in great detail is to mp3 encode them (or AAC encode which is usually better sounding). What bitrate? The maximum bitrate that removes the part one does not like and leaves the most of what they do like. Also, good mp3's can sound pretty darn good with a good DAC, although not nearly as good as good higher resolution formats if the recording is good. An mp3 encoder, or AAC encoder, by the nature of how they encode automatically remove things that are not musical and that people don't find essential to the music.
Using a low quality DAC because one fears one's music collection might not be well recorded is probably nothing to worry about. Most recordings no matter what kind or what genre of music are mastered to sound a good as possible given the artist's or the record company's wishes. Usually they sound good on a good DAC.
Also, I presume you wanted a decent DAC or you would not have purchased the one you did. Probably, you will like it a whole lot better if it sounds better than it does now. Most people do and wish they could afford something better than they have because they have heard better at a store or somewhere.
Say, that makes me think. Maybe you could go a hi-fi store and ask to listen to their very finest DAC. If your favorite music sounds bad, so bad you have to stop listening, then just maybe a better DAC isn't for you. Actually, that might be the best advice anyone could give you, because it is the only way you could hear for yourself in advance to decide what you want to do.
1. In the near field, you can hear whatever it truely sounds like.
2. No. Only ever happened to me once that I couldn't stand a CD because of the ugly sound quality. I did like the music though. The best solution was to listen to the mp3, which is why I bought the CD in the first place.
3. They sound like they sound, at least the DAC doesn't sound bad. I probably like it the least if the music is bad, the recording is bad, and the reproduction system is bad.
4. I find that has to do with other things. I dislike listening to satellite radio in my car because the SQ is so very bad. I noticed just how bad when I ripped some CDs to mp3 to play in the car. MP3 and AAC sound much better.
5. People vary a lot in how they hear things and what they find objectionable. I am aware that some people claim modern oversampling DACs sound bad, etc. I have to wonder if they ever heard a good one, not just one they thought was supposed to be good but found they didn't like. I don't like most DACs myself, of any kind. As for you individually, I don't know. I do know that usually the best fix I have found for music recorded in such a way as to leave it with ugly digital artifacts which I do not happen to like hearing in great detail is to mp3 encode them (or AAC encode which is usually better sounding). What bitrate? The maximum bitrate that removes the part one does not like and leaves the most of what they do like. Also, good mp3's can sound pretty darn good with a good DAC, although not nearly as good as good higher resolution formats if the recording is good. An mp3 encoder, or AAC encoder, by the nature of how they encode automatically remove things that are not musical and that people don't find essential to the music.
Using a low quality DAC because one fears one's music collection might not be well recorded is probably nothing to worry about. Most recordings no matter what kind or what genre of music are mastered to sound a good as possible given the artist's or the record company's wishes. Usually they sound good on a good DAC.
Also, I presume you wanted a decent DAC or you would not have purchased the one you did. Probably, you will like it a whole lot better if it sounds better than it does now. Most people do and wish they could afford something better than they have because they have heard better at a store or somewhere.
Say, that makes me think. Maybe you could go a hi-fi store and ask to listen to their very finest DAC. If your favorite music sounds bad, so bad you have to stop listening, then just maybe a better DAC isn't for you. Actually, that might be the best advice anyone could give you, because it is the only way you could hear for yourself in advance to decide what you want to do.
Last edited:
Markw4, are you endorsing the ES9028Pro board that Mikett said "board closer to the DAC3" in Post #744?
Q8 Ver2.2 ES9028 DAC ES9028PRO HiFi Audio Decoder Finished in case
It seems to be very reasonably priced. The output opamps are 5534DD and 5532DD. Mikett is talking some mods to the board.
Last edited:
kellau, My opinion is that the stock board we are talking about has to my ears rather poor SQ. Also, there is in fact a DAC-3 and AHB2 amplifier here I have for making subjective comparisons of SQ. To be clear, the Chinese DAC I have is the green pcb marked version 1.06. You can also see which one it is from looking at the photos I have posted.
IMHO, if fully modded the Chinese ES9038Q2M can come much closer to sounding like a DAC-3 than it ever can by playing around with plugging in different opamps or using transformer I/V outputs. However, it is never going to sound quite as clean, accurate, good, etc., as an actual DAC-3. That is physically impossible for various reasons. For one thing DAC-3 uses an ES9028PRO chip, not Q2M.
(Why does DAC-3 use a 28 and not 38? Probably because the extra output current was not needed for stereo operation. As is the 8 channels are combined into groups of 4 which results in the final 2 stereo channels.)
Another reason this DAC will not be as good is because adding a Spartan-6 DSP chip to this project would be too over the top and too expensive. That being said, a fully optimized Q2M DAC can potentially give almost the same performance as a very well implemented PRO chip design except Q2M noise is greater, and specified distortion is a very small 2dB worse.
Obviously, to get that level of performance takes a lot of work. I estimate probably the need for something over $200 to be invested in order to get the best performance we are likely going to be able to get in this project. Exactly how good is that performance? I don't know yet, I'm not done with mine. It is still a work in progress at the moment.
Once I get the master mode I2S mod outcome settled one way or the other, I can take my DAC to somewhere nearby and get some good measurements. In the meantime, IMHO, with all the mods except master mode I2S it sounds very good to me. Better than any other DAC I have heard anywhere near what this one cost to make.
Of course, I do have some labor into it but maybe a lot of labor has been saved from others having to do it since I went first and have described and taken photos of what I have found to work well, and which make easily audible improvements to my ears.
I also have described at least one mod I did and would probably make again or do something similar although it did not have a big effect, which was putting the 100MHz clock on its own 3.3v supply.
As much as humanly possible I try to tell it like it is. If a mod helps to my ears and I am not satisfied with the SQ without that particular mod then I say so. So far the list of mods is pretty short but there can be some work involved for each one. If one does not enjoy that kind of activity then maybe better not to DIY this DAC. Don't know what else to say.
For people who have more money than time but still looking for a bargain, a used DAC-1 might be a good choice. Although some people think DAC-1 sounds the same as DAC-3 I am not among them. They sound slightly different to me with DAC-3 being slightly better, slightly more detailed, and slightly smoother all at the same time. That opinion was formed after listening to the DAC-1 for several years. When the DAC-3 was first swapped in it was not hard to hear the difference. However, long term listening combined with abrupt change may be the most sensitive way to notice any difference. In short term listening it would probably be much harder. Anyway, maybe a used DAC-1 could be an attractive possibility for someone: used benchmark dac1 - Google Search
IMHO, if fully modded the Chinese ES9038Q2M can come much closer to sounding like a DAC-3 than it ever can by playing around with plugging in different opamps or using transformer I/V outputs. However, it is never going to sound quite as clean, accurate, good, etc., as an actual DAC-3. That is physically impossible for various reasons. For one thing DAC-3 uses an ES9028PRO chip, not Q2M.
(Why does DAC-3 use a 28 and not 38? Probably because the extra output current was not needed for stereo operation. As is the 8 channels are combined into groups of 4 which results in the final 2 stereo channels.)
Another reason this DAC will not be as good is because adding a Spartan-6 DSP chip to this project would be too over the top and too expensive. That being said, a fully optimized Q2M DAC can potentially give almost the same performance as a very well implemented PRO chip design except Q2M noise is greater, and specified distortion is a very small 2dB worse.
Obviously, to get that level of performance takes a lot of work. I estimate probably the need for something over $200 to be invested in order to get the best performance we are likely going to be able to get in this project. Exactly how good is that performance? I don't know yet, I'm not done with mine. It is still a work in progress at the moment.
Once I get the master mode I2S mod outcome settled one way or the other, I can take my DAC to somewhere nearby and get some good measurements. In the meantime, IMHO, with all the mods except master mode I2S it sounds very good to me. Better than any other DAC I have heard anywhere near what this one cost to make.
Of course, I do have some labor into it but maybe a lot of labor has been saved from others having to do it since I went first and have described and taken photos of what I have found to work well, and which make easily audible improvements to my ears.
I also have described at least one mod I did and would probably make again or do something similar although it did not have a big effect, which was putting the 100MHz clock on its own 3.3v supply.
As much as humanly possible I try to tell it like it is. If a mod helps to my ears and I am not satisfied with the SQ without that particular mod then I say so. So far the list of mods is pretty short but there can be some work involved for each one. If one does not enjoy that kind of activity then maybe better not to DIY this DAC. Don't know what else to say.
For people who have more money than time but still looking for a bargain, a used DAC-1 might be a good choice. Although some people think DAC-1 sounds the same as DAC-3 I am not among them. They sound slightly different to me with DAC-3 being slightly better, slightly more detailed, and slightly smoother all at the same time. That opinion was formed after listening to the DAC-1 for several years. When the DAC-3 was first swapped in it was not hard to hear the difference. However, long term listening combined with abrupt change may be the most sensitive way to notice any difference. In short term listening it would probably be much harder. Anyway, maybe a used DAC-1 could be an attractive possibility for someone: used benchmark dac1 - Google Search
Last edited:
I see that the used Benchmark DAC1 are available at about the same price as the new Sabre ES9028PRO/ES9038PRO boxes. However, I find it hard to believe a 10 year old Analog Devices AD1853 based DAC with NE5532 output buffer can sound close to a Benchmark DAC3. Yes, I know you don't believe that either.
My question is very simple. The ES9028PRO board that Mikett mentioned already has I-V buffer and regulated AVCC. It costs more than then ES9038Q2M with regulated PS. Is it a better starting point for a DIY DAC? I am talking about a beginner DIY DAC that can be further upgraded in the future. Not the ultimate DAC you are developing.
Last edited:
Not sure where the AD1853 comes from. Benchmark DAC-1 used an ES9018. When you say a used DAC-1 is about the same price as a new Sabre PRO box, I have to ask which boxes? Because there is a very good reason for used DAC-1 prices. The still sound better than most of the new boards because of the superior implementation. There are some application notes at Benchmark that explain some of the particulars of their technology. Not the same at all as most even new SABRE implementations. For example: Asynchronous Upsampling to 110 kHz - Benchmark Media Systems, Inc.
Regarding the ES9028 board you mention with regulated AVCC and I/V output, how well are those things implemented. I regulate my AVCC with an opamp with .00003% rated distortion performance Good enough for to meet ESS recommendations. How accurate is the AVCC regulation on the 9028 board and does it meet ESS specs? I use thin film resistors and COG caps for the IV and differential stages for ultra-low distortion, as ESS recommends (and stresses the importance of). How about the 9028 board? If anything less on that board and you want to meet ESS specifications to try and hit that -120dB distortion level, then you may have to rip out or not use parts that came with the 9028 board and mod those things. Then the 'golden' oscillator I have seen in some of the Chinese DAC photos and as an individual item on ebay is what is known in the trade as a 'standard oscillator,' same as what comes with the Q2M board. Despite possible appearances gold color and large size are not indicators of oscillator quality but they would like naive buyers to think so. In addition, so far as I know nobody has measured one of those as is, or has done comparative listening tests with something more or less up to Benchmark SQ level (any model Benchmark). From looking at what the Chinese designers choose to put in their more expensive boards the designs are still not optimized for sound quality and will need modification to sound like people probably expect they will be getting when them buy something that says SABRE DAC in the name.
In particular and with respect to that exact ES9028 board, there is no way for me to tell you exactly how it sounds, the quality of some of the components I can't see clearly in the pictures or from the markings, or what mods it would need to sound like a SABRE DAC can and should in terms of SQ. I would have to have one right here in front of me for awhile to do any of that. The only thing I can tell from the pictures and what schematics I have seen is that it probably would need a lot of work. I can't post a link here, but if you search around you might be able to find a service manual somewhere on the net for the original M-DAC. That board is pretty old and superseded by newer products, but if you look at the schematics in the back few pages of the service manual to see what they had to do with power supplies around the ES9018 chip to make it sound really good and you compare that to the power supplies in the Chinese 9028 boards and schematics they show, you might get some idea of the task you have in front of you. It's BIG.
Regarding the ES9028 board you mention with regulated AVCC and I/V output, how well are those things implemented. I regulate my AVCC with an opamp with .00003% rated distortion performance Good enough for to meet ESS recommendations. How accurate is the AVCC regulation on the 9028 board and does it meet ESS specs? I use thin film resistors and COG caps for the IV and differential stages for ultra-low distortion, as ESS recommends (and stresses the importance of). How about the 9028 board? If anything less on that board and you want to meet ESS specifications to try and hit that -120dB distortion level, then you may have to rip out or not use parts that came with the 9028 board and mod those things. Then the 'golden' oscillator I have seen in some of the Chinese DAC photos and as an individual item on ebay is what is known in the trade as a 'standard oscillator,' same as what comes with the Q2M board. Despite possible appearances gold color and large size are not indicators of oscillator quality but they would like naive buyers to think so. In addition, so far as I know nobody has measured one of those as is, or has done comparative listening tests with something more or less up to Benchmark SQ level (any model Benchmark). From looking at what the Chinese designers choose to put in their more expensive boards the designs are still not optimized for sound quality and will need modification to sound like people probably expect they will be getting when them buy something that says SABRE DAC in the name.
In particular and with respect to that exact ES9028 board, there is no way for me to tell you exactly how it sounds, the quality of some of the components I can't see clearly in the pictures or from the markings, or what mods it would need to sound like a SABRE DAC can and should in terms of SQ. I would have to have one right here in front of me for awhile to do any of that. The only thing I can tell from the pictures and what schematics I have seen is that it probably would need a lot of work. I can't post a link here, but if you search around you might be able to find a service manual somewhere on the net for the original M-DAC. That board is pretty old and superseded by newer products, but if you look at the schematics in the back few pages of the service manual to see what they had to do with power supplies around the ES9018 chip to make it sound really good and you compare that to the power supplies in the Chinese 9028 boards and schematics they show, you might get some idea of the task you have in front of you. It's BIG.
Last edited:
I got my information from the John Atkinson's review in Stereophile dated 2008. Are we talking about the same unit? I am not sure the ES9018 was released yet 10 years ago.
Benchmark DAC1 USB D/A processor & headphone amplifier | Stereophile.com
I found this used Benchmark DAC-1 at Audiogon. Wrong unit??JohnAtkinson said:
Last edited:
As I mentioned before I would get either a Topping D50 (dual ES9038Q2m) or a Hifimediy ES9038PRO before getting something like that and throwing even more money on mods. Topping and Hifime at least have a history and reputation for decent budget dacs. I have both right now and they sound very, VERY good. The Hifime in particular.
I have done my fair share of cheap ebay board modding in the past, but the lower price point of these new readily available units makes that option less and less viable, except as an educational exercise.
The Topping D50 has had some basic measurements done here. Not bad for $250 and no mods.
Alas the information that would be most interesting (to me anyway), is the stuff Mark can no longer talk about!
No, right unit. You got me there, I was mistaken about the DAC chip. I won't forget now though. Doh! Anyway, it is good sounding unit. If you read the Stereophile review of Benchmark DAC-3, on the second page down near the bottom it says they compared DAC-1 directly to DAC-3 and really couldn't be sure they could hear any difference: Benchmark DAC3 HGC D/A preamplifier-headphone amplifier Page 2 | Stereophile.com
As I said I think there is a difference, but if you read the Benchmark app note I linked to you will see how DAC-1 differs from typical implementations, even today. The high resale value for old units is for good reason, they are worth that much even old and used based on their SQ.
What people don't seem to get, and cdsgames has stressed this as well, is it isn't the chip that makes the SQ good nearly so much as it is the implementation. It is a lot of little details not one, or maybe a few, big-name-brand parts.
BTW, DAC-1 does sort of sound like it has 5532's in it, and DAC-3 sounds like it has what is has. I know some, probably the vast majority of people around DIYAudio probably find it ridiculous to think opamps can have an audible sound at least in well designed circuits. I don't want to get into any kind of argument about it. If there is an audible difference in this case it is very subtle, and most people probably would not hear any difference. To me, perceptually speaking, the mental experience is more that of a very fine texture sitting on top of the sound, ever-so-slightly rough like ultra-fine 2000 grit sandpaper with 5532 compared to smoother with 49720. Is it real or imagined? Can't say for sure, it is too much to accept without careful blind testing. I will withhold judgement until then.
Last edited:
Most likely this will be the final update on 'master mode I2s.' Still haven't head anything back regarding my datasheet question, but some experiments with settings and various measurements lead me to believe that running the DAC in 128_fs master mode and the SRC in slave mode is not going to work with either I2S or left-justified PCM formats. Why? I can't tell you. If I subsequently find out otherwise then I will return to this pursuit later.
For now I think I will move on to some further testing of those things that do work.
For now I think I will move on to some further testing of those things that do work.
You know the reason that many of us hanging around this DIY forum. In my experience, a DIY box most likely costs more than a retail offer. But the satisfaction of going through the process is priceless. I am learning a lot from Markw4 and others about what makes a DAC sing. But my goal is not the ultimate DAC.
My question is very simple. The ES9028PRO board that Mikett mentioned already has I-V buffer and regulated AVCC. It just need a power transformer. It costs more than then ES9038Q2M with regulated PS plus transformer. Is it a better starting point for a DIY DAC?
Last edited:
Sorry to hear about master mode Mark! I was really hoping for that to work.
I've now desided to use a LT3045 for each side of the dac chip(left/right). I plan to mount them underneath the board, and pull the + pin up trough the existing electrolytic pin holes. Found some measurements of this excact module and it was very good! LT3045 Low Noise RF Audio DAC ADC Linear Regulated Power Supply Module 15V 500mA | eBay.
The +pin distance should be 10-15mm from the chip so I hope the impedance will be extremely low between the chip and regulator.
I have a few questions regarding the bypass caps.. At what frequency do the dac chip draw current? I'm very aware of the impedance of pcb traces and capacitor leads at high frequency. I want good decoupling, but I wan't to avoid cap ringing between the large and small caps. I don't have any measurement tools to detect ringing(don't remember the correct term) so I hope someone could give me an educated guess on some cap values that would work. Do the cap operate at 100Mhz since that is the clock or does is draw current at audio frequencies?
Btw, from reading on Wiki I saw that 1nF ceramic had low inductance at 100Mhz.. Are smd film caps better than ceramic in bypassing because of less dielectric absorption?
From skimming trough the http://www.analog.com/media/en/technical-documentation/data-sheets/30451fa.pdf I got the impression that I don't need much more than 20-30uF bypassing capasitance?? Have any seen measurements on the LT3045 with large caps? Is has very good PSRR down to 10Hz so that makes me question the need for a larger capacitor. I've read that regulators might not like huge capacitors..
I've now desided to use a LT3045 for each side of the dac chip(left/right). I plan to mount them underneath the board, and pull the + pin up trough the existing electrolytic pin holes. Found some measurements of this excact module and it was very good! LT3045 Low Noise RF Audio DAC ADC Linear Regulated Power Supply Module 15V 500mA | eBay.
The +pin distance should be 10-15mm from the chip so I hope the impedance will be extremely low between the chip and regulator.
I have a few questions regarding the bypass caps.. At what frequency do the dac chip draw current? I'm very aware of the impedance of pcb traces and capacitor leads at high frequency. I want good decoupling, but I wan't to avoid cap ringing between the large and small caps. I don't have any measurement tools to detect ringing(don't remember the correct term) so I hope someone could give me an educated guess on some cap values that would work. Do the cap operate at 100Mhz since that is the clock or does is draw current at audio frequencies?
Btw, from reading on Wiki I saw that 1nF ceramic had low inductance at 100Mhz.. Are smd film caps better than ceramic in bypassing because of less dielectric absorption?
From skimming trough the http://www.analog.com/media/en/technical-documentation/data-sheets/30451fa.pdf I got the impression that I don't need much more than 20-30uF bypassing capasitance?? Have any seen measurements on the LT3045 with large caps? Is has very good PSRR down to 10Hz so that makes me question the need for a larger capacitor. I've read that regulators might not like huge capacitors..
What scares me about those LT3045 regulators on ebay is shown in one of their pictures on this page: LT3045 Low Noise RF Audio DAC ADC Linear Regulated Power Supply Module 15V 500mA | eBay
There is an LF noise corner that comes up pretty high at low frequencies. According the the LT3045 datasheet that can be minimized by making Cset the largest recommended value of 22uf. Don't know what is used on that module.
I will post the picture below. If that is the noise spectrum of what you will be getting I sure wouldn't use it.
On the other hand, wouldn't be too surprising if the ESS recommended opamp circuit is just as good at high audio frequencies as the LT3045 and any noise corner may be a lot lower with the filtering ESS recommends and with other proper attention to details such as layout and opamp rail power quality.
Sorry, but to answer your question, I have not measured the current spectrum but I would expect it probably goes up to at least the highest frequency audio you will be playing back, possibly higher due to switching transients or whatever.
Regarding your question about using a larger output cap with LT3045, the datasheet says larger than 10uf doesn't help much.
There is an LF noise corner that comes up pretty high at low frequencies. According the the LT3045 datasheet that can be minimized by making Cset the largest recommended value of 22uf. Don't know what is used on that module.
I will post the picture below. If that is the noise spectrum of what you will be getting I sure wouldn't use it.
On the other hand, wouldn't be too surprising if the ESS recommended opamp circuit is just as good at high audio frequencies as the LT3045 and any noise corner may be a lot lower with the filtering ESS recommends and with other proper attention to details such as layout and opamp rail power quality.
Sorry, but to answer your question, I have not measured the current spectrum but I would expect it probably goes up to at least the highest frequency audio you will be playing back, possibly higher due to switching transients or whatever.
Regarding your question about using a larger output cap with LT3045, the datasheet says larger than 10uf doesn't help much.
Attachments
Last edited:
The Chinese part may well be genuine although the overall implementation might not be as good as other higher-cost modules, we don't know at this point. A problem is if using a pre-built module it looks like you can't easily choose Cset unless you know how to do SMD rework. Not that another value for Cset would necessarily make it a good choice for AVCC.
LT3045 may be fine for regulating opamp rails because opamps typically tend to have better PSRR at lower frequencies. However, AVCC has zero PSRR at any frequency that we know of. Therefore LT3045 is probably not a suitable choice for that particular application, in particular due to its LF noise corner. Ideally, we would like the noise down at -120dB or lower all the way down do DC. Since we are not equipped to measure and evaluate the effects of AVCC power quality imperfections, certainly not within the scope of this one-off DIY project, the wisest choice is probably to do as ESS recommends.
LT3045 may be fine for regulating opamp rails because opamps typically tend to have better PSRR at lower frequencies. However, AVCC has zero PSRR at any frequency that we know of. Therefore LT3045 is probably not a suitable choice for that particular application, in particular due to its LF noise corner. Ideally, we would like the noise down at -120dB or lower all the way down do DC. Since we are not equipped to measure and evaluate the effects of AVCC power quality imperfections, certainly not within the scope of this one-off DIY project, the wisest choice is probably to do as ESS recommends.
Last edited:
Ok, thanks for the feedback! That graph above shows 500mAh current draw and I think someone mentioned earlier that the dac chips draws only 130-140mAh in total. So with 70mAh on each regulator the caps should have an easier job at low frequencies.
I hope someone can anwer the current draw frequency of the dac since I feel its essential to creating a good psu solution for the dac.
I followed Gmarsh decoupling technique for the Wiener 3116 in its own thread. Really informative! That chip amp which operates at 400kHz-1.2mHz. The right type of caps and placement is crusial at high frequencies. This bit of information is important for those who is chasing perfection
I assume this dac will draw current at frequencies comparable to other dacs.. Googling this matter surprisingly didn't give me an answer.
I hope someone can anwer the current draw frequency of the dac since I feel its essential to creating a good psu solution for the dac.
I followed Gmarsh decoupling technique for the Wiener 3116 in its own thread. Really informative! That chip amp which operates at 400kHz-1.2mHz. The right type of caps and placement is crusial at high frequencies. This bit of information is important for those who is chasing perfection
I assume this dac will draw current at frequencies comparable to other dacs.. Googling this matter surprisingly didn't give me an answer.Ultra low noise DC power supply
Ultra-low Noise Adjustable +/-15V DC Voltage Regulator Module, LT1963A LT3015. 6921407430679 | eBay
How sensitive is the ES9038Q2M to the +-15 volt power supply noise level? This 2-stage, LT1963A/LT3015 based supply claims a noise spec at 1/10 of the 317/337 based supply. Is it good and worth the extra cost?
Ultra-low Noise Adjustable +/-15V DC Voltage Regulator Module, LT1963A LT3015. 6921407430679 | eBay
Last edited:
The 15 volt rail gets further sliced and diced for the critical dac voltages, and should only be directly supplying the opamps, which both have excellent psrr and are being run at low gain, so do one better in rejecting rail noise.
In other words, upgrading this regulator would objectively improve the 15v rail, but practically do nothing to affect output. It's not a sensitive section of the circuit, so I'd leave it alone, especially to start.
In other words, upgrading this regulator would objectively improve the 15v rail, but practically do nothing to affect output. It's not a sensitive section of the circuit, so I'd leave it alone, especially to start.
That looks like it might be a good PS. If you get one you might want to look at the part numbers on the diodes and make sure they are fast-turnoff or ultra-fast turnoff types.
In particular, probably not a good idea to use a switcher upstream of the analog regulator.
Also, appropriate bypassing at the power pins at each IC to a low impedance ground plane is still needed even if incoming power quality is good.
Please also recall that the 3.3v regulator on the DAC board has poor load regulation, and both analog and digital devices are powered by that one 3.3v bus. Not so good. Many people have replaced the regulator with a better one. I chose to use separate 3.3v regulators for analog, digital, and clock (which is sort of both analog and digital). What the heck, they are cheap and easy insurance since we do not have a lot of test equipment to measure the minimum treatment needed, or at least we will assume we don't have that for this one-off DIY project.
Regarding whether or not to upgrade +-15 regulation, what are you using now? If it is clean 317/337 and snubbers or fast turnoff diodes are used it might very well be okay. The snubbers or fast diodes are just to prevent HF spikes from being generated by transformer parasitics. It is usually better to avoid making that stuff in the first place rather than trying to get rid of it later, at least for high performance circuits like very high SQ DACs.
Oops! Looks like we cross posted again. I am generally in agreement with what DPH said, my point was if you want to get a good regulator, the one you linked to looks very good. Very clean power is important, but it is HF hash that seems to cause more trouble than line regulation or LF ripple regulation at least for the +-15v power, and that is because, as DPH stated opamps have good PSRR at LF.
Last edited:
I have an ordinary +-15 regulator using the 7815/7915 IC pair and a KBL401 bridge rectifier. It has 2200 uF of capacitance on each rail.
It is an old kit from my bench storage and typical for audio opamp projects.