How good is your original digital source? Is it possible it sounds better with DSD and SRC due to a Jitter reduction by this Ak4137 board?
Did you try output same format to see if already an improvement? Or will it just pass the signal through as it is?
I am using a very good digital SPDIF source ( G-Sonos) so I wonder if there would be an improvement with the SRC board..
Did you try output same format to see if already an improvement? Or will it just pass the signal through as it is?
I am using a very good digital SPDIF source ( G-Sonos) so I wonder if there would be an improvement with the SRC board..
Probably there would be an improvement. Mikett said upsampling with SRC4392 made his PRO dac board sound better. For AK4137, it might have to do with how interpolation filtering is done for PCM but not nearly so much for DSD. It may to do with the internal oversampling ratio in the for DSD vs PCM that the chip uses. Don't know, but there are a number of possible reasons aside from jitter.
Tell you what, if you buy one and it doesn't help, I will refund what you paid by paypal or the method of your choice if you can send the AK4137 board to someone from this thread working on dac modding who might not be able to afford it or someone else that could use it. I will pay shipping for that. So, no risk to you at all.
Last edited:
Which pro chip are we talking about? Is is ES9028PRO or ES9038PRO, or an earlier one?
What other mods have been done to the dac, any? Which board is the PRO chip on?
I am asking the questions because Mikett reported that upsampling with SRC4392 helped his PRO chip board. He didn't have an AK4137 when he talked about it. Don't know if he has one now or not.
Also, everything in the data sheets for Q2M and PRO indicate they both treat PCM the same and treat DSD the same. There are different interpolation filters and DPLL bandwith settings for DSD vs PCM. DSD seems to be handled natively without conversion to PCM in both Q2M and PRO, so I am very curious about whatever is going on.
Last edited:
i was talking about 1kHz test tone THD using I2S for the unmod boards. so this is an exact technical measurement. another matter is what one hears.
no idea what internal processing for PCM Sabre PRO chips have, but obviously doing any kind of pre-processing will result in some audible differences, whatever the board is, and it is a matter of personal preferences. perhaps we need to distinguish between "it is better" and "I like it better".
Markw4,
Thank you very much for this offer - but frankly, I don´t see why you should pay for MY risk in improving MY DAC and instead also do my part of giving experiences to this Forum such as you already did a lot!
The Thing is, I have changed my clock to a NDK 50mHz clock already. So I don´t know how high I can do upsampling.
Same time I changed the clock I also changed the 47uF AVCC supply caps from electrolytics to polymer. I realized a more pronounced bass but also a less dynamic, less colored sound-Performance (not sure if this coloration was a Kind of distortion but it sounded more live-like before). So I wonder if the change was caused by the polymer cap or by the clock with its higher rising time or by lower clocking frequency which reduces AVCC power consumption and though might also have an Impact. I am very coriuos on your findings with film caps though..
I also have the buffer (schmitt Trigger) here recommended by cdsgames to reduce rise time but they are so tiny I was afraid to apply it to my circuit between clock and DAC where the resistor is in between. As this is no analog circuit and we have 50mHz I guess every Corner in the Signal path matters. Would you place it on top of the board or on bottom side under the copper plane shield, but with longer traces?
Thank you very much for this offer - but frankly, I don´t see why you should pay for MY risk in improving MY DAC and instead also do my part of giving experiences to this Forum such as you already did a lot!
The Thing is, I have changed my clock to a NDK 50mHz clock already. So I don´t know how high I can do upsampling.
Same time I changed the clock I also changed the 47uF AVCC supply caps from electrolytics to polymer. I realized a more pronounced bass but also a less dynamic, less colored sound-Performance (not sure if this coloration was a Kind of distortion but it sounded more live-like before). So I wonder if the change was caused by the polymer cap or by the clock with its higher rising time or by lower clocking frequency which reduces AVCC power consumption and though might also have an Impact. I am very coriuos on your findings with film caps though..
I also have the buffer (schmitt Trigger) here recommended by cdsgames to reduce rise time but they are so tiny I was afraid to apply it to my circuit between clock and DAC where the resistor is in between. As this is no analog circuit and we have 50mHz I guess every Corner in the Signal path matters. Would you place it on top of the board or on bottom side under the copper plane shield, but with longer traces?
Last edited:
Freezebox.
Is the NDK clock an "A" suffex part which indicates ultra low phase noise? If not, it might have ultra low phase noise or not. Prior to NDK offering the A parts, low phase noise was not a guaranteed parameter. Many people wanting to using them for low phase noise would buy a few because there were a number of random duds in a reel. They could meet published NDK specs, but were not low enough phase noise for audio use.
If you are sure you have a good clock then 50Mhz is probably still high enough since the dacs are capable of playing DSD 512 with a 100MHz clock. 11.2MHz is DSD 256, IIRC. I think it AK4137 should work for you. I just hope you have a good clock.
The clock buffers you have are the same size as TLC6655. I would try to mount in on the back of the ground plane underneath the clock next to the dac maybe on an adapter board unless enough pins could be soldered to ground to support it. It is awfully small to hand solder. Then you would need power for it. You could drill a couple of holes for clock in and clock out wires. If you can do it then you probably can do any SMD I have done. That LTC6655 I mounted on my 1st dac board ground plane was very very tiny for me to hand solder that way. As hard as anything I have done.
Or, you know a Crystek clock is only about $30 here including shipping (at least it is here). Now that we are standardizing more and experimenting less since we learned a lot about how to make a pretty good dac, I would be inclined just to get a new clock if you can afford the expense. After all this time and effort I would hate to see anyone end up with a dac that doesn't perform like we would want it to.
Is the NDK clock an "A" suffex part which indicates ultra low phase noise? If not, it might have ultra low phase noise or not. Prior to NDK offering the A parts, low phase noise was not a guaranteed parameter. Many people wanting to using them for low phase noise would buy a few because there were a number of random duds in a reel. They could meet published NDK specs, but were not low enough phase noise for audio use.
If you are sure you have a good clock then 50Mhz is probably still high enough since the dacs are capable of playing DSD 512 with a 100MHz clock. 11.2MHz is DSD 256, IIRC. I think it AK4137 should work for you. I just hope you have a good clock.
The clock buffers you have are the same size as TLC6655. I would try to mount in on the back of the ground plane underneath the clock next to the dac maybe on an adapter board unless enough pins could be soldered to ground to support it. It is awfully small to hand solder. Then you would need power for it. You could drill a couple of holes for clock in and clock out wires. If you can do it then you probably can do any SMD I have done. That LTC6655 I mounted on my 1st dac board ground plane was very very tiny for me to hand solder that way. As hard as anything I have done.
Or, you know a Crystek clock is only about $30 here including shipping (at least it is here). Now that we are standardizing more and experimenting less since we learned a lot about how to make a pretty good dac, I would be inclined just to get a new clock if you can afford the expense. After all this time and effort I would hate to see anyone end up with a dac that doesn't perform like we would want it to.
Hmm. Odd that they measure differently. Can you hear a difference with music at all just to double check? For my modded dac, the AK4137 makes it sound more like DAC-3, but it also sounds like it is reducing distortion too. If it is audible, it should show up on some test. A static 1kHz tone might not be a tough enough test though, not sure. May depend on how low distortion is to begin with. If a dac board is mostly unmodded, it may have enough distortion to swamp the difference.
Maybe an intermodulation distortion test would be more revealing? The only thing that should be different about a PRO dac would be that the output current is higher because the output resistance is lower. That makes the output noise lower, but otherwise distortion and DSD processing should be the same as with a Q2M with one exception that I doubt applies in your case. With PRO chips, at least with 9038PRO it is possible to change the interpolation filter settings so as to disable that filtering for all but two channels, and use the freed up resources to increase the number of PCM filter taps available. That might or might not make PCM closer to DSD in terms of distortion, although I don't see any specs saying that. Also, I think you would have to load your own coefficients for that to work. It is only intended if all 8 channels are being combined to run in two groups of 4 channels each. The channels are assumed to be paralleled to reduce noise for high end low noise stereo use.
Another possible test might be the jitter J-test which I think can be run with the free version of Arta. There is no analysis feature, but it can create the test signal. Something might show up with an fft showing some spurs, not sure how much haven't really dug into it.
Also, this discussion brings me back to thinking about the post about AD797 having less distortion that other opamps. If using an ES9038PRO, it produces too much current per channel to use LME49720 opamps for IV conversion. They would probably clip due to current limiting on the peaks. The recommended opamps would be OPA1611 or OPA1612 in that case.
Last edited:
yes, mobile and pro chips sound different. there might be several reasons for that, but I like DSD256 on both of them. on the other hand, despite lower THD, if I look at the 1kHz sinusoid right after I/V, then PCM delivers a pretty clean signal, in contrast, DSD has clear HF noise in it. regarding opamps, very good results have been reported with BUF3/transistor boosted AD744.
Actually in my latest listening sessions, I have found that there is a tradeoff. With the XMOS feeding the DAC, I prefer that combination over the SRC4392 now. One can easily hear that there is a difference between the XMOS and the SRC4392. Which is preferred comes down to personal preferences. If you have a history of listening to vinyl, the XMOS sounds better. The XMOS imparts some high end distortion but it does open up the soundstage with a little bit more with a more forward sounding character. That they sound different is undeniable....and using the same DAC and source except the SRC does some processing as well as the XMOS. If you've lost some HF hearing, you might well prefer the SRC 4392. This was done powering the SRC4392 off a 317 regulator for 5V. I will eventually try again this on possibly an LT3042 if circumstances permit and as the mods proceed over the winter. But the revelation is that with the same equipment, just how the file is presented, the difference in sound is quite large well....if you are actually listening carefully.
I thought we agreed SRC4392 added some distortion as used, and there was a trade off in terms of sound quality. The situation with AK4137 and conversion to DSD is different. Yes, DSD has has a higher noise floor than PCM. However, running ES9038Q2M in ASRC mode with a 100MHz clock sounds better and less distorted with upsampling to DSD than without it. This is assuming all other mods have been done to the dac boards so they are as clean and as accurate as possible before adding AK4137.
I think I have always said that DAC-3 runs with ASRC on, but with a 30MHz clock on a ES9028PRO and a 27MHz clock on SRC4392. They say they do that allow for a wider transition band for interpolation filtering. Between clocking differences and whatever filtering and or other processing in Spartan 6 it works really well.
We could buy a Spartan 6, but if I am not mistaken we can't buy those ultra low jitter clocks without buying a whole reel of each type. Obviously not happening in that case.
Also, Crane Song Solaris appears to be using the same clocking scheme as DAC-3, but they are using some lower jitter clocks that are hard make and there have been manufacturing delays.
I have looked at all the low jitter clocks available from the usual sources and they only carry the most popular frequencies for audio.
In addition, John Siau at some point straight out said that no dac chip on the market has good enough internal interpolation filters. The default for Sabre dacs is 128 taps with no calculation headroom to allow for intersample overs. It is possible with some PRO chips to get 256 taps, but using some ASIC someone could possibly get 2,048 taps at 32-bit float, or more.
The stuff with unobtainium clocks and fancy filters is beyond the scope of playing around with ES9038Q2M chips in a DIY context where we don't have a well equipped lab full of test equipment and manufacturers working to get our clock business with free samples and maybe custom development.
In addition, in the DIY context people still want to be able to play back and have very high sample rate DSD and PCM in the belief that it will offer improved sound quality. It doesn't in reality, if dacs are actually designed the best they can be. However, very few dacs are designed that way. DAC-3 is still considered state of the art for now and is one of a small number of dacs on Stereophile's recommended list. That we can come pretty close to that sound quality with off the shelf components and nothing more than a DVM and a scope is quite good.
Lastly, I would like to say it seems doubtful that PRO Sabre chips sound different from mobile except in terms of noise. What cannot help but sound different is the necessarily different surrounding circuitry such as output stage amplifiers and AVCC power supplies. LME49720's don't work for other than mobile chips since they can't handle high current of PRO chips. However, they and LME49710 remain the lowest distortion opamps one can buy, IMHO. Anything different is going to sound different, so let's everyone try to avoid mis-attribution as to causation.
I think I have always said that DAC-3 runs with ASRC on, but with a 30MHz clock on a ES9028PRO and a 27MHz clock on SRC4392. They say they do that allow for a wider transition band for interpolation filtering. Between clocking differences and whatever filtering and or other processing in Spartan 6 it works really well.
We could buy a Spartan 6, but if I am not mistaken we can't buy those ultra low jitter clocks without buying a whole reel of each type. Obviously not happening in that case.
Also, Crane Song Solaris appears to be using the same clocking scheme as DAC-3, but they are using some lower jitter clocks that are hard make and there have been manufacturing delays.
I have looked at all the low jitter clocks available from the usual sources and they only carry the most popular frequencies for audio.
In addition, John Siau at some point straight out said that no dac chip on the market has good enough internal interpolation filters. The default for Sabre dacs is 128 taps with no calculation headroom to allow for intersample overs. It is possible with some PRO chips to get 256 taps, but using some ASIC someone could possibly get 2,048 taps at 32-bit float, or more.
The stuff with unobtainium clocks and fancy filters is beyond the scope of playing around with ES9038Q2M chips in a DIY context where we don't have a well equipped lab full of test equipment and manufacturers working to get our clock business with free samples and maybe custom development.
In addition, in the DIY context people still want to be able to play back and have very high sample rate DSD and PCM in the belief that it will offer improved sound quality. It doesn't in reality, if dacs are actually designed the best they can be. However, very few dacs are designed that way. DAC-3 is still considered state of the art for now and is one of a small number of dacs on Stereophile's recommended list. That we can come pretty close to that sound quality with off the shelf components and nothing more than a DVM and a scope is quite good.
Lastly, I would like to say it seems doubtful that PRO Sabre chips sound different from mobile except in terms of noise. What cannot help but sound different is the necessarily different surrounding circuitry such as output stage amplifiers and AVCC power supplies. LME49720's don't work for other than mobile chips since they can't handle high current of PRO chips. However, they and LME49710 remain the lowest distortion opamps one can buy, IMHO. Anything different is going to sound different, so let's everyone try to avoid mis-attribution as to causation.
Last edited:
Who disagreed? except that the outboard SRC4392 is sounding very different with the same original file. Is the XMOS better or the SRC4392. I really don't know except what I hear in MY implementation. At this time in the state it is, I am prefer the XMOS to the SRC4392. But in Benchmarls implenetation both are inclusive of both. In my setup it is Either Or. and in that situation it could be a nice match.
What I have meant to do but for lack of time right now is to hook up my old Sonic Frontiers Jitterbug to the SRC4392 and see what comes out of that combination.
What I have meant to do but for lack of time right now is to hook up my old Sonic Frontiers Jitterbug to the SRC4392 and see what comes out of that combination.
I am also against hardware and software upsampling ... or convert to dsd. At studios 100Keur equipment or more try to deal with digital signal... i have studio master recordings recorded with different gear different software ...everything sonds different.
It is time that some of you try to use some serious gear and turntable as comparisson not just one dac and trim es9038q2m in that way. it is fine that you reveal how dac can be tuned into one sound direction. If that direction suits you fine if not is also nothing wrong.
Keep on tuning.
It is time that some of you try to use some serious gear and turntable as comparisson not just one dac and trim es9038q2m in that way. it is fine that you reveal how dac can be tuned into one sound direction. If that direction suits you fine if not is also nothing wrong.
Keep on tuning.
turntable
What is the price of a good turntable ?
What is the price of a good arm ?
What is the price of a good cartridge ?
What is the price of a good preamp ?
Too much for me
Serge
There is only one direction to work on improving a DAC, IMHO. That is towards accurately reproducing what is on a CD or other recording. The bits are there and they have one specific meaning only.
Also, comparing digital recordings to phonograph records won't work to find out what a CD should sound like. Records and CDs are mastered separately, often by different people using different equipment. Also, records are molded from stampers. When a stamper wears out a new one must be made by a mastering engineer. The next stamper version may sound different than the last one. Often different stampers are made by different mastering engineers using different equipment.
Anyway, my primary goal is to make a dac that is as accurate and as low distortion as possible given the resources available to do it. Why? Because very good dacs are expensive and most people can't afford one or justify the cost of one. It was painful for me to spend that much on a mere dac. I want people to have the best dacs possible at a cost as low as possible. To me, best means both accuracy at playing back the bits, and sound good in a musical way, no unpleasant distortion to the extent possible. So far, I think we are beating everything out there under $1,000 and some I am seeing lately from China costing more than $1,000. That is my opinion, and over time we will see how accurate it turns out to be. As more people fully mod dacs and compare the results with other dacs they and their friends can afford, I think they will be very happy with the results. I do stress however, that getting best results which are still not perfect requires careful attention to completing all mods properly.
I sure hope people will give it a try before they think up reasons why not to bother. Please folks, as you work on modding your dacs we would love to hear from you and see pics of your project. You may help motivate others to do the same as you and end up with something good they couldn't have otherwise.
I do want to end this post though by pointing out something that is part of the reality of what we are working on. It is true, IMHO, that different dacs, say, for example, on the Sterephile recommended equipment list, sound somewhat different. To me, that is a sign of where we are at with technology, at least with respect to dacs that are intended to as accurate as possible. That they sound different at all only means the best we can do is still not as good as we would like it to be. Over time hopefully things will improve, but good dacs are still hard to make for now. That's why you have to do some work to get one.
Last edited:
turntable
What is the price of a good turntable ?
What is the price of a good arm ?
What is the price of a good cartridge ?
What is the price of a good preamp ?
Too much for me
Serge
Buying all the pieces is one thing. That they will sound right, immediately after putting together. That is another matter. Remember setup is everything in a turntable situation. The cartridge alignment in all directions will sound different. Then you need to match cables and impedances to the cartridge and and and everything matters. Even the drink you're having at the time.
Perfect Sound Forever was the CD's promise...and digital. It was supposed to be easy. Pop in the CD and Ahhhhh..... WTF.
Hey occip,
Question for you please. Apparently the Arduino I2C library I have been using does not support multi-mastering so it only tells me the bus is busy rather than waiting for a chance to seize it. Could you tell us what hardware/software you are using for I2C access to ES9038Q2M board dac chip?
Thanks!
Serge and Mark, you didn*t uderstand me right.
I have not said that you should buy turntable because is better... ok i will reveal how my test was done at one audio meeting.
We took some songs that we have on all medias. On computer, on cd and on turntable.
We compare rpi streamer modified with 3 linear regulators all suplied with lipo as usb than trought spdif Alllo digione, than audio note cd player as transport trought spdif in es9038q2m (modified suplied with nimh accu pack, nanotransformer output and tube buffer) and also audio note dac , and turntable around 5k range..not ultimative high end.
Mark you are tuning your dac on your system on your listening envinorment, what is your main goal. That is excellent.
I wil just want in future that you concentarate on facts....choose some songs that everybody has in collection or can be recorded (can*t say download ) and for example.... at xxx song female vocal is not harsh anymore and is positioned little back and became more 3d ...not just it si better and better and better that is all we can read.
And we don*t know in what way. Maybe some of us don*t like change that was done and skip that mod, and goes along your way to his best sound... no hard feelings.
Btw you do a gret job for all diy members that could easier decide to start with modding dac (-s).
I have not said that you should buy turntable because is better... ok i will reveal how my test was done at one audio meeting.
We took some songs that we have on all medias. On computer, on cd and on turntable.
We compare rpi streamer modified with 3 linear regulators all suplied with lipo as usb than trought spdif Alllo digione, than audio note cd player as transport trought spdif in es9038q2m (modified suplied with nimh accu pack, nanotransformer output and tube buffer) and also audio note dac , and turntable around 5k range..not ultimative high end.
Mark you are tuning your dac on your system on your listening envinorment, what is your main goal. That is excellent.
I wil just want in future that you concentarate on facts....choose some songs that everybody has in collection or can be recorded (can*t say download
) and for example.... at xxx song female vocal is not harsh anymore and is positioned little back and became more 3d ...not just it si better and better and better that is all we can read.And we don*t know in what way. Maybe some of us don*t like change that was done and skip that mod, and goes along your way to his best sound... no hard feelings.
Btw you do a gret job for all diy members that could easier decide to start with modding dac (-s).
My clock is a SDA type, not SD. Good to know, it might work with AK4137 at 11,2MHz!
Regarding buffer installation I read even a sharp corner on a trace of a clock signal could provoke jitter, so not sure if wiring the buffer from bottom side would be the best choice or better installing a small ground plane on top side where the buffer could be soldered to?
But I also am thinking about buying the crystek 100MHz clock which has 2ns rise time compared to 8ns of my NDK. With buffer it would be even less than 2ns I guess..but as you say very hard to install it.
Andora76,
Every change I do is to reduce distortion and increase accuracy. Tubes and transformers are not allowed, as they increase distortion. Doing some of the necessary things makes the dac very revealing and resolving including of it's own shortcomings. The only fix for the shortcomings is not to obscure them with tube or transformer distortion, it is only allowed to get rid of them, not hide them. As the dac is made more resolving it will sound worse until the causes are removed then it will sound better than ever. None of the changes I make are to change the position of something in the sound stage. If something seems forward, it is distorted with harmonics. If something seems too far back it is blurred or hidden.
If you think about it there is linear and non-linear distortion in electronics. That's all there is. There is no sound stage or forward or back or anything like that. Those things are mental experiences only. Distortion is the physical property that is behind the mental experiences. However, the distortion produced by dacs is complex and different from nominally linear amplifiers. Some of the distortion mechanisms are non-stationary which means the distorting mechanism changes over time perhaps as a function dynamic changes from loud to soft, as is the case with state variable settling. FFTs of constant level 1kHz tones do not stimulate the mechanism to occur, and are thus useless for measuring that particular effect.
Also, for many reasons, I cannot organize the mods in terms of what they do to sound as a mental experience. They are organized in terms of what set of electrical components make sense to change at once for electrical reasons, not for perceptual reasons.
For the above reasons, if you want a description of how each mod changes the sound, and so you can skip some that make it more resolving before the doing a mod that makes the resolving sound good then you made a mistake. The resolving was done first for mechanical or electrical reasons, not for perceptual reasons. The sound will take advantage of being more resolving later when more mods are done.
Therefore, let me say it this way: There is only one mod, and it sounds almost perfectly just right. Not too forward, not too back, just right. That mod is all the mods at once. Just don't listen to the dac until you have finished that one big mod and you will like it, or if you don't then you can add tubes whatever at that time, not before. That's how the mod works, take it or leave it. I don't know any other way to do it. If I could get you to the right endpoint the way you want so you like not only the end point but at multiple checkpoints along the way, I would do it. But, I don't know how. The way it is perceptually is it only sounds its best at the end of the one big mod.
That may seem hard to believe, but I can't change your beliefs. If you were here I would show you so you could see for yourself and you could form your belief based on your own ears and observations.
All I can do from here is ask that you trust me and do what I have found works, the one big mod, and then if you don't like it, change it from there. I was hoping we would get 10 or 20 people to do the whole big full mod and I think we would see about 17 or 18 that would agree it sounds really good as is at the end point. You might see a couple that want to add .03% 2nd harmonic with some additional circuitry or maybe by mis-adjusting harmonic compensation a little. Fine. Just listen to the end first and take it from there. It will be the best experience by far for most people that way.
Every change I do is to reduce distortion and increase accuracy. Tubes and transformers are not allowed, as they increase distortion. Doing some of the necessary things makes the dac very revealing and resolving including of it's own shortcomings. The only fix for the shortcomings is not to obscure them with tube or transformer distortion, it is only allowed to get rid of them, not hide them. As the dac is made more resolving it will sound worse until the causes are removed then it will sound better than ever. None of the changes I make are to change the position of something in the sound stage. If something seems forward, it is distorted with harmonics. If something seems too far back it is blurred or hidden.
If you think about it there is linear and non-linear distortion in electronics. That's all there is. There is no sound stage or forward or back or anything like that. Those things are mental experiences only. Distortion is the physical property that is behind the mental experiences. However, the distortion produced by dacs is complex and different from nominally linear amplifiers. Some of the distortion mechanisms are non-stationary which means the distorting mechanism changes over time perhaps as a function dynamic changes from loud to soft, as is the case with state variable settling. FFTs of constant level 1kHz tones do not stimulate the mechanism to occur, and are thus useless for measuring that particular effect.
Also, for many reasons, I cannot organize the mods in terms of what they do to sound as a mental experience. They are organized in terms of what set of electrical components make sense to change at once for electrical reasons, not for perceptual reasons.
For the above reasons, if you want a description of how each mod changes the sound, and so you can skip some that make it more resolving before the doing a mod that makes the resolving sound good then you made a mistake. The resolving was done first for mechanical or electrical reasons, not for perceptual reasons. The sound will take advantage of being more resolving later when more mods are done.
Therefore, let me say it this way: There is only one mod, and it sounds almost perfectly just right. Not too forward, not too back, just right. That mod is all the mods at once. Just don't listen to the dac until you have finished that one big mod and you will like it, or if you don't then you can add tubes whatever at that time, not before. That's how the mod works, take it or leave it. I don't know any other way to do it. If I could get you to the right endpoint the way you want so you like not only the end point but at multiple checkpoints along the way, I would do it. But, I don't know how. The way it is perceptually is it only sounds its best at the end of the one big mod.
That may seem hard to believe, but I can't change your beliefs. If you were here I would show you so you could see for yourself and you could form your belief based on your own ears and observations.
All I can do from here is ask that you trust me and do what I have found works, the one big mod, and then if you don't like it, change it from there. I was hoping we would get 10 or 20 people to do the whole big full mod and I think we would see about 17 or 18 that would agree it sounds really good as is at the end point. You might see a couple that want to add .03% 2nd harmonic with some additional circuitry or maybe by mis-adjusting harmonic compensation a little. Fine. Just listen to the end first and take it from there. It will be the best experience by far for most people that way.
Last edited:
A sharp bend causes a change in characteristic impedance which causes reflections. There is inductance at the corners and capacitance changes where it gets closer and more distant from the ground plane too. How much trouble it is likely to cause depends in the wavelength of the highest frequencies that make the clock edges sharp. However the resistor in the line is an impedance change too with associated reflections. It depends on wavelengths what a reflection looks like. If the distance is short compared to a fraction of a quarter wavelength of the highest Fourier component then it just looks like a lumped cap, inductor, or resistor. Since you are starting out with a more of a sine wave coming out of your existing clock probably less reflection of that edge than after the buffer when it is sharpened up.
Anyway, I would say go for the Crystek clock. It works, it will be a standard installation, and it will save a lot of work that may need more fixing once you can test it. At some point I would probably want to try a lower frequency clock in order to be able to find one with less jitter. But, Katana uses lower frequency clocks and runs in master mode. As long as it doesn't sound way better than my dac, I don't think clocking is where my biggest problem is. Most likely the biggest problem is one my dac and Katana both have and that is the last little bit of distortion we just can't seem to get rid of. (Maybe my view will change when I hear Katana v1.2, have to wait and see.) For now that problem is what primarily separates both Q2M dacs and most other Sabre dacs on the market from DAC-3, and that problem seems most likely related to whatever is going on in the Spartan 6 that is not going on in all the other dacs.
So again, my vote or advice would be to just put a Crystek in there and get on with the rest of the one big mod I was just talking to Andora76 about.
Last edited: