It has been a while since I posted here since I didn't do any relevant audio projects since my LKS ES9018 and Lehman projects.
My tweaked LKS ES9018 project defected last summer because of an overheated ES9018 chip during summer heat in Hangzhou. I didn't bother to fix it for a long time and ran onboard audio (ES9018K2M) which eventually started bothering me. I started looking for a dac with a newer ES9038Q2M chip since I love the sound of my ES9018. When I was looking into the Topping D50 which has dual ES9038K2M in mono mode and excellent measurements on ASR, I ran into the DX3 Pro in their factory Taobao shop that just came out. Dual AKM AK4493 (upgraded AK4490, which is decent already). DSD512, PCM768, BT5.0, remote control AND powerful headphone amp! The measurements on ASR are near D50 but the featureset of the DX3 Pro is far more extensive. The D50 lacks BT, remote and headphone amp.
So after some consideration I ordered the DX3 Pro at 12.12 for 1085 RMB (€140) and inmediately ordered a 15V/1.5A LPS with it for 360 RMB (€45). Even though Amir his measurements on ASR showed that an LPS wouldn't improve sound too much compared to the original 1.0A switching brick. Later I found out that many devices had issues suddenly going into standby mode and topping themselves hinted on using another powersupply to see if the problem was solved. So far on all three of my units (later more on why I have three) I've had NO issues with that so far. So I'm glad I bought a solid powersupply anyway.
I hooked it up straight into my QMS Q5s and was really impressed with the amount of detail and imaging the unit produced. But there were some things I didn't like as much. The treble seemed a bit harsh and crisp out of the box and the bass was lacking in the lower frequencies. It was tight but not deep as the LKS ES9018. Also the soundstage is more narrow and it's less holographic. In general I'd say the ES9018 is more sophisticated and musically pleasing. and the AKM is more analytical and a more in your face with vocals. Don't get me wrong. The ES9018 is a highly tuned DAC carrying all kinds of expensive parts and opamps in class A. So I figured there must be more potential in the AKM. A couple more days of burning in smoothened out the edgy treble a bit but didn't bring out the bass in any way. This is when I decided to hook up my Lehman and it helped but I still felt there is more to be squeezed out of this unit.
And this is where I come to the part that I opened it up and planned to modify it.
In the next posts I'll share my modifications and plans to do so. I hope people get interested and join in. In another forum (which I won't call by name) a lot of people were already making modifications and encouraging eachother to try modifications but some elitarial upperclass started moaning about measurements and complaining about audiophile territory. Hostile environment. So I'm here hoping we can share and contribute and keep it nice and positive.
My tweaked LKS ES9018 project defected last summer because of an overheated ES9018 chip during summer heat in Hangzhou. I didn't bother to fix it for a long time and ran onboard audio (ES9018K2M) which eventually started bothering me. I started looking for a dac with a newer ES9038Q2M chip since I love the sound of my ES9018. When I was looking into the Topping D50 which has dual ES9038K2M in mono mode and excellent measurements on ASR, I ran into the DX3 Pro in their factory Taobao shop that just came out. Dual AKM AK4493 (upgraded AK4490, which is decent already). DSD512, PCM768, BT5.0, remote control AND powerful headphone amp! The measurements on ASR are near D50 but the featureset of the DX3 Pro is far more extensive. The D50 lacks BT, remote and headphone amp.
So after some consideration I ordered the DX3 Pro at 12.12 for 1085 RMB (€140) and inmediately ordered a 15V/1.5A LPS with it for 360 RMB (€45). Even though Amir his measurements on ASR showed that an LPS wouldn't improve sound too much compared to the original 1.0A switching brick. Later I found out that many devices had issues suddenly going into standby mode and topping themselves hinted on using another powersupply to see if the problem was solved. So far on all three of my units (later more on why I have three) I've had NO issues with that so far. So I'm glad I bought a solid powersupply anyway.
I hooked it up straight into my QMS Q5s and was really impressed with the amount of detail and imaging the unit produced. But there were some things I didn't like as much. The treble seemed a bit harsh and crisp out of the box and the bass was lacking in the lower frequencies. It was tight but not deep as the LKS ES9018. Also the soundstage is more narrow and it's less holographic. In general I'd say the ES9018 is more sophisticated and musically pleasing. and the AKM is more analytical and a more in your face with vocals. Don't get me wrong. The ES9018 is a highly tuned DAC carrying all kinds of expensive parts and opamps in class A. So I figured there must be more potential in the AKM. A couple more days of burning in smoothened out the edgy treble a bit but didn't bring out the bass in any way. This is when I decided to hook up my Lehman and it helped but I still felt there is more to be squeezed out of this unit.
And this is where I come to the part that I opened it up and planned to modify it.
In the next posts I'll share my modifications and plans to do so. I hope people get interested and join in. In another forum (which I won't call by name) a lot of people were already making modifications and encouraging eachother to try modifications but some elitarial upperclass started moaning about measurements and complaining about audiophile territory. Hostile environment. So I'm here hoping we can share and contribute and keep it nice and positive.
Attachments
-
2a4f3c52-33fd-4d78-9b18-29d9bff9924c.jpg -
WeChat Image_20190130235044.jpg -
WeChat Image_20190130235111.jpg -
00740bf0-bcab-4d08-8672-23f53a17cdbf.jpg -
3a6a15cd-307f-478a-8ca8-934c4fd50322.jpg -
f98916df-af40-4c51-a691-c65b78ecde49.jpg -
8a8be6a1-eaf9-4761-bfdb-66111a5d91ed.jpg -
5b203441-ff1a-417d-b8e9-cebc54ae4c1a.jpg -
30596d4e-0f73-4f4d-baf0-328ec3dac19e.jpg -
b343a978-8d9b-44fb-8514-8916ceef38ec.jpg
So when I tried to open the device I ran into some trouble not having a long hex key to reach the bolts in the front. Get this:
2mm x 20mm HEX key
Can reach the two bolts in the front panel from the back. then flip it over and pull out the board with cable.
So, first thing I wanted to try is to upgrade the opamp in the LPF. Replace the OPA1612 with the rediculously pricey Sparkos SS3602 (690 RMB €80). But I would have to convert the SOP8 socket to DIP8 first with limited space.
OPA1612
Sparkos SS3602
SOP8 to DIP8 adapter. Custom built for low profile
21mm. Will just fit with 22mm of space.
Then disaster struck. I had soldered in the new adapter at night after a dinner and a glass of whiskey or two.. and I broke it out of the socket by accident. don't ask me how. But that kind of ruined a good new board.
I contacted the factory and let them send me a new board and a new display. I also ordered a brand new original unit with that. To use as reference device for my modifications.
The new board went more smooth and careful
DIP8 Socket
SS3602 in socket
Barely fits. I put a bit of isolation tape on the case to be sure.
Result: I felt the sound became more smooth and detailed but the Sparkos opamp wasn't burned in yet and also sounded very dark. After a while of burning in the sound settled and the difference between standard OPA1612 and Sparkos were smaller than before. Maybe the Sparkos sounded a bit more organised and more effortless. I rolled back to a dip8 mounted OP1612 and started looking into other mods to compare to my reference DX3 Pro.
2mm x 20mm HEX key
Can reach the two bolts in the front panel from the back. then flip it over and pull out the board with cable.
So, first thing I wanted to try is to upgrade the opamp in the LPF. Replace the OPA1612 with the rediculously pricey Sparkos SS3602 (690 RMB €80). But I would have to convert the SOP8 socket to DIP8 first with limited space.
OPA1612
Sparkos SS3602
SOP8 to DIP8 adapter. Custom built for low profile
21mm. Will just fit with 22mm of space.
Then disaster struck. I had soldered in the new adapter at night after a dinner and a glass of whiskey or two.. and I broke it out of the socket by accident. don't ask me how. But that kind of ruined a good new board.
I contacted the factory and let them send me a new board and a new display. I also ordered a brand new original unit with that. To use as reference device for my modifications.
The new board went more smooth and careful
DIP8 Socket
SS3602 in socket
Barely fits. I put a bit of isolation tape on the case to be sure.
Result: I felt the sound became more smooth and detailed but the Sparkos opamp wasn't burned in yet and also sounded very dark. After a while of burning in the sound settled and the difference between standard OPA1612 and Sparkos were smaller than before. Maybe the Sparkos sounded a bit more organised and more effortless. I rolled back to a dip8 mounted OP1612 and started looking into other mods to compare to my reference DX3 Pro.
Attachments
-
WeChat Image_201901310025139.jpg -
WeChat Image_201901310025138.jpg -
WeChat Image_201901310025137.jpg -
WeChat Image_201901310025136.jpg -
WeChat Image_201901310025134.jpg -
WeChat Image_201901310025131.jpg -
WeChat Image_201901310025133.jpg -
WeChat Image_201901310025132.jpg -
WeChat Image_20190131002513.jpg -
WeChat Image_20190130221712.jpg
Last edited:
I got myself some nice transformers from a Chinese Taobao shop. The shop started cloning jensen transformers but has it's own products now which are considered better!
Ordered the R3 10k:10K with a little case. Decided to do the wiring myself with some shielded wiring.
Waiting for some PCB connectors to do it neatly.
I'll use them as output transformers and bypass the opamp eventually but first I'll try to squeeze everything out this opamp LPF.
If we look at the datasheet of the AK4493 we can see the following recommended schematics for LPF
having 100uf caps on the output instead of the 33uf tantalums that Topping uses. I have replaced them with 100uf Kemet KO-CAP polymers and put two of the 33uf that came free in the 10uf position at the opamp voltage rails.
I haven't listened to the result yet. I'll be testing it tomorrow against my original reference DX3 Pro with both running OPA1612 to see if it made a positive difference. Direct AB comparisson. I'll try to do it as blind as possible.
Further more I'll replace the tantalums around the DAC chips with polymer KO-CAPS and the electrolytes in the USB section with Panasonic 16SVPC100M.
After that I'll have a look at the powersupply and put some polymers in there. 5V and 3.3V.
Making sure ther base of the device is built on good quality components.
Ordered the R3 10k:10K with a little case. Decided to do the wiring myself with some shielded wiring.
Waiting for some PCB connectors to do it neatly.
I'll use them as output transformers and bypass the opamp eventually but first I'll try to squeeze everything out this opamp LPF.
If we look at the datasheet of the AK4493 we can see the following recommended schematics for LPF
having 100uf caps on the output instead of the 33uf tantalums that Topping uses. I have replaced them with 100uf Kemet KO-CAP polymers and put two of the 33uf that came free in the 10uf position at the opamp voltage rails.
I haven't listened to the result yet. I'll be testing it tomorrow against my original reference DX3 Pro with both running OPA1612 to see if it made a positive difference. Direct AB comparisson. I'll try to do it as blind as possible.
Further more I'll replace the tantalums around the DAC chips with polymer KO-CAPS and the electrolytes in the USB section with Panasonic 16SVPC100M.
After that I'll have a look at the powersupply and put some polymers in there. 5V and 3.3V.
Making sure ther base of the device is built on good quality components.
Attachments
Too bad you didn't come here a talk to us before you invested so heavily in what you have now. We have learned how to make ES9038Q2M sound exceptionally good. Don't know what else to say except if you have any remaining interest in the little Sabre chip, we have a thread with lots information.
Yeah I can do a sabre project after.
It's just for fun.. Not to reinvent the wheel or find the holy grail.
But just I'd be interested in doing a 9038 project also. If you could recommend me a good base platform that would help.
It's just for fun.. Not to reinvent the wheel or find the holy grail.
But just I'd be interested in doing a 9038 project also. If you could recommend me a good base platform that would help.
Well, you can see my test setup as it was a few days ago here: https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-373.html#post5675352
We have typically been starting out with a pretty minimal board because there are so many things that need fixing on almost any under $1,000 dac that its just easier to do it that way. A popular board that many people choose is like this: ES9038 Q2M I2S DSD Optical Coaxial Decoder DAC Headphone Output Audio Amplifier | eBay
We have typically been starting out with a pretty minimal board because there are so many things that need fixing on almost any under $1,000 dac that its just easier to do it that way. A popular board that many people choose is like this: ES9038 Q2M I2S DSD Optical Coaxial Decoder DAC Headphone Output Audio Amplifier | eBay
I have ordered a Khadas Tone board. to compare the ES9038 to my AK4493.
I'll have a listen and see what I find between these two.
And then decide if I'll do another 9038 project. So far I really like the sound signature of the AKM chip. VELVET sound and it performs superb with DSD.
I'll have a listen and see what I find between these two.
And then decide if I'll do another 9038 project. So far I really like the sound signature of the AKM chip. VELVET sound and it performs superb with DSD.
Please let us know as the Khadas Tone Board is well tested and liked in the audioscience forum. Would be interesting to get a subjective opinion as the audioscience forum cares for the objective measurements only. Very little on the subjective side.
I just tested the capacitor mods of the AKM output and the 2 small ones at the voltage rails of the opamp.
AB testing OPA1612 against each other. with just thoses caps changed.
Tested with DSD512 and PCM44.1
Conclusion: More bass wider sound stage and a more holographic sound. Meaning depth. Feels like it opened up more.
Definitely recommended!
Then I rolled in Sparkos SS3602 and the whole thing became more effortless and organic. Musical. Less clinical than the clean OPA1612.
Both mods make the device more sophisticated than the original. More controlled and more open and deeper sounding.
It's hard to put sound perception into words.
AB testing OPA1612 against each other. with just thoses caps changed.
Tested with DSD512 and PCM44.1
Conclusion: More bass wider sound stage and a more holographic sound. Meaning depth. Feels like it opened up more.
Definitely recommended!
Then I rolled in Sparkos SS3602 and the whole thing became more effortless and organic. Musical. Less clinical than the clean OPA1612.
Both mods make the device more sophisticated than the original. More controlled and more open and deeper sounding.
It's hard to put sound perception into words.
Attachments
That's why I left. We were sharing possible mods and experiences and then the measurement police came. contributing literally NOTHING. Just nagging about measurements just for the sake of nagging but they added literally NOTHING useful to the thread except their opinion on how it all should be.. it's tiring..
Yeah, too much emphasis on objective measurements. The intention is good but...love more subjective opinion and head to head comparisons along with the objective tests.
I'm making my tests as objective as possible by AB testing against an original reference unit with all the same conditions. Same cables and option to fast switch. It's not completely blind but it works pretty well.
I have the Topping dx3 and a es9038q2m , it is much easier to get good sound from modding the Topping as it is already all in one dac. And the Ak4493 does sound good to my ear. 🙂
I wait for you review.🙂
For small coupling caps, I prefer the Nichicon Muse ES. I also recommend you remove the ceramic cap decoupling the opamp. I remove both the ceramic and the tan cap and replace with Nichicon KZ and the mod give a less clinical sound.
Thing is, how good DACs sound has much more to do with implementation than it does with particular dac chips. That being said, Sabre DACs may have more configuration options available in the control registers (as compared to AK) that can be useful or not, depending on implementation.
I am curios, how we can connect Sabre DAC via I2C at ports 0x90; 0x92 and use native GUI (Windows PC) to control registers.
Attachments
Today I had a moment of realisation that I was listening to the modded DX3 for quite a while on a high volume with a random playist on my active speakers and suddenly realised that I wasn't fatigued by it and the sound was really smooth and precise and musical. Then I turned the volume down and listened a while on low volume and still the music came out equally precise and musically pleasing and effortless.
Then I rolled the opamp back to OPA1612 (from Sparkos SS3602) and the music became more messy and less controlled.
Then I switched back to the original DX3 Pro to do the next mod on the other one and I didn't have that same comfort level. The original DX3 has something anxious something out of line. Even though good it's not that sophisticated as the modified one. Also the bass is more lacking than with the modded one.
This is not a A/B comparison as I do normally witch a RCA switch and quickly switching USB input on the computer and restarting track and skipping to the same part. But just sharing the realisation that with the modded version music became music again and no fatigue which I do have with the original unit.
Even though the Sparkos SS3602 wouldn't be the first pick for LPF it works darn well in this position. I'd say this is mostly because it's operating in class A. Another test would be to put an CCS on the OPA1612 using 2 JFETS and see if I can come close to the performance of the Sparkos. But I remember trying to put OPA1612 in class A before and that didn't go so well. but later more on this.
For now I did my next mod:
Replacing the caps around the DAC chips with better quality ones and increasing the size a bit on some positions.
T520D107M010ATE018 KEMET KO-CAP 100uf 10V
T520B107M006ATE070 KEMET KO-CAP 100uf 6.3V
not my best soldering job but it did the trick.
After testing I found that a bit of solder hit the DAC output pins and this resulted in distortion on the left channel. After removing it all was well in the world again.
A/B test with OPA1612 and remembering the sound of before the mod with the same unit is a bit more difficult obviously.
By now the modded unit is a far better DAC already than the original one. by miles. The improvement of the caps on the DAC position brought not the same WOW effect as when I replaced the DAC output caps with 100uf or rolling the opamp but I'd say everything became a little more orderly and cleaner sounding. Hard to say. deminishing returns i'd say. But definitely no loss! I'll test more AB tomorrow.
One more thing I'd like to point out is upsampling tracks to 768K in PCM DOES improve the sound quality! it makes it more transparant and highs less edgy! Give it a try!
Foobar2000 resampler mod
Take the mod version and manually add 768K sample rate
Next mods will be looking at the powersupply.
So far I'm EXTREMELY happy with the mods I've done. The device is in another league now.
Then I rolled the opamp back to OPA1612 (from Sparkos SS3602) and the music became more messy and less controlled.
Then I switched back to the original DX3 Pro to do the next mod on the other one and I didn't have that same comfort level. The original DX3 has something anxious something out of line. Even though good it's not that sophisticated as the modified one. Also the bass is more lacking than with the modded one.
This is not a A/B comparison as I do normally witch a RCA switch and quickly switching USB input on the computer and restarting track and skipping to the same part. But just sharing the realisation that with the modded version music became music again and no fatigue which I do have with the original unit.
Even though the Sparkos SS3602 wouldn't be the first pick for LPF it works darn well in this position. I'd say this is mostly because it's operating in class A. Another test would be to put an CCS on the OPA1612 using 2 JFETS and see if I can come close to the performance of the Sparkos. But I remember trying to put OPA1612 in class A before and that didn't go so well. but later more on this.
For now I did my next mod:
Replacing the caps around the DAC chips with better quality ones and increasing the size a bit on some positions.
T520D107M010ATE018 KEMET KO-CAP 100uf 10V
T520B107M006ATE070 KEMET KO-CAP 100uf 6.3V
not my best soldering job but it did the trick.
After testing I found that a bit of solder hit the DAC output pins and this resulted in distortion on the left channel. After removing it all was well in the world again.
A/B test with OPA1612 and remembering the sound of before the mod with the same unit is a bit more difficult obviously.
By now the modded unit is a far better DAC already than the original one. by miles. The improvement of the caps on the DAC position brought not the same WOW effect as when I replaced the DAC output caps with 100uf or rolling the opamp but I'd say everything became a little more orderly and cleaner sounding. Hard to say. deminishing returns i'd say. But definitely no loss! I'll test more AB tomorrow.
One more thing I'd like to point out is upsampling tracks to 768K in PCM DOES improve the sound quality! it makes it more transparant and highs less edgy! Give it a try!
Foobar2000 resampler mod
Take the mod version and manually add 768K sample rate
Next mods will be looking at the powersupply.
So far I'm EXTREMELY happy with the mods I've done. The device is in another league now.
Attachments
You made clear that you like the ES9038 chip more and I've browsed through your project topic. truly impressive and I'll have a dive into it after I'm done squeezing the last bits out of this project for kicks.
Can we stay on topic? If you own a AK4493 board or have suggestions please share them. This is not a ES9038 board. I'm sorry. I'll be sure to chat you up when I start a ES9038 project.
Other than that my Khadas ES9038 will come in in maybe two weeks after spring festival. Which will be interesting to put against the AK4493 modified and unmodified.