Hi Nosian,
Transformer is dual primary 0,5+0,5:1 (1:1) type line transformer made by German Haufe company. Since we have a dual mono configuration, each dac chip is delivering two independent balanced outputs but carrying same signal. Thus I decided to directly wire these differential outputs to primary windings of transformer. With this configuration I was able to direct conversion of dual balanced to unbalanced signal. Unbalanced signal level is 2.1V rms.
Here is my topic asking how realistic is this type of connection:
http://www.diyaudio.com/forums/digital-line-level/261711-0-5-0-5-1-line-transformer-question.html
I actually don't recommend such unconventional wiring scheme, until being quite aware of what you are doing. Wrong/out-of-phase connection of dual primaries causes overloading dac chips.
I could have use 1:1 connection by paralleling outputs via 10K resistors (as what done on pcb) but I'm quite satisfied with the results then became lazy of trying other output options.
Regards.
AKM is not only company producing differential voltage-out dac chips. Cirrus Logic CS4397/8, Wolfson WM8741/2 are also highly regarded differential voltage output dac chips. ESS Sabre dacs also having Vout feature. One of the best thing about these chips is suitable of using a line transformer at outputs.
If you want to try a transformer, I don't recommend step up types like 1:5, 1:3, until you really need high gain at line levels, or your amp's gain ratio is low. You have to adjust your amp's gain accordingly otherwise your noise levels will rise dramatically.
1:1 600ohm types are considered as suitable ones at dac outputs. There are many companies producing such transformers. You can check out the listings at auction sites for reasonable prices too.
This is where I get them: http://www.ebay.de/itm/Haufe-Ubertrager-RK-736-1-1-Neumann-/181460116054
Regards.
Hello NinoSimona,
i finally managed to implement some of the mods you suggest earlier in the thread: the ground mod, the replacement of most of the capacitors using better ones (though Minishow mounted better electrolytic caps in my DAC, mostly ELNA and ROE) and the replacement of the 3 LT1085 regulators using MC7805s.
Unfortunately, just after the mod i noticed that right channel volume is now lower than left channel's. I made some measurement and found that left output of right AK4399 (the one nearer to regulators) is now 0,88V (that's bad!), while right output is 2,4/2,8V.
Left AK4399 has both outputs stuck to 2,4/2,8V, and it sounds good.
Also, VCML of right chip is 0,01V while VCML of left chip is 2,88V; VCMRs are 2,88V on both chips.
I really can't understand what could have gone wrong... i checked all caps i replaced and digital/analog power lines values and they're OK.
But my setup has 2 differences from the setup you suggest:
1) i used 100uF value for regulators output caps instead of 470uF (470 caps are too large in that position!)
2) i put all 47uF decoupling caps around the chips except the caps for AVDD that are 10uF on both chips, as 47uF caps wouldn't fit in that position.
Do you think one of these differences can be to blame for the issue?
What mistake did i make?
Thanks in advance to anybody could help...
i finally managed to implement some of the mods you suggest earlier in the thread: the ground mod, the replacement of most of the capacitors using better ones (though Minishow mounted better electrolytic caps in my DAC, mostly ELNA and ROE) and the replacement of the 3 LT1085 regulators using MC7805s.
Unfortunately, just after the mod i noticed that right channel volume is now lower than left channel's. I made some measurement and found that left output of right AK4399 (the one nearer to regulators) is now 0,88V (that's bad!), while right output is 2,4/2,8V.
Left AK4399 has both outputs stuck to 2,4/2,8V, and it sounds good.
Also, VCML of right chip is 0,01V while VCML of left chip is 2,88V; VCMRs are 2,88V on both chips.
I really can't understand what could have gone wrong... i checked all caps i replaced and digital/analog power lines values and they're OK.
But my setup has 2 differences from the setup you suggest:
1) i used 100uF value for regulators output caps instead of 470uF (470 caps are too large in that position!)
2) i put all 47uF decoupling caps around the chips except the caps for AVDD that are 10uF on both chips, as 47uF caps wouldn't fit in that position.
Do you think one of these differences can be to blame for the issue?
What mistake did i make?
Thanks in advance to anybody could help...
Yes it can.
Did you measure regulator output voltages both regulator side and dac side after mod?
Last edited:
Since my intangible memory is weak, at least I could say that when I switch to transformer during first experiments, I felt like vocals contain much more air, cymbal instruments have much better timbre and better dynamics overall. Once, I tried this setup with a dc-coupled LM3875 Gainclone amplifier which I built for a friend, I couldn't believe the timbre of cymbals with this setup. Now, I'm using an old Sansui amplifier until finishing my new amplifier project based on LM4702+Sanken darlingtons. I have to notice that my observation may relate with my setup as overall which uses external analog supplies using same LT1085's but based on jbau's experiments and military grade axial tantalums around dac chips.
Since transformers don't contain dc naturally, you can bypass dc-blocking caps on signal path for getting advantage of this feature (if you have speaker dc protection at outputs of course).
Last edited:
Of course..If you can afford for such pricey brand, you can also take a look at models from Jensen and Cinemag brands. But I suggest for getting reasonable priced/second hand ones from auction sites for experimenting. You may try to contact with the seller whom I bought mines from. There is email address on listing.
Last edited:
Yes, i did: very stable 5V current on Digital and Analogic. Power inputs into the Dac chips seems OK, i think the problem is elsewhere, maybe on other DAC regulation pins. I would have suspected caps values not to be well balanced between AVDD and others, or between regs output and current needed by chips caps (bigger -> more current stored), but if that should be to blame, why the left AK4399 works flawlessly and right one does not? Caps setup is just the same for both of them...
Maybe one of pins of chip disconnected from pcb. You can't see it by sight. Try to push the faulty chip with your finger. I observed similar problem after replacing chips with new ones. I figured out later that one of ground pins wasn't properly soldered.
I alos have to say that in a first attempt both channel of the right AK4399 were mute, showing 0,88V at all 4 outputs. But i measured DVDD pin and it was 2,4V: its decoupling capacitor had a bad solder problem, i fixed it and DVDD came back to 5V.
After that, right outputs for that chip started to output sound at 2,4/2,8V, but left outputs didn't - very strangely to me! I checked all other caps solder and re-soldered them, but this time i had no succes: i tested all soldering and now there is no bad solder, but left outputs stay at 0,88V while right outputs are Ok at 2,4/2,8V.
Can someone understand what's happening?
After that, right outputs for that chip started to output sound at 2,4/2,8V, but left outputs didn't - very strangely to me! I checked all other caps solder and re-soldered them, but this time i had no succes: i tested all soldering and now there is no bad solder, but left outputs stay at 0,88V while right outputs are Ok at 2,4/2,8V.
Can someone understand what's happening?
Hi Luca
100u output caps for MC7805CT are fine, I recommended 470u or bigger for LT1085.
Did you measure reference voltages (VREFL/VREFH) of the right dac? Also check if somehow the left channel of the right dac is improperly loaded, causing no output. And check the back of the board, each dac analog +5V must have it's own ground wire to its psu.
If this doesn't help, post us a picture of your board.
Regards
Nino
100u output caps for MC7805CT are fine, I recommended 470u or bigger for LT1085.
Did you measure reference voltages (VREFL/VREFH) of the right dac? Also check if somehow the left channel of the right dac is improperly loaded, causing no output. And check the back of the board, each dac analog +5V must have it's own ground wire to its psu.
If this doesn't help, post us a picture of your board.
Regards
Nino
Hello,
i measured VREFL and VREFH of right DAC at MKP caps feet, and they're OK (stable 5V), as is AVDD, that shares the same reg output.
The AVDD caps of both chips have a wire to the big reg cap ground as shown in the picture you posted about the gound mod.
About improper load to the right DAC, i don't know how to check that: what load could it have? What should i measure? ?
I tested all values both without feeding DAC inputs and while reproducing a song: they're just the same, so is even DAC output resistors voltage value:
Left AK4399 AOUTLN/AOUTLP 2,4/2,8V - AOUTRN/AOUTRP 2,4/2,8V;
Right AK4399 AOUTLN/AOUTLP 0,88V - AOUTRN/AOUTRP 2,4/2,8V.
I enclose a pair of pictures:
Hello, just to inform you that i solved the issue: there was a connection problem between VDDL trace and its capacitor pad.
Surely i have to say that this PCB quality is very low, desoldering and replacing the small caps was a real pain: many pads came away very shortly and some traces followed after a pair of attempt... i had to solder a pair of caps directly to the traces, that is a nightmare for all diyers, and fix everything using hot glue. I had to solder VDDL and VREFHL cap to the chip pins, by luck they're one next to the other.
It was a shock to me, as last PCB i assembled was a Glassware Audio product: an enormous quality difference. Glassware PCBs you can throw away, heat up a bit and treat very bad: they're really strong and well made!
Anyway, now the DAC works ok, but obviously i won't be able to make any more modification in the chip area.
Surely i have to say that this PCB quality is very low, desoldering and replacing the small caps was a real pain: many pads came away very shortly and some traces followed after a pair of attempt... i had to solder a pair of caps directly to the traces, that is a nightmare for all diyers, and fix everything using hot glue. I had to solder VDDL and VREFHL cap to the chip pins, by luck they're one next to the other.
It was a shock to me, as last PCB i assembled was a Glassware Audio product: an enormous quality difference. Glassware PCBs you can throw away, heat up a bit and treat very bad: they're really strong and well made!
Anyway, now the DAC works ok, but obviously i won't be able to make any more modification in the chip area.
Hi Luca,
Sorry for replying late. Glad you managed to solve the problem! I took a look at the pictures.
What strikes me, is that the capacitors you're using are so big. You don't need 25V caps for decoupling 5V: 6.3V types will suffice, they are much smaller in diameter.
Because the TVDD mod had so much influence on sound signature of the dac, I eventually changed the 1u polyester caps back to 100n PP, and the 100u Silmics at VREF I changed back to 47u Silmics. I like the sound better this way, the PP (Panasonic) bypasses combine great with the Silmics.
IMO the big Silmics for digital +5V are overkill, did you have better sound with bigger ones?
I did another mod to the +5V supply of this DAC, namely I transferred digital +5V to the other group, the same as which supplies +5V for MCU and 3.3V for AK4113. This way analog +5V, which needs the most current, will have it's own supply group. Impedance improves, better SQ. But you'll need to cut some tracks, I don't know you will be into this right now...
I agree, PCB isn't HQ... design either, but with some work and experiment, I have a nice sounding dac now, very rewarding in the end. Did the mods you did so far suit your needs? I hope so...
Regards
Nino
Sorry for replying late. Glad you managed to solve the problem! I took a look at the pictures.
What strikes me, is that the capacitors you're using are so big. You don't need 25V caps for decoupling 5V: 6.3V types will suffice, they are much smaller in diameter.
Because the TVDD mod had so much influence on sound signature of the dac, I eventually changed the 1u polyester caps back to 100n PP, and the 100u Silmics at VREF I changed back to 47u Silmics. I like the sound better this way, the PP (Panasonic) bypasses combine great with the Silmics.
IMO the big Silmics for digital +5V are overkill, did you have better sound with bigger ones?
I did another mod to the +5V supply of this DAC, namely I transferred digital +5V to the other group, the same as which supplies +5V for MCU and 3.3V for AK4113. This way analog +5V, which needs the most current, will have it's own supply group. Impedance improves, better SQ. But you'll need to cut some tracks, I don't know you will be into this right now...
I agree, PCB isn't HQ... design either, but with some work and experiment, I have a nice sounding dac now, very rewarding in the end. Did the mods you did so far suit your needs? I hope so...
Regards
Nino
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.