An update. I hope your mods have gone well Geoff (and anyone else).
Yesterday I replaced the analogue PS caps (C402/3, C76/77, C414/415 - Panasonic FC) and it makes a difference - clearer sound.
Today I swapped out the digital PS caps (C87, C89 - Pansonic FC, latter was side on, no chance of fitting FC's in the C424/425 spaces) and then the polarised decoupling capacitors (C426, C431, C130, C435, C25, C31, C35 - all Rubycon ZA). Again a clearer sound, more treble detail I think. I particularly noticed reverbs and echoes seem to come forward from the speakers.
I haven't yet swapped the non-polarised caps round the DACs and other digital IC's. But I have more Rubycon ZA's for this.
So far so good. The 4562's are a few days old so I look forward to them and the caps burning in. Soon I will ask to borrow a colleague's stock DACMagic to do an AB comparison.
Jon
Yesterday I replaced the analogue PS caps (C402/3, C76/77, C414/415 - Panasonic FC) and it makes a difference - clearer sound.
Today I swapped out the digital PS caps (C87, C89 - Pansonic FC, latter was side on, no chance of fitting FC's in the C424/425 spaces) and then the polarised decoupling capacitors (C426, C431, C130, C435, C25, C31, C35 - all Rubycon ZA). Again a clearer sound, more treble detail I think. I particularly noticed reverbs and echoes seem to come forward from the speakers.
I haven't yet swapped the non-polarised caps round the DACs and other digital IC's. But I have more Rubycon ZA's for this.
So far so good. The 4562's are a few days old so I look forward to them and the caps burning in. Soon I will ask to borrow a colleague's stock DACMagic to do an AB comparison.
Jon
Another update. I have now replaced almost all the electrolytics, including C424/425 by finding a smaller 2200uF 25V electrolytic. Changing the DAC digital supply decoupling caps has worked well, I used ZLG's in the end. The Wolfson datasheet (page 24) recommends 10uF for these caps and I could get 10uF ZLG's easily from Farnell. The polarity is clear on the schematic and the continuity mode on a multimeter quickly identifies the grounded side of the caps for the "-" connections.
It's taken 4 SMD op amp changes and over 30 electrolytics, plus the Maplin 9v power supply. So £70-80 and dozens of hours of DIY work. Better than freezing in the snow. Healthier and probably cheaper than the same time spent in pubs! But I am genuinely surprised by the improvement over the stock DACMagic, which is pretty good to start with. The sound is more detailed and smooth, with clear separation of instruments. Stereo width is very clear and there is more front to back depth, even some surround effects on some songs.
It's taken 4 SMD op amp changes and over 30 electrolytics, plus the Maplin 9v power supply. So £70-80 and dozens of hours of DIY work. Better than freezing in the snow. Healthier and probably cheaper than the same time spent in pubs! But I am genuinely surprised by the improvement over the stock DACMagic, which is pretty good to start with. The sound is more detailed and smooth, with clear separation of instruments. Stereo width is very clear and there is more front to back depth, even some surround effects on some songs.
Rikkert,
Contrary to what you report, I find connecting a Neutrik adapter _worse_ that the stock RCA outputs. This has partly to do with what, I believe, is the effect of the shorted (grounded) pin 3 on the adapter. I experimented with some XLR cable with disconnected pin 3 and the adapter. It seemed better. My reasoning is as follows:
If you ground pin 3, the NE5532 directly sees the AC coupling cap at the output, causing it to almost oscillate (massive overshoot at least). This will in turn modulate supply lines or otherwise couple into the signal line that you are actually using. This, to my ears, sounds like a loss of spatial resolution.
With pin 3 disconnected, I still find the sound inferior to that of the OP275 (which, as is quite commonly said by others, I find a P.O.$.). The top end of the 5532s sounds harsh and this seems to be balanced a bit by the OP275 (I guess this is the result of listening tests by CA...).
Since I had two OPA1612s available, I replaced the OP275s by those, and the sound finally did improve, I still hear the effect of the 5532s, but due to the summing, I guess the influence is mitigated a bit. I would replace the 5532s as well, but I don't have additional 1612s lying around at the moment.
But I'm already happier just replacing the OP275s by OPA1612s, not even touching the AC coupling caps or the noisy supply.
--
Greetz,
MatchASM
Labelled PCB
I agree about not grounding pin 3 on the XLR's. I would connect pin 1 to ground and take an unbalanced signal from pin 2.
I changed the 5532's to 4562 (U55/56/59/60 below) and the sound on the balanced output was clearer and less harsh. Changing the power supply and decoupling caps makes a big difference to the sound but it is about 30 capacitors...
I hope this photo and its annotations are helpful to others considering these modifications. I used Rubycon ZA and ZLG for the digital supply filtering and decoupling, which had as big an impact as the analogue capacitor changes to Panasonic FC's. Hopefully the "+" next to the electrolytics above will save someone else using the 8740 datasheet and a multimeter when replacing the 10uF NP's.
Please check any details on this diagram - I take no responsibility for the success or failure of your modifications. Good luck.
I agree about not grounding pin 3 on the XLR's. I would connect pin 1 to ground and take an unbalanced signal from pin 2.
I changed the 5532's to 4562 (U55/56/59/60 below) and the sound on the balanced output was clearer and less harsh. Changing the power supply and decoupling caps makes a big difference to the sound but it is about 30 capacitors...
I hope this photo and its annotations are helpful to others considering these modifications. I used Rubycon ZA and ZLG for the digital supply filtering and decoupling, which had as big an impact as the analogue capacitor changes to Panasonic FC's. Hopefully the "+" next to the electrolytics above will save someone else using the 8740 datasheet and a multimeter when replacing the 10uF NP's.
Please check any details on this diagram - I take no responsibility for the success or failure of your modifications. Good luck.
Last edited:
I found that soldering a 10uF tantalum across C419 (I did this on the backside of the PCB) removes the DAC's harshness at the high end of the spectrum, and it's a lot less work than replacing 30 elcos. I think this mod should be better than simply replacing the electrolytics by different brands of electrolytics, since elcos have inherently worse ESR/ESL performance at higher frequencies (I think the main concern with the DAC is intermodulation above the audio frequencies, due to the sigma-delta modulation and the 192kHz sample rate.) Beefing up the op-amp supplies might help also, but since most op-amps already have outrageously good PSRR, I expect the effect to be less strong.
It would be nice to have a schematic also... I was pleasantly surprised when I realised there were relays on the audio outputs and not simply (distorting) short-circuit transistors. Funny also how there are four dual relays, 1 for the RCA outs and 1 for each set of XLR pins...
The lack of schematic also caused my surprise when I tried plastic film capacitors as AC output coupling caps. At 10uF, the bass really suffers in the dacmagic, probably due to some resistors elsewhere on the PCB. And since the casing is rather small, there are two options:
1. Either cut a hole in the casing and insert massive MKPs or
2. Use higher-grade electrolytics (which I'm probably going to do, since I'm a bit lazy).
--
Greetz,
MatchASM
Last edited:
Parts list
Hi ggbaba, when you open up the box you can:
1. Replace the output capacitors which are C1/C2 and C40/C41 on the balanced outputs, C293 and C299 on the unbalanced outputs. These are 470uF 16 V, non-polarised electrolytic – I used Michicon Muse ES. Or just short them out by turning the PCB over and soldering a thin wire across each capacitor’s pins. However this DC couples the DACMagic output to the preamp or amp it is plugged into, so consider using a DC voltmeter first to measure the DC offsets.
2. Replace the power supply capacitors for the op amps/analogue circuits. I used Panasonic FC’s as follows:
a. C76, C77, C402, C403 are 1,000uF 50V
b. C414, C415 are 10uF 35V
c. C419 and C87 (display supply – could leave alone) are 1,000uF 35V I used 1,000uF 50V as 2a above
3. Replace the power supply capacitors for the DSP and DAC digital circuits. I used Rubycon ZA’s and ZLG’s:
a. C424, C425 and C89 are 2,200uF 25V – the first two need to be smaller e.g. ZLG to fit in
b. C8, C10, C25, C26, C31, C35, C45, C130, C426, C431, C435 are 10uF 35V
c. C103 is 100uF 16V
4. Replace the decoupling capacitors around the two WM8740 DAC’s. I used Rubycon ZLG’s:
a. C270, C275, C276 and C301, C303, C304 are all 10uF 35V
b. C271 and C302 are 47uF 35V – I used the DACMagic value and not the 10uF on the WM8740 datasheet
5. Replace the op amps, which are tiny surface mount SO8 packages and on very thin PCB tracks:
a. U55, U56, U59, U60 are NE5532D and can be upgraded to your preference - I used 4562MA’s
b. U57 and U61 are OP275GS which can be upgraded to 4562MA or your preferred op amps
I left C44, C400 and C401 in place. You can leave C87 as well. Hope this helps.
Hi ggbaba, when you open up the box you can:
1. Replace the output capacitors which are C1/C2 and C40/C41 on the balanced outputs, C293 and C299 on the unbalanced outputs. These are 470uF 16 V, non-polarised electrolytic – I used Michicon Muse ES. Or just short them out by turning the PCB over and soldering a thin wire across each capacitor’s pins. However this DC couples the DACMagic output to the preamp or amp it is plugged into, so consider using a DC voltmeter first to measure the DC offsets.
2. Replace the power supply capacitors for the op amps/analogue circuits. I used Panasonic FC’s as follows:
a. C76, C77, C402, C403 are 1,000uF 50V
b. C414, C415 are 10uF 35V
c. C419 and C87 (display supply – could leave alone) are 1,000uF 35V I used 1,000uF 50V as 2a above
3. Replace the power supply capacitors for the DSP and DAC digital circuits. I used Rubycon ZA’s and ZLG’s:
a. C424, C425 and C89 are 2,200uF 25V – the first two need to be smaller e.g. ZLG to fit in
b. C8, C10, C25, C26, C31, C35, C45, C130, C426, C431, C435 are 10uF 35V
c. C103 is 100uF 16V
4. Replace the decoupling capacitors around the two WM8740 DAC’s. I used Rubycon ZLG’s:
a. C270, C275, C276 and C301, C303, C304 are all 10uF 35V
b. C271 and C302 are 47uF 35V – I used the DACMagic value and not the 10uF on the WM8740 datasheet
5. Replace the op amps, which are tiny surface mount SO8 packages and on very thin PCB tracks:
a. U55, U56, U59, U60 are NE5532D and can be upgraded to your preference - I used 4562MA’s
b. U57 and U61 are OP275GS which can be upgraded to 4562MA or your preferred op amps
I left C44, C400 and C401 in place. You can leave C87 as well. Hope this helps.
Last edited:
Reply to MatchASM
Hi MatchASM I think the 10uF 35V NP caps are there to decouple the DACs' digital power supply rails, which I found out the hard way. Then I checked the WM8740 datasheet (page 24) to find these capacitors can be polarised electrolytics and 10uF is recommended. One professional upgrader uses Oscon SEPC's there, presumably because of their low impedance. I used Rubycon ZLC's instead.
I see your point about C419 and like the idea of trying a Tantalum in parallel, which I may try next time I open the DACMagic up. C419 is on the +5V supply rail to the analogue circuits in the WM8740 DACs. There are 4 other capacitors on this supply rail including the two 47uF 16V NP electrolytics (which I changed) and two 47nF 16V SMD's.
Hi MatchASM I think the 10uF 35V NP caps are there to decouple the DACs' digital power supply rails, which I found out the hard way. Then I checked the WM8740 datasheet (page 24) to find these capacitors can be polarised electrolytics and 10uF is recommended. One professional upgrader uses Oscon SEPC's there, presumably because of their low impedance. I used Rubycon ZLC's instead.
I see your point about C419 and like the idea of trying a Tantalum in parallel, which I may try next time I open the DACMagic up. C419 is on the +5V supply rail to the analogue circuits in the WM8740 DACs. There are 4 other capacitors on this supply rail including the two 47uF 16V NP electrolytics (which I changed) and two 47nF 16V SMD's.
Last edited:
Thanks Stormsonic, I already have the schematic. I thought the photo and upgrade parts list would help other people considering upgrades inside the box.
I agree with you that the DACMagic's DC offset is tiny compared to the 2,100 mV audio level. It's worth remembering that the amp or preamp it is connected to might not have DC blocking capacitors, or old ones that leak. Then there would be a DC voltage which could affect the DACMagic's op amps. Unlikely but worth a note of caution and a quick test with a Volt Meter.
I just took a look at the schematics. I noticed two things:
1. The effective loading on the NE5532 main amplifier is around 500 Ohms. That is way too low to get good linearity performance, even out of an LM4562/LME49720 or OPA1612/OPA2211. This would be an excellent location for a set of browndogged AD797s.
2. The relay RL6:B goes to L-OUT, whereas I think it should go to R+OUT. Is this a mistake in the diagram or a design (=manufacturing) error? Could someone check? I'm not upening up my dacmagic for the next couple of days...
--
Greetz,
MatchASM
1. The effective loading on the NE5532 main amplifier is around 500 Ohms. That is way too low to get good linearity performance, even out of an LM4562/LME49720 or OPA1612/OPA2211. This would be an excellent location for a set of browndogged AD797s.
2. The relay RL6:B goes to L-OUT, whereas I think it should go to R+OUT. Is this a mistake in the diagram or a design (=manufacturing) error? Could someone check? I'm not upening up my dacmagic for the next couple of days...
--
Greetz,
MatchASM
Schematic
1) Maybe I am missing something here? The 4562 has slew rate +/-20 V/uS and output current of +/-26mA, which is enough for a 1k or even 600R load to more than 192kHz if I have the maths right. The Browndogged AD797's would be quite something.
2) If it helps the photo has the relays from left to right: RL3, RL4, RL5, RL6. On the high res version I can see some of the PCB tracks on the top but not underneath, so I can't be sure what is connected to what. Maybe someone else can help us?
1) Maybe I am missing something here? The 4562 has slew rate +/-20 V/uS and output current of +/-26mA, which is enough for a 1k or even 600R load to more than 192kHz if I have the maths right. The Browndogged AD797's would be quite something.
2) If it helps the photo has the relays from left to right: RL3, RL4, RL5, RL6. On the high res version I can see some of the PCB tracks on the top but not underneath, so I can't be sure what is connected to what. Maybe someone else can help us?
Last edited:
If you look at THD measurements done by Samuel Groner, you will see that although to a 1st order all op-amps will drive 500 Ohms according to the slew rate spec, distortion does suffer nonetheless at higher frequencies. Be reminded: music is much more complex than a sinewave, so the audible differences might be much more severe than the measurement plot suggests.
The THD measurements can be found here.
Therefore, I suspect AD797s in this application will improve the sound big-time (at a few $s of course...).
--
Greetz,
MatchASM
Schematic intended for others who don't have it. Capacitor swaping and opamp rolling is hardly called an "upgrade".
Do some real upgrade work, like fixing digital input, new power supplys for 74VHC153, 74VHC32, 74HCU04, 74HCT14...chips, or adleast isolate them. Get rid of noisy DC-DC switcher inside, make new regulator for clock supply, right now there are ripples over 100mV there
Make new analog section & new power supply for analog section and fix a lot of other things. Been there, done that.
Where is this DC coming from? From preamp coming back to DAC? Then your preamp is faulty.
Last edited:
I disagree. Swapping capacitors and op-amps gives the highest ratio of audible improvement versus time spent, and I'll bet you there are a couple of people who refuse to spend time understanding the schematics and coming up with improvements. To such people, a parts list is helpful.
I agree that the digital supplies are not very good, especially the part that does the USB interfacing. Out-of-the-box, the USB interface is useless. I actually use a SB X-Fi HD USB to convert to optical SPDIF. It's quite nice, as the X-Fi has separate dedicated PLLs for 44.1kHz and for 48kHz.
But I bet the SPDIF jittter can be improved as well by shunt regulators and more aggressive decaps.
Removing AC coupling caps is like sex without condom: it's ok if you know your partner very well, but risky otherwise. I prefer to be able to plug my dacmagic into random strangers, so I'm keeping the caps. Maybe a bypass switch would be a nice idea.
Some preamps do have an input DC different from 0V. Not much wrong with that (although, of course, this means you need AC coupling). In my analogy, this is equivalent to a strange disease, so that's where the analogy goes wrong...
--
Greetz,
MatchASM
Alright. For fun, we could consider the constraint of fitting the additional board into the stock metal casing. Suddenly it becomes a bit more difficult, no?
--
Greetz,
MatchASM
I have done this with the Gigawork Dac board and it seems to be an improvement.
These are the boards I used.
I would use this as an outboard PS.
These are the boards I used.
An externally hosted image should be here but it was not working when we last tested it.
I would use this as an outboard PS.
Last edited:
Good point well made. And that's a very handy link, thank you.
This may be a daft question but is there any way we could reduce the loading on the NE5532's (or their replacements)?