If dacmagic uses a typical opamp output stage design, part of the task of the differential summing stage is to remove DC offset coming out of the dac outputs. If OPA275 was used for that purpose, then bypassing it could cause a problem. Easy enough to measure for an offset at the XLR output with a DVM.
The problem with optical SPDIF is usually mostly from jitter occurring in the optical cable. The capacitor is the least of the problems and is likely there to protect the SPDIF receiver from electrical damage. In other words, if the goal is to fix something that's hurting the sound its probably better to fix what's broke, and leave the good stuff alone. That's my two cents anyway.
The problem with optical SPDIF is usually mostly from jitter occurring in the optical cable. The capacitor is the least of the problems and is likely there to protect the SPDIF receiver from electrical damage. In other words, if the goal is to fix something that's hurting the sound its probably better to fix what's broke, and leave the good stuff alone. That's my two cents anyway.
Thanks for this, and this makes sense, but in current configuration the XLR output does not go anywhere near the OPA275 as I understand it, there are two 5532s per channel that provide L+ L- R+ R- and these go to the XLR output through coupling/isolating caps. They then combine in the OPA275 to provide the RCA output, what I am not sure of is why this provides a preferable solution to just taking the L+ and R+ from the 5532s.
I prefer the sound of the optical input to the USB, from a Mac and since there are 2 inputs for optical/SPDIF I figured that if I ever use a SPDIF I can use the one with the Cap left in place, and IF the optical signal goes through the cap I might as well remove one, as a straight through connection would logically sound better.
If the 5532s are used for I/V conversion of the dac chip output currents then most commonly in circuits like that there is a DC offset at their output pins. I would still suggest to check with a measurement.
The reason for having inverting and non-inverting outputs for each channel is so that a difference amplifier can be used to help cancel out common-mode distortions from the dac chip. If only the + outputs are used then that distortion would not be canceled. The cancelation, and some further filtering of dac output ultrasonic and RF noise from the dac chip usually takes place in a 'differential-summing MFB-filter' opamp stage. Seems likely that could be what the OPA275 is for. Reducing some distortion and filtering out some RF noise can be a good thing, especially so since downstream equipment (say, maybe a power amp) may not sound its best if come RF noise is being fed into it.
Regarding any DC blocking cap in a SPDIF stream, seems to me the worst thing the cap could do would be to affect jitter. It should have zero effect on the ones and zeros of the digital audio data. IME jitter can audibly affect dac sound some. That said, its well known that optical SPDIF cables are probably the weakest link in terms of adding jitter. The reason jitter can affect sound is that SPDIF can be thought of as a combination of digital data and analog timing reference information. When caps affect sound it is generally due to effects on analog signals, and on power supplies feeding analog signal handling circuitry. For the present discussion I am including RF clocking circuitry and or any RF phase locked loop circuitry as effectively more analog rather than digital.
The other thing is that it doesn't seem logical at all to me that removing every cap should result in better sound. Quite the contrary. There are caps that that it might help to bypass and or replace with better caps, and there are other caps that help improve make the sound good. In other words, its not so simple as your 'logic' would seem to dictate.
The reason for having inverting and non-inverting outputs for each channel is so that a difference amplifier can be used to help cancel out common-mode distortions from the dac chip. If only the + outputs are used then that distortion would not be canceled. The cancelation, and some further filtering of dac output ultrasonic and RF noise from the dac chip usually takes place in a 'differential-summing MFB-filter' opamp stage. Seems likely that could be what the OPA275 is for. Reducing some distortion and filtering out some RF noise can be a good thing, especially so since downstream equipment (say, maybe a power amp) may not sound its best if come RF noise is being fed into it.
Regarding any DC blocking cap in a SPDIF stream, seems to me the worst thing the cap could do would be to affect jitter. It should have zero effect on the ones and zeros of the digital audio data. IME jitter can audibly affect dac sound some. That said, its well known that optical SPDIF cables are probably the weakest link in terms of adding jitter. The reason jitter can affect sound is that SPDIF can be thought of as a combination of digital data and analog timing reference information. When caps affect sound it is generally due to effects on analog signals, and on power supplies feeding analog signal handling circuitry. For the present discussion I am including RF clocking circuitry and or any RF phase locked loop circuitry as effectively more analog rather than digital.
The other thing is that it doesn't seem logical at all to me that removing every cap should result in better sound. Quite the contrary. There are caps that that it might help to bypass and or replace with better caps, and there are other caps that help improve make the sound good. In other words, its not so simple as your 'logic' would seem to dictate.
Last edited:
The overall concept of side-stepping the OPA275 to take a singe +ve on each channel from the 5532 via the XLR outputs is coming from earlier opinion in the thread as a step forward in SQ, however my experience so far is to prefer the original output through the RCA, and I am close to returning to this after listening for a while.
I assume the output of each 5532 is 2.1V based upon the spec, (balanced 4.2V, RCA 2.1V) and so the OPA275 must receive + 2.1, and -2.1 and then output half of the difference somehow ?
Thanks again, all intriguing stuff. My experience so far is that if a cap is used in the signal to block DC then it almost always sounds better if it's removed, and this is usually achievable if in your own controlled system you know what it is connected to, and can prevent the output and the input both having a capacitor to block DC when you only need one.
Its not new that some people prefer the sound from only the + output from some dacs, and or maybe if used with reproduction systems that are tolerant of less noise filtering taking place in the dac, etc. One possible benefit from bypassing the last opamp is that it may sound like one less opamp is in the signal chain. In other words, some people say they don't like the sound of opamps, or maybe the sound of too many opamps.
Hey quick question the caps surrounding the two DACS are not marked as polar, and the recommended examples here still uses polar replacements.
It's the 10 uF 50V that came out C275, C303 C301 etc , and the 47 uF 16V see photo.
Can I put polar caps in, does it matter on the orientation or am I best to select some bipolar examples like the Nichicon ES.
Thanks in advance for the support
It's the 10 uF 50V that came out C275, C303 C301 etc , and the 47 uF 16V see photo.
Can I put polar caps in, does it matter on the orientation or am I best to select some bipolar examples like the Nichicon ES.
Thanks in advance for the support
Attachments
What do you think? And do you have a DMM? You could measure voltages to see what + and - are. They are decoupling caps mostly AFAIK so polar ones are to be preferred.
BTW C15 is missing?!? It should be there. Same value as C55 I think. A case of part rolling with damage?
good spot on the C15 missing, I will have to find the value and drop in a conventional cap here.
I have a DMM, and could measure the voltages on another DM device as I have two - I wonder why the original installation chose not to consider Polar caps here?
If you are worried about cap differences then replace the other SMD caps all for the same brand/type/value. Yes all 8 of them. Keep it simple and coherent. Consider it the risk that comes with stirring in equipment.
Don’t know about the polar/non polar as I haven’t seen the ones that were taken out. Best tip is to check the pins of the IC’s and check the datasheet to know what is what.
Don’t know about the polar/non polar as I haven’t seen the ones that were taken out. Best tip is to check the pins of the IC’s and check the datasheet to know what is what.
Last edited:
Found the spec on the Wolfson chip and since C15 is between pins 9 and 10 it's termed C4 in the specification sheet and therefore 0.1uF I think
https://www.farnell.com/datasheets/484663.pdf
It's a shame nobody seems to have a circuit diagram any more for the DacMagic, too old I guess!
https://www.farnell.com/datasheets/484663.pdf
It's a shame nobody seems to have a circuit diagram any more for the DacMagic, too old I guess!
Well then you have introduced non symmetry with parts. One out of 8 is then replaced for a random brand/type/value which is a true nightmare for cap rollers if I have understood them correctly. In this case the non symmetry is between channels. Horror!
BTW I would measure the value as maybe the manufacturer chose a different value.
Technical approach, you get used to it.
* the cap is between pin 9 and 10 but one of those is + and the other is GND so……
BTW I would measure the value as maybe the manufacturer chose a different value.
Technical approach, you get used to it.
* the cap is between pin 9 and 10 but one of those is + and the other is GND so……
Last edited:
So for C15 I can use a polar and put the +ve on pin9, and I float off C55 and copy with a non SMD Cap that matches for pure symmetry
Dear cap roller, C15 is a non polar SMD film cap AFAIK and closest to the ICs. A very good choice by the manufacturer. If you want to correct the introduced error then measure the other caps like C55, C54 and C308. I think they are all the same value but checking = knowing and knowing = not asking. Maybe it is a good idea to use SMD non polar film caps for those 4 caps again in the same brand/type/value/size in both channels. So no more RR (random rolling) but a technical choice making stuff right. Scary! If you are lucky then the missing cap is somewhere in the device. Take a good look and if you find it check/measure it. Would be the easiest option except then different solder will be used. Oh oh.
And you will have to get used to SMD as it is standard for at least 20 years now. Just do it. The plus side of matters is that there are then even more caps to roll.
* the caps indicated with a circle on the PCB are all electrolytic caps which are polar for decoupling usually. I don't know why the "+" has been omitted as one would think of bipolar caps but those don't make sense for decoupling duty. If you have kept the old ones you could take a picture. Strange things may have been done as this brand and designs sometimes have dubious choices in the cap area and fail more than normal.
** you can't go wrong with measuring the electrolytic cap pads as WM8740 only has positive power supplies so if you measure continuity between a pin and GND then it is the caps "-". Mark the pad outside its circle clearly visible with a "-" and make sure you solder the caps right in.
And you will have to get used to SMD as it is standard for at least 20 years now. Just do it. The plus side of matters is that there are then even more caps to roll.
* the caps indicated with a circle on the PCB are all electrolytic caps which are polar for decoupling usually. I don't know why the "+" has been omitted as one would think of bipolar caps but those don't make sense for decoupling duty. If you have kept the old ones you could take a picture. Strange things may have been done as this brand and designs sometimes have dubious choices in the cap area and fail more than normal.
** you can't go wrong with measuring the electrolytic cap pads as WM8740 only has positive power supplies so if you measure continuity between a pin and GND then it is the caps "-". Mark the pad outside its circle clearly visible with a "-" and make sure you solder the caps right in.
Last edited:
Thanks for the advice I now have the polarity on these decoupling caps measured and marked.
awesome, thanks so much
So the lost capacitor in question C15 is 47nF = 0.05uF, so half of the Wolfson
spec sheet at 0.1uF. I may try to swap the lost C15 with a Wima 0.1uF and see how this works
HI
No Wima in this location, please....
You should put a 47nF 16V Panasonic ECHU 1206 plastic film in there, same as C54/55/308 You allready have.
I like for C270/276/301/304 ( VMIDL/R ) nichicon UES . Try 22uF/25V.
For C275/303 (+5V digital ) panasonic FM/FR 100uF/25V work well ( double check polarity before....)
Use for C271/302 a nice audio grade nichicon of your choise ( polarity check and do not forget to check diameter before order )
In C419 position use an audio grade cap 1000uF/16V or 25V if fits, no low ESR please
Panasonic ECHU are easely danmaged if too much heat is applied
Have a nice day, amigo
No Wima in this location, please....
You should put a 47nF 16V Panasonic ECHU 1206 plastic film in there, same as C54/55/308 You allready have.
I like for C270/276/301/304 ( VMIDL/R ) nichicon UES . Try 22uF/25V.
For C275/303 (+5V digital ) panasonic FM/FR 100uF/25V work well ( double check polarity before....)
Use for C271/302 a nice audio grade nichicon of your choise ( polarity check and do not forget to check diameter before order )
In C419 position use an audio grade cap 1000uF/16V or 25V if fits, no low ESR please
Panasonic ECHU are easely danmaged if too much heat is applied
Have a nice day, amigo
47 nF = 0.047 uF. Maybe the manufacturer had issues with 0.1 uF film caps in parallel to the electrolytic caps for decoupling. Maybe the regulator was not happy with such low ESR connected and was the 47 nF/electrolytic cap combo best performing without regs oscillating . There sometimes are very good reasons that things are what they are as some engineers looked at it and probably tested and measured things. If you are the "I just do something" type then tube audio seems the way to go. Forgive me my cynical way of looking at matters but I fix stuff damaged by opamp/cap rollers that don't need schematics, measuring equipment or factual knowledge to know what performs best. They decide by ear to take away veils and are able to make "blacker than black" soundscapes with just some caps
often making "more oscillating than necessary" devices.The tip bicefalo gave is a good one. You could use 47nF 16V Panasonic ECHU 1206 for C15/54/55/308. I am not so sure about using ultra low ESR caps like Panasonic FM for C275 so I would advise to check the schematic what regulator is used and if it accepts ultra low ESR caps. The datasheet specifically mentions NP caps so I think the ESR thingie is actual. Non polar electrolytic caps have higher ESR than normal ones.
Having seen the schematic I am pretty sure upgrading the 7815/7915 and 7805 makes way more sense than rolling caps. With an external multi voltage PSU connected via a multi connector things will also improve. Then the internal switcher can go.
Last edited:
Thanks again for replies47 nF = 0.047 uF. Maybe the manufacturer had issues with 0.1 uF film caps in parallel to the electrolytic caps for decoupling. Maybe the regulator was not happy with such low ESR connected and was the 47 nF/electrolytic cap combo best performing without regs oscillating . There sometimes are very good reasons that things are what they are as some engineers looked at it and probably tested and measured things. If you are the "I just do something" type then tube audio seems the way to go. Forgive me my cynical way of looking at matters but I fix stuff damaged by opamp/cap rollers that don't need schematics, measuring equipment or factual knowledge to know what performs best. They decide by ear to take away veils and are able to make "blacker than black" soundscapes with just some caps
The tip bicefalo gave is a good one. You could use 47nF 16V Panasonic ECHU 1206 for C15/54/55/308. I am not so sure about using ultra low ESR caps like Panasonic FM for C275 so I would advise to check the schematic what regulator is used and if it accepts ultra low ESR caps. The datasheet specifically mentions NP caps so I think the ESR thingie is actual. Non polar electrolytic caps have higher ESR than normal ones.
Having seen the schematic I am pretty sure upgrading the 7815/7915 and 7805 makes way more sense than rolling caps. With an external multi voltage PSU connected via a multi connector things will also improve. Then the internal switcher can go.
So I am replacing the caps in here primarily because many have leaked, and they are old electrolytic - C15 was a casualty probably due to an old dry joint that got nudged as I did not have my soldering iron anywhere near it. I want to use the device again and if I get an improvement in SQ at the same time then that would be very nice.
Removing the SMPS and putting in an external PSU is interesting but for now I am just trying to get it working again.
I still have the original electrolytic that came out around the DACs and had chosen to swap them as I suspected that they might be the next to leak, I can of could of course not bother to change these and reinstall the original NP back in place.
Why is it so crazy to consider a Wima 0.1uF for C15 if I have one already - is this a technical or SQ/subjective advice?