I came onto the Class D bandwagon quite late with BRZ HiFi TPA3255 amp last year. Since then, I found diyaudio and I am grateful that I have learned a lot from many posters here.
To me, the attractiveness of 3e Audio SY-DAP1002 is in its choice of TPA3250 over TPA3255, and the inclusion of ADAU1701 programmable DSP chip. The both points are not at all widely accepted as a virtue, and are subject to debate. I will go into each, but in short, this 3250 sounds better than most 3255 for me.
Here's the initial review to start with:
The 3eAudio SY-DAP1002 arrived on March 9th here in Tokyo where I live. It came with Rev. 1.1 pcb, which means it was in a production batch after the initial 100 Rev. 1.0 boards with OPA1602 OpAmp were sold. Rev. 1.1 comes with DIP8 OpAmp sockets in place of the soldered SOIC OPA1602, and through-hole Elna Silmic II's replacing various through-hole and SMD capacitors.
As you can see, 3 big through-hole PS caps at C134, C150 and C151 came unpopulated, and what you might not see is the four "Background noise reduction filter" positions came fully populated with SMD chips. I was not sure if I'd get the Rev. 1.0 when I placed the order, and was ready to do this mod myself, so it was a nice surprise to find the amp totally silent at full volume with no signal from the listening position. The noise, the level of which does not change with volume position, is ever slightly audible with my ear at less than 1" away from a tweeter.
(Please ignore the CoilCraft info, that has nothing to do with DAP1002 layout.)
When compared to BRZ HiFi TPA3255 (unmodified, but with MUSES8920), the sound without any modification or aging was mediocre at best. There was a slight advantage or emphasis on mid range 'energy', but otherwise the BRZ HiFi with the sweet OpAmp outperformed it in almost every way.
3e Audio has not published the Sigma Studio .dspproj file yet for making the SY-DAP1002 volume control logarithmic instead of straight line increase, but one noticeable characteristic was the high level of gain at the lowest volume position, giving the impression like going from 3250 to the higher power 3255, when in reality I was used to 3255 of the BRZ to try the lower power 3250 of this 3e Audio. This later led me to believe the OpAmp is not used with unity gain, used probably at a gain of 2, just like in Aiyima TPA3251. This likely makes the amp more accommodating to exotic OpAmps that are not designed for audio which are more susceptible to oscillation, and is a huge designed-in advantage these two amps have over many others in 1. Wider OpAmp rolling choices, and 2. A lighter gain demand on the TI TPA chips at any given sound pressure level (SPL), in my opinion.
Anyway, the culprit was found as soon as I tried OPA1656 on this amp:
SY-DAP1002 Rev. 1.1 came with this SOIC 'N5532' on a SOIC to DIP conversion board.
Just by swapping the OpAmp to OPA1656, this 3e Audio amp outperformed the BRZ HiFi 3255 with MUSES8920 in almost every aspect of the sound quality. Probably the 'N5532' is a fake, but who would make a fake 'SOIC' NE5532 (produced mostly in DIP8 package in the Millions) for what purpose?? It really is a strange world.
Later I compared the two amps both with OPA1656, and the result became closer but did not change in the end. The largest difference was the midrange energy (or slightly more level of midrange 'punch' 3e Audio has) that made the music more involving and enjoyable. With the OPA1656, the both amps have excellent highs and resolution without being harsh, sterile or too analytical.
The unique parts of SY-DAP1002 that are shared with its TPA3255 version, SY-DAP2002. Top left is the STM32 32bit micro CPU by ST Micro that interprets the push button volume control knob switch (Power on = wake up from standby, power off = not a complete shut down but standby, bluetooth link-up, etc.), controls the LED ring around the knob, and more importantly, it detects if an established Bluetooth connection to a host is currently sending an audio signal. When the bluetooth link is idle for a while, it redirects the input from Bluetooth to RCA. Unlike many other amps with a bluetooth connectivity like the BRZ, once-connected smart phones do not need to switch off the bluetooth for the amp to restart accepting input from the RCA connector because of this. Lower left is Qualcomm QCC3008 chip for Bluetooth 5.0 with AptX, SBC and AAC codecs. With a long-ish antenna that comes with the amp, the bluetooth connectivity is quite good.
On the middle right is the ADAU1701 programmable DSP. The whitish square below it is the crystal, and the 8pin SOIC to the left of the crystal is FT24C64A EEPROM for AU1701 (STM32 has its program/data storage EEPROM built in). ADAU1701 and the EEPROM are accessible via 10pin USBi interface connector at the front of pcb, that requires a USBi programmer and an USB cable to a PC that runs Analog Devices' Sigma Studio DSP development environment (IDE) that can be downloaded for free.
I will be going into my background, the mods and eventually into the dsp programming, so stay tuned.
To me, the attractiveness of 3e Audio SY-DAP1002 is in its choice of TPA3250 over TPA3255, and the inclusion of ADAU1701 programmable DSP chip. The both points are not at all widely accepted as a virtue, and are subject to debate. I will go into each, but in short, this 3250 sounds better than most 3255 for me.
Here's the initial review to start with:
The 3eAudio SY-DAP1002 arrived on March 9th here in Tokyo where I live. It came with Rev. 1.1 pcb, which means it was in a production batch after the initial 100 Rev. 1.0 boards with OPA1602 OpAmp were sold. Rev. 1.1 comes with DIP8 OpAmp sockets in place of the soldered SOIC OPA1602, and through-hole Elna Silmic II's replacing various through-hole and SMD capacitors.
As you can see, 3 big through-hole PS caps at C134, C150 and C151 came unpopulated, and what you might not see is the four "Background noise reduction filter" positions came fully populated with SMD chips. I was not sure if I'd get the Rev. 1.0 when I placed the order, and was ready to do this mod myself, so it was a nice surprise to find the amp totally silent at full volume with no signal from the listening position. The noise, the level of which does not change with volume position, is ever slightly audible with my ear at less than 1" away from a tweeter.
(Please ignore the CoilCraft info, that has nothing to do with DAP1002 layout.)
When compared to BRZ HiFi TPA3255 (unmodified, but with MUSES8920), the sound without any modification or aging was mediocre at best. There was a slight advantage or emphasis on mid range 'energy', but otherwise the BRZ HiFi with the sweet OpAmp outperformed it in almost every way.
3e Audio has not published the Sigma Studio .dspproj file yet for making the SY-DAP1002 volume control logarithmic instead of straight line increase, but one noticeable characteristic was the high level of gain at the lowest volume position, giving the impression like going from 3250 to the higher power 3255, when in reality I was used to 3255 of the BRZ to try the lower power 3250 of this 3e Audio. This later led me to believe the OpAmp is not used with unity gain, used probably at a gain of 2, just like in Aiyima TPA3251. This likely makes the amp more accommodating to exotic OpAmps that are not designed for audio which are more susceptible to oscillation, and is a huge designed-in advantage these two amps have over many others in 1. Wider OpAmp rolling choices, and 2. A lighter gain demand on the TI TPA chips at any given sound pressure level (SPL), in my opinion.
Anyway, the culprit was found as soon as I tried OPA1656 on this amp:
SY-DAP1002 Rev. 1.1 came with this SOIC 'N5532' on a SOIC to DIP conversion board.
Just by swapping the OpAmp to OPA1656, this 3e Audio amp outperformed the BRZ HiFi 3255 with MUSES8920 in almost every aspect of the sound quality. Probably the 'N5532' is a fake, but who would make a fake 'SOIC' NE5532 (produced mostly in DIP8 package in the Millions) for what purpose?? It really is a strange world.
Later I compared the two amps both with OPA1656, and the result became closer but did not change in the end. The largest difference was the midrange energy (or slightly more level of midrange 'punch' 3e Audio has) that made the music more involving and enjoyable. With the OPA1656, the both amps have excellent highs and resolution without being harsh, sterile or too analytical.
The unique parts of SY-DAP1002 that are shared with its TPA3255 version, SY-DAP2002. Top left is the STM32 32bit micro CPU by ST Micro that interprets the push button volume control knob switch (Power on = wake up from standby, power off = not a complete shut down but standby, bluetooth link-up, etc.), controls the LED ring around the knob, and more importantly, it detects if an established Bluetooth connection to a host is currently sending an audio signal. When the bluetooth link is idle for a while, it redirects the input from Bluetooth to RCA. Unlike many other amps with a bluetooth connectivity like the BRZ, once-connected smart phones do not need to switch off the bluetooth for the amp to restart accepting input from the RCA connector because of this. Lower left is Qualcomm QCC3008 chip for Bluetooth 5.0 with AptX, SBC and AAC codecs. With a long-ish antenna that comes with the amp, the bluetooth connectivity is quite good.
On the middle right is the ADAU1701 programmable DSP. The whitish square below it is the crystal, and the 8pin SOIC to the left of the crystal is FT24C64A EEPROM for AU1701 (STM32 has its program/data storage EEPROM built in). ADAU1701 and the EEPROM are accessible via 10pin USBi interface connector at the front of pcb, that requires a USBi programmer and an USB cable to a PC that runs Analog Devices' Sigma Studio DSP development environment (IDE) that can be downloaded for free.
I will be going into my background, the mods and eventually into the dsp programming, so stay tuned.
Last edited:
After over 35 years into high-end audio, I believe the listening room resonance/reflection/absorption characteristics to be the most expensive and difficult item in the hobby as a modification requires an approval of my wife with strict limitations. I bought one of the first Digital Signal Processing (DSP) 'box' with 8 or so adjustable frequency bands in a parametric EQ filter (to be placed in between MC phono amp and pre-amp, or in between the pre-amp and main amp with RCA cables) to tailor the frequency response curve almost 30 years ago. The product was made by Denon or Panasonic, I forgot which, but no matter how much it cost, it was a cheap investment in my mind in light of what it promised, including the compensation for imperfect amplifiers and speakers in addition to room acoustics.
At the time, I did not have a reliable way to measure the room acoustics, so all the adjustment had to be done by my ears (not 100% sure but my speakers were the original Martin Logan Monolith then), and it took a long time for me to realize that 'life' in the music is lost in the DSP process no matter what the display on the DSP box says the frequency curve is. I regret selling the contraption without disassembling it, so I don't know what OpAmp, ADC and DAC were used in it, probably some of them were discreet circuits. The maker did publish the SNR figure, but not the THD on the box let alone the probably-horrendously-dismal sampling rate. No consumer electronics maker published a THD+N in a combined figure at the time I believe.
Deeply disillusioned, I didn't touch a DSP ever since until I saw this 3e Audio SY-DAP1002 at 73.66Euros + 9.66Euros shipping to Japan on AliExpress. I looked into this ADAU1701 DSP IC and the current state of DSP world, and could not believe:
1. Dayton Audio iMM-6 condenser microphone ($30 in Japan, discussed elsewhere on diyaudio) for measurements comes with free calibration file on each mic (the product comes with a serial number, and you can download the file for that particular product number free of charge).
2. Room EQ Wizard (REW) software is free, and is able to take reliable measurements with built in low to high sweeping audio tone generator, and a capability to compensate the measurements with DAC and mic calibration files.
3. This measurement file (after a bit tedious conversion) can be used in the Sigma Studio automatic parametric filter generator (single channel, two required in the IDE for L and R) to let it design a combination of filters theoretically best to counter the dips and peaks in the file, within the design parameters like number of filters I could specify.
4. This set of filters can be examined by my ears on the fly by playing music with the dsp/amp real time, and by my eyes/brain to adjust the specifics like peak or cut-off frequency point, the Q (steepness of the filter slope), level of boost (up or down), etc in numbers for each filter.
5. After the examination and manual adjustment to my liking, the set of filters can then be set on a single stereo parametric filter to be written onto the EEPROM for AU1701 on the amp. (I can leave the automatically generated two filters as is, but they are a heavy burden on the 1701 resources)
6. ADAU1701 apparently is AD's second generation of such ADC/DSP/DAC combo IC, and its power and features are impressive compared to the similarly low-cost alternatives in the past.
All at the unbelievably low price of $100 for the ADC/DSP/DAC/amplifier combo, $20 for the USBi programmer (the 3rd item on this linked page) and USB extension cable, and $30 for the mic plus an audio cable with 3.5mm 3 position female and male jacks on the ends.
I know it is theoretically much better to place the DSP in the digital stage of the chain from CD transport or server -> DAC -> Amplifier -> Speaker, so that the additional set of analog to digital conversion and digital to analog conversion in the ADAU1701 is omitted, but what such a chain possibility could beat or even come close to this price? It was too good a deal to pass up for my second crack at DSP in my life. With a lot of respect to the efforts by Analog Devices, yes, I do consider myself to be finally out of the "High End"
At the time, I did not have a reliable way to measure the room acoustics, so all the adjustment had to be done by my ears (not 100% sure but my speakers were the original Martin Logan Monolith then), and it took a long time for me to realize that 'life' in the music is lost in the DSP process no matter what the display on the DSP box says the frequency curve is. I regret selling the contraption without disassembling it, so I don't know what OpAmp, ADC and DAC were used in it, probably some of them were discreet circuits. The maker did publish the SNR figure, but not the THD on the box let alone the probably-horrendously-dismal sampling rate. No consumer electronics maker published a THD+N in a combined figure at the time I believe.
Deeply disillusioned, I didn't touch a DSP ever since until I saw this 3e Audio SY-DAP1002 at 73.66Euros + 9.66Euros shipping to Japan on AliExpress. I looked into this ADAU1701 DSP IC and the current state of DSP world, and could not believe:
1. Dayton Audio iMM-6 condenser microphone ($30 in Japan, discussed elsewhere on diyaudio) for measurements comes with free calibration file on each mic (the product comes with a serial number, and you can download the file for that particular product number free of charge).
2. Room EQ Wizard (REW) software is free, and is able to take reliable measurements with built in low to high sweeping audio tone generator, and a capability to compensate the measurements with DAC and mic calibration files.
3. This measurement file (after a bit tedious conversion) can be used in the Sigma Studio automatic parametric filter generator (single channel, two required in the IDE for L and R) to let it design a combination of filters theoretically best to counter the dips and peaks in the file, within the design parameters like number of filters I could specify.
4. This set of filters can be examined by my ears on the fly by playing music with the dsp/amp real time, and by my eyes/brain to adjust the specifics like peak or cut-off frequency point, the Q (steepness of the filter slope), level of boost (up or down), etc in numbers for each filter.
5. After the examination and manual adjustment to my liking, the set of filters can then be set on a single stereo parametric filter to be written onto the EEPROM for AU1701 on the amp. (I can leave the automatically generated two filters as is, but they are a heavy burden on the 1701 resources)
6. ADAU1701 apparently is AD's second generation of such ADC/DSP/DAC combo IC, and its power and features are impressive compared to the similarly low-cost alternatives in the past.
All at the unbelievably low price of $100 for the ADC/DSP/DAC/amplifier combo, $20 for the USBi programmer (the 3rd item on this linked page) and USB extension cable, and $30 for the mic plus an audio cable with 3.5mm 3 position female and male jacks on the ends.
I know it is theoretically much better to place the DSP in the digital stage of the chain from CD transport or server -> DAC -> Amplifier -> Speaker, so that the additional set of analog to digital conversion and digital to analog conversion in the ADAU1701 is omitted, but what such a chain possibility could beat or even come close to this price? It was too good a deal to pass up for my second crack at DSP in my life. With a lot of respect to the efforts by Analog Devices, yes, I do consider myself to be finally out of the "High End"
Attach the images in the normal way and then copy the link/s and paste it into the message. I've just inserted one image here. Add the IMG tags as shown. Sizing images can be a problem though.
Attachments
Thank you. I want to embed a thumbnail of an image in the body of a message. It doesn't matter if the image is attached or is an url somewhere else.
Trial embedding of attached image in the body of post as thumbnail:
[thumbnail]https://www.diyaudio.com/forums/attachment.php?attachmentid=936049&stc=1&d=1616798716[/thumbnail]
Obviously the above does not work as there is no tag like [thumbnail]. Also, [raw] tag does not appear to work as described in
Wiki Formatting And Syntax - diyAudio Guides
This works (ignore "HTML Code:"):
but it displays the image in full size like this:
And the following do not work:
[image|936049|caption=As described in diyaudio guides][/image]
[image|1616798716|size=small][/image]
[image|link=https://www.diyaudio.com/forums/attachment.php?attachmentid=936049|align=center|caption=This format should work according to the above diyaudio guide, and probably it did at one time][/image]
![url]](/community/proxy.php?image=http%3A%2F%2F%5Burl%5Dhttps%3A%2F%2Ffiles.diyaudio.com%2Fforums%2Fimages%2Fsmilies%2Fconfused.gif%5B%2Furl%5D&hash=2167e6f9cdea7bc145e633e2118e2081)
Is there a complete list of ++CURRENTLY VALID++ Wiki or xenForo tags for DiyAudio Forum somewhere?
`
Trial embedding of attached image in the body of post as thumbnail:
[thumbnail]https://www.diyaudio.com/forums/attachment.php?attachmentid=936049&stc=1&d=1616798716[/thumbnail]
Obviously the above does not work as there is no tag like [thumbnail]. Also, [raw] tag does not appear to work as described in
Wiki Formatting And Syntax - diyAudio Guides
This works (ignore "HTML Code:"):
HTML:
[img]https://www.diyaudio.com/forums/attachment.php?attachmentid=936049&stc=1&d=1616798716][/img]And the following do not work:
[image|936049|caption=As described in diyaudio guides][/image]
[image|1616798716|size=small][/image]
[image|link=https://www.diyaudio.com/forums/attachment.php?attachmentid=936049|align=center|caption=This format should work according to the above diyaudio guide, and probably it did at one time][/image]
Is there a complete list of ++CURRENTLY VALID++ Wiki or xenForo tags for DiyAudio Forum somewhere?
`
I know its not much help at the moment but hopefully all these image issues will resolve when we move to XF, something that really is well under way now.
Quite a bit of the old functionally of the forum doesn't work as it used in years gone by but I'm afraid I'm no expert on the technical reasons why.
Quite a bit of the old functionally of the forum doesn't work as it used in years gone by but I'm afraid I'm no expert on the technical reasons why.
OK, continuing on with the review:
The volume control knob is the only means of normal control on this DAP1002 amp. Because of the LED ring surrounding the knob, it has the appearance of being a stepping rotary switch, but it's a regular pot with smooth rotation, probably sending a variable voltage to the CPU and ADAU1701. I am not that sure, but the volume control seems to be done at AU1701 dsp without changing the gain in TPA3250 amp.
In place of a power switch, the knob can be pushed in to send click-and-hold signals to the CPU, like short single click = Power On, click and hold for 3 seconds = Power Off.
Two points that might make this amp better than Aiyima A04 for our purposes:
1. The case has about 34mm space above the PCB, so I can install 33mm tall parts on the board.
2. The PCB seems to take quite a bit of thermal abuse without issues.
Another point rather unique to SY-DAP1002 is the inclusion of power supply (PS) in the case, but this feature is shared by BRZ HiFi 3255. However, BRZ's PS is an universal model that is used by many other chinese amplifiers, adjusted to 40V, bolted to the case with the main pcb board cut off to accommodate the space for PS in the case.
On the other hand, 3e Audio designed a "Quasi-Resonant" power supply dedicated for this amp using a Low-Dropout regulator at 30V as a part of the main board. I don't know what Quasi-Resonant actually means, but I measured AC320Vrms at one point on the PS circuit when feeding the low eastern japanese 100V/50Hz house current/voltage, and I would think:
At the least - PS is designed specifically for the requirement of this amp.
At the most - PS shares the frequency signal from the crystal so that the switching power generation frequency of the PS partly or wholly coincides (resonates) with the current demand frequency of TPA3250 PWM stage (at 600kHz?).
(3e Audio SY-DAP2002 (TPA3255) uses basically the same PS at a higher voltage. Aiyima A04 measured at about 50W peak output in the ASR test because of TPA3251 thermal protection limiting the power.)
Main board underside before the mod (top side pic is at the beginning of this thread):
TPA3250 has its thermal pad facing down (mounted on the top side), so the PCB is the primary means of heat dissipation. Heatsink on the top side being rather small is not much of a relevant issue.
SY-DAP1002 antenna on the right, BRZ's on the left. Both with the cap off.
The volume control knob is the only means of normal control on this DAP1002 amp. Because of the LED ring surrounding the knob, it has the appearance of being a stepping rotary switch, but it's a regular pot with smooth rotation, probably sending a variable voltage to the CPU and ADAU1701. I am not that sure, but the volume control seems to be done at AU1701 dsp without changing the gain in TPA3250 amp.
In place of a power switch, the knob can be pushed in to send click-and-hold signals to the CPU, like short single click = Power On, click and hold for 3 seconds = Power Off.
Two points that might make this amp better than Aiyima A04 for our purposes:
1. The case has about 34mm space above the PCB, so I can install 33mm tall parts on the board.
2. The PCB seems to take quite a bit of thermal abuse without issues.
Another point rather unique to SY-DAP1002 is the inclusion of power supply (PS) in the case, but this feature is shared by BRZ HiFi 3255. However, BRZ's PS is an universal model that is used by many other chinese amplifiers, adjusted to 40V, bolted to the case with the main pcb board cut off to accommodate the space for PS in the case.
On the other hand, 3e Audio designed a "Quasi-Resonant" power supply dedicated for this amp using a Low-Dropout regulator at 30V as a part of the main board. I don't know what Quasi-Resonant actually means, but I measured AC320Vrms at one point on the PS circuit when feeding the low eastern japanese 100V/50Hz house current/voltage, and I would think:
At the least - PS is designed specifically for the requirement of this amp.
At the most - PS shares the frequency signal from the crystal so that the switching power generation frequency of the PS partly or wholly coincides (resonates) with the current demand frequency of TPA3250 PWM stage (at 600kHz?).
(3e Audio SY-DAP2002 (TPA3255) uses basically the same PS at a higher voltage. Aiyima A04 measured at about 50W peak output in the ASR test because of TPA3251 thermal protection limiting the power.)
Main board underside before the mod (top side pic is at the beginning of this thread):
TPA3250 has its thermal pad facing down (mounted on the top side), so the PCB is the primary means of heat dissipation. Heatsink on the top side being rather small is not much of a relevant issue.
SY-DAP1002 antenna on the right, BRZ's on the left. Both with the cap off.
Thermal management of an audio amp is important not only for performance and longevity, but also for the effect on my day to day life. I used Pass Lab X3 for decades with its 200W idle power consumption, effectively having a 200W electric heater in my living room, as I don't have a dedicated listening room - a cost of living in Tokyo. X3 takes hours to reach its optimal operating temperature, so I installed a thermo-couple thermometer to monitor it, and the amp was regularly switched on year-round even in the middle of Summer. If I seriously wanted to enjoy music with a quality sound, I needed to plan ahead to switch on the power, or had to keep it on all the time, just like with a big tube amp.
This thermal management is a complicated subject, and BRZ HiFi excels in this field in my opinion. BRZ HiFi 3255 (does not have a fan) has the top and bottom of the case well vented, and I can feel a slightly warm airflow with the palm of my hand over the top after about 20 minutes of normal use.
Note the four mounting screws and slightly visible aluminum underside of the universal PS frame/heatsink on this BRZ HiFi 3255. Four relatively large polyurethane(?) 'feet' in the corners create a considerable 8.5mm of space underneath the amp for airflow.
This thermally well designed case may not be good enough without a fan for the thermal dissipation of TPA3255 (thermal pad facing up - so a heatsink on top is the primary means of heat dissipation) if it is driven at 48-51V PVDD. But BRZ HiFi limits it at 40Vdc, and the parts inside are always kept at a comfortable "slightly to moderately warm to the touch" level.
TPA3250 with a higher power efficiency, lower heat dissipation and the thermal pad facing down, is a different animal. I was initially skeptical, but the no-vent design of the 3e Audio SY-DAP1002 (PVDD=30V) is finally making sense to me after many days of constant use. The heat generated by TPA3250 is transferred to the PCB, then to the cast-in rails on the side of the case, then dissipated to the ambient air by the case itself acting as a heatsink, which never got hotter than "slightly warm to the touch" level so far. Although SY-DAP2002 case with TPA3255 is longer front to back thus it could be a heftier heatsink if the thermal pad was facing down, installing a fan on an unvented case does not make sense to me.
If the face-down thermal pad dictated a thicker copper, better quality PCB in the design, I would consider this to be a major advantage of TPA3250 over TPA3251/TPA3255 in a majority of chinese amplifiers. TI application note on modern thermal management is at https://www.ti.com/lit/an/spra953c/spra953c.pdf?ts=1616907719303
Front panel has a sturdy 8mm thickness, allowing the mounting posts for volume control pcb to be fixed on the back side of it without the ends of the post protruding to the front surface. Small rubber feet create about 5mm space underneath, 3mm under the front panel.
This aluminum case is well designed in the detail, like the meeting surface (3.5mm thickness) of the upper and the bottom half shells having a front-to-back length-wise convex 'ridge' on one part, and a receiving concave 'trough' on the other for a snug fit.
Back panel does not have a fan.
`
This thermal management is a complicated subject, and BRZ HiFi excels in this field in my opinion. BRZ HiFi 3255 (does not have a fan) has the top and bottom of the case well vented, and I can feel a slightly warm airflow with the palm of my hand over the top after about 20 minutes of normal use.
Note the four mounting screws and slightly visible aluminum underside of the universal PS frame/heatsink on this BRZ HiFi 3255. Four relatively large polyurethane(?) 'feet' in the corners create a considerable 8.5mm of space underneath the amp for airflow.
This thermally well designed case may not be good enough without a fan for the thermal dissipation of TPA3255 (thermal pad facing up - so a heatsink on top is the primary means of heat dissipation) if it is driven at 48-51V PVDD. But BRZ HiFi limits it at 40Vdc, and the parts inside are always kept at a comfortable "slightly to moderately warm to the touch" level.
TPA3250 with a higher power efficiency, lower heat dissipation and the thermal pad facing down, is a different animal. I was initially skeptical, but the no-vent design of the 3e Audio SY-DAP1002 (PVDD=30V) is finally making sense to me after many days of constant use. The heat generated by TPA3250 is transferred to the PCB, then to the cast-in rails on the side of the case, then dissipated to the ambient air by the case itself acting as a heatsink, which never got hotter than "slightly warm to the touch" level so far. Although SY-DAP2002 case with TPA3255 is longer front to back thus it could be a heftier heatsink if the thermal pad was facing down, installing a fan on an unvented case does not make sense to me.
If the face-down thermal pad dictated a thicker copper, better quality PCB in the design, I would consider this to be a major advantage of TPA3250 over TPA3251/TPA3255 in a majority of chinese amplifiers. TI application note on modern thermal management is at https://www.ti.com/lit/an/spra953c/spra953c.pdf?ts=1616907719303
Front panel has a sturdy 8mm thickness, allowing the mounting posts for volume control pcb to be fixed on the back side of it without the ends of the post protruding to the front surface. Small rubber feet create about 5mm space underneath, 3mm under the front panel.
This aluminum case is well designed in the detail, like the meeting surface (3.5mm thickness) of the upper and the bottom half shells having a front-to-back length-wise convex 'ridge' on one part, and a receiving concave 'trough' on the other for a snug fit.
Back panel does not have a fan.
`
De rien.
I have not tried it, but I would think OPA1602 is a very good OpAmp, without having much probability of it being a fake. I wanted to try different OpAmps, but when I placed the order after waiting for as long as I could in the latter half of February, I was not sure if I'd receive the v.1.0 or the next version.
I was, and I still am, not sure if I could hear the difference between OPA1602 and OPA1656. With either of these two OpAmps, I believe this SY-DAP1002 is a huge value for money, because of the relative quality in its power supply (catering to the lower voltage demand of TPA3250) and the output filters.
I have not tried it, but I would think OPA1602 is a very good OpAmp, without having much probability of it being a fake. I wanted to try different OpAmps, but when I placed the order after waiting for as long as I could in the latter half of February, I was not sure if I'd receive the v.1.0 or the next version.
I was, and I still am, not sure if I could hear the difference between OPA1602 and OPA1656. With either of these two OpAmps, I believe this SY-DAP1002 is a huge value for money, because of the relative quality in its power supply (catering to the lower voltage demand of TPA3250) and the output filters.
For the modifications to improve the sound quality, I owe a lot to diyaudio posters especially Claude. His patience, tenacity and the analytical mind, combined with the willingness to share the findings, saved me at least tens, probably hundreds of hours in research.
I was willing to do a lot of try-and-error in steps to see what works and what doesn't when I was about to embark on the fun challenge, but I realized probably 60 to 80% of them have already been done by him. Of course each product has it's own strengths and weaknesses, and a particular mod that worked in one amp may not work in another. But the general principle of "it is better to bypass a capacitor with another" and "the lower the impedance in a capacitor, the better they sound" are seemingly applicable universally.
I say 'impedance' because ESR (Equivalent Series Resistance) is just one part of what determines the resistance to let the current flow in/out of a capacitor, and the total resistance in the chain of ESR - EPR - ESL in series is the whole picture, and this aggregate resistance is called the 'impedance' (indicated as Z). EPR = Equivalent Parallel Resistance, sometimes indicated as Rp. ESL = Equivalent Series Inductance.
Granted we don't usually deal with high frequencies in the MHz or GHz range, so the ESL may not matter much, but when we look at the ESR graph of a Murata 0.1uF 25V X7R cap as an example :
we might think the low 0.05 - 0.06 Ohm to be the resistance at the 450-600kHz switching frequency of TPA3250-3255.
But in reality, it is at a whopping 3 - 4 Ohms as indicated in the graph below:
The frequency at the bottom of the valley of the above blue line (about 8.5MHz) is called the self resonant frequency of a capacitor, and this valley moves to the left (to a lower frequency) as the capacitance of a cap increases, in general.
It makes sense to use the cap at this frequency if a suitable cap is available in voltage rating, capacitance, quality, price and most importantly, size.
450 - 600kHz operating frequency is not the only consideration in trying to remove the voltage fluctuation, and especially noise better. But in light of more and more manufacturers publishing the impedance graph on their products, I wanted to add this impedance aspect to the findings of Claude on Aiyima A04, which are proven in experience (on that particular amp).
In any event, I decided to do the mods in one shot after deciding on what caps to use, so I do not have step by step results. Sorry for my laziness, and thank you, Claude.
Top view after the mod:
The only visible differences are the addition of three 1500uF 35V Nippon Chemi-con "Low Z" 30mm tall KZH series cap
EKZH350ELL152MK30S
and the OPA1656's with 4.7uF 35V Vishay/Sprague Tantalum cap on top
199D475X9035CA2
All of them have bypass caps on the bottom side of PCB.
Bottom view after the mod:
As I decided to go with SMD, small ones are difficult to see. There are ten 10uF 35V Elna Silmic IIs on the board (2 near RCA jack, 8 around the OpAmps) that are 5mm in diameter with 3mm lead spacing. As the through hole leads are so closely together, I used "Length and Width Reversed" SMD caps with 1.25mm length and 2.0mm width for the bypass, which have lower ESR than the regular 2.0mm length 1.25mm width cap, and these are the difficult-to-see small ones:
LLL216R71E104MA01L
Other bypass caps are:
GQM22M5C2H101JB01L 100pF 500V for inside cap of output LC
GRM32DR72E104KW01L 0.1uF 250V for outside cap of output LC
GCJ32EC71H106KA01K 10uF 50V for the output blocking cap
and for the three 2200uF 35V caps as well as for the three added 1500uF 35V caps.
The reason for the above high voltage ratings is that I wanted certain lengths to fit in between the leads of the caps on the other side for a clean installation. If I wanted a longer length, they tended to come with a higher voltage rating.
`
I was willing to do a lot of try-and-error in steps to see what works and what doesn't when I was about to embark on the fun challenge, but I realized probably 60 to 80% of them have already been done by him. Of course each product has it's own strengths and weaknesses, and a particular mod that worked in one amp may not work in another. But the general principle of "it is better to bypass a capacitor with another" and "the lower the impedance in a capacitor, the better they sound" are seemingly applicable universally.
I say 'impedance' because ESR (Equivalent Series Resistance) is just one part of what determines the resistance to let the current flow in/out of a capacitor, and the total resistance in the chain of ESR - EPR - ESL in series is the whole picture, and this aggregate resistance is called the 'impedance' (indicated as Z). EPR = Equivalent Parallel Resistance, sometimes indicated as Rp. ESL = Equivalent Series Inductance.
Granted we don't usually deal with high frequencies in the MHz or GHz range, so the ESL may not matter much, but when we look at the ESR graph of a Murata 0.1uF 25V X7R cap as an example :
we might think the low 0.05 - 0.06 Ohm to be the resistance at the 450-600kHz switching frequency of TPA3250-3255.
But in reality, it is at a whopping 3 - 4 Ohms as indicated in the graph below:
The frequency at the bottom of the valley of the above blue line (about 8.5MHz) is called the self resonant frequency of a capacitor, and this valley moves to the left (to a lower frequency) as the capacitance of a cap increases, in general.
It makes sense to use the cap at this frequency if a suitable cap is available in voltage rating, capacitance, quality, price and most importantly, size.
450 - 600kHz operating frequency is not the only consideration in trying to remove the voltage fluctuation, and especially noise better. But in light of more and more manufacturers publishing the impedance graph on their products, I wanted to add this impedance aspect to the findings of Claude on Aiyima A04, which are proven in experience (on that particular amp).
In any event, I decided to do the mods in one shot after deciding on what caps to use, so I do not have step by step results. Sorry for my laziness, and thank you, Claude.
Top view after the mod:
The only visible differences are the addition of three 1500uF 35V Nippon Chemi-con "Low Z" 30mm tall KZH series cap
EKZH350ELL152MK30S
and the OPA1656's with 4.7uF 35V Vishay/Sprague Tantalum cap on top
199D475X9035CA2
All of them have bypass caps on the bottom side of PCB.
Bottom view after the mod:
As I decided to go with SMD, small ones are difficult to see. There are ten 10uF 35V Elna Silmic IIs on the board (2 near RCA jack, 8 around the OpAmps) that are 5mm in diameter with 3mm lead spacing. As the through hole leads are so closely together, I used "Length and Width Reversed" SMD caps with 1.25mm length and 2.0mm width for the bypass, which have lower ESR than the regular 2.0mm length 1.25mm width cap, and these are the difficult-to-see small ones:
LLL216R71E104MA01L
Other bypass caps are:
GQM22M5C2H101JB01L 100pF 500V for inside cap of output LC
GRM32DR72E104KW01L 0.1uF 250V for outside cap of output LC
GCJ32EC71H106KA01K 10uF 50V for the output blocking cap
and for the three 2200uF 35V caps as well as for the three added 1500uF 35V caps.
The reason for the above high voltage ratings is that I wanted certain lengths to fit in between the leads of the caps on the other side for a clean installation. If I wanted a longer length, they tended to come with a higher voltage rating.
`
The result of the mod was quite noticeable. I am not that confident with my aging ears that I would have been able to assess the improvement/degradation accurately if the mods were evaluated on each smaller step. The psychological bias of "I spent this much effort/time/money, so it must sound better" really is a formidable monster in my mind.
But the mod done in one shot, the difference was obvious. I previously stated that DAP1002 has more midrange 'energy' when compared to BRZ HiFi 3255, but now I realize I should have described it as "midrange resolution", and this set of mods extended this resolution further down in the frequency scale. "Tighter bass" may be a more common description, but this is more than that.
As 'resolution' is often used to describe the fidelity in higher frequencies, 'energy' may be an easier concept to grasp, but may be this is simpler described to just having a "higher fidelity" in the mid/lower ranges in addition to the highs that was already quite good before the mod. This realization is rather shocking for me as I used to regard the sound quality of BRZ HiFi 3255 to be quite good, most probably (daniboun might concur) favorably comparable to other 3255 amps with its "all film capacitor" construction of the input pre/tone control stages:
The above pic is of BRZ HiFi 3255 after the modification of 1. OpAmp swap, 2. Speaker output cable change to silver plated 14ga. lines, 3. Addition of ground wire, and 4. Addition of solid silver lines to PS-to-mainboard 40V connection. I couldn't detect any difference in sound quality on the latter three mods at the SPL levels I use with 6R 90dB Revel Concerta2 F35.
Because TPA3255 requires a higher PVDD than 3250, the primary PS caps are rated at 63V for the total of 470 x 4 = 1880uF on the BRZ. (Plus may be 2 or 4 of 470uF 50V SMD caps on the mainboard, but I can't tell.)
On the other hand, DAP1002 drives TPA3250 at 30V, so the PS caps are rated at 35V for the total of 2200 x 3 = 6600uF before the mod, 6600 + 1500 x 3 = 11,100uF after the mod. The 2,200uF caps looked like Elna Silmic II, but they are not. I might consider swapping them out for a better cap, but I'm 90% sure I won't hear any difference. But the difference of 1800uF vs. 11000uF is significant. It may be an overkill for a Class D, but I consider this to be an advantage 3250 holds against 3255.
DAP1002 Rev.1.0 had a total of 470 x 3 + 2200 x 3 = 8,010uF, but apparently they had a quality control issue with the 470uF SMD caps, and they are removed on my Rev.1.1
The Coilcraft chokes were replaced by square-ish chokes (Würth?) with '100' printed on them. Is it reasonable to assume they are 100uH, I wonder.
Another difference this DAP1002 mod resulted in is a bit difficult to describe. The soundstage depth became a bit deeper, but more noticeable was the impression that the entire soundstage got 'quieter'. There was no audible noise before and after, so it is not a reduction of noise, but the impression is especially the background on complex music like orchestra became quieter so any (actually many) note/sound at an extremely low level that was not audible before became audible not only in the highs, but also in mids.
The above evaluations comparing before and after were all done with OPA1656 with the 4.7uF Vishay Tantalum cap on top (in addition to the 10uF Silmic II's on the board as OpAmp PS caps). After the mod, sound quality difference to BRZ 3255 (also with OPA1656 and Vishay, tone control switched off) became very obvious. I don't miss my 200W electric heater much any more.
`
But the mod done in one shot, the difference was obvious. I previously stated that DAP1002 has more midrange 'energy' when compared to BRZ HiFi 3255, but now I realize I should have described it as "midrange resolution", and this set of mods extended this resolution further down in the frequency scale. "Tighter bass" may be a more common description, but this is more than that.
As 'resolution' is often used to describe the fidelity in higher frequencies, 'energy' may be an easier concept to grasp, but may be this is simpler described to just having a "higher fidelity" in the mid/lower ranges in addition to the highs that was already quite good before the mod. This realization is rather shocking for me as I used to regard the sound quality of BRZ HiFi 3255 to be quite good, most probably (daniboun might concur) favorably comparable to other 3255 amps with its "all film capacitor" construction of the input pre/tone control stages:
The above pic is of BRZ HiFi 3255 after the modification of 1. OpAmp swap, 2. Speaker output cable change to silver plated 14ga. lines, 3. Addition of ground wire, and 4. Addition of solid silver lines to PS-to-mainboard 40V connection. I couldn't detect any difference in sound quality on the latter three mods at the SPL levels I use with 6R 90dB Revel Concerta2 F35.
Because TPA3255 requires a higher PVDD than 3250, the primary PS caps are rated at 63V for the total of 470 x 4 = 1880uF on the BRZ. (Plus may be 2 or 4 of 470uF 50V SMD caps on the mainboard, but I can't tell.)
On the other hand, DAP1002 drives TPA3250 at 30V, so the PS caps are rated at 35V for the total of 2200 x 3 = 6600uF before the mod, 6600 + 1500 x 3 = 11,100uF after the mod. The 2,200uF caps looked like Elna Silmic II, but they are not. I might consider swapping them out for a better cap, but I'm 90% sure I won't hear any difference. But the difference of 1800uF vs. 11000uF is significant. It may be an overkill for a Class D, but I consider this to be an advantage 3250 holds against 3255.
DAP1002 Rev.1.0 had a total of 470 x 3 + 2200 x 3 = 8,010uF, but apparently they had a quality control issue with the 470uF SMD caps, and they are removed on my Rev.1.1
The Coilcraft chokes were replaced by square-ish chokes (Würth?) with '100' printed on them. Is it reasonable to assume they are 100uH, I wonder.
Another difference this DAP1002 mod resulted in is a bit difficult to describe. The soundstage depth became a bit deeper, but more noticeable was the impression that the entire soundstage got 'quieter'. There was no audible noise before and after, so it is not a reduction of noise, but the impression is especially the background on complex music like orchestra became quieter so any (actually many) note/sound at an extremely low level that was not audible before became audible not only in the highs, but also in mids.
The above evaluations comparing before and after were all done with OPA1656 with the 4.7uF Vishay Tantalum cap on top (in addition to the 10uF Silmic II's on the board as OpAmp PS caps). After the mod, sound quality difference to BRZ 3255 (also with OPA1656 and Vishay, tone control switched off) became very obvious. I don't miss my 200W electric heater much any more.
`
Yes, indeed. 100uH is way out of line, and it most likely is a 10uH dual (10 with zero 0).