Thank you. I would think this amp letting the signal pass through the DAC in 1701 even when there is no EQ filter programmed in the DSP, is a reasonable assumption. I don't know how you evaluated the sound quality of 1701's DAC, but "together with the ADC and DSP preceding the DAC" ADAU1701 as a whole in a signal path sounds a cut worse than PCM5102 is something I can (may be even Analog Devices can) wholeheartedly accept.
Doesn't this make the fact DAP1002 sounds better than BRZ HiFi 3255 all the more amazing? (That is before and after the mods. Granted, may be only to my ears, but I bought two BRZ HiFi 3255 before.)
I accepted analog to digital conversion (CD) a long time ago, which made me an outcast in the field of high-end audio. And room acoustics has been the most difficult and expensive issue to tackle. I feel Analog Devices is onto something in making a DSP of reasonable quality available at an unbelievably low cost.
"High end audio in the future will have a DSP built in." may make me an outcast again, but that is something I have become believing in. Imagine an audio system making automatic adjustments to room acoustics.
`
You may be right, and you may not be. Of course everyone is entitled to his/her opinion, and I respect that.
If I needed to process many channels for a studio equipment, I might need ADAU1462 or 1467 for their I/O channels. If I need to do a very complicated processing per sample, then 1452, 1453, 1463 or 1466 may make more sense for their high instructions per sample and the high core frequency.
ADAU1472 has an impressive 144 pins coming out of the IC, as opposed to 1701's 48. And some of them are 4 times bigger than 1701 (7 x 7mm vs. 14 x 14mm).
Many of these cost more than 1701 and we might think higher price equals better quality, but the price difference is small and are mostly for special functionality and I/O channels, which are not used for stereo speaker/amp/room acoustics compensation.
I know it is theoretically better to place a DSP before the digital signal gets converted to analog for an audio amp, however, 1701 is not designed for that. 1701 is a ADC-DSP-DAC all in one single IC to be placed in an analog equipment like an audio amp.
In a way, the argument that an additional A/D and D/A conversion needs to be avoided is similar to the audio purists' opinion back in the days of CD infancy when they argued that analog to digital conversion for making a CD and the digital to analog conversion at the CD player cannot be better than no A/D conversion at all for sound quality. In a way they were correct, but today we all (except a few die-hard vinyl aficionados) would disregard such an opinion.
Most, if not all Bluetooth ICs have a set of ADC and DAC built in because audio signal comes into it in digital and goes out in analog for receiving music, and vice versa for sending voice. We know there are better DACs out there, but we don't even think about using a separate DAC when using Bluetooth, because it's a reasonable trade off in getting the convenience of it. (PCM5102A may well be incomparably better in sound quality than the DAC in a Qualcomm bluetooth chip.)
To me, room acoustics is a huge issue in enjoying a high fidelity music reproduction. For that, I'm willing to give up some SNR and/or THD, and I intend to find out how that trade off is going to pan out with ADAU1701, and it's looking very good so far. I might start insisting that a DSP needs to reside in a DAC box or a music file server in the end though...
`
If I needed to process many channels for a studio equipment, I might need ADAU1462 or 1467 for their I/O channels. If I need to do a very complicated processing per sample, then 1452, 1453, 1463 or 1466 may make more sense for their high instructions per sample and the high core frequency.
ADAU1472 has an impressive 144 pins coming out of the IC, as opposed to 1701's 48. And some of them are 4 times bigger than 1701 (7 x 7mm vs. 14 x 14mm).
Many of these cost more than 1701 and we might think higher price equals better quality, but the price difference is small and are mostly for special functionality and I/O channels, which are not used for stereo speaker/amp/room acoustics compensation.
I know it is theoretically better to place a DSP before the digital signal gets converted to analog for an audio amp, however, 1701 is not designed for that. 1701 is a ADC-DSP-DAC all in one single IC to be placed in an analog equipment like an audio amp.
In a way, the argument that an additional A/D and D/A conversion needs to be avoided is similar to the audio purists' opinion back in the days of CD infancy when they argued that analog to digital conversion for making a CD and the digital to analog conversion at the CD player cannot be better than no A/D conversion at all for sound quality. In a way they were correct, but today we all (except a few die-hard vinyl aficionados) would disregard such an opinion.
Most, if not all Bluetooth ICs have a set of ADC and DAC built in because audio signal comes into it in digital and goes out in analog for receiving music, and vice versa for sending voice. We know there are better DACs out there, but we don't even think about using a separate DAC when using Bluetooth, because it's a reasonable trade off in getting the convenience of it. (PCM5102A may well be incomparably better in sound quality than the DAC in a Qualcomm bluetooth chip.)
To me, room acoustics is a huge issue in enjoying a high fidelity music reproduction. For that, I'm willing to give up some SNR and/or THD, and I intend to find out how that trade off is going to pan out with ADAU1701, and it's looking very good so far. I might start insisting that a DSP needs to reside in a DAC box or a music file server in the end though...
`
the topic here is @Yiba sharing his finding and modification contribution to other diyer.
if you can make it happen with ADAU1452-67 and external DACs and sell it within 100USD,i think many people here are very interesting.
Haha, thank you for your comment 3eaudio, but I like and appreciate receiving comments from drMordor. He says so much with so few words. There are many talented people on diyaudio, and he is one of them.
I will be grateful if you could confirm or deny my assumptions and 'guesses' on things I cannot be sure like:
1. DAP1002/2002 PS shares the frequency signal from the crystal with the rest of the chips on the board, so its power generation frequency coincides with TPA3250/3255 power demand frequency.
2. ADAU1701 converts analog signal to digital and then back to analog even when there is no EQ filter programmed in the DSP.
Of course if you feel the answers may end up disclosing your secrets, just stating 'confidential' is perfectly OK.
Thanks
`
I will be grateful if you could confirm or deny my assumptions and 'guesses' on things I cannot be sure like:
1. DAP1002/2002 PS shares the frequency signal from the crystal with the rest of the chips on the board, so its power generation frequency coincides with TPA3250/3255 power demand frequency.
2. ADAU1701 converts analog signal to digital and then back to analog even when there is no EQ filter programmed in the DSP.
Of course if you feel the answers may end up disclosing your secrets, just stating 'confidential' is perfectly OK.
Thanks
`
Hey! I do not in the least belittle your amplifiers, they are great!
But we strive for the best)))
Yiba, thanks))) No secrets)))
My reticence is due to the fact that I cannot correctly build phrases in English)))
1 I didn't understand the question
2 Yes.
It is necessary to ask the respected 3E in the next revision to use an analog volume control, and put a small relay to switch to direct mode, that is, bypassing 1701
My reticence is due to the fact that I cannot correctly build phrases in English)))
1 I didn't understand the question
2 Yes.
It is necessary to ask the respected 3E in the next revision to use an analog volume control, and put a small relay to switch to direct mode, that is, bypassing 1701
Last edited:
Thank you very much for your comment drMordor, please don't worry about your English, I also learned English as a foreign language. My question #1 means:
1. A switching PS operates on the switching frequency (normally RC oscillator or crystal).
2. Class D amps also operate on a switching frequency in converting an audio signal into PWM (Pulse Width Modulation) pulse. TPA325x uses a selectable frequency in 450-600kHz range.
3. By making the two frequencies the same (or x2, x4, x8 times) and making the timing the same by sharing a crystal on the PCB, isn't it possible to meet the demand frequency with the supply frequency? This would tremendously decrease the load on voltage regulators and PS capacitors, no?
4. Is this how DAP1002 PS is designed/made? (This was/is a question for 3eaudio)
`
I see, thanks.
BRZ HiFi 3255 uses two of such a relay to:
1. Avoid power on/off pop.
2. Letting a signal from QCC3003 select input from/to Bluetooth/RCA.
However, DAP1002 solved the both issues (it does not have any power on/off pop at all) without using relays, and I consider that to be an elegant solution.
As DAP1002 sounds better than the BRZ, with the 1701 ADC/DAC constantly on, I don't see much meaning to bypassing it, unless there are other reasons than sound quality. DAP1002 is not expensive, and I could buy some other amp if I don't like the ADAU1701 DSP.
Also, there is an update coming up to implement a logarithmic volume control curve in a Sigma Studio .dspproj file, and I want to wait and see how it performs before asking for an analog volume control.
The relays 'sound' nice and give a bit of 'expensive' impression to using the BRZ3255
`
BRZ HiFi 3255 uses two of such a relay to:
1. Avoid power on/off pop.
2. Letting a signal from QCC3003 select input from/to Bluetooth/RCA.
However, DAP1002 solved the both issues (it does not have any power on/off pop at all) without using relays, and I consider that to be an elegant solution.
As DAP1002 sounds better than the BRZ, with the 1701 ADC/DAC constantly on, I don't see much meaning to bypassing it, unless there are other reasons than sound quality. DAP1002 is not expensive, and I could buy some other amp if I don't like the ADAU1701 DSP.
Also, there is an update coming up to implement a logarithmic volume control curve in a Sigma Studio .dspproj file, and I want to wait and see how it performs before asking for an analog volume control.
The relays 'sound' nice and give a bit of 'expensive' impression to using the BRZ3255
`
It is a rule of thumb that power switching at low voltages of class d-amps works best in the ballpark of 500kHz, while high voltage smps more around 100kHz. Nonetheless off-line full resonant 1kW-converters with 1Mhz clock rate have been built successfully about 15yrs ago, so this is not strictly impossible to be done. Personally I doubt that such a syncrone system is really beneficial.
Last edited:
Thanks a lot. So it is theoretically possible to do what I proposed at 500kHz. Although most TPA325x implementations use 600kHz, this is good to know.
Taking the theory one step further,
1. Power supply provides a constant 500kHz square wave current at 50% duty cycle at the voltage governed by the input Vrms (averaged for a few ms. or longer)
2. Instead of 'creating' the PWM pulse from scratch, the amp (operating on the same 500kHz timing) just reduces and increases the duty cycle by dumping the unwanted current to a capacitor, and getting the needed current from that same capacitor and from a regular PS power.
3. As the amp is not creating the PWM from scratch, coming up with a near-perfect PWM pulse by 'shaping' according to the input audio signal should be easier.
Am I onto something? Couldn't this enable a much more efficient and higher fidelity amplification? Could I name this a Class DY?
`
Taking the theory one step further,
1. Power supply provides a constant 500kHz square wave current at 50% duty cycle at the voltage governed by the input Vrms (averaged for a few ms. or longer)
2. Instead of 'creating' the PWM pulse from scratch, the amp (operating on the same 500kHz timing) just reduces and increases the duty cycle by dumping the unwanted current to a capacitor, and getting the needed current from that same capacitor and from a regular PS power.
3. As the amp is not creating the PWM from scratch, coming up with a near-perfect PWM pulse by 'shaping' according to the input audio signal should be easier.
Am I onto something? Couldn't this enable a much more efficient and higher fidelity amplification? Could I name this a Class DY?
`
Class DY amp v.1.1
DAP1002 after the mod, before DSP programming, lets me enjoy Sam Jones (b) in Cannonball Adderley & Miles Davis "Autumn Leaves". Before, the joy came from Saxophone and Trumpet.
I should be going back to DAP1002, but here's an easier to understand version of Class DY explanation with some corrections while the idea is fresh:
1. During the course of mod/evaluation cycles, I realized power supply to create the PWM pulse to be a common bottleneck in today's Class D.
2. To remedy that, instead of supplying the PVDD, power supply provides a 500kHz square wave current at 50% duty cycle at the voltage governed by the input Vrms x Volume level. This pulse is different from the output of ordinary Class D in its pulse widths, but roughly equal in amplitude/voltage. (50% duty cycle means 50% of the time on, 50% of the time off)
3. The amp creates an ordinary Class D PWM pulse from the input signal, but in voltage only. This is a lot easier than creating it with the speaker driving current.
4. It compares the PS pulse (50% duty cycle) with the Class D pulse (variable duty cycle).
5. When the duty cycle of Class D pulse is smaller than 50%, it just reduces the width of the square of the PS pulse.
6. When the Class D pulse is wider than 50%, it increases the width.
7. Amplitude/voltage is also adjusted if it is incorrect.
8. NFB is provided to these adjustments, and the amount of addition and subtraction should roughly equal over time in my untrained mind.
9. The result is fed to the ordinary Class D output filter.
10. As the amp is not creating the PWM from scratch, coming up with a near-perfect PWM pulse according to the input audio signal should be easier.
What do you think?
`
DAP1002 after the mod, before DSP programming, lets me enjoy Sam Jones (b) in Cannonball Adderley & Miles Davis "Autumn Leaves". Before, the joy came from Saxophone and Trumpet.
I should be going back to DAP1002, but here's an easier to understand version of Class DY explanation with some corrections while the idea is fresh:
1. During the course of mod/evaluation cycles, I realized power supply to create the PWM pulse to be a common bottleneck in today's Class D.
2. To remedy that, instead of supplying the PVDD, power supply provides a 500kHz square wave current at 50% duty cycle at the voltage governed by the input Vrms x Volume level. This pulse is different from the output of ordinary Class D in its pulse widths, but roughly equal in amplitude/voltage. (50% duty cycle means 50% of the time on, 50% of the time off)
3. The amp creates an ordinary Class D PWM pulse from the input signal, but in voltage only. This is a lot easier than creating it with the speaker driving current.
4. It compares the PS pulse (50% duty cycle) with the Class D pulse (variable duty cycle).
5. When the duty cycle of Class D pulse is smaller than 50%, it just reduces the width of the square of the PS pulse.
6. When the Class D pulse is wider than 50%, it increases the width.
7. Amplitude/voltage is also adjusted if it is incorrect.
8. NFB is provided to these adjustments, and the amount of addition and subtraction should roughly equal over time in my untrained mind.
9. The result is fed to the ordinary Class D output filter.
10. As the amp is not creating the PWM from scratch, coming up with a near-perfect PWM pulse according to the input audio signal should be easier.
What do you think?
`
Going back to DAP1002, there was one more LARGE difference the mod resulted in. It is on the accommodation for exotic OpAmps, which I mentioned before to be an advantage Aiyima A04 and DAP1002 (and other amps with OpAmp gain of higher than 1) have.
LT1364 OpAmp has earned a special status among the Japanese opamp rollers for its speed, resolution, extreme tightness in the bass, and soundstage width/depth, but also with its finicky demand for avoiding oscillation (though it is rated as "unity gain stable") and supply current.
1364 consumes 6-7mA per amp for the dual (it comes in dual or quad) total of 12-14mA, which results in a very high thermal dissipation for a DIP-8. I don't feel comfortable in using it without a small heatsink on this IC, so I chose LT1361 instead as a rolling candidate partly because a heatsink makes placing a cap on top difficult.
LT1364 specs:
1000V/µs Slew Rate
50ns Settling Time to 0.1%
70MHz Gain Bandwidth
9nV/√Hz Input Noise Voltage
LT1361 is a bit less extreme version in the same series of opamps. It consumes 10mA max with a lot less heat dissipation.
LT1361 specs:
800V/µs Slew Rate
60ns Settling Time to 0.1%
50MHz Gain Bandwidth
9nV/√Hz Input Noise Voltage
with a slightly better-looking overshoot characteristics:
However, LT1361 does not work in BRZ3255 or the unmodified DAP1002, showing an extremely fine resolution and tight bass, but with harsh highs sounding way too analytical and sterile for the enjoyment of music. I tried many different caps on top in search for a solution:
On the far side are OPA1656 with WIMA MKS2 1.0µF 63V
MKS2C041001F00JSSD, the near side are the LT1361 with Nippon Chemi-con PSC series 470µF 16V "super low ESR" aluminum polymer APSC160ELL471MJB5S for a temporary trial.
Finally, I found the LT1361 + WIMA MKS2 1.0µF combination works in DAP1002 after the modifications. And boy, this sounds GOOD.
The super analytical character is there, but it completely lost the harshness and rough edges, with the resolution extended to low, low frequencies, with super deep soundstage on top of the width that might be wider than OPA1656.
I did not believe when radelius said something to the effect that OPA1656 sounds colored, but after listening to LT1361 + WIMA and compared in this modified DAP1002, now I believe his ears may be spot on.
Compared to LT1361, the largest difference is OPA1656's low frequency range sounding a lot muddied in relative terms (on my DAP1002 after the mod with Revel F35, OPA1656 also with the same WIMA 1.0µF film on top). It is difficult to make LT1361 work in unity gain (I couldn't in my BRZ3255), so it cannot be a safe recommendation, but I highly recommend trying it to die-hard OpAmp rollers. I am very glad we have OPA1656 as a benchmark accepted by many, so that we could easily tell if it is working well.
In my room, on my ears, with my speakers, the mods + LT1361 exceeded the sound quality of my Luxman CL-88 tube pre-amp + Pass Lab X3. Sorry Nelson, but you are welcome to come listen to this if you have a plan to come to Tokyo. I have found myself to be in Dick Olsher school of First Watt several years ago, after the X3 served me extremely well for many many years.
`
LT1364 OpAmp has earned a special status among the Japanese opamp rollers for its speed, resolution, extreme tightness in the bass, and soundstage width/depth, but also with its finicky demand for avoiding oscillation (though it is rated as "unity gain stable") and supply current.
1364 consumes 6-7mA per amp for the dual (it comes in dual or quad) total of 12-14mA, which results in a very high thermal dissipation for a DIP-8. I don't feel comfortable in using it without a small heatsink on this IC, so I chose LT1361 instead as a rolling candidate partly because a heatsink makes placing a cap on top difficult.
LT1364 specs:
1000V/µs Slew Rate
50ns Settling Time to 0.1%
70MHz Gain Bandwidth
9nV/√Hz Input Noise Voltage
LT1361 is a bit less extreme version in the same series of opamps. It consumes 10mA max with a lot less heat dissipation.
LT1361 specs:
800V/µs Slew Rate
60ns Settling Time to 0.1%
50MHz Gain Bandwidth
9nV/√Hz Input Noise Voltage
with a slightly better-looking overshoot characteristics:
However, LT1361 does not work in BRZ3255 or the unmodified DAP1002, showing an extremely fine resolution and tight bass, but with harsh highs sounding way too analytical and sterile for the enjoyment of music. I tried many different caps on top in search for a solution:
On the far side are OPA1656 with WIMA MKS2 1.0µF 63V
MKS2C041001F00JSSD, the near side are the LT1361 with Nippon Chemi-con PSC series 470µF 16V "super low ESR" aluminum polymer APSC160ELL471MJB5S for a temporary trial.
Finally, I found the LT1361 + WIMA MKS2 1.0µF combination works in DAP1002 after the modifications. And boy, this sounds GOOD.
The super analytical character is there, but it completely lost the harshness and rough edges, with the resolution extended to low, low frequencies, with super deep soundstage on top of the width that might be wider than OPA1656.
I did not believe when radelius said something to the effect that OPA1656 sounds colored, but after listening to LT1361 + WIMA and compared in this modified DAP1002, now I believe his ears may be spot on.
Compared to LT1361, the largest difference is OPA1656's low frequency range sounding a lot muddied in relative terms (on my DAP1002 after the mod with Revel F35, OPA1656 also with the same WIMA 1.0µF film on top). It is difficult to make LT1361 work in unity gain (I couldn't in my BRZ3255), so it cannot be a safe recommendation, but I highly recommend trying it to die-hard OpAmp rollers. I am very glad we have OPA1656 as a benchmark accepted by many, so that we could easily tell if it is working well.
In my room, on my ears, with my speakers, the mods + LT1361 exceeded the sound quality of my Luxman CL-88 tube pre-amp + Pass Lab X3. Sorry Nelson, but you are welcome to come listen to this if you have a plan to come to Tokyo. I have found myself to be in Dick Olsher school of First Watt several years ago, after the X3 served me extremely well for many many years.
`
Hi Yiba, thanks a lot for sharing.
(think caps leg lengh should not differ too much when comparing)
MKS2 OK : Tried MKS4 ? (1uF/100v LS7.5)
Last edited:
Obvious. Tech already there. Will become mandatory...
iLoud MTM
Nubert nuPro X RC Series - The next evolutionary step in active loudspeaker systems
Thank you for your comments, douede. A cap leg length matters when Equivalent Series Inductance (ESL) comes into play in MHz and GHz range. On that Aluminum Polymer cap, the 470µF capacitance is way out of ordinary for cap bypassing, and I used the cap for only a few minutes "after making the lengths uniform". I first thought I wouldn't even need to cut the longer leg, but uniform lengths was more convenient for pushing it in and pulling it out I have not tried the MKS4 yet.
I didn't know about this ARC System, or the Nubert X. Thanks! Looks like Analog Devices is already selling a not-so-inconsequential number of Sigma DSP chips.
I have not tried the MKS4 yet.I didn't know about this ARC System, or the Nubert X. Thanks! Looks like Analog Devices is already selling a not-so-inconsequential number of Sigma DSP chips.
Having established an unexpectedly high baseline without the Post Filter Feed Back(PFFB), now it's time for fiddling with the Digital Signal Processing (DSP). I am very happy because if the DSP does not work out, I can always come back to this line where I know I can enjoy music a lot without the DSP. DSP has many uses like noise cancellation for headphones, changing a girl's voice into a man's voice over the phone for anonymity, etc., but we are mostly interested in two fields:
-Correction of Amp and Speaker Imperfections-
Tone control, loudness control (this actually is for the correction of human sensitivity to certain frequencies), and multiband equalizer have been with us for a long time, and they have largely been regarded as detrimental to music reproduction in high-end audio. But the users of inexpensive stereo system know their benefit by ear, and even the highest quality audio equipment is very far from being perfect.
-Correction of Room Acoustics-
Any room has characteristics in sound reflection, absorption and resonance. Unlike a church that is used for recording with the same acoustic character for organ, choir or chamber music, our listening environment has to deal with a vast variety of acoustics from an outdoor concert in Central Park, a noisy Jazz club in Amsterdam, a historic orchestra venue in Berlin, and to a live/dead studio setup in London.
Although it is commonly described as "live feeling", the acoustic characteristic of the recording environment is normally different to our listening environment, so a hifi music reproduction is always low fidelity in that regard. But I don't know why ASR does not publish its listening/evaluation environment acoustics data yet (please excuse my ignorance if they have).
`
Thanks to Analog Devices, state of the art DSP has become available to us at a very low cost, and I am going to see how ADAU1701 performs with the above in mind.
The first step in using a DSP should be the measurement of my room to see if there are bad reflection/absorption/resonance in certain frequency ranges. If the target is the correction of speaker imperfections, the measurement should obviously be on the speakers. But every speaker performs differently in a different room, so it makes sense to measure the room/speaker with the testing sound coming out of that particular set of speakers located in the positions they are normally used.
In order to measure a sound, I need a microphone. But no mic can be perfect and if the mic has a bad frequency response, then the measurement will be bad as well. What I need to use instead is a good mic with a precise frequency response data. Many mics for professional use in the past had a 'representative' or 'typical' frequency response curve published on the particular model, but that is not good enough for our purposes. Instead, the measured data needs to be high resolution (many measured frequency points) on that particular mic, and such a mic with data used to cost a fortune in the past. With the data file, the imperfections of the mic can be calibrated out from the room/speaker measurements by the measuring software.
Dayton Audio iMM-6 Condenser Microphone costs about $32 and comes in this can. The 3.5mm Stereo extension cable needs to be long enough for the PC to listening position distance. In my case, the convenient-to-use PC is the music/video server PC mounted on the back of a TV in between the speakers, so it's rather long. The TRRS (4 position) to TRS (3 position) converter on the lid may not be needed, but is convenient to change the direction of the mic at the end of the cable. The PC, or a DAC attached to it, needs to be connected to the amp to send the test signal out.
Each mic comes with a serial number like this, and the calibration file for that particular number is downloaded for free by clicking on "Download Measurement Microphone Calibration Files" in the middle of Dayton Audio - iMM-6 iDevice Calibrated Measurement Microphone and entering the number.
A condenser mic needs a small amount of electricity to operate, and it's provided by a small battery in the mic, or by something called Phantom Power (DC12, 24 or 48V) on 3 prong balanced mic cable for pro use. But iMM-6 does not need either of those. Here are some info, but a PC 3.5mm mic input jack has a small (2.5V) bias voltage between the base ring (neg) and the tip (pos) of the 3 position jack according to the old Sound Blaster standard, and this works with iMM-6. drMordor might yell at me, but if the PC internal sound card is not good enough for you, something cheap like Sound Blaster "Play! 3" ($12?) is far better and it works to drive the mic and to receive the signal from it.
`
-Correction of Amp and Speaker Imperfections-
Tone control, loudness control (this actually is for the correction of human sensitivity to certain frequencies), and multiband equalizer have been with us for a long time, and they have largely been regarded as detrimental to music reproduction in high-end audio. But the users of inexpensive stereo system know their benefit by ear, and even the highest quality audio equipment is very far from being perfect.
-Correction of Room Acoustics-
Any room has characteristics in sound reflection, absorption and resonance. Unlike a church that is used for recording with the same acoustic character for organ, choir or chamber music, our listening environment has to deal with a vast variety of acoustics from an outdoor concert in Central Park, a noisy Jazz club in Amsterdam, a historic orchestra venue in Berlin, and to a live/dead studio setup in London.
Although it is commonly described as "live feeling", the acoustic characteristic of the recording environment is normally different to our listening environment, so a hifi music reproduction is always low fidelity in that regard. But I don't know why ASR does not publish its listening/evaluation environment acoustics data yet (please excuse my ignorance if they have).
`
Thanks to Analog Devices, state of the art DSP has become available to us at a very low cost, and I am going to see how ADAU1701 performs with the above in mind.
The first step in using a DSP should be the measurement of my room to see if there are bad reflection/absorption/resonance in certain frequency ranges. If the target is the correction of speaker imperfections, the measurement should obviously be on the speakers. But every speaker performs differently in a different room, so it makes sense to measure the room/speaker with the testing sound coming out of that particular set of speakers located in the positions they are normally used.
In order to measure a sound, I need a microphone. But no mic can be perfect and if the mic has a bad frequency response, then the measurement will be bad as well. What I need to use instead is a good mic with a precise frequency response data. Many mics for professional use in the past had a 'representative' or 'typical' frequency response curve published on the particular model, but that is not good enough for our purposes. Instead, the measured data needs to be high resolution (many measured frequency points) on that particular mic, and such a mic with data used to cost a fortune in the past. With the data file, the imperfections of the mic can be calibrated out from the room/speaker measurements by the measuring software.
Dayton Audio iMM-6 Condenser Microphone costs about $32 and comes in this can. The 3.5mm Stereo extension cable needs to be long enough for the PC to listening position distance. In my case, the convenient-to-use PC is the music/video server PC mounted on the back of a TV in between the speakers, so it's rather long. The TRRS (4 position) to TRS (3 position) converter on the lid may not be needed, but is convenient to change the direction of the mic at the end of the cable. The PC, or a DAC attached to it, needs to be connected to the amp to send the test signal out.
Each mic comes with a serial number like this, and the calibration file for that particular number is downloaded for free by clicking on "Download Measurement Microphone Calibration Files" in the middle of Dayton Audio - iMM-6 iDevice Calibrated Measurement Microphone and entering the number.
A condenser mic needs a small amount of electricity to operate, and it's provided by a small battery in the mic, or by something called Phantom Power (DC12, 24 or 48V) on 3 prong balanced mic cable for pro use. But iMM-6 does not need either of those. Here are some info, but a PC 3.5mm mic input jack has a small (2.5V) bias voltage between the base ring (neg) and the tip (pos) of the 3 position jack according to the old Sound Blaster standard, and this works with iMM-6. drMordor might yell at me, but if the PC internal sound card is not good enough for you, something cheap like Sound Blaster "Play! 3" ($12?) is far better and it works to drive the mic and to receive the signal from it.
`
Last edited: