I just built an "AK phonostage" , it started on a thread over at AK and now there are boards (and kits I believe) available for the stage and PSU. I only got the phonostage board and did the rest myself.
I built a stand alone 15VDC dual power supply that uses XLR for the power out. I did this so I could use it to power more than one device and also because I prefer to have the PSU separate from the signal electronics, used XLR because it can handle low voltage and looks good. I basically copied Jim Hagerman's Bugle 2 power supply, here is a link to that build:
Dual Power Supply Photos by cytovette | Photobucket
I already have a Cornet 2 phonostage that I built myself and had been tweaking for 2 years, just had the itch to build something and after looking at the schematic I realized that I already had 95% of the parts. My build looks different because of that, in order to fit the big capacitors (like orange drops). I had to use both sides and be inventive. I really like that the gain and input impedance are adjustable design, I actually built adjustable input impedance into my Cornet because I like that feature. Here is a link to that build:
AK Phonostage Photos by cytovette | Photobucket
Where I got the board: https://hackaday.io/project/4788-muffsy-phono-preamp-pp-2
As far as the sound? My 2 channel system consists of; Yaqin MS-20L integrated amp, Technics SL-1200M3D with cardas tonarm wire going directly to RCA jacks, KAB fluid damper and AT120E cartridge with ATN150 stylus, 1981 Klipsch La Scalas with ALK crossovers and two Emotiva Ultra 10 subs.
The AK phonostage is dead quiet, something I miss with a tube phonostage. The sound is clean and detailed. I used to have a Musical Fidelity V-LPS and this beats it, however it does not beat the Cornet 2. Comparing it to the Cornet 2 may not be quite fair as I spent $500+ building that not counting tubes. The key difference is that the Cornet has a wider soundstage and slightly better imaging, but then it cost 10 times as much to build.
In my system it sounded bright and lacking a little bass at first, Klipsch speakers tend to do that, I raised the gain from 40 to 44 and cut the input impedance from 47k to 33k and pretty much did the trick.
Overall this is not the best phonostage out there but it sounds very good and sounds better than anything in it's price range. More than worth the cost and should be a relatively easy build for less experienced builders. I was a licensed electronics tech in the 70s tube days and spend my freee time tweaking my equipment now.
BTW I have no affiliation with the designer or seller, just reporting my findings.
I built a stand alone 15VDC dual power supply that uses XLR for the power out. I did this so I could use it to power more than one device and also because I prefer to have the PSU separate from the signal electronics, used XLR because it can handle low voltage and looks good. I basically copied Jim Hagerman's Bugle 2 power supply, here is a link to that build:
Dual Power Supply Photos by cytovette | Photobucket
I already have a Cornet 2 phonostage that I built myself and had been tweaking for 2 years, just had the itch to build something and after looking at the schematic I realized that I already had 95% of the parts. My build looks different because of that, in order to fit the big capacitors (like orange drops). I had to use both sides and be inventive. I really like that the gain and input impedance are adjustable design, I actually built adjustable input impedance into my Cornet because I like that feature. Here is a link to that build:
AK Phonostage Photos by cytovette | Photobucket
Where I got the board: https://hackaday.io/project/4788-muffsy-phono-preamp-pp-2
As far as the sound? My 2 channel system consists of; Yaqin MS-20L integrated amp, Technics SL-1200M3D with cardas tonarm wire going directly to RCA jacks, KAB fluid damper and AT120E cartridge with ATN150 stylus, 1981 Klipsch La Scalas with ALK crossovers and two Emotiva Ultra 10 subs.
The AK phonostage is dead quiet, something I miss with a tube phonostage. The sound is clean and detailed. I used to have a Musical Fidelity V-LPS and this beats it, however it does not beat the Cornet 2. Comparing it to the Cornet 2 may not be quite fair as I spent $500+ building that not counting tubes. The key difference is that the Cornet has a wider soundstage and slightly better imaging, but then it cost 10 times as much to build.
In my system it sounded bright and lacking a little bass at first, Klipsch speakers tend to do that, I raised the gain from 40 to 44 and cut the input impedance from 47k to 33k and pretty much did the trick.
Overall this is not the best phonostage out there but it sounds very good and sounds better than anything in it's price range. More than worth the cost and should be a relatively easy build for less experienced builders. I was a licensed electronics tech in the 70s tube days and spend my freee time tweaking my equipment now.
BTW I have no affiliation with the designer or seller, just reporting my findings.
Hi cyto, and thanks for reviewing my phono stage. I'm glad that you like it, and I really enjoyed seeing the pictures.
I've run the circuit through Spice, and it shows that the (simulated) RIAA accuracy is +/-0.023 dB from 1-33.000 Hz. For 1-100.000 Hz, it's +/-0.4 dB.
I have written quite a lot of information about the phono stage on hackaday.io, if anyone's interested.
I've run the circuit through Spice, and it shows that the (simulated) RIAA accuracy is +/-0.023 dB from 1-33.000 Hz. For 1-100.000 Hz, it's +/-0.4 dB.
I have written quite a lot of information about the phono stage on hackaday.io, if anyone's interested.
Did you get this curve by subtracting the exact RIAA curve from the Spice output,
or by using an inverse RIAA network at the input and simulating the response?
I used this solution to simulate the RIAA signal:
http://www.diyaudio.com/forums/analog-line-level/74983-spice-transfer-function-riaa-testing.html
If you used an inverse RIAA network at the input for simulation, the hf rise is due to that network,
not your phono circuit. The inverse RIAA network cannot have an accurate high frequency response,
and its output into the phono stage levels off at a certain point, instead of continuing to increase.
This causes your simulated hf rise. A primary advantage of the passive equalization approach is the
accurate hf response.
Last edited:
Thanks for the explanation rayma.
If you do the simulation in a different way, without the inverse RIAA network,
the results will be even better, especially at hf.
Now for some noise measurements of the Muffsy Phono Preamp.
The noise floor of the Muffsy Phono Preamp PP-3, powered by the Muffsy Hifi Dual Power Supply, is well below -100 dB. The average of all measured points, from 20-20.000 Hz is -107.74 dB and the highest single harmonic is at 92.11 dB (2 kHz).
The measurements are taken with a Rigol DG1022 function generator feeding a 1kHz/5mVpp signal into the phono stage. The PP-3's output signal is recorded using a Behringer UCA202 soundcard, and the FFT is generated from that recording. 16.384 measurement points have been recorded, almost 1 measurement per Hz.
The measurements are done without any calibration, on a relatively cheap USB sound card with a whole range of error sources that will provide less than optimal values. It's also recorded using the standard Windows mixer.(A QuantAsylum QA400 is on its way though...)
Notice that there is a 50 Hz distortion in the graph (with harmonics). That's unfortunately from the power cable that is running across the input of the PP-3 inside the B-0509 cabinet.
There is also a slope from 10-200 Hz. This slope is identical when I use the Behringer sound card to measure the Rigol DG1022 directly. It probably stems from the DC blocking high-pass filter in the sound card.
Finally, the abrupt roll-off at 20 kHz is the bandwidth of the sound card. It can only measure up to that level.
Here are some numbers:
]The average level of all measured points is -107.75 dB (20-20.000 Hz), including the 50 Hz distortion and the less than ideal measurements up to 200 Hz. This does not include the steep roll-off at frequencies above 20 kHz.
The highest peak is at 100 Hz, coming in at 88.6 dB. That value can be disregarded as it comes from my power cabling running to close to the phono stage input
Harmonics:
1st (2 kHz): -92.11 dB
2nd (3 kHz): -100.38 dB
3rd (4 kHz): -98.85 dB
The highest harmonic peak comes in at -92.11 dB, again disregarding the 50 Hz noise and its harmonics.
All the noise that's measured lies well below -100 dB
The noise floor of the Muffsy Phono Preamp PP-3, powered by the Muffsy Hifi Dual Power Supply, is well below -100 dB. The average of all measured points, from 20-20.000 Hz is -107.74 dB and the highest single harmonic is at 92.11 dB (2 kHz).
The measurements are taken with a Rigol DG1022 function generator feeding a 1kHz/5mVpp signal into the phono stage. The PP-3's output signal is recorded using a Behringer UCA202 soundcard, and the FFT is generated from that recording. 16.384 measurement points have been recorded, almost 1 measurement per Hz.
The measurements are done without any calibration, on a relatively cheap USB sound card with a whole range of error sources that will provide less than optimal values. It's also recorded using the standard Windows mixer.(A QuantAsylum QA400 is on its way though...)
Notice that there is a 50 Hz distortion in the graph (with harmonics). That's unfortunately from the power cable that is running across the input of the PP-3 inside the B-0509 cabinet.
There is also a slope from 10-200 Hz. This slope is identical when I use the Behringer sound card to measure the Rigol DG1022 directly. It probably stems from the DC blocking high-pass filter in the sound card.
Finally, the abrupt roll-off at 20 kHz is the bandwidth of the sound card. It can only measure up to that level.
Here are some numbers:
]The average level of all measured points is -107.75 dB (20-20.000 Hz), including the 50 Hz distortion and the less than ideal measurements up to 200 Hz. This does not include the steep roll-off at frequencies above 20 kHz.
The highest peak is at 100 Hz, coming in at 88.6 dB. That value can be disregarded as it comes from my power cabling running to close to the phono stage input
Harmonics:
1st (2 kHz): -92.11 dB
2nd (3 kHz): -100.38 dB
3rd (4 kHz): -98.85 dB
The highest harmonic peak comes in at -92.11 dB, again disregarding the 50 Hz noise and its harmonics.
All the noise that's measured lies well below -100 dB
Measurements, for real
I just got a QuantAsylum QA400 audio analyzer and it just takes the guesswork out of it all. I thought I'd share my findings with you.
The measurements are with a 1 kHz sine in and an output of 1 Vrms from the CNC, with a bandwidth from 10 to 20.000 Hz. As it stands, I'd need a decent inverse RIAA circuit to measure through the whole bandwidth.
The phono stage is equipped with OPA2134 and +/- 15V PSU (LM317/337). No out of the ordinary components, just standard 1% metal film resistors and Wima MKS2 caps.
THD: 0.00190%
THD+N: 0.01013%
SNR: 80,6 dB
IMD: 0.0195% (ITU-R 19+20 kHz)
Dynamic Range: >100 dB (THD+N: 1%)
RIAA Compliance: +/- 0.025 dB
The dynamic range can't be determined without an attenuator, as the QA400 goes into clipping long before the THD+N increases by more than 0.01%.
I'll measure crosstalk as well, when I find time for it.
-skrodahl
Audiobah TPA3116 || ESP P88 || ProJect Debut Carbon Espirit DC || Muffsy Phono Preamp PP-3 || RuneAudio / RaspBerry Pi / HifiBerry DAC || Dynaco A25
I just got a QuantAsylum QA400 audio analyzer and it just takes the guesswork out of it all. I thought I'd share my findings with you.
The measurements are with a 1 kHz sine in and an output of 1 Vrms from the CNC, with a bandwidth from 10 to 20.000 Hz. As it stands, I'd need a decent inverse RIAA circuit to measure through the whole bandwidth.
The phono stage is equipped with OPA2134 and +/- 15V PSU (LM317/337). No out of the ordinary components, just standard 1% metal film resistors and Wima MKS2 caps.
THD: 0.00190%
THD+N: 0.01013%
SNR: 80,6 dB
IMD: 0.0195% (ITU-R 19+20 kHz)
Dynamic Range: >100 dB (THD+N: 1%)
RIAA Compliance: +/- 0.025 dB
The dynamic range can't be determined without an attenuator, as the QA400 goes into clipping long before the THD+N increases by more than 0.01%.
I'll measure crosstalk as well, when I find time for it.
-skrodahl
Audiobah TPA3116 || ESP P88 || ProJect Debut Carbon Espirit DC || Muffsy Phono Preamp PP-3 || RuneAudio / RaspBerry Pi / HifiBerry DAC || Dynaco A25
Worth noting:
I have measured the SNR according to this article: Signal-to-noise Ratio (SNR), Dynamic Range, and Noise
This is the Signal-to-noise Ratio with the inputs shorted, and relative to my 1 Vrms/0 dBV signal: 112.83 dB.
The dynamic range (measured at 1% THD+N) will then be even greater than this. Sadly, this is about as much I can squeeze out of the 24-bit ADC's capabilities.
I have measured the SNR according to this article: Signal-to-noise Ratio (SNR), Dynamic Range, and Noise
This is the Signal-to-noise Ratio with the inputs shorted, and relative to my 1 Vrms/0 dBV signal: 112.83 dB.
The dynamic range (measured at 1% THD+N) will then be even greater than this. Sadly, this is about as much I can squeeze out of the 24-bit ADC's capabilities.
Last edited:
I get that by measuring and matching them.
EDIT:
I have not been able to actually measure the linearity, so I'm relying on the simulations that I have described earlier in this thread.
Last edited:
Crosstalk measurements are in, at -75.58 dB. I've also done some work to get a proper test environment, and THD fell to below 0.0008 dB. At these levels, the readings are fluctuating somewhat. The presented specs are a decent average representation of the measurements:
THD: 0.00190%
THD+N: 0.01013%
SNR: 112.83 dB (Inputs shorted, relative to 0 dBV/1 Vrms)
Crosstalk: -75.58 dB
IMD: 0.0195% (ITU-R 19+20 kHz)
Dynamic Range: >110 dB (THD+N: 1%)
THD: 0.00190%
THD+N: 0.01013%
SNR: 112.83 dB (Inputs shorted, relative to 0 dBV/1 Vrms)
Crosstalk: -75.58 dB
IMD: 0.0195% (ITU-R 19+20 kHz)
Dynamic Range: >110 dB (THD+N: 1%)
Having had several months now to play with this phono stage I thought I would post an update:
I did a little opamp rolling and ended up with an OPA2134 in the 1st position and LME4972 in position #2. Putting the LME4972 in seems to enhance the soundstage slightly.
As I said in my first post I also have a Hagerman Cornet 2 that I built several years ago. I have been going back and forth between this and the Cornet 2 a lot. I find myself preferring the opamp bases phono stage to the Cornet 2 most of the time.
Also as I said before the largest difference in the two was that the Cornet had a better soundstage. After adjusting the placement of my speakers and using the LME4972 in position #2 the gap has narrowed significantly.
I did a little opamp rolling and ended up with an OPA2134 in the 1st position and LME4972 in position #2. Putting the LME4972 in seems to enhance the soundstage slightly.
As I said in my first post I also have a Hagerman Cornet 2 that I built several years ago. I have been going back and forth between this and the Cornet 2 a lot. I find myself preferring the opamp bases phono stage to the Cornet 2 most of the time.
Also as I said before the largest difference in the two was that the Cornet had a better soundstage. After adjusting the placement of my speakers and using the LME4972 in position #2 the gap has narrowed significantly.
I'm really happy that you found an opamp configuration that suites you and your system cyto, although I wouldn't fully expect it to surpass the Cornet. At least not if you prefer tubes.
I did some measurements of the RIAA equalization. Except for ~0.2 to 0.3 dB higher level in the bass (150 Hz and below) and a +0.11 dB deviation at 2 kHz, the biggest deviation through the whole audio band was +0.04 dB. This is of course with the caps measured to be less than 1% away from what it says on the component, and matched for both channels.
I did some measurements of the RIAA equalization. Except for ~0.2 to 0.3 dB higher level in the bass (150 Hz and below) and a +0.11 dB deviation at 2 kHz, the biggest deviation through the whole audio band was +0.04 dB. This is of course with the caps measured to be less than 1% away from what it says on the component, and matched for both channels.
I get that by measuring and matching them.
EDIT:
I have not been able to actually measure the linearity, so I'm relying on the simulations that I have described earlier in this thread.
To what level of accuracy did you need to match the resistors and capacitors to get these results?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.