Dear Joachim
Rüdiger (Onvinyl) just spotted a mistyping of yours that all of us have not spotted.
Instead of typing “I tried to phone you”, you typed “I tried to phono you”.
No big deal, but since you have devoted hundred of hours writing about phono preamplifiers at this site, Rüdiger presented it as a credit to your activities.
Enjoy
Best Regards
George
Today i did some finishing touches on the low Z MPP. I consider that work now done. All circuits work fine and have been measured and listened too. What is missing is PCB´s and building and listening to the Salas Shunt Regulators. Salas, it´s a shame. I have not found the time to build some. I was too bussy to prepare for and work on the CES but i have seen some beatifull layout work that my friend Brian has done in the states. We found some SMD subtitudes for the bipolars and some FETs and will use zerro recovery diodes i think from a company called Cree.
So i present here the first finished version. It is a capacitive type with the 75usec timeconstant done transimpedance in the headamp. It has the advantage of simplicity and it totally DC coupled. The OPAMP i use is an LT1468. It is fast, has good precission, low noise and is one of the lowest distortion OPAMPs on the market. It has a folded cascode topology too. The servo is an OPA134. The servos i use are very slow to prevent effects in the audio range so before you switch on your system give it a New York minute to stabilise. I have selected the transistors in the headamp for Hfe and the input
transistors and the current mirrors for Ube too to get the least amount of current through the cartridge. I supply the headamp and the RIAA stage with seperate double mono LC audio low noise powersupplies anhanced by 27mH coils and big Elcaps with bypass. I decouple all OPAMPS with 220uF Pannasonic FM and 0.1uF Röderstein MKP lokaly.
Trim DC offset of the headamp to under 1mV with the trimmers. I thermocoupled the input transistors and the current mirrors. Bias is very stable and i have seen somewhere that Scott Wurzer said that one Bias source is enough and even better so i did not try any more sofisticated bias shemes on the input. Noise is very low (around 0.5nV qHz) with the ultra low noise ROHM transistors i had choosen. There are a lot of substitutes and Syn08 has a whole list somewhere here on the web. Noise could be lowered still by 3dB with bypassing and low pass filtering the input transistors with 100 Ohm resistors and 2200uF low Z elcaps but i did not do that in my quest for DC coupling. I recomment a bias of around 3mA on the input and 8mA on the folded cascodes. Bias on the folded cascodes should be set with a FET that has around 20mA IDss. You can experiment with higher values if you think that brings an advantage. I used red LEDs because they need the lowest voltage to glow and i have not much voltage on them. I did not try any other colour though.
Offset voltage of the headamp is somewhat dependend on the cartridge impedance beacuse there is still a VERY small current flowing through the cartidge. I emulated my 5.5Ohm Titan i with a 5.6Ohm resistor from input to ground (also emulating contact resistance : 0.1Ohm) during offset trimming.
So i present here the first finished version. It is a capacitive type with the 75usec timeconstant done transimpedance in the headamp. It has the advantage of simplicity and it totally DC coupled. The OPAMP i use is an LT1468. It is fast, has good precission, low noise and is one of the lowest distortion OPAMPs on the market. It has a folded cascode topology too. The servo is an OPA134. The servos i use are very slow to prevent effects in the audio range so before you switch on your system give it a New York minute to stabilise. I have selected the transistors in the headamp for Hfe and the input
transistors and the current mirrors for Ube too to get the least amount of current through the cartridge. I supply the headamp and the RIAA stage with seperate double mono LC audio low noise powersupplies anhanced by 27mH coils and big Elcaps with bypass. I decouple all OPAMPS with 220uF Pannasonic FM and 0.1uF Röderstein MKP lokaly.
Trim DC offset of the headamp to under 1mV with the trimmers. I thermocoupled the input transistors and the current mirrors. Bias is very stable and i have seen somewhere that Scott Wurzer said that one Bias source is enough and even better so i did not try any more sofisticated bias shemes on the input. Noise is very low (around 0.5nV qHz) with the ultra low noise ROHM transistors i had choosen. There are a lot of substitutes and Syn08 has a whole list somewhere here on the web. Noise could be lowered still by 3dB with bypassing and low pass filtering the input transistors with 100 Ohm resistors and 2200uF low Z elcaps but i did not do that in my quest for DC coupling. I recomment a bias of around 3mA on the input and 8mA on the folded cascodes. Bias on the folded cascodes should be set with a FET that has around 20mA IDss. You can experiment with higher values if you think that brings an advantage. I used red LEDs because they need the lowest voltage to glow and i have not much voltage on them. I did not try any other colour though.
Offset voltage of the headamp is somewhat dependend on the cartridge impedance beacuse there is still a VERY small current flowing through the cartidge. I emulated my 5.5Ohm Titan i with a 5.6Ohm resistor from input to ground (also emulating contact resistance : 0.1Ohm) during offset trimming.
Next is the inductive active version. It is very importand in this version that the DC offset of the input amp is low (under 1mV) to make the 100kOhm servo resistor work strong enough. In case you get problems you can lower it to 50kOhm or even 20kOhm. That will influence somewhat the teble response over 20kHz because the servo resistor is theoreical in parallel with the inductor resistor combination. Response will rise over 20kHz the lower the servo resistor is. This can be compensated for with a bigger cap in the transimpedance section of the headamp. Some people may even prefer that rise because it creates a certain airiness that may be artificial but very attractive. I changed the coil to 27mH and the resistor to 9 Ohm (that includes 4 Ohm from the coil). With that combination of parts RIAA was + - 0.15dB from 20 to 20Khz provided you use part with better then 1% accuracy. I can supply coils if you like to experiment.
The ones i have developped are very accurate and can be supplied in matched pairs. They have to be shielded and taken away from transformers for lowest hum. I will get some coils in MU metal quite soon that will solve that issue but unfortunately they are quite expensive. I have listened to this version the most because it has a very attractive, transparent midrange and very accurate tonal balance without any spittyness in the treble.
The ones i have developped are very accurate and can be supplied in matched pairs. They have to be shielded and taken away from transformers for lowest hum. I will get some coils in MU metal quite soon that will solve that issue but unfortunately they are quite expensive. I have listened to this version the most because it has a very attractive, transparent midrange and very accurate tonal balance without any spittyness in the treble.
Now comes the third version. Inductive active RIAA with passive shelvig filter. This version sounds the most transparent and it seems that the center of information has been slightly shifted upward especially with a 10nF silver mica in the shelve. Why this is i can only speculate because RIAA is accurate too.
DC offset can be a problem with this version beacuse it has enormous DC gain. One solution is a coupling cap between headamp and RIAA as shown here or a second servo on the second OPAMP. I recommend a precission FET opamp for the first stage. I used a OPA134 here not because i think it is the best but because i had it around. An OPA827 or ADA4627 in that position has 10dB lower distortion at 20kHz theoretically but didtortion is very low anyway because of the falling response of the inductive stage.
By the way i used an OPA134 in the active inductive RIAA too, this time biased into class A with a simple 3.3kOhm resistor to the filtered minus supply. It sounded great so i did not try constant current sources or cascodes.
DC offset can be a problem with this version beacuse it has enormous DC gain. One solution is a coupling cap between headamp and RIAA as shown here or a second servo on the second OPAMP. I recommend a precission FET opamp for the first stage. I used a OPA134 here not because i think it is the best but because i had it around. An OPA827 or ADA4627 in that position has 10dB lower distortion at 20kHz theoretically but didtortion is very low anyway because of the falling response of the inductive stage.
By the way i used an OPA134 in the active inductive RIAA too, this time biased into class A with a simple 3.3kOhm resistor to the filtered minus supply. It sounded great so i did not try constant current sources or cascodes.
So what´s next ?
I will start to work on High Z Headamps and i have decided to put the Hawsford Cascode Bipolar version on the side for some weeks beause i am very eager to explore FETs.
I have all the parts i need and will start tomorrow.
I have also another idea for the ECHO El Cheepo that lowers distortion and noise and will give it quite credible performance for very low cost and ease of availlability. I hope i can make up for the Echo Extreme desaster somewhat although i got it working with 8 BC550C in parallel now. Anyway, asymmetric designs can be a bit unsafe so i will not explore that further (sorry again, but they can sound great especially in transparency and dynamics).
I will start to work on High Z Headamps and i have decided to put the Hawsford Cascode Bipolar version on the side for some weeks beause i am very eager to explore FETs.
I have all the parts i need and will start tomorrow.
I have also another idea for the ECHO El Cheepo that lowers distortion and noise and will give it quite credible performance for very low cost and ease of availlability. I hope i can make up for the Echo Extreme desaster somewhat although i got it working with 8 BC550C in parallel now. Anyway, asymmetric designs can be a bit unsafe so i will not explore that further (sorry again, but they can sound great especially in transparency and dynamics).
This will be the basic structure of the MPP High Z Fet headamp. I have boiled it down to the minimum. Still it should work well because the bipolar version is very similiar and works to my satisfaction. The FETs are NOS Toshiba 2SK170BL and 2SJ74BL. My goal is 0.5nV/qHz noise and distortion under -100dB. How i will do the DC trimming i will see.
Joachim,
your final low-Z input stage is that far away from what I tried, but doesn't contain the fet-CCS of rabeyrolles. I wonder, if that is a reason for your better DC-behaviour. I used the THAT340 chip for the input quad, but still. For the cascodes, I tried both 2sc1775/2sa870a bjts and 2sk982/2sj148 mosfets that sounded better than the BC's.
It will be interesting to me, which input stage style you'll like more, low-Z or high-Z.
Rüdiger
your final low-Z input stage is that far away from what I tried, but doesn't contain the fet-CCS of rabeyrolles. I wonder, if that is a reason for your better DC-behaviour. I used the THAT340 chip for the input quad, but still. For the cascodes, I tried both 2sc1775/2sa870a bjts and 2sk982/2sj148 mosfets that sounded better than the BC's.
It will be interesting to me, which input stage style you'll like more, low-Z or high-Z.
Rüdiger
DC offset stability is really very good in the final design. I get it down to 1mV and less and it stays there. I know there are better and newer transistors then the BC´s and when i find the time i will try them out. Still i got -110dB distortion at cruising level.
The Rabeyrolles has quite high noise and does not allow easy bias changes like the MPP Low Z does.
The Rabeyrolles has quite high noise and does not allow easy bias changes like the MPP Low Z does.
Joachim,
A question. In Post 729 you give the scema of you latest non inductove headamp, and very elegant it is too.
But what is the purpose of the 2.2uS roll off in the output?
This is the direct opposite of the "Neuman" trick I pioneered, which would require the cap to be in series with the load R to flatten the response above 50kHz, not roll it off more than even RIAA requests?
If it's not a mistaken attempt at the "Neumann" 50kHz bend, then what's it for?
Regards, Allen
A question. In Post 729 you give the scema of you latest non inductove headamp, and very elegant it is too.
But what is the purpose of the 2.2uS roll off in the output?
This is the direct opposite of the "Neuman" trick I pioneered, which would require the cap to be in series with the load R to flatten the response above 50kHz, not roll it off more than even RIAA requests?
If it's not a mistaken attempt at the "Neumann" 50kHz bend, then what's it for?
Regards, Allen
High Allen !
Did i miss something ?
The way i see it : R8 x C3 = 340 x 0.22 = 74,8 usec (ca. 75 usec)
R13 x C6 = 3300 x 0.1 = 330 usec that should be ideally 318 usec but takes into acount that R14 is in parallel to C6
R14 x C6 = 31800 x 0.1 = 3180 usec
with this values accurate to 1% i measured RIAA precission from 20 to 20kHz with + - 0.15 dB. maybe i am blind but i can not see the 2.2usec pole in this design.
Did i miss something ?
The way i see it : R8 x C3 = 340 x 0.22 = 74,8 usec (ca. 75 usec)
R13 x C6 = 3300 x 0.1 = 330 usec that should be ideally 318 usec but takes into acount that R14 is in parallel to C6
R14 x C6 = 31800 x 0.1 = 3180 usec
with this values accurate to 1% i measured RIAA precission from 20 to 20kHz with + - 0.15 dB. maybe i am blind but i can not see the 2.2usec pole in this design.