We can speak nearly all languages, but not hochdeutsch.
Es lebe der Kini!
Regards
Sam
I was quite bussy that last two days but i got some good things done.
We are moving into a new building and i was able to setup part of my lab.
I also build a working prototype of the High Z Headamp just running on 9V alkaline blocks. The circuit as it where did not work because the single Led made the input starving on current and the cascodes did not conduct. Going to a string of 3 red 10mA LED´s solved the problem. The input transistors run now on 5MA each for a total of 10mA and the cascode runs on 11mA with the BF246A running on 8mA. I was not able to find enough selected pairs for the FET´s so i decided to take them straight from the pakage. I ended up with 130mV offset behind the cascodes (very stable to +-1mV) and 0.02% mostly second harmonic at 1 VRMS including the Capacitive Active RIAA. With trimming the offset at the right place i should be able to lower distorion too and DC coupling will be posible. For the moment i am using some industrial 2uF MKP caps.
Noise was very low at -145dB using a 5 ohm resistive devider as substitude for my 5.5 Ohm Titan cartridge. That translates into a dynamic range of 85dB referenced to 1mV and 79dB referenced to 0.5mV. I could improve that by 4dB with 8 transistors in parallel and slightly higher bias but the noise is subjectively extremely low and has a very "soft" quality so i thought this is more then good enough.
I also did a small chage to the bias arangement of the cascodes i raised the value of R8 and R9 and used a slightly bigger cap for C1 and C2 for improved filtering. Current gain at 700 is quite high in the cascode so these resistors could even be bigger. I will try that later. I used a 470 Ohm metal film as loading at the input but will try other values and makes and tell you if i can hear or measure any difference. I also have an idea for a bias servo that whould substitude the BF246A with a resistive ladder and an opamp.
RIAA precission was a straight line slightly falling to -0.5dB at 20kHz, the way i like it in my not treble shy system with the values i show here. I will do a version precise to +- 0.1dB if you like a totally flat response later.
We are moving into a new building and i was able to setup part of my lab.
I also build a working prototype of the High Z Headamp just running on 9V alkaline blocks. The circuit as it where did not work because the single Led made the input starving on current and the cascodes did not conduct. Going to a string of 3 red 10mA LED´s solved the problem. The input transistors run now on 5MA each for a total of 10mA and the cascode runs on 11mA with the BF246A running on 8mA. I was not able to find enough selected pairs for the FET´s so i decided to take them straight from the pakage. I ended up with 130mV offset behind the cascodes (very stable to +-1mV) and 0.02% mostly second harmonic at 1 VRMS including the Capacitive Active RIAA. With trimming the offset at the right place i should be able to lower distorion too and DC coupling will be posible. For the moment i am using some industrial 2uF MKP caps.
Noise was very low at -145dB using a 5 ohm resistive devider as substitude for my 5.5 Ohm Titan cartridge. That translates into a dynamic range of 85dB referenced to 1mV and 79dB referenced to 0.5mV. I could improve that by 4dB with 8 transistors in parallel and slightly higher bias but the noise is subjectively extremely low and has a very "soft" quality so i thought this is more then good enough.
I also did a small chage to the bias arangement of the cascodes i raised the value of R8 and R9 and used a slightly bigger cap for C1 and C2 for improved filtering. Current gain at 700 is quite high in the cascode so these resistors could even be bigger. I will try that later. I used a 470 Ohm metal film as loading at the input but will try other values and makes and tell you if i can hear or measure any difference. I also have an idea for a bias servo that whould substitude the BF246A with a resistive ladder and an opamp.
RIAA precission was a straight line slightly falling to -0.5dB at 20kHz, the way i like it in my not treble shy system with the values i show here. I will do a version precise to +- 0.1dB if you like a totally flat response later.
Today i build the secong channel of the High Z Headamp and it worked right away just fine. This time i used only one pair of modified LC Audio low noise PSUs, so no double-double mono this time, only double mono. Hiss and Hum is extremely low even slightly better then the Low Z MPP. Now i have of cause ideas to improve the Low Z version.
The sound i got after some cable swapping and grounding experiments is quite exiting. Whereas the Low Z Headamp majors on bass slamm, tightness and raw dynamics the High Z version has a more suble sound reminding me on tubes. The bass is a little softer and does not go so deep (subjectively) but there is a lot of tonal coulor, space and microdetail. Midrange is very open and transparent. Placement on the stage is a bit
less focussed but with a lot of acoustic information. Treble is not spitty at all and noise from the record is low. The majority of raw dynamics are a bit shifted from the bass to the midrange. Height, depth and width of the soundstge are quite similiar to the Low Z version. I experimented with the extremely expensive Wire World Platinum Eclipse from Tonearm to Headamp but found that sound ultimative a bit strange and went back to my home brewn Chaos-Fractal cable. It´s hard to express in words but i missed the freely floating, unfatiquing quality of my own creation although i heard some improvement in precission. Between Headamp and RIAA the Wire World Platinum Eclipse did a much better job. It sounded quite neutal there. Maybe it is not fully run in or there is some interaction because of high capacitance i do not know. By the way i decided on a 4.7kOhm 1 W metalfilm resistor as cartridge loading. I made a seperate input jack were i can insert other values and makes easily so i will do more experimenting soon. I plan also to try the Rosi with transimpedance 75 usec pole to directly compare to the High Z Headamp.
The sound i got after some cable swapping and grounding experiments is quite exiting. Whereas the Low Z Headamp majors on bass slamm, tightness and raw dynamics the High Z version has a more suble sound reminding me on tubes. The bass is a little softer and does not go so deep (subjectively) but there is a lot of tonal coulor, space and microdetail. Midrange is very open and transparent. Placement on the stage is a bit
less focussed but with a lot of acoustic information. Treble is not spitty at all and noise from the record is low. The majority of raw dynamics are a bit shifted from the bass to the midrange. Height, depth and width of the soundstge are quite similiar to the Low Z version. I experimented with the extremely expensive Wire World Platinum Eclipse from Tonearm to Headamp but found that sound ultimative a bit strange and went back to my home brewn Chaos-Fractal cable. It´s hard to express in words but i missed the freely floating, unfatiquing quality of my own creation although i heard some improvement in precission. Between Headamp and RIAA the Wire World Platinum Eclipse did a much better job. It sounded quite neutal there. Maybe it is not fully run in or there is some interaction because of high capacitance i do not know. By the way i decided on a 4.7kOhm 1 W metalfilm resistor as cartridge loading. I made a seperate input jack were i can insert other values and makes easily so i will do more experimenting soon. I plan also to try the Rosi with transimpedance 75 usec pole to directly compare to the High Z Headamp.
Well, i traded that cable for a pair of Sonics Amerigo speakers i designed. Davis Salz is using my speakers and i am using his cable. I was his importer in the 90th so we developed a great admiration. I could not efford that cable if i had to buy it in a shop and i whould not do it even if i could. I just have luck to be in the privileged position to get samples from some makers because they trust my expertise and listening ability. It´s a shame that we are so far away. I could imagine that you find the sound i get at home enjoyable. It´s precise and correct and transports a lot of emotion nevertheless.
Today i did some first listening tests with cartridge loading. I used the High Z Fet headamp. I replaced the 4.7kOhm 1W metal film with a 910 Ohm non magnetic Tantal (NOS). The sound got more transparent and fluid but it was not night and day. If that is the result of lower impedance or other type of resistor i do not know. I have some Dale RN60D antimagnetic in 1kOhm and Caddock MK132 in 1 kOhm that are closer in value to the Tantal so i will try them tomorrow.
I can not let my hands from Low Z headamps so i tryed to get the best out of the Rosi.
I raised the voltage to 12 V from a Gel Accu and changed the decoupling a bit. That lowered outputimpedance and i designed a fitting Transimpedance 75usec filter for the capacitive RIAA. The sound i got is so crasy good that i was flabbergasted. Very dramatic and BIG with very good tonal balanance. I immediately thought that i should look at the MPP low Z again and did some first improvements.
Looking carefully at the datasheet of the 2SB737 and the 2SD786 i found that Syn08 is right. Those super low Rbb transistors benefit from higher bias. Looking at 10Hz, 100Hz and 1kHz i thought that something a little under 10mA whould be a good compromise. I also decided to give the cascodes a bit more class a and ended up with 7.5mA on the input resistors and 13.5mA in the cascodes. I raised PSU voltage a bit to +-10V. More is not posible with the circuit i use now because the base-emitter diode is then braking down in the cascodes. I will look later if with another dimensioning higher voltage is posible. I do not think it will be very different but the trimmers alow even more bias on the input transitors. I think much more then 10mA makes no sense. The circuits works ok in the lab but i did not find the time to listen.
I can not let my hands from Low Z headamps so i tryed to get the best out of the Rosi.
I raised the voltage to 12 V from a Gel Accu and changed the decoupling a bit. That lowered outputimpedance and i designed a fitting Transimpedance 75usec filter for the capacitive RIAA. The sound i got is so crasy good that i was flabbergasted. Very dramatic and BIG with very good tonal balanance. I immediately thought that i should look at the MPP low Z again and did some first improvements.
Looking carefully at the datasheet of the 2SB737 and the 2SD786 i found that Syn08 is right. Those super low Rbb transistors benefit from higher bias. Looking at 10Hz, 100Hz and 1kHz i thought that something a little under 10mA whould be a good compromise. I also decided to give the cascodes a bit more class a and ended up with 7.5mA on the input resistors and 13.5mA in the cascodes. I raised PSU voltage a bit to +-10V. More is not posible with the circuit i use now because the base-emitter diode is then braking down in the cascodes. I will look later if with another dimensioning higher voltage is posible. I do not think it will be very different but the trimmers alow even more bias on the input transitors. I think much more then 10mA makes no sense. The circuits works ok in the lab but i did not find the time to listen.
I build the second channel of the Low Z MPP 2.0 and it works just fine. I listened breefly and the noise is audibly lower (ear on speaker). I can apreciate that. Collector potential is at 4.7 V around half of the supply voltage. That shoud optimise dynamic range.
I got a bit competitive now and my next project is a low Z MPP Headamp that cracks the 0.3nVqHz barier. I will drink a glas of wine now and draw the circuit tomorrow.
I got a bit competitive now and my next project is a low Z MPP Headamp that cracks the 0.3nVqHz barier. I will drink a glas of wine now and draw the circuit tomorrow.
Congratulation Joachim.
I'm very interested in knowing how you compare the NOS tantalum (not to say Shinkoh) vs. the DALE RN60 or Caddock MK132, but I'm not sure if the 1k is a good match for your Titan.
Unfortunately, I have not experience with that cartridge, but anyway I would suggest you
to try a MC transformer (some 20dB or less) and rise the loading resistor to 47k. Of course,
you probably don't need the extra gain, but maybe the influence of the input resistor will be less audible (maybe I'm wrong, since higher resistors produce usually higher noise...).
BTW, try to compare 470R vs 1k vs 47k loading of the same brand and let us know.
Ciao!
Tino
I'm very interested in knowing how you compare the NOS tantalum (not to say Shinkoh) vs. the DALE RN60 or Caddock MK132, but I'm not sure if the 1k is a good match for your Titan.
Unfortunately, I have not experience with that cartridge, but anyway I would suggest you
to try a MC transformer (some 20dB or less) and rise the loading resistor to 47k. Of course,
you probably don't need the extra gain, but maybe the influence of the input resistor will be less audible (maybe I'm wrong, since higher resistors produce usually higher noise...).
BTW, try to compare 470R vs 1k vs 47k loading of the same brand and let us know.
Ciao!
Tino
Yes Tino. I will try that. I also got some YAM2 resistors today and there are other candidates. I tried a lot of transformers, even very expensive and exotic japanese ones. Although the noise can be very low i found some resolution problems especially in the high treble even in the best ones, so momentaryly i do not plan to use transformers. I have idias to lower the noise even further and i think an active solution can be as quit or even better.
What i did today is listen to the MPP Low Z 2.0 and i am very satisfied. The sound has elements on the Rosi ( punsh, drama, sheer grunt) and the High Z Fet version ( warms, tonal colours etc.). The king is dead, long live the king.
What i did today is listen to the MPP Low Z 2.0 and i am very satisfied. The sound has elements on the Rosi ( punsh, drama, sheer grunt) and the High Z Fet version ( warms, tonal colours etc.). The king is dead, long live the king.
Today i will do more listening tests and work on the Salas shunts and the Ultra Low Noise Low Z MPP.
I will compare the Low Z MPP 2.0 to the High Z Fet MPP and also try different loading resistors on the High Z.
I have drawn a cicuit diagram of the Ultra Low Noise Low Z MPP Headamp. Noise should be around 0.3nVqHz. Better performane at room temperature is nearly imposibe so that is the boundary of what is doable unless i whould find a way to cool down the input transistors. I am pretty happy with the 0.5nVqHz - 0.6nVqHz performance i get with my latest circuits. There is absolutely no noise audible at the listening seat with my 86dB sensitive speakers at ear splitting volume so this is more an accademic exercise. I sit very close to the speakers, maybe 1.8m from them so in a typical environment (speakers at 2.5m) you could crank the volume even more without audible noise. Anyway, some of you may have cartridges with very low output and impedance and speakers with very high sensitivity and still you do not want to hear noise when you crank the volume. For such an environment the circuit shown here is the cure.
So when your cartridge is over 0.5mV there will not be any major improvment but there are some really challanging specimen around. Here is a small list that could give my older circuits a hard time with high sensitivity speakers in decending order of output:
Ortofon MC10 Super: 0.3mV 3 Ohm
Lyra Helicon SL : 0.22mV 3 Ohm
Audio Note 102VOH : 0.15mV Ohm ?
Audio Note 10-M : 0.12mV Ohm ?
Fidelity Research FR1 MK111 : 0.09mV 1 Ohm
Audio Note IO Ltd. : 0.04mV Ohm ? !!!!!!
P.S. : how does the Ultra Low Noise Low Z MPP work:
1: I lowpass filter the reference voltage from the current mirrors and shunt the noise to ground with a cap. That should give an improvement of 3dB.
2: i improved the current mirrors with helper tansistors for better regulation and precission.
3: i parallel four ultra low noise bipolars. That gives an advantage of 3dB too.
4: i raised the bias to 20mA in the inputstage for least voltage noise.
ceveat:
with this ultra low Z design the cartridge should have less then 10 Ohm impedance and low inductance to avoid treble loss. Send me the data of your cart and i could give you advice if it works.
6dB noise reduction is theoretical so expect more something like 4.5dB that is still worthwile.
P.P.S. The input transistors i will use are 2SA1085 and 2SC2597 because they have the best combination of low noise and high Hfe.
I will compare the Low Z MPP 2.0 to the High Z Fet MPP and also try different loading resistors on the High Z.
I have drawn a cicuit diagram of the Ultra Low Noise Low Z MPP Headamp. Noise should be around 0.3nVqHz. Better performane at room temperature is nearly imposibe so that is the boundary of what is doable unless i whould find a way to cool down the input transistors. I am pretty happy with the 0.5nVqHz - 0.6nVqHz performance i get with my latest circuits. There is absolutely no noise audible at the listening seat with my 86dB sensitive speakers at ear splitting volume so this is more an accademic exercise. I sit very close to the speakers, maybe 1.8m from them so in a typical environment (speakers at 2.5m) you could crank the volume even more without audible noise. Anyway, some of you may have cartridges with very low output and impedance and speakers with very high sensitivity and still you do not want to hear noise when you crank the volume. For such an environment the circuit shown here is the cure.
So when your cartridge is over 0.5mV there will not be any major improvment but there are some really challanging specimen around. Here is a small list that could give my older circuits a hard time with high sensitivity speakers in decending order of output:
Ortofon MC10 Super: 0.3mV 3 Ohm
Lyra Helicon SL : 0.22mV 3 Ohm
Audio Note 102VOH : 0.15mV Ohm ?
Audio Note 10-M : 0.12mV Ohm ?
Fidelity Research FR1 MK111 : 0.09mV 1 Ohm
Audio Note IO Ltd. : 0.04mV Ohm ? !!!!!!
P.S. : how does the Ultra Low Noise Low Z MPP work:
1: I lowpass filter the reference voltage from the current mirrors and shunt the noise to ground with a cap. That should give an improvement of 3dB.
2: i improved the current mirrors with helper tansistors for better regulation and precission.
3: i parallel four ultra low noise bipolars. That gives an advantage of 3dB too.
4: i raised the bias to 20mA in the inputstage for least voltage noise.
ceveat:
with this ultra low Z design the cartridge should have less then 10 Ohm impedance and low inductance to avoid treble loss. Send me the data of your cart and i could give you advice if it works.
6dB noise reduction is theoretical so expect more something like 4.5dB that is still worthwile.
P.P.S. The input transistors i will use are 2SA1085 and 2SC2597 because they have the best combination of low noise and high Hfe.
Hi Brianco ! I think then you really need the Low Z Ultra low Noise MPP. To get that performance with Fets i whould need to parallel at least 8 and use very low value degeneration.
Today my audiophile friend came and we compared the Low Z MPP 2.0 to the High Z Fet version. First we listened to the Low Z and sound was gorgeous. Big, real, certainly goosebump teritory. We then switched to the Hogh Z Fet and soundquality dropped by a considerable margin. More detached and artificial. We decided to do a coffee pause and after half an hour sound got much better. The High Z needs some warm up, that for SURE. Still not 100% satisfied we replaced the 910 Ohm Tantalum resistors with 100 Ohm YAM2 resistors. The sound got better still. More extended in the bass and treble and more alive. If that is the result of downloading the cartridge or the better sound of the resistor i do not know. I have some problems getting different types with the same impedance. Anyway, i know from the old times that the YAM2 is hard to beat. It´s a pitty that it is not made any more. Thank´s Brianco for sending me some. later the day we went back to the Low Z 2.0 and both of us prefered the sound over the Fet version. More real, more punsh, more fun, everything. We definetly prefer the Low Z version in my system. I still give not up on the Fet version. It is brand new and does not have the benefit of several month of development work on the Low Z version. With other cartridges that have higher impedance it may have some advantages, namely lower current noise.
Today my audiophile friend came and we compared the Low Z MPP 2.0 to the High Z Fet version. First we listened to the Low Z and sound was gorgeous. Big, real, certainly goosebump teritory. We then switched to the Hogh Z Fet and soundquality dropped by a considerable margin. More detached and artificial. We decided to do a coffee pause and after half an hour sound got much better. The High Z needs some warm up, that for SURE. Still not 100% satisfied we replaced the 910 Ohm Tantalum resistors with 100 Ohm YAM2 resistors. The sound got better still. More extended in the bass and treble and more alive. If that is the result of downloading the cartridge or the better sound of the resistor i do not know. I have some problems getting different types with the same impedance. Anyway, i know from the old times that the YAM2 is hard to beat. It´s a pitty that it is not made any more. Thank´s Brianco for sending me some. later the day we went back to the Low Z 2.0 and both of us prefered the sound over the Fet version. More real, more punsh, more fun, everything. We definetly prefer the Low Z version in my system. I still give not up on the Fet version. It is brand new and does not have the benefit of several month of development work on the Low Z version. With other cartridges that have higher impedance it may have some advantages, namely lower current noise.
Since 3 days i am fighting with the Ultra Low Noise MPP. One problem that comes up when i use high current in the inputstage is that it is very hard to adjust the current through the cartridge to zerro (ideally). So today i redesigned the powersupply to floating like in the Rosi. The circuits playes and no current flows in the cartridge but i have some humm problems. I hope i can sort it out. Another problem is that i need big coupling caps now but i have an idea to solve that too. Give me some time.