Slowly, you and your devoted followers are going to understand the design decisions in my HPS 4.1. I won't spoil your fun in the process 
You are a tough guy but you never spoiled my fun. On the contrary. whithout you this thread whould be more boring. Anyway, i am trying to catch up and in the quest for even better technical data there are even some circuits in the works (horrors of horrors) with feedback to the input stage. I hope they are innovative or at least interesting.
I was thinking more about your difficulties in adjusting the input bias current. It's indeed very difficult if you are trying to adjust it in a gain stage and without an embedded servo loop
This was one of the reasons why HPS4.1 uses input followers...
Last edited:
My DSL was down for one week so i could not post anything. Nevertheless i was bussy.
First i was not totally satisfied with the High Z Fet Headamp. Liquid and warm sounding it was but i missed some openness and dynamics and rhythm. I thought that this circuit (and the Low Z MPP too) is ideal for a bias servo that also makes the output cap obsolete.
I installed it and the sound improved a lot in terms of freshness and imediacy. If that is the result of the obsolete cap ( i used an old idustrial 2uF MKP) or the better DC balance i do not know but the jump in soundquality was unexpectedly big. Then i changed the 100 Ohm YAM2 resistor to a 1kOhm Caddock MK132. Again a big improvement in transparency. Maybe the slightly bright sound of the Caddock ( i had tried them in other circuits with less success, the sound always got a bit bright and sterile) complements the liquid and warm sound of the FET MPP well.
Brianco has send me some 3kOhm YAM2 so i will try them out too. Maybe the 100 Ohm was just too much loading the cartridge, anyway it sounded better then the 1kOhm nonmagnetic Tantals. Puzeling, now i am where i do not want to be, experimenting with cartridge loading. One little problem remains : when i crank the volume to insane levels ( at leat 20dB louder then i am used to listen) i can see a small and very low frequency moving of the bass cones, so i have to look if i can improve the stability of the Bias Servo. That should be a not unsurmountable problem.
I was also successfull to lower the noise (hiss) in my low Z designs. I can present here the Ultra Low Noise version of the Rosi and the Ultra Low Noise Low Z MPP. Both have such low noise that it is hard to keep the humm as low. A lot can be said about the quite challanging development work and i will aleborate more on the circuits if you ask me questions. The Ultra Low Noise Rosi was ready first and got maybe 2 hours of listening to some reference tracks. Does it sound better then the older MPP Low Z ?
Yes and no. On one hand the "old" MPP with a noise level of ca. 0.5nV/qHz is so quit in my system that i can not hear the noise in the listening seat at high volume. On the other hand i thought that the ULN Rosi had a different way of handling dynamics. Somehow the energy was shifted more into the deeper regions and in that region (800 Hz down) i could hear more dynamic range. For example when a female voice was singing and the ochestry set in the effect was more pronounced. "Light" intruments like strings profitted the most and sounded very creamy and fleshed out. Interestingly treble resolution was still fine but not in the same class as the standart MPP. If that is the effect of the even lower input impedance, the higher Bias current or the use of other components ( i am running out of high precission RIAA caps, i simply used a selected MKT with a parallel MKP in the ULN Rosi) i do not know yet.
Finally i have an Ultra Low Noise version of the Low Z MPP on the testbech. I have not listened to it so far but the measurements are promissing.
After i have optimised all three circuits i will end my adventure into unbalanced, low feedback ( they still have current feadback and in the case of the Rosi, lokal feedback) circuits. All sound to my full satisfaction and after the last stones have been turned i will produce some PCB´s. They can be made very low noise, wide bandwidth and low distortion ( -100dB) and there is no cocern about low level resolution. Anything measuring better needs global feedback and i will investigate if that sounds any better or worse. The topology i have choosen is the discreet intrumentation amplifier and i will tell you more tomorrow.
The future: My commercial phonostage for Spiral Groove uses intrumentation techniques ( with OP´s) with great results so what i whould like to develop here is the folowing : Balanced input with noise less then 0.5nV/qHz, balanced RIAA inductive and capacitive, balanced output, input stage discreet.
First i was not totally satisfied with the High Z Fet Headamp. Liquid and warm sounding it was but i missed some openness and dynamics and rhythm. I thought that this circuit (and the Low Z MPP too) is ideal for a bias servo that also makes the output cap obsolete.
I installed it and the sound improved a lot in terms of freshness and imediacy. If that is the result of the obsolete cap ( i used an old idustrial 2uF MKP) or the better DC balance i do not know but the jump in soundquality was unexpectedly big. Then i changed the 100 Ohm YAM2 resistor to a 1kOhm Caddock MK132. Again a big improvement in transparency. Maybe the slightly bright sound of the Caddock ( i had tried them in other circuits with less success, the sound always got a bit bright and sterile) complements the liquid and warm sound of the FET MPP well.
Brianco has send me some 3kOhm YAM2 so i will try them out too. Maybe the 100 Ohm was just too much loading the cartridge, anyway it sounded better then the 1kOhm nonmagnetic Tantals. Puzeling, now i am where i do not want to be, experimenting with cartridge loading. One little problem remains : when i crank the volume to insane levels ( at leat 20dB louder then i am used to listen) i can see a small and very low frequency moving of the bass cones, so i have to look if i can improve the stability of the Bias Servo. That should be a not unsurmountable problem.
I was also successfull to lower the noise (hiss) in my low Z designs. I can present here the Ultra Low Noise version of the Rosi and the Ultra Low Noise Low Z MPP. Both have such low noise that it is hard to keep the humm as low. A lot can be said about the quite challanging development work and i will aleborate more on the circuits if you ask me questions. The Ultra Low Noise Rosi was ready first and got maybe 2 hours of listening to some reference tracks. Does it sound better then the older MPP Low Z ?
Yes and no. On one hand the "old" MPP with a noise level of ca. 0.5nV/qHz is so quit in my system that i can not hear the noise in the listening seat at high volume. On the other hand i thought that the ULN Rosi had a different way of handling dynamics. Somehow the energy was shifted more into the deeper regions and in that region (800 Hz down) i could hear more dynamic range. For example when a female voice was singing and the ochestry set in the effect was more pronounced. "Light" intruments like strings profitted the most and sounded very creamy and fleshed out. Interestingly treble resolution was still fine but not in the same class as the standart MPP. If that is the effect of the even lower input impedance, the higher Bias current or the use of other components ( i am running out of high precission RIAA caps, i simply used a selected MKT with a parallel MKP in the ULN Rosi) i do not know yet.
Finally i have an Ultra Low Noise version of the Low Z MPP on the testbech. I have not listened to it so far but the measurements are promissing.
After i have optimised all three circuits i will end my adventure into unbalanced, low feedback ( they still have current feadback and in the case of the Rosi, lokal feedback) circuits. All sound to my full satisfaction and after the last stones have been turned i will produce some PCB´s. They can be made very low noise, wide bandwidth and low distortion ( -100dB) and there is no cocern about low level resolution. Anything measuring better needs global feedback and i will investigate if that sounds any better or worse. The topology i have choosen is the discreet intrumentation amplifier and i will tell you more tomorrow.
The future: My commercial phonostage for Spiral Groove uses intrumentation techniques ( with OP´s) with great results so what i whould like to develop here is the folowing : Balanced input with noise less then 0.5nV/qHz, balanced RIAA inductive and capacitive, balanced output, input stage discreet.
I thought my last post will get very long when i ad expanations but bit by bit i will do that now.
Rosi Ultra Low Noise:
I substituted the input transitors with 2SD786/2SB737 and raised the bias to 13mA. In this circuit a compromise has to be struck between noise and distortion. Raising the bias raises second harmonic and lowers third. The value chosen gives 0.002% third and 0.006% second at 300mV out at high frequencies. At cruising level the distortion is swapped by noise so this is a very low distortion circuit inspite of the opvious simplicity.
If you can not find the low Rbb ROHM transistors you can substitude for (in order of preference but performance should be very similiar):
Hitachi 2SC2547/2SA1085 (NOS)
Toshiba 2SC3329/2SA1315 (NOS)
in full production:
Sanyo 2SC3504/2SA1433
Sanyo 2SC3601/2SA1407
for a little bit more noise:
Fairchild KSC3503/KSA1381 (Rbb ca. 20 Ohm)
you can of cause go back to the standard Rosi and use the abundant and cheep 2N4401/4403
The noise with the low Rbb devices is around 0.3nV/qHz. This is so low that it is difficult to keep the humm level down to the same subjective contribution. I had succes with triple groudplane construction.
The output transistors are low level darlingtons. I used BC516/517. Others like MPSA12/65 can be substituted.
Adjust R14 for 3V over R15/16 for maximising dynamic range.
The circuit is very well suited to DIY. No sorting is necessarry for good operation although it helps to select the input trasistors for equal Hfe (+-10% is fine) to get the least amount of distortion.
No current flows into the cartridge because the PSU is floating. No DC at the output eather because of AC coupling. Use good quality caps of your taste here. I use a bank of 4 Rifa PHE450 4.7uF 250V with great results. The loss of detail is mimimal compared to DC coupling with this caps. I whould say that 2 to 5% of extreme overtones are swalowed replaced by a somewhat brighter presentation that gives the impression of even more detail between 10 and 15kHz but this is at the boundary of what i whould describe as audible and whould certainly be swapped in a double blind presentation.
You could experiment with a bank of 10 1uF caps in parallel with even higher voltage rating for even less loss or shunt with low value teflon cap etc. I am perfectly happy with my solution so in the here and now the case is settled.
For PSU i use 12V lead Gel accumulators shunted with 10.000uF elcap and 1uF MKT.
Some people may prefer the "sound" of other accumulators like Alkaline, Lithium etc.
I am perfectly happy with the result i get with the Lead cell.
Rosi Ultra Low Noise:
I substituted the input transitors with 2SD786/2SB737 and raised the bias to 13mA. In this circuit a compromise has to be struck between noise and distortion. Raising the bias raises second harmonic and lowers third. The value chosen gives 0.002% third and 0.006% second at 300mV out at high frequencies. At cruising level the distortion is swapped by noise so this is a very low distortion circuit inspite of the opvious simplicity.
If you can not find the low Rbb ROHM transistors you can substitude for (in order of preference but performance should be very similiar):
Hitachi 2SC2547/2SA1085 (NOS)
Toshiba 2SC3329/2SA1315 (NOS)
in full production:
Sanyo 2SC3504/2SA1433
Sanyo 2SC3601/2SA1407
for a little bit more noise:
Fairchild KSC3503/KSA1381 (Rbb ca. 20 Ohm)
you can of cause go back to the standard Rosi and use the abundant and cheep 2N4401/4403
The noise with the low Rbb devices is around 0.3nV/qHz. This is so low that it is difficult to keep the humm level down to the same subjective contribution. I had succes with triple groudplane construction.
The output transistors are low level darlingtons. I used BC516/517. Others like MPSA12/65 can be substituted.
Adjust R14 for 3V over R15/16 for maximising dynamic range.
The circuit is very well suited to DIY. No sorting is necessarry for good operation although it helps to select the input trasistors for equal Hfe (+-10% is fine) to get the least amount of distortion.
No current flows into the cartridge because the PSU is floating. No DC at the output eather because of AC coupling. Use good quality caps of your taste here. I use a bank of 4 Rifa PHE450 4.7uF 250V with great results. The loss of detail is mimimal compared to DC coupling with this caps. I whould say that 2 to 5% of extreme overtones are swalowed replaced by a somewhat brighter presentation that gives the impression of even more detail between 10 and 15kHz but this is at the boundary of what i whould describe as audible and whould certainly be swapped in a double blind presentation.
You could experiment with a bank of 10 1uF caps in parallel with even higher voltage rating for even less loss or shunt with low value teflon cap etc. I am perfectly happy with my solution so in the here and now the case is settled.
For PSU i use 12V lead Gel accumulators shunted with 10.000uF elcap and 1uF MKT.
Some people may prefer the "sound" of other accumulators like Alkaline, Lithium etc.
I am perfectly happy with the result i get with the Lead cell.
Just curious, have you evaluated/experimented with/measured these devices by yourself?
Last edited:
Now for the DC couple Fet Head Amp.
I did not change that circuit much. I substituted the contant current Fet with two 470 Ohm resistors for 9mA into the Led´s and added a (conventional ?) non inverting bias servo. I still have to work on stability because at insane levels a small and slow movement of the cones is visible. The sound made quite a jump into a more fresh and open direction. The reasons are unclear but i have the old industrial coupling cap under suspicion.
The Fets at the input run on 5mA each and are the trustworthy 2SK170/2SJ74 Toshibas that the nice gentlemen at CES donated. They are from the B type and have an IDss of around 8mA. They are now NOS and especially the 2SJ74 is hard to get. Linear Systems produce the LSK170 substitute and promissed a matching P channel. I hope for the best. Noise in this cicuit is around 0.5nV/qHz and sounds very "mellow". Not only voltage noise is low but also current noise so high to medium output MCs should work well too and the exotic top Grados will be happy too i gess.
No current flows into the cartridge and the input is adjustabe in impedance over a very wide margin. This is a mixed blessing i learned. I am still swapping resistors for best result. The circuit is totally DC coupled and easy to build. As i already mentioned the sound is quite a bit different from the Low Z types. A bit rounder and warmer. Very liquid, nearly "oily" like a nice warm bath in salt from the dead ocean ( i am running out of words). The soundstaging is very high, quite deep (i do not get carvernous depth in my sytem due to idiosycracies in my room) and extremely wide after the modification. Focus is as good as with the Low Z designs. The only thing i miss is the slam and drama in the bass i get from the low Z designs. This is maybe the function of the totaly different loading. It could effect didtortion and tracking. Anyway, for a person that loves tubes for what they are good at (tonal balance, tonal shadings, little fatique, fluidity, a kind of human tough, space) this is the semiconductor alternative.
I could not find matched pairs of the Fets so Iss was between 6 to 10mA. The degenaration resistors (10 Ohm) help but IDss matching whould certainly help to supress second harmonic so this could be the road to a slighty more "neutral" sound.
I did not change that circuit much. I substituted the contant current Fet with two 470 Ohm resistors for 9mA into the Led´s and added a (conventional ?) non inverting bias servo. I still have to work on stability because at insane levels a small and slow movement of the cones is visible. The sound made quite a jump into a more fresh and open direction. The reasons are unclear but i have the old industrial coupling cap under suspicion.
The Fets at the input run on 5mA each and are the trustworthy 2SK170/2SJ74 Toshibas that the nice gentlemen at CES donated. They are from the B type and have an IDss of around 8mA. They are now NOS and especially the 2SJ74 is hard to get. Linear Systems produce the LSK170 substitute and promissed a matching P channel. I hope for the best. Noise in this cicuit is around 0.5nV/qHz and sounds very "mellow". Not only voltage noise is low but also current noise so high to medium output MCs should work well too and the exotic top Grados will be happy too i gess.
No current flows into the cartridge and the input is adjustabe in impedance over a very wide margin. This is a mixed blessing i learned. I am still swapping resistors for best result. The circuit is totally DC coupled and easy to build. As i already mentioned the sound is quite a bit different from the Low Z types. A bit rounder and warmer. Very liquid, nearly "oily" like a nice warm bath in salt from the dead ocean ( i am running out of words). The soundstaging is very high, quite deep (i do not get carvernous depth in my sytem due to idiosycracies in my room) and extremely wide after the modification. Focus is as good as with the Low Z designs. The only thing i miss is the slam and drama in the bass i get from the low Z designs. This is maybe the function of the totaly different loading. It could effect didtortion and tracking. Anyway, for a person that loves tubes for what they are good at (tonal balance, tonal shadings, little fatique, fluidity, a kind of human tough, space) this is the semiconductor alternative.
I could not find matched pairs of the Fets so Iss was between 6 to 10mA. The degenaration resistors (10 Ohm) help but IDss matching whould certainly help to supress second harmonic so this could be the road to a slighty more "neutral" sound.
I have not measured the Sanyos and the Fairchilds but i trust your recommendation and measurements on your website. I have mentioned before in this tread that you are the one that came up with this choices. Maybe you missed parts of the tread. I do not take any pride to have dicovered them. Picasso ones said " i am a thieve" but nevertheless came up with some original matirial. I have ample experience with the ROHMs, Toshibas and Hitachis and still think they are the best choise because their high Hfe lowers base current and distortion. Important in my current feedback, low z designs.
The Ulta Low Noise Low Z MPP:
I am using now a double row of 2SD786/2SB737 with 22mA bias fed from adjustable, low pass filtered current mirrors with current feedback (the 5 Ohm resistors) for best bias stability and shunting of the noise of the current mirrors. Noise is now at around 0.3nV/qHz. Aproximately 4.5dB better then the standart MPP. Any more paralleling will not help beause the law of diminishing returns sets in. I am curious what incredible low noise Syn08 will get with his peltier cooling project. Noise is now so low that it is a challenge not to swap the noise with hum. Anyway i am good with avoiding hum so this should not be a bottleneck. Noise is around 10dB better then with a low noise OP like the AD797 or LT1115 for that matter, so this circuit plays in a different class noisewise. Still 10dB is only half the volume in terms of subjective loudness so i do not expect a miracle. Nevertheless it´s fun to try my best. The loading resistors of the input pair got quite low because the LEDs keep the voltage over them always at around 5V. That means that the noise of the cascode transistors can not be ingnored any more. That is the reason i am using two in parallel now. At the collector current chosen ( around 13mA) that pair has a base spread resistance of around 70 Ohm worsening the noise by aproximately 1.5dB. If you can not tolerate that you can use 2N4001/4003 (Rbb = 40 Ohm) or KSC3503/KSA1381 (Rbb = 18 Ohm). I prefer the BC550C/BC560C because of their high Hfe and you could use 4 in parallel. Clever people like Syn08 may have a better idea here. One bold move whould be to use the ROHMs in that position too. Case settled. My experience with transistor types is limitted and i shurely have to catch up what happened the last 10 years. Where i have good knowlage is with OP amps but that does not help much here. The constant voltage (5V) over the collector resistors makes it difficult to measure the input bias directly. One solution is to measure the voltage over the 49 Ohm base resistors and multiply that value with Hfe ( around 500 in that case). The value you get then devided by 49 gives the collector current. Set it at 22mA and the cascodes will then run on 13mA automatically.
I also did the bias servo trick so the circuit is now DC coupled. Use a low offset Fet OP here. In askending order of cost : TL071, OPA124, AD711, Opa827, OPA627 etc. I hope that a big discussion about the sound of servos will not follow. I heard an improvement in the FET version with the AD711 compared to the cap but already a friend of mine has called up and said that he whould much prefer the "sound" of the OPA627 in that position. And he is a German HF Dipl.Ing. !!! that builds high precission measurement instruments.
One concern is DC through the cartridge.
I use the following strategy making this stage not very well suited for volume manufacture but perfect for DIY where time is not the issue and tweeking is fun.
First i connect a resistor of 5 Ohm parallel to the input mimiking my 5.5 Ohm Cartridge. then i adjust the trimmers in the current mirror for 22mA on each line by measuring the voltage over the 49 Ohm resistors. DC offset can then be minimised by asymetric adjustments of the current mirrors. Having selected the input transistors for Hfe and ideally Ube helped but is not mandatory. When you get it down to under 0.05mV everything is fine soundwise. Bias is rock solid, so this value is stable. Just for fun i thermocoupled the input transistors with a teflon strap and heat conduction paste.
I am using now a double row of 2SD786/2SB737 with 22mA bias fed from adjustable, low pass filtered current mirrors with current feedback (the 5 Ohm resistors) for best bias stability and shunting of the noise of the current mirrors. Noise is now at around 0.3nV/qHz. Aproximately 4.5dB better then the standart MPP. Any more paralleling will not help beause the law of diminishing returns sets in. I am curious what incredible low noise Syn08 will get with his peltier cooling project. Noise is now so low that it is a challenge not to swap the noise with hum. Anyway i am good with avoiding hum so this should not be a bottleneck. Noise is around 10dB better then with a low noise OP like the AD797 or LT1115 for that matter, so this circuit plays in a different class noisewise. Still 10dB is only half the volume in terms of subjective loudness so i do not expect a miracle. Nevertheless it´s fun to try my best. The loading resistors of the input pair got quite low because the LEDs keep the voltage over them always at around 5V. That means that the noise of the cascode transistors can not be ingnored any more. That is the reason i am using two in parallel now. At the collector current chosen ( around 13mA) that pair has a base spread resistance of around 70 Ohm worsening the noise by aproximately 1.5dB. If you can not tolerate that you can use 2N4001/4003 (Rbb = 40 Ohm) or KSC3503/KSA1381 (Rbb = 18 Ohm). I prefer the BC550C/BC560C because of their high Hfe and you could use 4 in parallel. Clever people like Syn08 may have a better idea here. One bold move whould be to use the ROHMs in that position too. Case settled. My experience with transistor types is limitted and i shurely have to catch up what happened the last 10 years. Where i have good knowlage is with OP amps but that does not help much here. The constant voltage (5V) over the collector resistors makes it difficult to measure the input bias directly. One solution is to measure the voltage over the 49 Ohm base resistors and multiply that value with Hfe ( around 500 in that case). The value you get then devided by 49 gives the collector current. Set it at 22mA and the cascodes will then run on 13mA automatically.
I also did the bias servo trick so the circuit is now DC coupled. Use a low offset Fet OP here. In askending order of cost : TL071, OPA124, AD711, Opa827, OPA627 etc. I hope that a big discussion about the sound of servos will not follow. I heard an improvement in the FET version with the AD711 compared to the cap but already a friend of mine has called up and said that he whould much prefer the "sound" of the OPA627 in that position. And he is a German HF Dipl.Ing. !!! that builds high precission measurement instruments.
One concern is DC through the cartridge.
I use the following strategy making this stage not very well suited for volume manufacture but perfect for DIY where time is not the issue and tweeking is fun.
First i connect a resistor of 5 Ohm parallel to the input mimiking my 5.5 Ohm Cartridge. then i adjust the trimmers in the current mirror for 22mA on each line by measuring the voltage over the 49 Ohm resistors. DC offset can then be minimised by asymetric adjustments of the current mirrors. Having selected the input transistors for Hfe and ideally Ube helped but is not mandatory. When you get it down to under 0.05mV everything is fine soundwise. Bias is rock solid, so this value is stable. Just for fun i thermocoupled the input transistors with a teflon strap and heat conduction paste.
Joachim,
your comprehensive input is well received here. Wow!
When I have finished my pipelined projects, (I am working quite a bit slower than you) I will try your low Z mpp.
Your descriptions of the fet stage compared to the low z mpp and rosi do not surprise me! I credit this to the fets! I have yet to find a stage so open, both super good resolution and mellow at the same time (as a fet stage) combined with the tightness in the bass region my version of the phonoclone does provide (and my rabeyrolles had to a degree).
I'm having hopes with the HPS 3.1 I'm building at the moment, because of the tight feedback arrangement. (Nature uses feedback as well, so it can't be all wrong...)
I had the same subjective listining impressions with a fet follower vs. a fancy diamond follower. The latter doesn't have the fet's resolution but punches the bass region.
And I am sorry to say, that I always liked a 627 best as servo amp...
Rüdiger
your comprehensive input is well received here. Wow!
When I have finished my pipelined projects, (I am working quite a bit slower than you) I will try your low Z mpp.
Your descriptions of the fet stage compared to the low z mpp and rosi do not surprise me! I credit this to the fets! I have yet to find a stage so open, both super good resolution and mellow at the same time (as a fet stage) combined with the tightness in the bass region my version of the phonoclone does provide (and my rabeyrolles had to a degree).
I'm having hopes with the HPS 3.1 I'm building at the moment, because of the tight feedback arrangement. (Nature uses feedback as well, so it can't be all wrong...)
I had the same subjective listining impressions with a fet follower vs. a fancy diamond follower. The latter doesn't have the fet's resolution but punches the bass region.
And I am sorry to say, that I always liked a 627 best as servo amp...
Rüdiger
well Rüdiger, i will try other OPs for the servo. Usually i use OPA134 but you have to consider that the bias servo is not in the feedback path so i can not imagine how a music signal could pass through it but as my mother said "ich habe schon Pferde vor der Apotheke kotzen sehen" and she always claimed that she has witnessed the unlikely unicorn in the form of a "Kugelblitz". She is quite a character.
Anyway, i am happy to have convinced you about the merits of the Low Z MPP and look very much forward how you like the sound.
I will usher into feedback soon. I have no problem with the theoretical advantages but my arbitter is soundquality so when the measurements are "good enough" a circuit with inferiour measurements can sound very good as my Hiraga Optime adventure has shown not to mention tubes. How much distortion is audible and can certain kinds of distortion "enhance" the sound? When i read the work of Geddes for example we may need a reassasment. Again, i like low distortion and when i can make a circuit with very good measurements THAT SOUNDS GREAT i am happy. My view is that the human ear is not a simple microphone but whenever i try to elaborate on that theorem i get hatemail. That is the reason i stay out at the moment from the cable thread. Without my carefully chosen cables my system whould not sound the way it sounds but sadly i can not "prove" that in a way that convices the flat eathers. What can not be can not be, thats the argument i can not beat. I trust my ears more then my instruments and this is not a popular view in times where 44kHz 16Bit audio is considered "perfect" and "non distiguishable" from any system with more resolution.
A small anecdote ends my theatrise today. The plan was to get rid of the stability problem in the bias servo but i was not in the mood to solder. Instead i got involved into a two hour listening session. I had fired up the Ultra Low Noise Rosi with the capacitive RIAA and something was wrong. The soundstage was unspecific and the bass overblown. I had measured the electronics many times so i was sure that nothing was wrong there. Any time i change a piece of elctronics i run into similar problems. I have to rearange my system. Do you know that problem too ? Has it to do with burn in or is it a psychological problem or something else on the quantum level ( one change changes everything)?
This time i decided to change the physical speaker placement, the placement of my listening seat and i played with the balance control because my room is a bit asymetric from left to right. i spare you the details but suddenly the system started to sing like a bird and i heard things i never heard before. For example on "Kansas City Milkman" from the strangely recorded Level42 album i could hear a piano on the right channel that was 3 meters behind the right speaker and ocupied a space of around 4qmm. Then it was mixed to the midle and increased to a considarable size. I swear, i never heard such a spacial effect with digital and i have have heard a lot of systems up to the latest studio standarts. I record my own band with double resolution DSD so i have hands on experience.On "The Dude" i had the impression that my cealing was a magic mirror!
That is the fascination of analog. There seems no resolution limit.
Anyway, i am happy to have convinced you about the merits of the Low Z MPP and look very much forward how you like the sound.
I will usher into feedback soon. I have no problem with the theoretical advantages but my arbitter is soundquality so when the measurements are "good enough" a circuit with inferiour measurements can sound very good as my Hiraga Optime adventure has shown not to mention tubes. How much distortion is audible and can certain kinds of distortion "enhance" the sound? When i read the work of Geddes for example we may need a reassasment. Again, i like low distortion and when i can make a circuit with very good measurements THAT SOUNDS GREAT i am happy. My view is that the human ear is not a simple microphone but whenever i try to elaborate on that theorem i get hatemail. That is the reason i stay out at the moment from the cable thread. Without my carefully chosen cables my system whould not sound the way it sounds but sadly i can not "prove" that in a way that convices the flat eathers. What can not be can not be, thats the argument i can not beat. I trust my ears more then my instruments and this is not a popular view in times where 44kHz 16Bit audio is considered "perfect" and "non distiguishable" from any system with more resolution.
A small anecdote ends my theatrise today. The plan was to get rid of the stability problem in the bias servo but i was not in the mood to solder. Instead i got involved into a two hour listening session. I had fired up the Ultra Low Noise Rosi with the capacitive RIAA and something was wrong. The soundstage was unspecific and the bass overblown. I had measured the electronics many times so i was sure that nothing was wrong there. Any time i change a piece of elctronics i run into similar problems. I have to rearange my system. Do you know that problem too ? Has it to do with burn in or is it a psychological problem or something else on the quantum level ( one change changes everything)?
This time i decided to change the physical speaker placement, the placement of my listening seat and i played with the balance control because my room is a bit asymetric from left to right. i spare you the details but suddenly the system started to sing like a bird and i heard things i never heard before. For example on "Kansas City Milkman" from the strangely recorded Level42 album i could hear a piano on the right channel that was 3 meters behind the right speaker and ocupied a space of around 4qmm. Then it was mixed to the midle and increased to a considarable size. I swear, i never heard such a spacial effect with digital and i have have heard a lot of systems up to the latest studio standarts. I record my own band with double resolution DSD so i have hands on experience.On "The Dude" i had the impression that my cealing was a magic mirror!
That is the fascination of analog. There seems no resolution limit.
My idea is that measurable distortion is just the indication of how well explored a certain circuit with certain semis or valves is. I don't think that we hear the differences of the lowered distortion per se, but the circuit working better. Geddes I am sure can easily demonstrate how deaf we can be in induced harmonic percentages. He is acoustics. In electronics, when you manage your circuit for half the distortion, it is simply 6dB better, or one order more correct. That one has to manifest in resolution somehow. Its info is more orderly. There can be a 0.1% limit to get a valve CCT or 0.001% limit to get a FET CCT, you got to reach it. Then, when you will compare properly executed different CCTs for some devices, you can have preference irrespectively of the absolute value. For several reasons. In a system context that is.
I understand that well and you have alaborated on that before. Sadly better technical performance often comes wth complication e.g. more parts that ad their character. Of cause the flat eathers have no problem with 100 OP amps and 500 caps and pots in the chain. "they are inaudible" if i can improve a circuit without added complexity i whould be a fool not to try it. but look at a circuit like the Ros in it´s various disguises. Lower the noise and second harmonic goes up and third goes down. What authority tells you which compromise sounds the best? Why do you allow valves to distort more ? because that is what is duable ? i do not understand that rationate.
I used the word ''manage'' on purpose, as opposed to ''force''. So if you still can get the SAME topology circuit to distort less, it always resolves more in my humble experience. If you force it with more or different stuff you are no longer comparing the same systems. Say you have a woofer (example since you are a speaker man) in a certain finished speaker that you like best. And you get to manage its spider better. Nothing else. If you liked that speaker better than another that already had less harmonic at same SPL due to better power handling for instance, you may like it even more now, although you haven't reached the other's absolute distortion yet. That's your valve VS your OP AMP. The authority is your imagination from your collected musical experience and taste. Is what satisfies you. Ear is a machine, its interpreter (the brain) is an impostor. I.e. different strokes for different folks. Right or wrong, all design decisions manifest, and its a matter of perspective so to put them in a balanced context. That's the ''art'' part, IMHO. All else is EE and is in the books. Just a matter of being thorough, knowledgeable, hard working and experienced, and you can master the EE part. Many here have. Not me of course. Generaly speaking. Now the art part can be silly or essential depending on the audience. Its what pleases the creator, can't control more. You just let the bug loose out there and either succeeds for popularity or dies.
Talk abour spiders. They are the biggest source of distortion in a speakers and Prof.Dr.Klippel has devoted the best part of his life to research that phenomenon.
Talking about distortion in tube amps my friends at Test Factory, an independent institute in Germany that also works for Audio and Streoplay magazine found a distortion mechanism in poweramps that happens in the first 100msec after an impulse. Most tube ams and amps with low or no overal feedback did much better on this test then amps with high lokal feedback. Ironically Prof.Dr.Hawsford found similar (but not same) resolution problems with high feedback circuits in very low level stages. For me the doors are still open. We do not understand everything hopefully or i whould quit my job. When a 100,-€ Blue Ray player is perfect i see no reason to work in this bussiness.
The hardcore scientist whoud of cause argue that this subjective debate has already setled in the 60th so the only raison déntre is to make things cheaper and smaller.
Brave new world.
Talking about distortion in tube amps my friends at Test Factory, an independent institute in Germany that also works for Audio and Streoplay magazine found a distortion mechanism in poweramps that happens in the first 100msec after an impulse. Most tube ams and amps with low or no overal feedback did much better on this test then amps with high lokal feedback. Ironically Prof.Dr.Hawsford found similar (but not same) resolution problems with high feedback circuits in very low level stages. For me the doors are still open. We do not understand everything hopefully or i whould quit my job. When a 100,-€ Blue Ray player is perfect i see no reason to work in this bussiness.
The hardcore scientist whoud of cause argue that this subjective debate has already setled in the 60th so the only raison déntre is to make things cheaper and smaller.
Brave new world.
If everything was the same this forum would not exist. Or even better, it would still exist just for the shake of making them different to break boredom.
Depends on the audience as I wrote earlier. If its not a hobby, the current state of mass commercial devices is super adequate. See the happy people on their mp3 players in the subway. We are audio casualties, me for one I admit. Once bitten, forever smitten. At least we exchange our smarts here, for the monkey on our back not to cost an arm and a leg.
Depends on the audience as I wrote earlier. If its not a hobby, the current state of mass commercial devices is super adequate. See the happy people on their mp3 players in the subway. We are audio casualties, me for one I admit. Once bitten, forever smitten.
At least we exchange our smarts here, for the monkey on our back not to cost an arm and a leg.
Isn´t that strange that soundquality does not matter any more for the majority of "users".
Sometimes i feel like an alien from another planet. What concerns me more is that a lot of modern designers care only about technical excelence ( ore perfect mediocrity for that matter) that is quite easy to get with folowing the established rules. Listening is then condemned as subjective and non scientific. Ironically they make the money and the musically sensitive minority is acused for cheating.
How does a designer that already build the "subjectively transparent" amplifier justify his next project ? AH, i have the answer: it´s smaler and cheaper !
Sometimes i feel like an alien from another planet. What concerns me more is that a lot of modern designers care only about technical excelence ( ore perfect mediocrity for that matter) that is quite easy to get with folowing the established rules. Listening is then condemned as subjective and non scientific. Ironically they make the money and the musically sensitive minority is acused for cheating.
How does a designer that already build the "subjectively transparent" amplifier justify his next project ? AH, i have the answer: it´s smaler and cheaper !
It was always the same. The majority is never ''connoisseur'', in no sector. Does not mean that because beer is king, there is no fine wine cult all along. Its just specialist audio is not a new thing anymore. Back in its ''golden era'' was something popular to have. Now its embedded in humanity's collective mind as something old and time consuming. Back then music mattered also. The LP collection on weekends, and the cherishing, and the musical movements and the politics and the dreams. Now its stress and jets and computers and economy and fast fast fast. Most people can't have the time for finer things in life. Not even to acquire a general idea of them. Only the ones with solved economics. Thus specialized audio It is now something mainly for the ''refined'' commercially. Goes with the plush house, the car, and the Rolex. Hence the rich clueless customers targeting and the ''high-end'' cult VS the good old much more geek ''hi-fi'' folk. Back then was more rational just because it was targeting more people. So you got more engineering based stuff and more value. DIYers of today are the lamenting die hards of the hi-fi then, IMHO. That is why many get enraged with audio bubble speak or extravagant spending on parts without engineering proof. Its natural reflex for snake oil under the door crack maybe creeping in the last resort. Although they sometimes yield to sirens and some non magnetic Dales crop up. Its not that you may not speak honestly about what you do and hear, but there are so many out there that don't, and got well off of it, that the guard is high in the hardcore. Enjoy your ride.
Its not that you may not speak honestly about what you do and hear, but there are so many out there that don't, and got well off of it, that the guard is high in the hardcore. Enjoy your ride.