I made a basic noise calculation of the Low Z INA ( if it works ?).
It whould be around 0.73nV/qHz in an ideal world.
The contribution of the cartridge alone ( i measure 5.5 Ohm including the cabes) is 0.365nV/qHz so this configuration gives a noise figure of 6dB, very close to the physical boundary of a balanced stage.
If that happens in reality i doupt.
It whould be around 0.73nV/qHz in an ideal world.
The contribution of the cartridge alone ( i measure 5.5 Ohm including the cabes) is 0.365nV/qHz so this configuration gives a noise figure of 6dB, very close to the physical boundary of a balanced stage.
If that happens in reality i doupt.
Dynamic range can be expressed by signal/noise ratio and that is more common to the
audiophile without technical training. It is somewhat like HP in a car. The higher, the better.
I did some basic calculation to put that in perspective.
A 5.5 Ohm resistor alone ( say, my Titan i cartridge) has a noise figure of 0.365nV/qHz.
Measured over a bandwidth of 20kHz that transfers into a SN of 79.8dB. So with a headamp that is theoretically noiseless ( such a thing does not exist) this whould also be the boundary on dynamic range when we define the noise as perceptual boundary.
In reality it is bit different and humans and especially sea animals like a sharc are supurb to pick up information under the noise. An english Prof. for electronics engineering told me that a tone of say
3.5kHz ( right in the middle of the ears highest sensitivity) can be picked up 30dB under pink noise. Well, i did not do a test like that so far. I am just comunicating was was told to me.
The Low Z stage i posted has a noise figure of 0.73nV/qHz so over a bandwidth of 20kHz that transfers into a SN of 74dB.
Even the best stage i ever saw ( Syn08 HS4 with slightly under 0.3nV/qHz) whould ad some noise in this case around 2.5dB so SN with that stage and the Titan is 77.3dB, some 3.3dB better. This is not much considering the logarithmic nature of the ear. This is not halve as loud in any way. I think it was defined in the Elloit Audio Pages as " a little bit".
I will in no way critisise syn08s work. It´s a tremendoes tour de fource.
The only other person that impressed me as much concerning knowlage about noise is Bernhard Vogel. His book "The Sound of Silence" is an enyoyable read.
I prefer balanced stage out of other reasons.
They will always have a 3dB noise disadvantage but i hope my theatrise has put that in perspective to the real world situation.
P.S. Peaks of 20dB are there on some records so that whould improve dynamic range of my setup to 94dB. Well this levels do not happen in all frequency ranges at ones but when you read page 213 of JLH´s book "Valve & Transistor Audio Amplifiers" you will see that the maximum a modern helium cooled head can cut ( actually it can cut even stronger amplitudes but then it leaves the maximum permitted cutter excursion of 0.005cm and some serious groove jumping whould happen) is right there in the upper midrange where the ear is most sensitive. No wonder that subjective dynamic range from Vinyl can be huge.
I replicate the text here for compleetness :
" Walten considered that, under optimum conditions, and at a 7.5" groove diameter, a groove modulation velocity equivalent to +30dB at 2kHz, with respect to 1cm/sec, was the maximum which whould be cut."
Practical limitations set that to 25dB in a broad area around 500Hz and 8kHz.
The most famous extrem cut is a Shure Broth. test record that has bells cut at +25dB at 10kHz.
Note that Walton used 1cm/sec as reference so that data has to be scaled to make it valuabe for the more usual 5cm/sec.
audiophile without technical training. It is somewhat like HP in a car. The higher, the better.
I did some basic calculation to put that in perspective.
A 5.5 Ohm resistor alone ( say, my Titan i cartridge) has a noise figure of 0.365nV/qHz.
Measured over a bandwidth of 20kHz that transfers into a SN of 79.8dB. So with a headamp that is theoretically noiseless ( such a thing does not exist) this whould also be the boundary on dynamic range when we define the noise as perceptual boundary.
In reality it is bit different and humans and especially sea animals like a sharc are supurb to pick up information under the noise. An english Prof. for electronics engineering told me that a tone of say
3.5kHz ( right in the middle of the ears highest sensitivity) can be picked up 30dB under pink noise. Well, i did not do a test like that so far. I am just comunicating was was told to me.
The Low Z stage i posted has a noise figure of 0.73nV/qHz so over a bandwidth of 20kHz that transfers into a SN of 74dB.
Even the best stage i ever saw ( Syn08 HS4 with slightly under 0.3nV/qHz) whould ad some noise in this case around 2.5dB so SN with that stage and the Titan is 77.3dB, some 3.3dB better. This is not much considering the logarithmic nature of the ear. This is not halve as loud in any way. I think it was defined in the Elloit Audio Pages as " a little bit".
I will in no way critisise syn08s work. It´s a tremendoes tour de fource.
The only other person that impressed me as much concerning knowlage about noise is Bernhard Vogel. His book "The Sound of Silence" is an enyoyable read.
I prefer balanced stage out of other reasons.
They will always have a 3dB noise disadvantage but i hope my theatrise has put that in perspective to the real world situation.
P.S. Peaks of 20dB are there on some records so that whould improve dynamic range of my setup to 94dB. Well this levels do not happen in all frequency ranges at ones but when you read page 213 of JLH´s book "Valve & Transistor Audio Amplifiers" you will see that the maximum a modern helium cooled head can cut ( actually it can cut even stronger amplitudes but then it leaves the maximum permitted cutter excursion of 0.005cm and some serious groove jumping whould happen) is right there in the upper midrange where the ear is most sensitive. No wonder that subjective dynamic range from Vinyl can be huge.
I replicate the text here for compleetness :
" Walten considered that, under optimum conditions, and at a 7.5" groove diameter, a groove modulation velocity equivalent to +30dB at 2kHz, with respect to 1cm/sec, was the maximum which whould be cut."
Practical limitations set that to 25dB in a broad area around 500Hz and 8kHz.
The most famous extrem cut is a Shure Broth. test record that has bells cut at +25dB at 10kHz.
Note that Walton used 1cm/sec as reference so that data has to be scaled to make it valuabe for the more usual 5cm/sec.
I am suffering from a particular bad case of overloadis maximus and noisea.
I had to do some maesurements with my poor home equipment to asure me that the Low Z version i posted here whould work.
I tried all kind of records. Old, new, badly worn and 45s. I measured voltage after the shunt stage differentially before the 5.1kOhm resistors of the diff stage.
That was the most reliable point and gave the highest raedings.
I sampled the output and integrated it with a time constant of 1sec. Measuring after the output of the diff stage the readings where around 30% lower, so i have captured some high frequency content at least. I will repeat that measurements with my 250MHz analog scope when i am back in the factory to capture high frequency transients.
The biggest reading was 60mV on "March to the Scaffold" on 45 RR known for extrem crescendos. That is some 41.5 dB higher then the 0.5mV @ 5cm/sec average of my
Titan i cartridje. Most records that sounded very dynamic subjectively like the reissue of "Crime of the Century" from Supertramp hovered around 20mV maximum.
I was ready to go.
I build some cap jumpers with 2200uF Nichicon CL bypassed with 1uF BC components MKT. After implementing the change to Low Z the circuit worked fine right away.
I was a bit surprised. How is the sound ? Geil !
Very big and robust.
I have to listen more but that Discreet INA has already won my heart.
P.S. just for compleetness and not to scare the cartridge current sensitive i measured the voltage across the cartridge terminals. Nill with my home equipment. I will measure again with my Sennheiser high impedance meter but i can asure you that it is under 15uV, and there is the canceling action going on.
Also the low frequency limit of the 2200uF caps is a bit better then i calculated because the cartridge impedance adds for a total of 15,5 Ohm and 1100uF.
One advantage of the Low Z stage is that you can set AC and DC gain seperately if you wish.
This circuit starts to fascinate me. What a ride ! Thanks to all that contributed.
I had to do some maesurements with my poor home equipment to asure me that the Low Z version i posted here whould work.
I tried all kind of records. Old, new, badly worn and 45s. I measured voltage after the shunt stage differentially before the 5.1kOhm resistors of the diff stage.
That was the most reliable point and gave the highest raedings.
I sampled the output and integrated it with a time constant of 1sec. Measuring after the output of the diff stage the readings where around 30% lower, so i have captured some high frequency content at least. I will repeat that measurements with my 250MHz analog scope when i am back in the factory to capture high frequency transients.
The biggest reading was 60mV on "March to the Scaffold" on 45 RR known for extrem crescendos. That is some 41.5 dB higher then the 0.5mV @ 5cm/sec average of my
Titan i cartridje. Most records that sounded very dynamic subjectively like the reissue of "Crime of the Century" from Supertramp hovered around 20mV maximum.
I was ready to go.
I build some cap jumpers with 2200uF Nichicon CL bypassed with 1uF BC components MKT. After implementing the change to Low Z the circuit worked fine right away.
I was a bit surprised. How is the sound ? Geil !
Very big and robust.
I have to listen more but that Discreet INA has already won my heart.
P.S. just for compleetness and not to scare the cartridge current sensitive i measured the voltage across the cartridge terminals. Nill with my home equipment. I will measure again with my Sennheiser high impedance meter but i can asure you that it is under 15uV, and there is the canceling action going on.
Also the low frequency limit of the 2200uF caps is a bit better then i calculated because the cartridge impedance adds for a total of 15,5 Ohm and 1100uF.
One advantage of the Low Z stage is that you can set AC and DC gain seperately if you wish.
This circuit starts to fascinate me. What a ride ! Thanks to all that contributed.
i told you about my long cable between RIAA and preamp.
I thought it could be interesting to make that connecion balanced.
The INA has a balanced input and the 75usec part of the RIAA is balanced too, so doing also the rest balanced is logical.
The simple way whould be to intruduce a passive RIAA and make the output with a fully balanced high quality chip like the OPA1632.
Exactly that was done by Russ White in his Retro stage.
So implementing this solution after the input whould work well.
The passive RIAA has a loss of 40dB @ 20kHz so the input stage needs more gain then that to avoid a noise panalty.
The INA has great freedom in gain just by changing one resistor and the Low Z version
(here shown with only 4 input transistors) has a gain of 45.5dB already with my Titan i.
With this higher gain comes an overload penalty but i had asured me that this is not a practical problem.
Looking at the shematic, i do not have a symbol for a balanced in balanced out chip so i took a modified symbol of an opamp.
If you are in doupt look up the retro thread.
I also made a slight modification in the second shematic.
The Retro uses only one cap for the upper portion of the RIAA. This works only perfect if the input stage is tightly balanced. Otherwise the RIAA could be screwed up.
The solution in that case is to split the RIAA and reference the caps to ground.
This comes with a penalty in the form that now two bigger caps are needed instead of one smaller. That raises cost and size.
If it brings any benefit, i have to try.
A plethora of other balanced RIAAs are posible and i will elaborate on them later.
My favourite whould be to do the 75usec passive and the rest balanced active.
A balanced inductive version is in the makes too.
I also have to design a balanced input receiver because my preamp is unbalanced.
I already have a good idea.
I thought it could be interesting to make that connecion balanced.
The INA has a balanced input and the 75usec part of the RIAA is balanced too, so doing also the rest balanced is logical.
The simple way whould be to intruduce a passive RIAA and make the output with a fully balanced high quality chip like the OPA1632.
Exactly that was done by Russ White in his Retro stage.
So implementing this solution after the input whould work well.
The passive RIAA has a loss of 40dB @ 20kHz so the input stage needs more gain then that to avoid a noise panalty.
The INA has great freedom in gain just by changing one resistor and the Low Z version
(here shown with only 4 input transistors) has a gain of 45.5dB already with my Titan i.
With this higher gain comes an overload penalty but i had asured me that this is not a practical problem.
Looking at the shematic, i do not have a symbol for a balanced in balanced out chip so i took a modified symbol of an opamp.
If you are in doupt look up the retro thread.
I also made a slight modification in the second shematic.
The Retro uses only one cap for the upper portion of the RIAA. This works only perfect if the input stage is tightly balanced. Otherwise the RIAA could be screwed up.
The solution in that case is to split the RIAA and reference the caps to ground.
This comes with a penalty in the form that now two bigger caps are needed instead of one smaller. That raises cost and size.
If it brings any benefit, i have to try.
A plethora of other balanced RIAAs are posible and i will elaborate on them later.
My favourite whould be to do the 75usec passive and the rest balanced active.
A balanced inductive version is in the makes too.
I also have to design a balanced input receiver because my preamp is unbalanced.
I already have a good idea.
front end amp + passive RIAA yield a 5.5 db gain, thus destroying
the first stage low noise ratio...
the second stage, with 60db gain , must have an EIN no more than
5.5 db higher than the first stage to preserve the signal/noise ratio
provided by the latter....
since the front end is in the vicinity of 0.5 nV/sqrt hz , the second
stage must have no more than 1 nV/sqrt hz EIN....
the first stage low noise ratio...
the second stage, with 60db gain , must have an EIN no more than
5.5 db higher than the first stage to preserve the signal/noise ratio
provided by the latter....
since the front end is in the vicinity of 0.5 nV/sqrt hz , the second
stage must have no more than 1 nV/sqrt hz EIN....
Hello Wahab ! How whould you distribute the gain to get the best SN ? This circuit has a wide choice of gain. It is not my favourite anyway. I would rather go for a full active version or split the RIAA into 75usec passive and 318usec and 3180usec active.
I simply did not find the time to draw other options or optimise the passive version.
Today i got a visit from Martina Schöner from Garrard / Loricraft.
We listened to the Low Z INA. Just for fun we tried if she could hear a difference when i bias the output stage into Class - A . I used the most simpe topology in the form of a 3.3kOhm resistor from output to the minus or the plus supply. Usually the resistor is connected to the minus supply with the rationate that NPN transistors work better then PNP typs. See a discussion on the Tangentsoft webpage.
I was especially interested in the connection to the plus supply because i could not find a circuit on the web that uses that option.
She prefered the plus connection, second was no connection and the looser was the minus connection. She is a woman 10 years younger then me, trained in classical singing so she should have a good ear.
She claimed to hear differences in the 3 dimensional imaging and amount of information. The plus connection gave the most dynamic, spacious and informative sound she told me.
I was a bit surprised about that and looked up the OP134 spec sheet. What struck me was that PSR was 30dB better over 1kOhm in the negative supply.
Maybe that is an explanation and not so much the Class-A drive. The OPA134 drives a 33kOhm load, so low distortion should be ensured without biasing.
That 3.3kOhm resistor gives around 3.6mA. The OPA134 now has to drive 33kOhm in parallel to 3.3kOhm. That is 3kOhm so it can deliver 10.8V in Class - A into that load.
I thought that is a good compromise. A fet constant current source, a fet cascode, or a bipolar current mirror could be used for higher dynamic impedance, but as i said, it was just for fun but adjucating.
Today i got my long awaited Mu-Metal shielded RIAA coils after my specification from
Sowter. They look fine but where quite expensive (autsch...).
I will try the inductive RIAA again and will find out if those eliminate the last trace of hum. I will use the High Z MPP for that. Today i got some non magnetic Tantals from Brianco so now i can compare 1kOhm Tantal to 47kOhm Tantal.
Sorry Allen that this experiment had to wait for so long but i do not want to compare apples to oranges. I found earlier that the type of resistor is audible and not only the value. I will try to get some 47kOhm Z foil to but then i really have to make a break in component swapping and listen more to music.
I simply did not find the time to draw other options or optimise the passive version.
Today i got a visit from Martina Schöner from Garrard / Loricraft.
We listened to the Low Z INA. Just for fun we tried if she could hear a difference when i bias the output stage into Class - A . I used the most simpe topology in the form of a 3.3kOhm resistor from output to the minus or the plus supply. Usually the resistor is connected to the minus supply with the rationate that NPN transistors work better then PNP typs. See a discussion on the Tangentsoft webpage.
I was especially interested in the connection to the plus supply because i could not find a circuit on the web that uses that option.
She prefered the plus connection, second was no connection and the looser was the minus connection. She is a woman 10 years younger then me, trained in classical singing so she should have a good ear.
She claimed to hear differences in the 3 dimensional imaging and amount of information. The plus connection gave the most dynamic, spacious and informative sound she told me.
I was a bit surprised about that and looked up the OP134 spec sheet. What struck me was that PSR was 30dB better over 1kOhm in the negative supply.
Maybe that is an explanation and not so much the Class-A drive. The OPA134 drives a 33kOhm load, so low distortion should be ensured without biasing.
That 3.3kOhm resistor gives around 3.6mA. The OPA134 now has to drive 33kOhm in parallel to 3.3kOhm. That is 3kOhm so it can deliver 10.8V in Class - A into that load.
I thought that is a good compromise. A fet constant current source, a fet cascode, or a bipolar current mirror could be used for higher dynamic impedance, but as i said, it was just for fun but adjucating.
Today i got my long awaited Mu-Metal shielded RIAA coils after my specification from
Sowter. They look fine but where quite expensive (autsch...).
I will try the inductive RIAA again and will find out if those eliminate the last trace of hum. I will use the High Z MPP for that. Today i got some non magnetic Tantals from Brianco so now i can compare 1kOhm Tantal to 47kOhm Tantal.
Sorry Allen that this experiment had to wait for so long but i do not want to compare apples to oranges. I found earlier that the type of resistor is audible and not only the value. I will try to get some 47kOhm Z foil to but then i really have to make a break in component swapping and listen more to music.
The last two days i was playing with my Sowter coils. I rebuild the inductive RIAA and did some tweaking. I like to DC couple, but with this circuit it is not that easy. The inductive stage has a lot of DC gain (10000 / 10 = 60dB) so the servo needs some authority. Making the resistor in series with the servo small changes frequency response because the servo resistor is in parallel with the coil. I found 100kOhm a good compromise. It barely changes the response of the circuit. Lower resistors like that whould lead to raised treble. For example a 10kOhm resistor whould give 0.6dB more output at 20kHz. This is not much but i am a perfectionist. To make the 100kOhm resistor work the Opamp in the first stage has to be a precission Fet Opamp. I decided on the OPA134. If in doupt take the tighter specked OP132. OPA827 or ADA4627 could be subtituted. They have 10dB more gain at 20kHz so should give even lower distortion.
Anyway, distortion is low in that stage because amplification is falling with 6dB octave and the next stage has virtually no loading. Just for fun i biased the OPA134 into Class-A with a 3.3kOhm resistor to the plus supply. Read my earliers posts for the reason why. I just got an e-mail from Nigel Pearson of Loricraft that has a lot of experience with that tweak. After reading that i must say that i barely have scratched the surface.
Maybe i can persuade Nigel to post about that and be more specific.
In one of his phonostages he has 3 OPs in parallel. One tied to plus, one tied to minus and one not biased into Class-A. He should talk about the reasons because i can imagine that some of you find that very esoteric or even disturbing and i realy do not like to destroy your beleave system.
Anyway, i also shifted the 2.2usec pole between the input OP and output OP.
With the 100kOhm resistor one servo swings 3.5V and the other 5.5V.
The maximum they can swing with the +-14V PSU is 12.5V so the headamp should not have much more then 5mV DC offset ( with a small safety margin) unless considarable DC appears at the output lowering dynamic range and disturbing any DC coupled stage that follows it.
Now the good news : the Mu-Metal encased Sowters are totally hum free even when i do not earth the case. They have long wire and i play them right in the open with 2mm Multicontact banana jacks. Of cause i experimentd with eathing the case and the advantage was extremely little. Unfortunately they cost a furtune but i can understand now why Sowter products are used in the hundreds in prestigious professional boards and effects. Great job Brian, i am totally happy.
Because the circuit does not like DC i used the Ultra Low Noise Rosi. In the tweak mood i bypassed the Rifa450 output caps with NOS Röderstein KP1832. They have 12.5nF and 1500V. Caps in that quality are simply not made in Germany any more but i heard from RTI, one of the biggest makers of audiophile caps in the states, that we still make the best foils.
see this link for some pictures : http://www.steinmusik.de/audiofino/bauteile-
kondensatoren.html
The Ulta Low Noise Rosi is the quietest stage i made so far and holds 0.4nV/qHz even down to 10Hz. In combination with the inductive RIAA noise is a non issue.
Soundwise the low impedance Rosi and the inductive RIAA pack a lot of meat on the bones. Again that special open quality in the midrange was obvious from the inductive stage and the Rosi simply packs a punsh.
When you ask me what i prefer : the Rosi-Inductive or the INA-Capacitive i found one comparison:
The Rosi-Inductive is a master butcher that presents his wares in a nicely lit glas case and the INA.Capacitive is a precission surgeon hat lets you watch the laser surgery on a high tech Oled screen.
Anyway, distortion is low in that stage because amplification is falling with 6dB octave and the next stage has virtually no loading. Just for fun i biased the OPA134 into Class-A with a 3.3kOhm resistor to the plus supply. Read my earliers posts for the reason why. I just got an e-mail from Nigel Pearson of Loricraft that has a lot of experience with that tweak. After reading that i must say that i barely have scratched the surface.
Maybe i can persuade Nigel to post about that and be more specific.
In one of his phonostages he has 3 OPs in parallel. One tied to plus, one tied to minus and one not biased into Class-A. He should talk about the reasons because i can imagine that some of you find that very esoteric or even disturbing and i realy do not like to destroy your beleave system.
Anyway, i also shifted the 2.2usec pole between the input OP and output OP.
With the 100kOhm resistor one servo swings 3.5V and the other 5.5V.
The maximum they can swing with the +-14V PSU is 12.5V so the headamp should not have much more then 5mV DC offset ( with a small safety margin) unless considarable DC appears at the output lowering dynamic range and disturbing any DC coupled stage that follows it.
Now the good news : the Mu-Metal encased Sowters are totally hum free even when i do not earth the case. They have long wire and i play them right in the open with 2mm Multicontact banana jacks. Of cause i experimentd with eathing the case and the advantage was extremely little. Unfortunately they cost a furtune but i can understand now why Sowter products are used in the hundreds in prestigious professional boards and effects. Great job Brian, i am totally happy.
Because the circuit does not like DC i used the Ultra Low Noise Rosi. In the tweak mood i bypassed the Rifa450 output caps with NOS Röderstein KP1832. They have 12.5nF and 1500V. Caps in that quality are simply not made in Germany any more but i heard from RTI, one of the biggest makers of audiophile caps in the states, that we still make the best foils.
see this link for some pictures : http://www.steinmusik.de/audiofino/bauteile-
kondensatoren.html
The Ulta Low Noise Rosi is the quietest stage i made so far and holds 0.4nV/qHz even down to 10Hz. In combination with the inductive RIAA noise is a non issue.
Soundwise the low impedance Rosi and the inductive RIAA pack a lot of meat on the bones. Again that special open quality in the midrange was obvious from the inductive stage and the Rosi simply packs a punsh.
When you ask me what i prefer : the Rosi-Inductive or the INA-Capacitive i found one comparison:
The Rosi-Inductive is a master butcher that presents his wares in a nicely lit glas case and the INA.Capacitive is a precission surgeon hat lets you watch the laser surgery on a high tech Oled screen.
Thank you very much. That is ecouraging.
Tomorrow i will post some balanced RIAA´s based on the Discreet INA.
I design with my subconciesness. When you do something very intense the brain works in the backround but who am i talking too ?
What really fascinates me is that this is a never ending story.
It opened up a new world for me and i am really enyoying it.
Tomorrow i will post some balanced RIAA´s based on the Discreet INA.
I design with my subconciesness. When you do something very intense the brain works in the backround but who am i talking too ?
What really fascinates me is that this is a never ending story.
It opened up a new world for me and i am really enyoying it.
Saturday afternoon project
I thought about a little project i can make on saturday afternoon with parts i have.
It´s the INA with Balanced Shunt-Passive RIAA all in one go Unbalanced Out.
I can spare the capacitive RIAA that way. For the output chip i will try the LT1468 instead
of the OPA134.
I thought about a little project i can make on saturday afternoon with parts i have.
It´s the INA with Balanced Shunt-Passive RIAA all in one go Unbalanced Out.
I can spare the capacitive RIAA that way. For the output chip i will try the LT1468 instead
of the OPA134.
OT
Have you seen Jan's site re. intuitive decision making etc?
http://www.linearaudio.nl/musings-1.htm
I suspect that the trick is to learn to trust intuition .
Have you seen Jan's site re. intuitive decision making etc?
http://www.linearaudio.nl/musings-1.htm
I suspect that the trick is to learn to trust intuition .
My project was successful. I was a bit nervous but also exited because this is the first balanced RIAA i build. I had designed some on paper and did the 75usec part balanced but not AIO ( all in one).
It worked right away.
Some elaboration is posible and leads to better performance.
I am posting the circuit to that i listen momentarily.
It has a servo added. The DC offset was around 50mV without servo. No problem in a system where a cap follows. If you want to go "naked" use the servo. It performs better too because it creates a virtual ground which isolates the transimpedance RIAA from ground current polution.
See the 3.3kOhm resistor from output to - supply. It makes the LT1648 biased into class A. I found this better then wihout.
The next step whould be a balnced in - balanced out RIAA and the new Burr-Brown BI-BO chips are perfect for that. I ordered some and will test the circuit.
It worked right away.
Some elaboration is posible and leads to better performance.
I am posting the circuit to that i listen momentarily.
It has a servo added. The DC offset was around 50mV without servo. No problem in a system where a cap follows. If you want to go "naked" use the servo. It performs better too because it creates a virtual ground which isolates the transimpedance RIAA from ground current polution.
See the 3.3kOhm resistor from output to - supply. It makes the LT1648 biased into class A. I found this better then wihout.
The next step whould be a balnced in - balanced out RIAA and the new Burr-Brown BI-BO chips are perfect for that. I ordered some and will test the circuit.
I am listening for a while now to the Low Z INA and i am very happy. The only thing i did not try is the buffering of the feedback resistors to lower noise. The Low Z version will not profit much from that because the cartridge is in parallel noisewise. So noise impedance goes down from 3.55 Ohm to 2.62 Ohm. An improvement of only around 0.1nV/qHz.
I felt that the commercial linestage i am using limited the result so i designed a no compromise linestage. It has quite amasing Data. I optimised for transient response so
frequency response is down only -3dB at 25Mhz, limmited by an output Zobel.
It is a transimpedance class-A design with fast buffer in the feedback loop.
Two of my engineer friends that saw the circuit thought that i whould never get it stable but it is totally stable now. Noise and distortion are nill too and i build a special balance and volume control that works in shunt mode with a Charcroft Z foil resistor.
I went to great pains that crosstalk is very low up to high frequencies.
The calculation of the gain and control structure took the the best of two weeks.
I also made sure that common mode distortions are avoided because i decided on series feedback. I am currently back engineering what i did and then will post the circuit.
Now moving into balanced i need a Balanced-In , Unbalnced - Out converter.
There are very good integrated solutions from Burr-Brown for example but i have a lot of Opamps around and like to roll my own.
I will use the Superbal topology that has the advantage that input impedance on hot and cold are equal for both differential and common mode.
i found this idea in D.Selfs book "Self on Audio".
I am looking forward to read his new book. He promissed a lot of new discoveries.
We are living in exiting times for audio.
I felt that the commercial linestage i am using limited the result so i designed a no compromise linestage. It has quite amasing Data. I optimised for transient response so
frequency response is down only -3dB at 25Mhz, limmited by an output Zobel.
It is a transimpedance class-A design with fast buffer in the feedback loop.
Two of my engineer friends that saw the circuit thought that i whould never get it stable but it is totally stable now. Noise and distortion are nill too and i build a special balance and volume control that works in shunt mode with a Charcroft Z foil resistor.
I went to great pains that crosstalk is very low up to high frequencies.
The calculation of the gain and control structure took the the best of two weeks.
I also made sure that common mode distortions are avoided because i decided on series feedback. I am currently back engineering what i did and then will post the circuit.
Now moving into balanced i need a Balanced-In , Unbalnced - Out converter.
There are very good integrated solutions from Burr-Brown for example but i have a lot of Opamps around and like to roll my own.
I will use the Superbal topology that has the advantage that input impedance on hot and cold are equal for both differential and common mode.
i found this idea in D.Selfs book "Self on Audio".
I am looking forward to read his new book. He promissed a lot of new discoveries.
We are living in exiting times for audio.
Today a friend came by and we listened to the Low Z INA. He thought that the sound was very good but somehow a bit to intense for him. Just for fun a substituted the mix of NE5534 and OP27 with 4 OPA134. They work in this circuit without stability problems.
My friend found the sound warmer and more spacious. He said that the sound got "0.5m wider". Strange isn´t it that he could give that exact value. I myself do not think that OPs
have a "sound". It matters very much how you use them so just swapping OPs does compare aplles to oranges. Later in the evening i listened alone and found the sound less precise and tight. For my taste the NE5534 works wonderfull in transimpedance mode.I think it is also known to be able to drive a tough load whereas the OP134 is less robust. See the catalog data and Samuel Groners survay. It´s still a goody, even today.
The OPA134 may have had some trouble with current delivery and in other test i made
it gave a very attractive but somehow distorted treble. D.Self is not a fan of that OP eather beacuse it has higher distortion as the NE5534 and cost quite a bit more.
Anyway, my INA will leave the house soon and play in a system of another hifi buddy.
I will build him a compleet preamp with remote control and line buffer stage.
Now is the time to build an INA for myself. I will build a fully balanced version that has both High Z and Low Z input. See my diagrams with my component choices. As resistors i will use the good old Dale RS55 0.1% antimagnetic. The PSU caps are from Frolyt, an innovative east german maker. For RIAA i use LCR styroflex and Röderstein MKP1837. They are compact and come in 1% tolerance. I tried the Röderstein as coupling caps in front of my poweramp and can not hear any loss of detail at all.
Maybe i get old.
My friend found the sound warmer and more spacious. He said that the sound got "0.5m wider". Strange isn´t it that he could give that exact value. I myself do not think that OPs
have a "sound". It matters very much how you use them so just swapping OPs does compare aplles to oranges. Later in the evening i listened alone and found the sound less precise and tight. For my taste the NE5534 works wonderfull in transimpedance mode.I think it is also known to be able to drive a tough load whereas the OP134 is less robust. See the catalog data and Samuel Groners survay. It´s still a goody, even today.
The OPA134 may have had some trouble with current delivery and in other test i made
it gave a very attractive but somehow distorted treble. D.Self is not a fan of that OP eather beacuse it has higher distortion as the NE5534 and cost quite a bit more.
Anyway, my INA will leave the house soon and play in a system of another hifi buddy.
I will build him a compleet preamp with remote control and line buffer stage.
Now is the time to build an INA for myself. I will build a fully balanced version that has both High Z and Low Z input. See my diagrams with my component choices. As resistors i will use the good old Dale RS55 0.1% antimagnetic. The PSU caps are from Frolyt, an innovative east german maker. For RIAA i use LCR styroflex and Röderstein MKP1837. They are compact and come in 1% tolerance. I tried the Röderstein as coupling caps in front of my poweramp and can not hear any loss of detail at all.
Maybe i get old.
What is the rational behind this? The common mode and differential source impedance of a cartridge (connected in true differential mode) are quite different. I would think you want to maximize the ratio of differential to common mode gain at the output (in the ideal case no common mode gain). You might want the commom mode impedances equal for cancellation, but I see no reason to make them equal to the differential impedance.
Last edited: