Today i made more listening tests to loading resistors. The good news first : The 300 Ohm
TX resistor sounded much better then the 1Kohm. The thinness is gone replaced by a totally natural tonal balance. So again the message is that the value of the loading resistor is critical, maybe even more then the type provided you use a "good" one.
To describe the difference between the TX and the Dale RS65C ( sorry, i named it RS65E before but it is a C type in fact) is much harder. Both resistors sound better than all others i tried including NOS Tantal, Mil.Spec. Wirewounds, Caddock 132 and Dale RS55 unmagnetic. Tonal balance and resolution of micro detail is very similiar in both cases.
The TX sounding dead neutral and the Dale very slightly warmer. Concerning speed music sounds very fast with the TX and the Dale not far behind. I whould say that the Dale has the edje on older material like Hellen Merill in Tokyo and Katy Lied by Steely Dan. It is slighty more convincing on thoose tracks. Somehow it sounds more relaxed and magical. The focus is more on the music and less on the sound if you understand what i like to say. On modern Electropop and Funk like Electribe 101 and Nona Hedricks
the TX provides the more exciting ride. Listening to this kind of music on high volume left my breathless. No time for analysis.
The biggest differences apeared in terms of 3 dimensional soundstage presentation. The TX projected voices more forward and less high. The Dale had amasing height projection. On Hellen Merill i had the feeling that she was singing 5 meters away and 2 meters over my head. I can not say what i more accurate but i certainly enyoyed the Dale more in that regard. It gave a kind of hyperrealism that i only experience seldom in very high resolution Audio. It can not be compared to the life experience and seems to be an artifact of reproduced music, still it did not distract from enyoying the music, quite the oposite. I had that experience when i tested resistors for speakers and the good old Welvyn W22 was the most spectacular in that regard. I have shown that effect to several people that visited me and it never failed to impress. Why that is so i have no idea. Later the day i will download even more. I have a 138 Ohm TX and a 150 Ohm Dale RS65C for that. I also will listen to classical music first with the 300 Ohm resistors.
One remark : I whould rate both resistors to be in he same quality class so the Dale is a huge bargain when you have the space to mount it. Some TX could be thrown in for good effect if you like a more "Matter of Fact" sound. For comparison of size i made a photo of the tested resistors side by side.
TX resistor sounded much better then the 1Kohm. The thinness is gone replaced by a totally natural tonal balance. So again the message is that the value of the loading resistor is critical, maybe even more then the type provided you use a "good" one.
To describe the difference between the TX and the Dale RS65C ( sorry, i named it RS65E before but it is a C type in fact) is much harder. Both resistors sound better than all others i tried including NOS Tantal, Mil.Spec. Wirewounds, Caddock 132 and Dale RS55 unmagnetic. Tonal balance and resolution of micro detail is very similiar in both cases.
The TX sounding dead neutral and the Dale very slightly warmer. Concerning speed music sounds very fast with the TX and the Dale not far behind. I whould say that the Dale has the edje on older material like Hellen Merill in Tokyo and Katy Lied by Steely Dan. It is slighty more convincing on thoose tracks. Somehow it sounds more relaxed and magical. The focus is more on the music and less on the sound if you understand what i like to say. On modern Electropop and Funk like Electribe 101 and Nona Hedricks
the TX provides the more exciting ride. Listening to this kind of music on high volume left my breathless. No time for analysis.
The biggest differences apeared in terms of 3 dimensional soundstage presentation. The TX projected voices more forward and less high. The Dale had amasing height projection. On Hellen Merill i had the feeling that she was singing 5 meters away and 2 meters over my head. I can not say what i more accurate but i certainly enyoyed the Dale more in that regard. It gave a kind of hyperrealism that i only experience seldom in very high resolution Audio. It can not be compared to the life experience and seems to be an artifact of reproduced music, still it did not distract from enyoying the music, quite the oposite. I had that experience when i tested resistors for speakers and the good old Welvyn W22 was the most spectacular in that regard. I have shown that effect to several people that visited me and it never failed to impress. Why that is so i have no idea. Later the day i will download even more. I have a 138 Ohm TX and a 150 Ohm Dale RS65C for that. I also will listen to classical music first with the 300 Ohm resistors.
One remark : I whould rate both resistors to be in he same quality class so the Dale is a huge bargain when you have the space to mount it. Some TX could be thrown in for good effect if you like a more "Matter of Fact" sound. For comparison of size i made a photo of the tested resistors side by side.
Attachments
When you scroll down the PDF you see transitor types and instuctions to trimm the circuit. After the trimm distortion is minimum so is DC offset. Theoretical that circuit can swing 2V with -100dB of distortion without global feedback. In praxis i measure around -90dB and that seems to be the boundary of what is doable without NFB. When you want better measured performance, see my INA developments.
Actually the basic circuit did not change much but i worked on the biasing and trimming. The circuit as is now has superb PSU rejection and can be direct coupled to the next stage.
Circuit in post 62 has a shunt feedback stucture and works well too. The stucture i recently posted has the 75usec part as transimpedance and the midrange-bass section active in the second stage. I think it sound good because i first reduce the trebe content in the 75usec transimpedance structure so high frequency junk from ticks and pops that come from scratches and dirt do not tax the second stage. Subjectively noise from the record is spacially seperated from the music so this circuit has a subjecively clean presentation of the music. Listening to old, worn records is posible with this circuit with minimal distraction. Tick and pops also sound very "short" in duration.
Actually the basic circuit did not change much but i worked on the biasing and trimming. The circuit as is now has superb PSU rejection and can be direct coupled to the next stage.
Circuit in post 62 has a shunt feedback stucture and works well too. The stucture i recently posted has the 75usec part as transimpedance and the midrange-bass section active in the second stage. I think it sound good because i first reduce the trebe content in the 75usec transimpedance structure so high frequency junk from ticks and pops that come from scratches and dirt do not tax the second stage. Subjectively noise from the record is spacially seperated from the music so this circuit has a subjecively clean presentation of the music. Listening to old, worn records is posible with this circuit with minimal distraction. Tick and pops also sound very "short" in duration.
I've overlooked your marks at the end of the document. For T5/T8 I cannot find a BF446A JFET. Did you mean BF246A? In addition have you ever tried a lower beta transistor like the BC547A and BC557A for T6/T7? A higher beta results in a lower early voltage so y22 is also lower. As y22 is quite non-linear a high beta transistor has a higher distortion level when used as cascode transistor / common base circuit. At least in theory. I've never confirmed that in a real circuit as I ever used the lower beta types like the BC547A instead of a BC547C in this application.
Sorry, some mistakes came up. Yes, it is an RN65C. You can get it at Mousser.
The P/N is 71-RN65C1500F for the 150 Ohm for example.
yes, the Fet is a BF246A.
Yes, i tried lower beta in my Balanced INA desing. I used 2N4403, 2N4401. Anyway, i am very happy with the performance with that circuit. Distortion is around -90dB at cruising level. Other people sugested other transistors then the BC550C/ BC560C too but it is simply imposible to try everything. Do you have any sugestion ?
The P/N is 71-RN65C1500F for the 150 Ohm for example.
yes, the Fet is a BF246A.
Yes, i tried lower beta in my Balanced INA desing. I used 2N4403, 2N4401. Anyway, i am very happy with the performance with that circuit. Distortion is around -90dB at cruising level. Other people sugested other transistors then the BC550C/ BC560C too but it is simply imposible to try everything. Do you have any sugestion ?
I think BC546 - BC550 are all the same chips nowadays. The only difference is the beta between them. BC546 or similar do not perform very well on higher currents. 10 - 20mA is the max. current before they degrade in parameters. ZTX450 and ZTX550 (Zetex / Diodes) are medium power transistors and have a very linear beta over quite a large current range. BD139/140 are also not bad for cascoding.
I think the ZTX214 is the same as BC214 and ZTX384 is BC384. When I remember my old days right, the BC384/BC214 series is replaced with the BC549/BC559. So I think ZTX214/ZTX384 is roughly the same as BC560/BC550. Do you have some BC639/BC640 on the hand? These parts are very similar to the BD139/BD140.
If you want some very special you can try the 2SA1381/2SC3503 from Fairchild (very low in and output capacitances, high early voltage)
Farnell stocks the ZTX series, 2SA1381/2SC3503 are available through Mouser (ZTX450/550 as well).
If you want some very special you can try the 2SA1381/2SC3503 from Fairchild (very low in and output capacitances, high early voltage)
Farnell stocks the ZTX series, 2SA1381/2SC3503 are available through Mouser (ZTX450/550 as well).
First i build a lot of Low Z versions with bipolars. Maybe 10 different versions. I later whent to Fets because Ube is a very weak function of idlle current and it was hard (nearly imposible) to get zerro DC into the cartridge without coupling cap. The balanced version of the Low Z MPP ( one of the first circuits i posted) does not have that problem provided i select for Ube. From the High Z MPP i build maybe five. The circuit was not changed much but i worked a lot on optimising bias and trimmimg distortion and offset. Then i went to discreet INA design. Transimpedance circuits ( the ones i researched first)
have a lot of slamm and drahma so my first High Z versions did have a very nice tone but i missed some raw dynamics. Later iterations did not have that problem any more.
The INAs i made have even less distortion and have both, raw energy and lot´s of micro detail. I also made very good experiences with the sound of inductive RIAAs.
I am currently working on an inductive, fully balanced INA that hopefully is a culmination in sound of what i did before but make no mistake. Even the simple High Z MPP Capacitive RIAA has a very fine sound that is rarely heard in commercial offerings.
I find most commercial stages too sterile or the opposite, too difuse.
Anyway, the more experienced i get i find that i have to virtually build the whole system myself to get the sound i like but that was always the case with loudspaekers since i can remember. My dream when i was 16 was to combine the bass of the Klipschorn with the midrange of the Quad and the treble of a Kelly Ribbon if you know what i mean. I am very picky when it comes to sound.
have a lot of slamm and drahma so my first High Z versions did have a very nice tone but i missed some raw dynamics. Later iterations did not have that problem any more.
The INAs i made have even less distortion and have both, raw energy and lot´s of micro detail. I also made very good experiences with the sound of inductive RIAAs.
I am currently working on an inductive, fully balanced INA that hopefully is a culmination in sound of what i did before but make no mistake. Even the simple High Z MPP Capacitive RIAA has a very fine sound that is rarely heard in commercial offerings.
I find most commercial stages too sterile or the opposite, too difuse.
Anyway, the more experienced i get i find that i have to virtually build the whole system myself to get the sound i like but that was always the case with loudspaekers since i can remember. My dream when i was 16 was to combine the bass of the Klipschorn with the midrange of the Quad and the treble of a Kelly Ribbon if you know what i mean. I am very picky when it comes to sound.
I started to design a discrete OPamp for the second stage of the Capacitive High Z MPP.
I have some nice LSK389C double Fets and they motivated to do something like this.
Actually it´s my first discrete OPamp design. I am not very experienced with differential stages because i mostly design parallel symmetric when i do discrete and use good quality Integrated OPamps for diffential work.
The philosophy of this design is simplicity and a very short, straight forward signal path.
It is not without sofistication but much simpler then something like a discrete AD797, that i admire much. My design has not even an output buffer but i use medium power transistors with very high idle current. The circuit shows the concept but may still contain some errors. If you, dear reader have any idea for improvement or other sugestions this is most wellcome.
I have some nice LSK389C double Fets and they motivated to do something like this.
Actually it´s my first discrete OPamp design. I am not very experienced with differential stages because i mostly design parallel symmetric when i do discrete and use good quality Integrated OPamps for diffential work.
The philosophy of this design is simplicity and a very short, straight forward signal path.
It is not without sofistication but much simpler then something like a discrete AD797, that i admire much. My design has not even an output buffer but i use medium power transistors with very high idle current. The circuit shows the concept but may still contain some errors. If you, dear reader have any idea for improvement or other sugestions this is most wellcome.
Wow, thats really an impressive number of versions. I'm building maybe one or two versions after that I've most often other ideas.
I'm working on something similar as I/V stage for a DAC. The concept itself is nearly the same although I'm using BJTs instead of FETs. I think you don't need the curent mirror from T7/T8 as this only boosts the gain of the amp. In my design I'm using a current mirror at the output and a second folded cascode similar to T3. In addition I'm using current sources for the drain of your two JFETs. The base of this transistor based current source can be based from your first LED of your LED string.
When you are using local feedback or a servo you can also omit R9.
Your stage comes close to the discrete AD797. That also has a second cascode and a current mirror in the second stage. In the AD797 the current mirror is bootstraped.
You say that in my stage the current mirror raises gain but that is exactly what i want plus better balance in the input stage but that can be adjusted with the trimmer too.
You use current sources that also raise gain when i am right. Anyway, i will try to implement your sugestions in another design. Thank you for contributing.
You say that in my stage the current mirror raises gain but that is exactly what i want plus better balance in the input stage but that can be adjusted with the trimmer too.
You use current sources that also raise gain when i am right. Anyway, i will try to implement your sugestions in another design. Thank you for contributing.