arthur said:
This is a picture of one "amp-module"
This is used BOTH in RIAA and LINE.
Normally one adjusts each stage, so that it does an optimal job.
The needs of an RIAA is Quite different
from a LINE-amp.
For example OP-amps used in HIFI Phono & MIC-input amps
must have special qualities.
WHILE LINE amps will need other Qualities, such as LINE-driver OUTPUT,
for eventual multiple low loads.
This Stage is a SIMPLE diff JFET input amp - OP amp style.
We can see what transistors are used.
Output is not PUSH-PULL.
But a single Emitter-follower Output transistor.
This stage is exactly copied into 4. Providing Stereo RIAA/Line amps.
Voltage supply is +-35 VOLT. This is probably Max, for the JFETs used.
/halo
Attachments
arthur said:
As far as I know this is pretty much the same as what I recall was commercially called "La Solstice" or something similar which I built in the mid 80's. It was originally published in Europe in L'Audiophile in France AFAIK. The exception is that that one did not have the low boost. My particular implementation was published in the Swedish magazine Audio & Video (long gone since) and the components were sold by a company called Tesserakt Media Förlag (previously something else and long gone since) in Stockholm (Previously Nässjö, the owner, Anders Edenholm, now occasionally writes for the Swedish magazine Hifi & Musik). They used (as I have mentioned in another thread) tantalum resistors everywhere and the famous Shizuki capacitor as interstage de-coupling. The RIAA caps were copper foil/polystyrene (I think polystyrene). The difficult part to build this amp is the selecting and matching of the transistors. They need to be carefully selected and matched according to the circuit requirement. The trimmers are there for offset adjust and which is somewhat tricky as both trimmers affect eachother and need to be trimmed together while the amp heats up which takes about half an hour. The trimmers were Cosmos low thermal drift and the volume pot was Cosmos (or Copal maybe) as well I think. The PSU was quite chunky for a pre-amp and existed in many version (at least 3 I recall) and they all used (in Sweden) a Schaffner line filter (model 632 in the 80's maybe?) a dual C or R core transformer (much like the trafos sold by Selctronic in France http://www.selectronic.fr/), some special diodes and some very nice caps and a choke or resistors depending on version. One version used a LT317AH/LT337AH pre-regulator (not the ordinary LM317/337 but the AH (AFAIK) which was much better version and now discontinued). Switches were normally ELMA and internal wiring was done with Isoda multi-metal wire.
As far as sound goes I am convinced that this is the best pre-amp I have ever heard or built so far. I used all the recommended components and I used it with different amps (among others a pair of KlangFilm 18 W mono-blocks from an old theater in the south of Sweden. I refurbished them and they played extremely nice, later sold to a friend of Josef Svalander). I also took the pre-amp to the Hifi-Klubben shop (a chain of shops in Sweden once very good with reasonable prices) and compared it to the top-of-the-line Denon and Nad available at that time. The guy in the shop (not the owner Ivar at that time but his young helper) and I both got the same impression (which is unexpected since he was selling ready-made stuff) and that was that the difference was huge and that is was the brand stuff that was the looser. IMHO the sound from this pre-amp was SO much more relaxed and easy going than the comparably sterile and industrial, cold and mechanical (?) sounding Denons. It was not just a small difference but VERY big.
I say, if you can find details on the matching/selction criteria for the transistors then build it. If not then try, it is just a handful of components of which none are very expensive. Start with the amp and if you get that working build a very nive PSU for it, it is needed to get the best out of it. All the transistors in each diff pair should be glued together and preferably fitted with a small piece of copper. The original used a small piece of copper tube I seem to recall and this was squeezed around the pair.
I later sold it to a guy from Norway on an ad in Hifi & Musik. He called me early in the morning and asked if it was still available, I said yes and he replied that he'd be at my place in 10 hours. He arrived in a rusty old Porsche and listened to the pre-amp for a few second paid the cash and left for home. He came from Bergen in Norway which was just about 1000 kilometers away. He did 2000 kilometers in one day to buy it so he probably also thought it was good amplifier...
Sorry for the irrelevant information.
Re: Re: kaneda preamp
OK, we have someting in common. I wrote for this magazine too.
http://home5.swipnet.se/~w-50719/hifi/qro/startsida_qro.html
Norwegians are cool! 2000 km (1250 miles) in one day!
UrSv said:
OK, we have someting in common. I wrote for this magazine too.
http://home5.swipnet.se/~w-50719/hifi/qro/startsida_qro.html
Norwegians are cool! 2000 km (1250 miles) in one day!
Attachments
Re: Re: Re: kaneda preamp
Yes I know ;-) I HAVE visited your homepage...
Yes, definitely. Norwegians are cool. That in my book would be not very unlikely for a Norwegian but very unlikely for a Swede. I once had some friends that went on a bus trip to the alps for skiing with a Norwegian that brought nothing but a toothbrush and a VISA card.
Audio & Video had loads of nice projects.
peranders said:
Yes I know ;-) I HAVE visited your homepage...
Yes, definitely. Norwegians are cool. That in my book would be not very unlikely for a Norwegian but very unlikely for a Swede. I once had some friends that went on a bus trip to the alps for skiing with a Norwegian that brought nothing but a toothbrush and a VISA card.
Audio & Video had loads of nice projects.
Yes, norwegians are good people
And very nice friends, too
Not only Sweden & Denmark (Vifa, Peerless, Scanspeak)
Danish Sound Technology - DST
but Norway has a lot of good Audio-links
Andiha audio - A Norwegian GREAT audio amp site
------------------------------
This Kaneda Amp is interestingly as an Discrete Amplifier.
I imagine it will benefit from good regulated/filtered Power Supply
as it does not use ANY active current source
Only uses Resistors to set the currents and workingpoints.
Simple - but good!
halo - has almost got the material at home
to build his own Kaneda-clone
And very nice friends, too
Not only Sweden & Denmark (Vifa, Peerless, Scanspeak)
Danish Sound Technology - DST
but Norway has a lot of good Audio-links
Andiha audio - A Norwegian GREAT audio amp site
------------------------------
This Kaneda Amp is interestingly as an Discrete Amplifier.
I imagine it will benefit from good regulated/filtered Power Supply
as it does not use ANY active current source
Only uses Resistors to set the currents and workingpoints.
Simple - but good!
halo - has almost got the material at home
to build his own Kaneda-clone
Hmmm..
Built this one a couple of years back, just for the kicks....
The RIAA section used the dual 2SK30, and the line section used singles, as this was the original idea. The different transistor types were claimed to have sonic differences optimized to the job.
The RIAA section balanced out perfectly, but even after several sessions of matching and rematching again, I never got the line section into DC balance....
It is still sitting on that shelf.........
BTW--Norwegians ARE cool, and the further north, the cooler they get.....
Built this one a couple of years back, just for the kicks....
The RIAA section used the dual 2SK30, and the line section used singles, as this was the original idea. The different transistor types were claimed to have sonic differences optimized to the job.
The RIAA section balanced out perfectly, but even after several sessions of matching and rematching again, I never got the line section into DC balance....
It is still sitting on that shelf.........
BTW--Norwegians ARE cool, and the further north, the cooler they get.....
I'm a little bit allergic of designs which are impossible to build without selecting parts. Sometimes it's more or less impossible to get it right. I remember when I made my sub which had JFET input stage. It took me 100-150 fets to find two descent pairs! Not very good as a DIY project unless you want to buy lots of transistors and/or pay someone to match for for.
Cult Status
Hi ,
the Kaneda has incredible high cult status. Word is that it only sounds good with those oldfashioned hard to get FET's like 2SK30 etc.
I do prefer 25912 or 2SK389 dual monolitic.
And a simple emittor follower as ouputstage....?
The double diamond buffer LH0003 style is more to my liking.
Folded cascode sounds better than this circuit and....
The situation is similar to tubes in a way. Only those hard to get NOS extremely expensive ones will bring you to audio Nirvana. 
Hi ,
the Kaneda has incredible high cult status. Word is that it only sounds good with those oldfashioned hard to get FET's like 2SK30 etc.
I do prefer 25912 or 2SK389 dual monolitic.
And a simple emittor follower as ouputstage....?
The double diamond buffer LH0003 style is more to my liking.
Folded cascode sounds better than this circuit and....
The situation is similar to tubes in a way. Only those hard to get NOS extremely expensive ones will bring you to audio Nirvana.
Kaneda was and is a prolific writer, and has published many, many designs in the pages of MJ. The one posted here is a very early design - he subsequently shifted to constant-current sinks feeding the input differential, Widlar current mirrors summing the second-stage differential, push-pull outputs, and somewhat later, constant-current output (only for preamps and phono stages, not for power amps). He was a stickler for components, and precisely specified what semiconductors, trim pots, resistors, capacitors, electrolytics and so on should be used. Kaneda also put much effort into power supplies, and experimented with battery operation, AC mains, switching regulators, series regulators, push-pull regulators, and so on.
Today, most of his designs consist of simpler tube-SS hybrid circuits on AC power - a concession, I believe, to the lowered design and engineering abilities of Japan's current DIY community.
I have not built any of Kaneda's later hybrid designs, but based on his 80's-vintage designs, I agree with Elso, that a folded-cascode has greater potential to sound good than a Kaneda-style dual-differential. But as always, careful, intelligent engineering is the key, and a well-executed dual-differential will almost certainly measure and sound better than an indifferently produced folded-cascode.
For a second opinion, I believe that Hartmut Quaschik converted his Kaneda to folded-cascode operation a few years back, although I don't have the particulars of the circuit that Hartmut ultimately used. From memory, Hartmut claimed that the results were a mixed bag - in some ways the folded cascode was preferable, in other ways he preferred the dual-differential.
regards, jonathan carr
Today, most of his designs consist of simpler tube-SS hybrid circuits on AC power - a concession, I believe, to the lowered design and engineering abilities of Japan's current DIY community.
I have not built any of Kaneda's later hybrid designs, but based on his 80's-vintage designs, I agree with Elso, that a folded-cascode has greater potential to sound good than a Kaneda-style dual-differential. But as always, careful, intelligent engineering is the key, and a well-executed dual-differential will almost certainly measure and sound better than an indifferently produced folded-cascode.
For a second opinion, I believe that Hartmut Quaschik converted his Kaneda to folded-cascode operation a few years back, although I don't have the particulars of the circuit that Hartmut ultimately used. From memory, Hartmut claimed that the results were a mixed bag - in some ways the folded cascode was preferable, in other ways he preferred the dual-differential.
regards, jonathan carr
The circuit that I like is whatever happens to be most suited for the job at hand.
Folded-cascodes have certain topological advantages that translate into ease of stabilization, good bandwidth, and fast slew-rate. However, they have some clear disadvantages that you should keep in mind.
Discrete-device folded cascode topologies tend to top out at about 100dB open-loop gain max, while you can push a dual-differential to at least 150dB. If you want to make a single-stage active RIAA phono amp with a target closed-loop gain of 65dB, a folded-cascode is not the right choice.
Due to the biasing of the folded cascode (unless you are using a complementary scheme), the clipping behaviour is highly asymmetric. The side of the circuit with the folded cascode will clip long before the other side, meaning that utilization efficiency of the voltage rails ain't good. If you want a circuit that can swing lots of volts from within limited supply rail voltages, a folded-cascode is not a good choice.
It is far easier to minimize the voltages placed across the input devices with a dual-differential than it is with a folded cascode. BJTs are not much of a problem, but due to gate leakage currents, N-channel JFETs usually need to be used well inside their maximum Vds ratings unless you can deal with relatively high amounts of DC offset. The situation is somewhat better for P-channel JFETs, but in general, if you want any serious voltage swing at all from a folded cascode, cascoding the input devices is de rigor. Put another way, the reasons for cascoding the input devices of a folded cascode and a dual-differential are not the same.
In my experience, folded cascodes are more tolerant of power-supply quality than dual-differentials, but differences in grounding practices are more obvious with folded cascodes. And folded cascodes still retain an Achilles heel when it comes to PSRR - the summing current mirror. I believe that it is possible to minimize this PSRR issue by selecting a device with the appropriate parameters, but you are almost certain to require access to SPICE files - data sheets won't suffice.
Speaking of, does anyone know where there is a good collection of SPICE files for small and medium-signal BJTs made by Toshiba, Sanyo, Hitachi, NEC and other Japanese manufacturers?
regards, jonathan carr
Folded-cascodes have certain topological advantages that translate into ease of stabilization, good bandwidth, and fast slew-rate. However, they have some clear disadvantages that you should keep in mind.
Discrete-device folded cascode topologies tend to top out at about 100dB open-loop gain max, while you can push a dual-differential to at least 150dB. If you want to make a single-stage active RIAA phono amp with a target closed-loop gain of 65dB, a folded-cascode is not the right choice.
Due to the biasing of the folded cascode (unless you are using a complementary scheme), the clipping behaviour is highly asymmetric. The side of the circuit with the folded cascode will clip long before the other side, meaning that utilization efficiency of the voltage rails ain't good. If you want a circuit that can swing lots of volts from within limited supply rail voltages, a folded-cascode is not a good choice.
It is far easier to minimize the voltages placed across the input devices with a dual-differential than it is with a folded cascode. BJTs are not much of a problem, but due to gate leakage currents, N-channel JFETs usually need to be used well inside their maximum Vds ratings unless you can deal with relatively high amounts of DC offset. The situation is somewhat better for P-channel JFETs, but in general, if you want any serious voltage swing at all from a folded cascode, cascoding the input devices is de rigor. Put another way, the reasons for cascoding the input devices of a folded cascode and a dual-differential are not the same.
In my experience, folded cascodes are more tolerant of power-supply quality than dual-differentials, but differences in grounding practices are more obvious with folded cascodes. And folded cascodes still retain an Achilles heel when it comes to PSRR - the summing current mirror. I believe that it is possible to minimize this PSRR issue by selecting a device with the appropriate parameters, but you are almost certain to require access to SPICE files - data sheets won't suffice.
Speaking of, does anyone know where there is a good collection of SPICE files for small and medium-signal BJTs made by Toshiba, Sanyo, Hitachi, NEC and other Japanese manufacturers?
regards, jonathan carr
Re: Cult Status
Thank you Elso, you said it before me. I have always thought, even in the 80's, that Kaneda seemed to be too good to be true, not really bad but not very advanced if we talk circuit solution. Now I hear that Mr Kanega added some "candy" in his design.
Has the Kaneda amp really earned it's good reputation? Just wondering.
Elso Kwak said:
Thank you Elso, you said it before me. I have always thought, even in the 80's, that Kaneda seemed to be too good to be true, not really bad but not very advanced if we talk circuit solution. Now I hear that Mr Kanega added some "candy" in his design.
Has the Kaneda amp really earned it's good reputation? Just wondering.
Jonathan,
Thanks for your insight of the differences between the folded cascode and dual differential. You raise a several good points particularly about power supply quality.
My first experience with the dual differential was the Kenwood L-07 amplifiers and the Hitachi app. notes for front end design for use with their mosfets. These designs always seemed to be on the verge of instability and required a heoric amount of compensation. Maybe I took the easy way out and should have persued them further.
I did come to the couclusion that the second differential in the dual differential did not work as well with a current source and a current mirror worked better, probably due to current starvation of the next stage ( especially in the case of power amps.)
Any further thoughts?
Jam
Thanks for your insight of the differences between the folded cascode and dual differential. You raise a several good points particularly about power supply quality.
My first experience with the dual differential was the Kenwood L-07 amplifiers and the Hitachi app. notes for front end design for use with their mosfets. These designs always seemed to be on the verge of instability and required a heoric amount of compensation. Maybe I took the easy way out and should have persued them further.
I did come to the couclusion that the second differential in the dual differential did not work as well with a current source and a current mirror worked better, probably due to current starvation of the next stage ( especially in the case of power amps.)
Any further thoughts?
Jam
Jam:
>My first experience with the dual differential was the Kenwood L-07 amplifiers.<
The amps that allowed you to extend the NFB loop out to the speakers? Toshiba had amps with similar technology around the same time (last part of the 1970s, early 1980s), under the Aurex brand. Good idea if you were making a self-powered subwoofer, but not so hot for mid-range and treble duties. In general, this type of scheme requires heavier phase compensation to stabilize the amplifier than would otherwise be required. I don't think the Kenwoods are good or typical examples of dual differential circuits.
>Hitachi app. notes for front end design for use with their mosfets<
Which are pretty mundane circuits by today's standards.
>Maybe I took the easy way out and should have persued them further.<
It probably would have been worthwhile to pursue dual-differential designs further - but as for pursuing designs like the Hitachi application notes, or the Kenwood "feedback amok" monsters, it likely would have been a waste of your time.
>The second differential in the dual differential did not work as well with a current source and a current mirror worked better, probably due to current starvation of the next stage (especially in the case of power amps.)<
Is it possible to be more specific? Do you mean that your second differentials worked better with a current-mirror feeding the emittors instead of a normal current-source? Or are you referring to the signal-summing current-mirror after the second differential?
regards, jonathan carr
>My first experience with the dual differential was the Kenwood L-07 amplifiers.<
The amps that allowed you to extend the NFB loop out to the speakers? Toshiba had amps with similar technology around the same time (last part of the 1970s, early 1980s), under the Aurex brand. Good idea if you were making a self-powered subwoofer, but not so hot for mid-range and treble duties. In general, this type of scheme requires heavier phase compensation to stabilize the amplifier than would otherwise be required. I don't think the Kenwoods are good or typical examples of dual differential circuits.
>Hitachi app. notes for front end design for use with their mosfets<
Which are pretty mundane circuits by today's standards.
>Maybe I took the easy way out and should have persued them further.<
It probably would have been worthwhile to pursue dual-differential designs further - but as for pursuing designs like the Hitachi application notes, or the Kenwood "feedback amok" monsters, it likely would have been a waste of your time.
>The second differential in the dual differential did not work as well with a current source and a current mirror worked better, probably due to current starvation of the next stage (especially in the case of power amps.)<
Is it possible to be more specific? Do you mean that your second differentials worked better with a current-mirror feeding the emittors instead of a normal current-source? Or are you referring to the signal-summing current-mirror after the second differential?
regards, jonathan carr
Jam:
"Current starvation" could mean different things, depending on whether you are using a MOSFET or BJT output stage, and in case of the latter, the hfe of the devices, and whether you are using some sort of Darlington or compound transistor arrangement. But as long as you aren't trying to drive a low hfe BJT directly from the collectors of the second differential, and as long as you push enough current through the first and second differentials (you may have to tolerate some extra noise), you should not have any major problems.
I usually use current sources to feed both differentials. To some extent, the amount of current from a current source will change according to the voltage across it. Also, the output impedance of a current source isn't all that high. For these reasons, I cascode any current source intended for an application that I consider to be critical. The cascode bases are fixed, and ensure that the current from the current sources remains constant under all conditions.
I agree that a current mirror works well on the collectors of the second differential. I prefer a current mirror to resistors in this location, and I prefer a Wilson to a Widlar. But even a Wilson has its problems (manifested as poor PSRR). Because the underlying cause is the Early voltage of the devices, the problem remains the same whether the Wilson mirror is used with a dual differential or a folded-cascode. And this is why I was asking if anyone knew of a comprehensive collection of SPICE files.
regards, jonathan carr
"Current starvation" could mean different things, depending on whether you are using a MOSFET or BJT output stage, and in case of the latter, the hfe of the devices, and whether you are using some sort of Darlington or compound transistor arrangement. But as long as you aren't trying to drive a low hfe BJT directly from the collectors of the second differential, and as long as you push enough current through the first and second differentials (you may have to tolerate some extra noise), you should not have any major problems.
I usually use current sources to feed both differentials. To some extent, the amount of current from a current source will change according to the voltage across it. Also, the output impedance of a current source isn't all that high. For these reasons, I cascode any current source intended for an application that I consider to be critical. The cascode bases are fixed, and ensure that the current from the current sources remains constant under all conditions.
I agree that a current mirror works well on the collectors of the second differential. I prefer a current mirror to resistors in this location, and I prefer a Wilson to a Widlar. But even a Wilson has its problems (manifested as poor PSRR). Because the underlying cause is the Early voltage of the devices, the problem remains the same whether the Wilson mirror is used with a dual differential or a folded-cascode. And this is why I was asking if anyone knew of a comprehensive collection of SPICE files.
regards, jonathan carr
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.