I'm able to answer my questions:
1) Elna caps, bypassed with good films.
2) SemiSouth shut down last year, so unfortunately PASS-SIT-1's (or an equivalent) will most likely never see the light of day.
High-tech manufacturer SemiSouth Laboratories closes after decade-plus run - Mississippi Business Journal
It is interesting to note that "Power Integrations" is/was parent company to both SemiSouth and Lovoltech. It also seems that PI no longer offers the power JFETs developed by Lovoltech and SemiSouth.
Here's a photo of one just before it went off to CES.
I could really cry, that SONY, which never supported their classic V-FET products after 1990 and totally forgot them afterwards, now let Papa Pass
source each and every remaining spare part (V-FET) still in existence in the world to build a "new" demonstration amplifier, which hopefully explodes some day by driving a Wilson Audio Sasha into clipping, while playing Cassandra Wilson.SONY/Pass, you just burnt spare parts for a hundred Sony TA-8650 ??
Did you ever hear the magic sound of the power section of a Sony TA-8650 ? No, obviously not ! Otherwise you would not do that !
From my standpoint, the ultimate commemorative V-FET amp is a Sony TA-8650. This was the first amplifier it started with in 1974. And this was done at a time, when Mr. Pass still thought about the Japanese in general as "Copycats".
If SONY/Pass needs some classic V-FET amp for a comparison to the V-FET Monster, I can deliver. I own a few TA-8650, TA-N7B and any other classic V-FET(SIT) amplifier ever commercially produced.
What about a shootout at the next Highend fair in Germany at the booth of Pass Lab's importer or at SONY's booth ?
I can bring any classic V-FET amplifier with me and then we can see, what the ultimate V-FET/SIT amp is !
What do you think about that Mr. Pass ?
Now I need a big, big bottle of Wodka
.
don't be silly
same circuit , but in proper physical layout , not compromised with need to put entire integrated in one box , with decent PSU - will blow (your , or anyone else's) regular 8650 .......
let alone big honking welder sized PSU , connected to ultra-simple low nfb circuit , carefully tuned to exact THD spectra ..... which Pass Sony SIT amp is
in 1974 , I was first school grade boy ....... but ,since, had enough time to stop being fan boy ........
same circuit , but in proper physical layout , not compromised with need to put entire integrated in one box , with decent PSU - will blow (your , or anyone else's) regular 8650 .......
let alone big honking welder sized PSU , connected to ultra-simple low nfb circuit , carefully tuned to exact THD spectra ..... which Pass Sony SIT amp is
in 1974 , I was first school grade boy ....... but ,since, had enough time to stop being fan boy ........
You should not believe everything, what your Master tells you ! So I understand you correctly that you stopped being a fan boy. repeating, what other's marketing blah tells you ?
Some objective arguments:
A bank of >=30 double V-FETs per channel is a heavy load to be driven at audio frequencies and far away to be properly loaded at any impedance higher than 0.5 Ohms.
On the other hand side the Sony V-FET amps had probably too few transistors per channel, but at that time 8Ohms was still a low impedance, so this was OK. To drive these rather few power transistors with beefy drivers was a good idea anyhow.
The result was an extremely stable soundstage and a "lightfooted" presentation the Krells and Thresholds a decade later could still only dream of.
The key to V-FET sound is not the V-FET characteristic itself, it is the possibilities in amp topology it gives you. SONY used these possibilities.
I have an internal SONY technical report at hand, which shows THD spectra of a TAN-8550 with and without overall feedback applied. The open loop THD was less than 0.1%, far below the claimed 2% of the Pass design using the switch V-FETs. You can be sure, that three pairs 2SK60/2SJ18 pair at rank 58 was much better matched (at least at that time), than you could do with the last remaining stock of 2SK82/2SJ28 (which were made for switching applications !)
Even without knowing the precise Pass circuit, I can nevertheless argue, that it is certainly like with all simplistic Pass designs, that the huge non linear Miller capacitance of the output devices is a dominating part of the (local) feedback path or is at least not driven with enough authority. This blurrs up sound stage for higher frequencies and kills all virtues of the V-FETs.
The "proper layout argument" has to be relativated, since even a TA-8650 power stage runs up to several Watts in pure Class A, so influences of magnet field distortions come up only at higher powers and their characteristic is much smoother (This is one advantage of the V-FET architectures). PP-Class B (running at 0.5A idle) designs with complementary V-FETs are much more forgiving regarding layout than any other Class B design.
Yes, the pre amp section is heavily compromized in the TA-8650, but it is mechanically totally and properly separated from the power amp section. The amp even runs with physically removed pre amp section. Regarding the sound, I talked about the power amplifier with the pre section bypassed.
If a welder like PSU and dozens of power transistors per channel and welder like layout were really a quality mark for good sound, then a Krell MRA would beat any other amplifier in the world, which it doesn't, because it's a better welder than a audio power amp.
A well designed PSU has a shielding around the transformer (See TA-8650) and uses a softer core.
And really:
How could be a 2014 welder like looking red painted tank be a hommage to a Ferrari Berlinetta Boxer from 1974 ? Because it is red painted ?
It might be, that the tank makes a louder "boom", but the Ferrari is a "grande machina" with a sound to die for ... NP should know this better than I do ...
Try it !
Some objective arguments:
A bank of >=30 double V-FETs per channel is a heavy load to be driven at audio frequencies and far away to be properly loaded at any impedance higher than 0.5 Ohms.
On the other hand side the Sony V-FET amps had probably too few transistors per channel, but at that time 8Ohms was still a low impedance, so this was OK. To drive these rather few power transistors with beefy drivers was a good idea anyhow.
The result was an extremely stable soundstage and a "lightfooted" presentation the Krells and Thresholds a decade later could still only dream of.
The key to V-FET sound is not the V-FET characteristic itself, it is the possibilities in amp topology it gives you. SONY used these possibilities.
I have an internal SONY technical report at hand, which shows THD spectra of a TAN-8550 with and without overall feedback applied. The open loop THD was less than 0.1%, far below the claimed 2% of the Pass design using the switch V-FETs. You can be sure, that three pairs 2SK60/2SJ18 pair at rank 58 was much better matched (at least at that time), than you could do with the last remaining stock of 2SK82/2SJ28 (which were made for switching applications !)
Even without knowing the precise Pass circuit, I can nevertheless argue, that it is certainly like with all simplistic Pass designs, that the huge non linear Miller capacitance of the output devices is a dominating part of the (local) feedback path or is at least not driven with enough authority. This blurrs up sound stage for higher frequencies and kills all virtues of the V-FETs.
The "proper layout argument" has to be relativated, since even a TA-8650 power stage runs up to several Watts in pure Class A, so influences of magnet field distortions come up only at higher powers and their characteristic is much smoother (This is one advantage of the V-FET architectures). PP-Class B (running at 0.5A idle) designs with complementary V-FETs are much more forgiving regarding layout than any other Class B design.
Yes, the pre amp section is heavily compromized in the TA-8650, but it is mechanically totally and properly separated from the power amp section. The amp even runs with physically removed pre amp section. Regarding the sound, I talked about the power amplifier with the pre section bypassed.
If a welder like PSU and dozens of power transistors per channel and welder like layout were really a quality mark for good sound, then a Krell MRA would beat any other amplifier in the world, which it doesn't, because it's a better welder than a audio power amp.
A well designed PSU has a shielding around the transformer (See TA-8650) and uses a softer core.
And really:
How could be a 2014 welder like looking red painted tank be a hommage to a Ferrari Berlinetta Boxer from 1974 ? Because it is red painted ?
It might be, that the tank makes a louder "boom", but the Ferrari is a "grande machina" with a sound to die for ... NP should know this better than I do ...
Try it !
Last edited:
I could really cry, that SONY, which never supported their classic V-FET products after 1990 and totally forgot them afterwards, now let Papa Pass
...From my standpoint, the ultimate commemorative V-FET amp is a Sony TA-8650. This was the first amplifier it started with in 1974. And this was done at a time, when Mr. Pass still thought about the Japanese in general as "Copycats".
To clarify a couple points, I did leave a thousand pairs or so on the table at
circuitdiy.com out of consideration for the rest of the planet.
Also, as I did not have a public presence in 1974, you would be speculating
about what I thought about the Japanese in general.
To the contrary, while I have encountered some imitation and one egregious
rip-off, my opinion of the Japanese audio industry has been one of respect
and admiration, and the VFET amps I built for Sony were intended to reflect
that.
Hello Lord,
In high school I wrote many book reports without ever reading the assigned book, thinking it more clever and a better use of time to read cliffs notes instead. Looking back now I realize this was technically cheating, and that while I may have made the grade I actually robbed myself of many opportunities to enjoy fine pieces of literature in the manner their authors intended. Guess we all make mistakes
Delecoy
In high school I wrote many book reports without ever reading the assigned book, thinking it more clever and a better use of time to read cliffs notes instead. Looking back now I realize this was technically cheating, and that while I may have made the grade I actually robbed myself of many opportunities to enjoy fine pieces of literature in the manner their authors intended. Guess we all make mistakes
Delecoy
Another correction: The big version is balanced, and not all of the devices
are VFETs - there are also supply regulators.
So instead of 30 Vfets, each output polarity has 6 pairs, or 12 devices, and
you will be aware that they only have to slew half as fast as well, making the
performance the equivalent of an unbalanced 3 pair in terms of the effective
capacitance.
had my hands in few TA
brought them from not just dead but butchered , to very alive
so , spare me , and keep rant like that for some FiFi Forum
I could really cry, that SONY, which never supported their classic V-FET products after 1990 and totally forgot them afterwards, now let Papa Pass
Jon, you are free to buy some (or the whole remaining batch) yourself if you wish to save them for a more noble cause ....
2SK82 2SJ28 P-CH and N-CH VFET From Sony
Lots of errors in Jon Lord's posts.
Nelson uses 2SJ28/2SK82, Sony used 2SJ18/2SK60 which are half current, lower voltage and higher capacitance per drain current amp. 24 pairs of J28/L82 would at most fix 6 of Sony's 8650/8550 amps assuming 2x J28/K28 for 3x J18/K60 and careful modification of the driver stages, because there are differences between the original and J28/K82 pairs to be accounted for (most notably mu compensation since the Sony amps don't use regulated power supplies).
Speaking of class B in terms of VFETs running at ~0.5A idle current is at the very least inaccurate. Most of the open loop low distortion achieved by VFET output stages is due to their 'soft' cut-off (read: not really possible to completely cut off without getting reverse breakdown between G and S!), resulting in a very wide and soft crossover region (if it can even be called that). As Nelson has already explained, the near-square law characteristic of the VFETs results in something close to hyperbolic operation (I deliberately avoid saying it's class A or for that matter B), where neither side of a complementary output stage ever turns completely off even at a considerably higher current than twice the idle current. Here the VFETs have a considerable advantage, any distortion due to crossover remains confined to low order harmonics. Also, with some attention to th design, clipping is soft.
VFETs also have a rather unexpected property - although their capacitances are non-linear just like all semiconductors (you can view the G-S reverse biassed 'diode' as a big varicap diode!), they are in general somewhat less nonlinear than other semis. One reason is the same as the wide 'crossover area' - it is spread along a mich higher voltage variation of Vgs. For sane idle currents and relatively high Vds normal operating conditions put it well outside significantly nonlinear regions. Sony VFETs in particular are slightly differently made than other varieties, because they did not design for maximum Gm. This sacrifice, however, makes the devices so much dependent on geometry that intrinsic differences like charge mobility between P and N doped silicon stop being the major factor. As a result, the Sony VFETs are far more complementary than one would expect from a semiconductor device, and this also includes the capacitance nonlinearity. To a large extent it cancels out on a complementary output stage, which is a big plus. Hence, even a lot of pairs in parallel, while not an easy load, are a predictable load, which makes driver circuit design much easier.
Nelson uses 2SJ28/2SK82, Sony used 2SJ18/2SK60 which are half current, lower voltage and higher capacitance per drain current amp. 24 pairs of J28/L82 would at most fix 6 of Sony's 8650/8550 amps assuming 2x J28/K28 for 3x J18/K60 and careful modification of the driver stages, because there are differences between the original and J28/K82 pairs to be accounted for (most notably mu compensation since the Sony amps don't use regulated power supplies).
Speaking of class B in terms of VFETs running at ~0.5A idle current is at the very least inaccurate. Most of the open loop low distortion achieved by VFET output stages is due to their 'soft' cut-off (read: not really possible to completely cut off without getting reverse breakdown between G and S!), resulting in a very wide and soft crossover region (if it can even be called that). As Nelson has already explained, the near-square law characteristic of the VFETs results in something close to hyperbolic operation (I deliberately avoid saying it's class A or for that matter B), where neither side of a complementary output stage ever turns completely off even at a considerably higher current than twice the idle current. Here the VFETs have a considerable advantage, any distortion due to crossover remains confined to low order harmonics. Also, with some attention to th design, clipping is soft.
VFETs also have a rather unexpected property - although their capacitances are non-linear just like all semiconductors (you can view the G-S reverse biassed 'diode' as a big varicap diode!), they are in general somewhat less nonlinear than other semis. One reason is the same as the wide 'crossover area' - it is spread along a mich higher voltage variation of Vgs. For sane idle currents and relatively high Vds normal operating conditions put it well outside significantly nonlinear regions. Sony VFETs in particular are slightly differently made than other varieties, because they did not design for maximum Gm. This sacrifice, however, makes the devices so much dependent on geometry that intrinsic differences like charge mobility between P and N doped silicon stop being the major factor. As a result, the Sony VFETs are far more complementary than one would expect from a semiconductor device, and this also includes the capacitance nonlinearity. To a large extent it cancels out on a complementary output stage, which is a big plus. Hence, even a lot of pairs in parallel, while not an easy load, are a predictable load, which makes driver circuit design much easier.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.