I am just expressing my opinion that to best judge the sonics of the VFET hi eff speaker is probably better. Ultimately, you play the speakers you have and prefer. Where did I say he was wrong? It's a free country, maybe, XRX, and everyone else, can do whatever they want!
Last edited:
Listening Impressions: Sony VFET pt 1
My recently completed VFET amp (#055) has been occupying the pedestal in my system for a week already, and has been keeping me busy with early morning, mid morning, late evening and mid late afternoon listening sessions.
This amp does take a couple days to run in before it really starts to sound its best. So getting to know the amp's signature sound is a process best savored over time.
I already have a fairly decent collection of DIY FirstWatt clone amps, which include the M2x, Aleph J & F6, plus a few variations on the ACA. The Sony VFET (SE P-type) certainly exhibits a common lineage with these, including the Edcor gain stage and a dynamic CCS loaded output stage. It also has a character all its own. As other Nelson Pass designs have consistently delivered, the Sony VFET has a smooth and balanced tonal character, one that is faithful to reproducing the detailed harmonics of voice, acoustic and electronically amplified instruments. It also preserves what some Flat Earthers call pace, rhythm and timing, which invites a deeper emotional involvement with the musical reproduction. The special VFET transistor in the output stage, plus that little bit of iron in the front end make this amp particularly easy on the ears. Since there are also front end circuits available which forgo iron as the gain stage, and provide such with silicon, it will eventually be interesting to hear how the all transistor front ends alter the presentation.
As the amp continued to run in, the inner dynamic details of recordings became more apparent. Musical instrument attack, decay and hall (or studio) reverb became more lifelike. Extended playing also improved the sense of ease with which louder and more dynamic passages were presented. This is where the amp can be said to deliver higher performance than is common for its weight class.
Within the soundstage as presented, instruments have good placement, including vertical as well as front-to-back. One of the more distinct features of this amp is its forward presentation. The musical performance feels as if it has been brought into the listening room. Of the other amps, only my Aleph J gives a similar presentation. If there is a weakness to the Sony VFET, it is that the soundstage is confined mainly between the centerlines of my speakers. The old Vandersteen 2C speakers in my system are demanding in this respect, and have required extra effort to deliver full channel separation and the more expansive soundstage that comes with that. It is likely that other speakers may be less demanding, particularly in the lower octaves.
With that said, the rest of my system has undergone quite a few upgrades over the years, including extra attention to the power supplies for the source components. The picture below shows the mix of analog and digital source equipment that was used to evaluate the VFET amp. Note that the large chassis in the bottom two shelves of the rack are separate outboard power supplies – one for my preamp and the other for my CD player. Each of these is full dual-mono, and capable of delivering enough current to make small power amps pretty happy. In this setting, the SMPS brick powering both channels of the Sony VFET is the suspected weak link. Not the amplifier circuit.
My experience with the various ACA builds suggested that there was a fairly simple thing to try. So I added a 440 uF, 50V aluminum organic polymer (Kemet p/n A759MX447M1HAAE028) bypass capacitor to the power inputs of each output stage card. Not the big motor run cap that I threatened to use in an earlier post, but a better value that posed no real threat of oscillation. This simple addition had immediate benefits to the musical presentation. Instruments had a more 3D sense of placement both front-to-back as well as vertical, and hall / studio ambiance now extended beyond the speakers, essentially making the big Vandersteens disappear. That is no mean feat. The other benefit was to musical content in the lower octaves, where the foundation of a performance rests. Bass lines were now more noticeable and articulate.
So I consider that little experiment an indication of progress in the right direction. There was no audible hiccup from the SMPS, nor did I expect one, given that the direct load was only changed from 1000 uF to 1880 uF. Still, best to take small steps for now. The next thing was to compare with one of my larger amps, the F6. To be described in my next post..
My recently completed VFET amp (#055) has been occupying the pedestal in my system for a week already, and has been keeping me busy with early morning, mid morning, late evening and mid late afternoon listening sessions.
This amp does take a couple days to run in before it really starts to sound its best. So getting to know the amp's signature sound is a process best savored over time.I already have a fairly decent collection of DIY FirstWatt clone amps, which include the M2x, Aleph J & F6, plus a few variations on the ACA. The Sony VFET (SE P-type) certainly exhibits a common lineage with these, including the Edcor gain stage and a dynamic CCS loaded output stage. It also has a character all its own. As other Nelson Pass designs have consistently delivered, the Sony VFET has a smooth and balanced tonal character, one that is faithful to reproducing the detailed harmonics of voice, acoustic and electronically amplified instruments. It also preserves what some Flat Earthers call pace, rhythm and timing, which invites a deeper emotional involvement with the musical reproduction. The special VFET transistor in the output stage, plus that little bit of iron in the front end make this amp particularly easy on the ears. Since there are also front end circuits available which forgo iron as the gain stage, and provide such with silicon, it will eventually be interesting to hear how the all transistor front ends alter the presentation.
As the amp continued to run in, the inner dynamic details of recordings became more apparent. Musical instrument attack, decay and hall (or studio) reverb became more lifelike. Extended playing also improved the sense of ease with which louder and more dynamic passages were presented. This is where the amp can be said to deliver higher performance than is common for its weight class.
Within the soundstage as presented, instruments have good placement, including vertical as well as front-to-back. One of the more distinct features of this amp is its forward presentation. The musical performance feels as if it has been brought into the listening room. Of the other amps, only my Aleph J gives a similar presentation. If there is a weakness to the Sony VFET, it is that the soundstage is confined mainly between the centerlines of my speakers. The old Vandersteen 2C speakers in my system are demanding in this respect, and have required extra effort to deliver full channel separation and the more expansive soundstage that comes with that. It is likely that other speakers may be less demanding, particularly in the lower octaves.
With that said, the rest of my system has undergone quite a few upgrades over the years, including extra attention to the power supplies for the source components. The picture below shows the mix of analog and digital source equipment that was used to evaluate the VFET amp. Note that the large chassis in the bottom two shelves of the rack are separate outboard power supplies – one for my preamp and the other for my CD player. Each of these is full dual-mono, and capable of delivering enough current to make small power amps pretty happy. In this setting, the SMPS brick powering both channels of the Sony VFET is the suspected weak link. Not the amplifier circuit.
My experience with the various ACA builds suggested that there was a fairly simple thing to try. So I added a 440 uF, 50V aluminum organic polymer (Kemet p/n A759MX447M1HAAE028) bypass capacitor to the power inputs of each output stage card. Not the big motor run cap that I threatened to use in an earlier post, but a better value that posed no real threat of oscillation. This simple addition had immediate benefits to the musical presentation. Instruments had a more 3D sense of placement both front-to-back as well as vertical, and hall / studio ambiance now extended beyond the speakers, essentially making the big Vandersteens disappear. That is no mean feat. The other benefit was to musical content in the lower octaves, where the foundation of a performance rests. Bass lines were now more noticeable and articulate.
So I consider that little experiment an indication of progress in the right direction. There was no audible hiccup from the SMPS, nor did I expect one, given that the direct load was only changed from 1000 uF to 1880 uF. Still, best to take small steps for now. The next thing was to compare with one of my larger amps, the F6. To be described in my next post..
Attachments
I believe the XO has to be near 650 Hz to have 5 octaves with one driver and five with the other. (20–40, 40-80, 80-160, 160-320, 320-640)
dave
your amp, your tweaks, your ears
most important thing Pa is giving us in all these years is not Plethora^N funny gadgets, but approach - "have an Idea - go and try, hear what you got, measure, think"
motor run could not induce oscillations, same as it is not having anything with PSU ripple, whatever Freq. is in question
is it beneficial to your ears/brain combo - you know
so, experiment till the end of time and have fun, with only one condition - do not ruin anything of Cute Papapackage (and we all know that's implicit)
I have also experienced similar improvements in soundstage when driving a stereo amp from a common PSU. Adding extra rail capacitance helps to improve stereo channel separation. This can dramatically improve stereo soundstage.
The Mean Well SMPS has an upper limit on the amount of capacitance that can be added. However, this limit can be mitigated with a soft start circuit that slowly opens up the current between the SMPS and the cap bank. Usually through a largish value NTC (circa 30ohms cold) and then after say 2 to 5 seconds a solid state relay (SSR) bypasses the NTC, at which point the caps have already charged and the SMPS is not presented with a huge in-rush that usually would cause it to go into hiccup mode.
I might try this and report back, and of course, will make the circuit design and Gerbers freely available.
.. and in comparison to my F6
The F6 was the most recent of the Nelson Pass designs that I built before starting to experiment with large hockey puck-like critters. It is, perhaps, more overbuilt* than is common amongst the usual DIY amps discussed in these forums. I chose this amp as a comparison to the Sony VFET because it was one I had hacked to present a little more H2 in its harmonic output spectrum. And it uses a little bit of iron for its gain stage, Jensen instead of Edcor, but they both do nice things with music. To some extent, my F6 is still a work in progress, but since it has reached a point of great enjoyment, I have left it alone for over a year so far.
Like the Sony VFET, my F6 exhibits inky blackness in the background until music starts playing. After that several differences make themselves apparent. The extra gain and extra power on tap do make themselves known as soon as musical recordings hit their crescendos and other more prominent passages. Not really a fair comparison, as this F6 was built to hit even harder than other 25W class A builds. The soundstage fills the entirety of the rear wall of my living room. Depending on the recording, instruments may appear centered at the speaker locations and beyond, with hall and studio ambiance extending well beyond the instruments themselves. The extra space around instruments brings them into the room, yet seemingly recessed behind the speakers. That seems to be a feature of the hacks I did to the output stage. The soundstage from the F6 has enough front-to-back extension that the rear wall of my room seems to disappear along with my speakers.
Finally, the bass content is such that the lowest octave of bass guitar, as well as synthesized bass instrumentation has a full and realistic presentation. I happen to dabble in bass guitar, so I know what one sounds like in my living room. Preserving the fundamental frequencies of bass instruments in a larger space is difficult, and rewarding when it is achieved. I have mentioned the importance of bass instrumentation being key to the foundation of a musical performance. When the F6 is opened up, the foundation is there, exactly as is called for.
For the record, my Aleph J performs every bit as well as the F6 in soundstage and bass performance. It also has an overbuilt dual-mono PSU, just a different kind. And it is an Aleph J.
Time to loop back to the Sony VFET and uncover a little more of its potential..
* My F6 used FQH44N10 Mosfets running off +/– 26V rails, biased at 1.85 Amps each. PSUs were a pair 300VA transformers driving xrk971's SLBs with extra bulk capacitance. Oh, and a Diamond Buffer front end inspired by Mark J's Austin card.
The F6 was the most recent of the Nelson Pass designs that I built before starting to experiment with large hockey puck-like critters. It is, perhaps, more overbuilt* than is common amongst the usual DIY amps discussed in these forums. I chose this amp as a comparison to the Sony VFET because it was one I had hacked to present a little more H2 in its harmonic output spectrum. And it uses a little bit of iron for its gain stage, Jensen instead of Edcor, but they both do nice things with music. To some extent, my F6 is still a work in progress, but since it has reached a point of great enjoyment, I have left it alone for over a year so far.
Like the Sony VFET, my F6 exhibits inky blackness in the background until music starts playing. After that several differences make themselves apparent. The extra gain and extra power on tap do make themselves known as soon as musical recordings hit their crescendos and other more prominent passages. Not really a fair comparison, as this F6 was built to hit even harder than other 25W class A builds. The soundstage fills the entirety of the rear wall of my living room. Depending on the recording, instruments may appear centered at the speaker locations and beyond, with hall and studio ambiance extending well beyond the instruments themselves. The extra space around instruments brings them into the room, yet seemingly recessed behind the speakers. That seems to be a feature of the hacks I did to the output stage. The soundstage from the F6 has enough front-to-back extension that the rear wall of my room seems to disappear along with my speakers.
Finally, the bass content is such that the lowest octave of bass guitar, as well as synthesized bass instrumentation has a full and realistic presentation. I happen to dabble in bass guitar, so I know what one sounds like in my living room. Preserving the fundamental frequencies of bass instruments in a larger space is difficult, and rewarding when it is achieved. I have mentioned the importance of bass instrumentation being key to the foundation of a musical performance. When the F6 is opened up, the foundation is there, exactly as is called for.
For the record, my Aleph J performs every bit as well as the F6 in soundstage and bass performance. It also has an overbuilt dual-mono PSU, just a different kind. And it is an Aleph J.
Time to loop back to the Sony VFET and uncover a little more of its potential..
* My F6 used FQH44N10 Mosfets running off +/– 26V rails, biased at 1.85 Amps each. PSUs were a pair 300VA transformers driving xrk971's SLBs with extra bulk capacitance. Oh, and a Diamond Buffer front end inspired by Mark J's Austin card.
Sony VFET
Having done some modest experimentation with the VFET amp, and also having looked at a partial goal for further achievement it was now time to try a little more.
First, however, with this special amp the overriding directive is to do no harm.
So the Kemet 440 uF caps were removed and replaced a pair of Panasonic FC series 1000 uF, 50V caps that I save for such special occasions. And that was all.
That was all that was really necessary. While it has been said that the VFET output stage really should present no dynamic load to the power supply, adding 1000 uF of local supply decoupling really does wonders. Way out of proportion to what one might expect. Or perhaps it is the relative consistency of current draw in the output stage that lets just a little extra capacitance have more leverage than expected.
There is now all the width, depth and height to the soundstage that I have enjoyed with my 25W amps. Bass extension now supports the low octave of a bass guitar, the fundamentals not just the harmonics. Really quite lovely.
Most importantly, the amp is simply delightful to listen to. In some ways, it now reminds my of my Aleph J, slightly different character, but the magic is there.
Where next? This is it for a while. If I were inclined to tweak the signal path, I might replace C2 on each of the FE cards with the Kemet organic polymer caps that I removed from the OS, and replace C2 in the OS cards with Nichicon 1500 uF, 16V aluminum organic polymer. But.. not gonna.
I suspect that Mark J's input card designs may be fun for some folks to try. It so happens that in addition to different types of circuits, they also will have the effect of isolating the FE cards very well from the OS and from each other. So there's that. I'm intrigued by Bulwark, so I might try that one. Later.
Having done some modest experimentation with the VFET amp, and also having looked at a partial goal for further achievement it was now time to try a little more.
First, however, with this special amp the overriding directive is to do no harm.
So the Kemet 440 uF caps were removed and replaced a pair of Panasonic FC series 1000 uF, 50V caps that I save for such special occasions. And that was all.
That was all that was really necessary. While it has been said that the VFET output stage really should present no dynamic load to the power supply, adding 1000 uF of local supply decoupling really does wonders. Way out of proportion to what one might expect. Or perhaps it is the relative consistency of current draw in the output stage that lets just a little extra capacitance have more leverage than expected.
There is now all the width, depth and height to the soundstage that I have enjoyed with my 25W amps. Bass extension now supports the low octave of a bass guitar, the fundamentals not just the harmonics. Really quite lovely.
Most importantly, the amp is simply delightful to listen to. In some ways, it now reminds my of my Aleph J, slightly different character, but the magic is there.
Where next? This is it for a while. If I were inclined to tweak the signal path, I might replace C2 on each of the FE cards with the Kemet organic polymer caps that I removed from the OS, and replace C2 in the OS cards with Nichicon 1500 uF, 16V aluminum organic polymer. But.. not gonna.
I suspect that Mark J's input card designs may be fun for some folks to try. It so happens that in addition to different types of circuits, they also will have the effect of isolating the FE cards very well from the OS and from each other. So there's that. I'm intrigued by Bulwark, so I might try that one. Later.
Yes, it is now me. In the last two days I matched up 1600 irfp140's for Aleph J's after 800 2SJ74's.
15+ years ago, I snapped up a small stash of the Toshiba JFETs while they were still available from reputable / authorized dealers. Then serialized and measured them all... my colleagues thought I was nuts.
Now I don't feel like such a weirdo.
Still, matching parts is a lot of work... a few hundred is plenty, but 1600 + 800!!
Now that's devotion!Some tips if you decide to do some matching of your own... build a little test jig, it goes much faster. I found it was necessary to handle parts with tweezers to avoid body heat transfer affecting the measurements. ZIF socket for quick cycle time & low stress on the part leads. I used a simple push-button activated one-shot timer chip to apply the test pulse and trigger the bench meter for consistent timing / self-heating effects from part to part... overkill perhaps, but easy enough to wire up. If you really want to go over the top, hook up GPIB or RS-232 from the meter, and log your results straight to the computer without typos!
The end result is worth the effort...
Attachments
This is a good read, thanks! I appreciate that, on top of your subjective perceptions (in which I implicitly trust given your success on the ACA mods, but have not heard by myself at all), you give technical objective reasons behind your ideas.
It goes well above my understanding of electronics and I would hope someone else would corroborate your views, but I appreciate you taking the time to share them. I learn a lot with this, if only for theoretical reasons.
Rafa.
Talk about "sharing the porn". You have a fortune's worth there, I have yet to purchase my first pair at $50!
That is impressive!