A set of good pictures like this can help make a crowd-sourced DIY assembly guide.
Yes please post lots of hi-rez pics so I don't screw up.😀
jeff
The new Bulwark FE boards are coming along quite nicely. I used a 1.0 uF Polycarbonate cap at C1 and a pair of 100 uF Elna Silmic caps at C2 and C3. These tend to need a run-in period and reward the listener for their patience. I am also using film caps at C7 and C9, as my experience with the Dreadnought boards was that ceramics just don’t sound good.
These were considerably easier to build compared to the Dreadnought boards, and are also easier to visually inspect once built. I think this amp likes having some iron in the signal path, or maybe that’s just me.
These were considerably easier to build compared to the Dreadnought boards, and are also easier to visually inspect once built. I think this amp likes having some iron in the signal path, or maybe that’s just me.
After making a gain adjustment to the Bulwark boards, I've found what I consider to be a sweet spot. I was shooting for +3dB of gain instead of unity, so I set R22 to 2.2k, R21 to 4.75k and C10 to 47uF (Nichicon ES, 50V). Beyond simply increasing the gain, this adjustment seems to open up the clarity of the FE stage. More details are noticeable, without the overall tonal balance sounding too bright. This change has also added crispness to percussion instruments, which is something I appreciate. It seems that the Edcor responds well to a hotter input signal
As I have been listening over the last few days, the dual Elna Silmic caps on the output have broken in, along with the film input cap. The overall effect is still very smooth, but with more detail. The clarity of this FE is such that I'm tempted to do some capacitor 'rolling' between the input and output coupling caps, but I will control myself until I'm completely out of other ideas to try. After all, I have another pair of bare Bulwark PCBs waiting to test other ideas and an alternate signal transformer.
As the easier of the two alternate front ends I've built so far, the Bulwark boards are my recommendation for those who want to try something different with their VFET lottery amps. These boards should work well with both P-type and N-type output stages. Remember that the N-type does not require the FE to invert the signal, and wire accordingly.
As I have been listening over the last few days, the dual Elna Silmic caps on the output have broken in, along with the film input cap. The overall effect is still very smooth, but with more detail. The clarity of this FE is such that I'm tempted to do some capacitor 'rolling' between the input and output coupling caps, but I will control myself until I'm completely out of other ideas to try. After all, I have another pair of bare Bulwark PCBs waiting to test other ideas and an alternate signal transformer.
As the easier of the two alternate front ends I've built so far, the Bulwark boards are my recommendation for those who want to try something different with their VFET lottery amps. These boards should work well with both P-type and N-type output stages. Remember that the N-type does not require the FE to invert the signal, and wire accordingly.
......It seems that the Edcor responds well to a hotter input signal........
well, without searching back for exact schematic, considering that if Edcor is there, it is still connected in same way as in Papa's origin FE ......... if I presume that you're listening on same SPL ......... Edcor is still getting same signal, whatever gain is of preceding stage
As the Edcor is wired the same way as in the original FE, a simple 1:5 gain element, you are correct. I am still at a loss to understand how reconfiguring the preceding silicon gain stage for an extra 3dB gives the level of improvement that it does. It is more than just a little louder / less rotation of the volume knob..
Along similar lines, I also went back to my Dreadnought boards and make a couple more small tweaks to those. The Dreadnoughts are now singing very smoothly. I replaced the last of the ceramic caps in a feedback signal path with film, and changed to a polycarbonate cap for one of the signal coupling caps. So the Dreadnoughts have responded well to some gentle cap rolling.
Note: I also reduced the gain of these boards slightly, so they now match the gain I have in the Bulwark FE boards. This makes it easier to compare the two, and I like where the overall gain is for both at 17dB total.
Along similar lines, I also went back to my Dreadnought boards and make a couple more small tweaks to those. The Dreadnoughts are now singing very smoothly. I replaced the last of the ceramic caps in a feedback signal path with film, and changed to a polycarbonate cap for one of the signal coupling caps. So the Dreadnoughts have responded well to some gentle cap rolling.
Note: I also reduced the gain of these boards slightly, so they now match the gain I have in the Bulwark FE boards. This makes it easier to compare the two, and I like where the overall gain is for both at 17dB total.
Last edited:
changes you're hearing are in function of rest of the circuit
you either made drive of Edcor worse or better but , what's most important, you like it more
you either made drive of Edcor worse or better but , what's most important, you like it more
The Bulwark FE continues to deliver wonderful music. It strikes an excellent balance between smoothness and detail, with an airy upper end that reveals much that lies in the recording. So much so, that one may wonder if it is the same CD.
As I was installing the extra resistors and capacitor to boost the gain, I had a small value film cap in mind to bypass R22 if necessary. At this point it doesn't seem necessary. Some recordings are just a little bright as they were produced. Others are not. I've listened to more CDs over the last couple days, and the Bulwark cards remain revealing, yet fairly so. I think the pair of 100 uF Elna Silmic caps at C2 and C3 is a good choice. They did take a while to settle in. I used the Toshiba TTA004B in place of the KSA1220AY in all locations, so I can confirm that they work well in this application.
As I was installing the extra resistors and capacitor to boost the gain, I had a small value film cap in mind to bypass R22 if necessary. At this point it doesn't seem necessary. Some recordings are just a little bright as they were produced. Others are not. I've listened to more CDs over the last couple days, and the Bulwark cards remain revealing, yet fairly so. I think the pair of 100 uF Elna Silmic caps at C2 and C3 is a good choice. They did take a while to settle in. I used the Toshiba TTA004B in place of the KSA1220AY in all locations, so I can confirm that they work well in this application.
I wondered about this myself. My conclusion is that the Edcor itself is as fast as it can be, and the rest of the circuit may be adjusted to improve the apparent speed of the front end. Both the Dreadnought and Bulwark cards needed some adjustments to let them better "get out of the way" to allow the output stage to sound as it was fully capable. I have written about the changes I made along the way, so they are scattered over several posts going back a few weeks. The Dreadnought circuit needed a little more help, and the Bulwark was fairly easy....
May I humbly ask for the sake of learning what kind of means can be used to "speed up" a transformer?
Learning everyday, many thanks in advance
Guess what? Of the two, the Bulwark now has a slight edge in detail and clarity over the Dreadnought. This was somewhat surprising, as the Dreadnought initially sounded like the faster of the two after the first set of adjustments.
On the original article about the VFET, Papa Nelson pointed out that the FE introduces some resonant ringing. The output stage alone is a much cleaner square wave.
Have any of the builders of the "other" front ends, with or without the edcors, have measured if this ringing is more prominent or less?
I fully appreciate Tungsten's subjective opinions about his modifications, but that is the result of his "entire" setup: CD player, speakers, AC line cleanness, even perhaps cables (not sure I buy much into that, but hey, it's part of the system).
Is there some way to make some objective comparisons other than the true-and-tested "build it yourself and see what you think"?
Thanks,
Rafa.
Have any of the builders of the "other" front ends, with or without the edcors, have measured if this ringing is more prominent or less?
I fully appreciate Tungsten's subjective opinions about his modifications, but that is the result of his "entire" setup: CD player, speakers, AC line cleanness, even perhaps cables (not sure I buy much into that, but hey, it's part of the system).
Is there some way to make some objective comparisons other than the true-and-tested "build it yourself and see what you think"?
Thanks,
Rafa.
You can adjust the Edcor ringing yourself, if you have a capacitor substitution box, resistor substitution box, signal generator, oscilloscope, and DC power supply. The secondary termination network for the Edcor, which Nelson Pass chose (220pF series 10K), may not be to your liking, in which case . . . . . Do It Yourself.
You are free to apply whichever decision criteria appeal to you, of course. If I remember correctly, the last time a DIYA member undertook this project, it was on the Edcor secondary damping network of the M2x amplifier. That member decided to optimize the frequency domain response; another choice would have been to optimize the time domain response. The white paper on Jensen Transformers' website suggested the latter, if I remember correctly. Jensen's goal when selecting a transformer secondary damping network, was to get as close as possible to a second order Bessel response in the time domain.
Students with excellent memory will recall from their EE undergraduate days, that the Bessel response does include a wee bit of overshoot ("ringing"). So maybe you don't want Bessel after all. The Jensen Transformers people know a LOT about transformers but they don't know what YOU might desire.
You are free to apply whichever decision criteria appeal to you, of course. If I remember correctly, the last time a DIYA member undertook this project, it was on the Edcor secondary damping network of the M2x amplifier. That member decided to optimize the frequency domain response; another choice would have been to optimize the time domain response. The white paper on Jensen Transformers' website suggested the latter, if I remember correctly. Jensen's goal when selecting a transformer secondary damping network, was to get as close as possible to a second order Bessel response in the time domain.
Students with excellent memory will recall from their EE undergraduate days, that the Bessel response does include a wee bit of overshoot ("ringing"). So maybe you don't want Bessel after all. The Jensen Transformers people know a LOT about transformers but they don't know what YOU might desire.
I feel I very often say something that is interpreted quite the other way around. I hope this is due to English not being my first language, although I sometimes feel people enjoy twisting my meaning fully knowing what I meant.
Just to try to say it again: I was not “complaining” about the ringing, merely pointing to a paragraph that Papa wrote himself in the article. I was wondering if some of the other FE cards (with or without the edcors) had a different behavior and if someone had gone through the process of measuring, maybe because it’s an interesting subject.
Not only is the “secondary termination network for the Edcor, which Nelson Pass chose” to my liking, I have praised it as the most musical thing I have ever heard. Also, I think that anyone that has read a single post of mine knows I could not, even if my life depended on it, start changing those values and reach a working FE. And I think that if there are 3 people in this entire forum that can “improve” a NP design, it’s being optimistic.
Where I really don’t understand the replies that come my way, is the need, when I ask: “has someone measured X”, to always reply: “do it yourself”. Maybe someone already did and have insight to give. If I have said: “I think that cap X would sound much better than NPs solution”, sure, call me out on it, make me look like an idiot.
I merely asked if someone had measured if this behavior is present in other options. Why does that read as if I were attacking the design, is beyond me. In my language, that question does not imply what was assumed here. Maybe in English asking a question implies a critique to the subject being asked about?
🙁
Rafa.
Just to try to say it again: I was not “complaining” about the ringing, merely pointing to a paragraph that Papa wrote himself in the article. I was wondering if some of the other FE cards (with or without the edcors) had a different behavior and if someone had gone through the process of measuring, maybe because it’s an interesting subject.
Not only is the “secondary termination network for the Edcor, which Nelson Pass chose” to my liking, I have praised it as the most musical thing I have ever heard. Also, I think that anyone that has read a single post of mine knows I could not, even if my life depended on it, start changing those values and reach a working FE. And I think that if there are 3 people in this entire forum that can “improve” a NP design, it’s being optimistic.
Where I really don’t understand the replies that come my way, is the need, when I ask: “has someone measured X”, to always reply: “do it yourself”. Maybe someone already did and have insight to give. If I have said: “I think that cap X would sound much better than NPs solution”, sure, call me out on it, make me look like an idiot.
I merely asked if someone had measured if this behavior is present in other options. Why does that read as if I were attacking the design, is beyond me. In my language, that question does not imply what was assumed here. Maybe in English asking a question implies a critique to the subject being asked about?
🙁
Rafa.
Last edited:
your question was completely valid
shunting RC being in function of xformer loading, same as in function of xformer feed (to lesser degree, if we assume that all FE iterations are having low enough Rout)
whatever FE iteration, if there is xformer ditto in front of OS ( and that is not changing), Papa's RC applies - if you prefer Papa's approach
shunting RC being in function of xformer loading, same as in function of xformer feed (to lesser degree, if we assume that all FE iterations are having low enough Rout)
whatever FE iteration, if there is xformer ditto in front of OS ( and that is not changing), Papa's RC applies - if you prefer Papa's approach
Rafa,
Don't take it personally - your english is as good as mine, quite possibly 'better' as I always have to edit mine down to the essentials. I can usually replace a paragraph with a sentence and be clearer for it.
The absence of the facial cues available during a conversation has been blamed for a lot of misunderstanding (and vituperation) on the net. It certainly contributes.
We tend to fill in the gaps for ourselves, add our own interpretation, and then hit the keyboard instantly with no time for reflection. Which is why a hostile post often reveals a lot about the offended party......
On the bright side, I've never used the Ignore List on this forum and there's more generosity and goodwill here than I've come across anywhere else.
Don't take it personally - your english is as good as mine, quite possibly 'better' as I always have to edit mine down to the essentials. I can usually replace a paragraph with a sentence and be clearer for it.
The absence of the facial cues available during a conversation has been blamed for a lot of misunderstanding (and vituperation) on the net. It certainly contributes.
We tend to fill in the gaps for ourselves, add our own interpretation, and then hit the keyboard instantly with no time for reflection. Which is why a hostile post often reveals a lot about the offended party......
On the bright side, I've never used the Ignore List on this forum and there's more generosity and goodwill here than I've come across anywhere else.
Hey Rafa
Sometimes it seems as you have stated but really it was an invitation to put on your FAB hat and learn stuff. Often this path has been useful to the poster you think is picking on you. Maybe they are just giving you a nudge to try stuff because that is where the learning is hiding, Just words from this old teacher to think about.
Sometimes it seems as you have stated but really it was an invitation to put on your FAB hat and learn stuff. Often this path has been useful to the poster you think is picking on you. Maybe they are just giving you a nudge to try stuff because that is where the learning is hiding, Just words from this old teacher to think about.
NP changed the component values of the Edcor secondary termination network, at least three times between July 2020 and April 2021. I know this because I tracked and implemented each of those changes in the early prototypes of the Front End cards called Scourge and Bulwark.
What this means is: it's perfectly OK to tinker with the Edcor termination network. Nelson did, so can you. The network isn't carved in stone, it's a work in progress. Maybe the next improvement will come from the club of VFET amp builders, which would be really cool.
What this means is: it's perfectly OK to tinker with the Edcor termination network. Nelson did, so can you. The network isn't carved in stone, it's a work in progress. Maybe the next improvement will come from the club of VFET amp builders, which would be really cool.
I have been waiting to see if anyone else had built and listened to a pair of Bulwark FE cards. So far I haven't seen any other mention of them. There has only been one other builder that briefly mentioned the Dreadnought cards on one of the other threads. I don't recall there being much of a description or listening impressions for those. I would be happy to see other builders trying these and other variations of the front end.
Both the Bulwark and the original NP front end use the same secondary termination network that Edcor recommends for their transformer. I haven't experimented with these values, as that would require hooking up a proper signal generator as well as my scope to examine the square wave response. I usually enjoy listening tests more than looking at scope traces, but sometimes the scope tells us things we won't discover in other ways.
As my last few posts in this forum have indicated, the circuits themselves, including some of the capacitors, have played a greater role in the sound of the Dreadnought and Bulwark front ends than I expected. And the differences that I have been hearing are unlikely to be the things that will show up on simple scope traces. Music is a far more complex test signal than most ever use when looking at results on test equipment.
Both the Bulwark and the original NP front end use the same secondary termination network that Edcor recommends for their transformer. I haven't experimented with these values, as that would require hooking up a proper signal generator as well as my scope to examine the square wave response. I usually enjoy listening tests more than looking at scope traces, but sometimes the scope tells us things we won't discover in other ways.
As my last few posts in this forum have indicated, the circuits themselves, including some of the capacitors, have played a greater role in the sound of the Dreadnought and Bulwark front ends than I expected. And the differences that I have been hearing are unlikely to be the things that will show up on simple scope traces. Music is a far more complex test signal than most ever use when looking at results on test equipment.
- Home
- Amplifiers
- Pass Labs
- DIY Sony VFET Builders thread