Whilst I wait for my parts kit and chassis shipments, I have started reading through the article and watching some of Nelson's BAF talk videos on YouTube.
Looking at the kit amp schematic, I am thinking it might not be too hard to convert that current source transistor into a more active participant in the output... perhaps we can get this puppy converted over to "full DEF" without too much trouble?
Forgive me if there are others already in discussion along the same lines - alas, the other various VFET threads are already so long that I haven't a hope of reading through them all! So, I thought it would be best to start a separate thread, and constrain the topic here to just the topic of DEF conversion modifications for the kit.
This will not be a project for the beginners - at least not until these ideas have been tested and fettled. I plan to build it as intended first, to get a baseline set of measurements and get familiar with how it sounds, and then start experimenting... so for now, these ideas are merely napkin scribbles, but I wanted to put them out for others to noodle on and chime in with their thoughts.
OK... so how might we get that IRF250 a little more involved??
The trick with DEF is usually to match up Q1 and Q2 so that the bias current lands where we want it. In this case, we don't have matched FETs to play with, but... we do have this lovely opto-isolated bias scheme, which looks perfect for some creative adaptation. The obvious move here is to relocate the current sensing resistors, R1 and R2 to the drain side of Q2, along with R6 and the LED side of Q3. We'll have to watch the AC current through the LED - may need a filter cap there to steady it up, but the time constant will have to be faster than R7/C2, I think.
We may or may not want to keep a little source resistance on Q2 just to start with, so the empty socket where R1 used to be could be stuffed with something like 0.47Ω for initial experimentation.
Next, we need a way to couple the AC drive signal into the gate of Q2. Alas, C2 is firmly clamping the AC gate voltage to the source. We'll have to add a high-value resistor such as 100K in series between R7 / C2 and R3, and bring another coupling cap from the input. Now the input source will see R9 and our new resistor in parallel - so, something close to 50kΩ impedance, which may nor may not be too much loading for the front end (most likely it'll be OK) ... But if necessary, those two resistors could be changed to, say, 200kΩ.
Et Voilà! I think that is about all we need to get the top FET working alongside the VFET in DEF mode...
Thoughts? Opinions? Am I a lunatic, or is this idea worth pursuing?
Looking at the kit amp schematic, I am thinking it might not be too hard to convert that current source transistor into a more active participant in the output... perhaps we can get this puppy converted over to "full DEF" without too much trouble?
Forgive me if there are others already in discussion along the same lines - alas, the other various VFET threads are already so long that I haven't a hope of reading through them all! So, I thought it would be best to start a separate thread, and constrain the topic here to just the topic of DEF conversion modifications for the kit.
This will not be a project for the beginners - at least not until these ideas have been tested and fettled. I plan to build it as intended first, to get a baseline set of measurements and get familiar with how it sounds, and then start experimenting... so for now, these ideas are merely napkin scribbles, but I wanted to put them out for others to noodle on and chime in with their thoughts.
OK... so how might we get that IRF250 a little more involved??
The trick with DEF is usually to match up Q1 and Q2 so that the bias current lands where we want it. In this case, we don't have matched FETs to play with, but... we do have this lovely opto-isolated bias scheme, which looks perfect for some creative adaptation. The obvious move here is to relocate the current sensing resistors, R1 and R2 to the drain side of Q2, along with R6 and the LED side of Q3. We'll have to watch the AC current through the LED - may need a filter cap there to steady it up, but the time constant will have to be faster than R7/C2, I think.
We may or may not want to keep a little source resistance on Q2 just to start with, so the empty socket where R1 used to be could be stuffed with something like 0.47Ω for initial experimentation.
Next, we need a way to couple the AC drive signal into the gate of Q2. Alas, C2 is firmly clamping the AC gate voltage to the source. We'll have to add a high-value resistor such as 100K in series between R7 / C2 and R3, and bring another coupling cap from the input. Now the input source will see R9 and our new resistor in parallel - so, something close to 50kΩ impedance, which may nor may not be too much loading for the front end (most likely it'll be OK) ... But if necessary, those two resistors could be changed to, say, 200kΩ.
Et Voilà! I think that is about all we need to get the top FET working alongside the VFET in DEF mode...
Thoughts? Opinions? Am I a lunatic, or is this idea worth pursuing?
You can use other components for trying and building anything you like and want.
Converting this particular, special and prestige kit to any other thing is a shame.
IMHO
Converting this particular, special and prestige kit to any other thing is a shame.
IMHO
Oh, I don't know. There seems to be no compunction about changing out Nelson's original front end board to any of a half dozen varieties, for example. Does that change the spirit of the project? It will certainly change the sound of the amp. So... what about the output stage? If we can improve upon the basic circuit provided, is that not simply making the most of these rare VFETs, allowing them to shine their brightest and bring the maximum musical joy into our lives?
I like to think that as a true DIY project, one is expected to "make it their own". Rest assured, however, that these special VFETs will receive all of the respect and care they deserve.
I like to think that as a true DIY project, one is expected to "make it their own". Rest assured, however, that these special VFETs will receive all of the respect and care they deserve.
Thinking a bit more about the opto-isolated bias circuit, perhaps the sense resistors should sit behind a big power filter cap, so that the voltage driving the bias LED doesn't experience large AC swings. They'll serve double-duty there as well, contributing to the power rail smoothing.
Another quick note on a minor detail I overlooked initially... the gate of Q2 is bootstrapped by C2, so whatever RC we use to couple our input signal to the gate will likewise be bootstrapped, raising it's AC impedance as seen by the front end... so increasing the value of R9 may not be all that important.
One question which is now foremost in my mind is whether or not the provided IRF250 will make a good DEF pairing with the VFET, or if I should seek out an alternative type in TO-3 package. Haven't looked at the data sheets yet (or read through any of the many threads here to start getting an idea of what qualities would make for a good pairing), so I'm hoping someone will chime in on this aspect.
One question which is now foremost in my mind is whether or not the provided IRF250 will make a good DEF pairing with the VFET, or if I should seek out an alternative type in TO-3 package. Haven't looked at the data sheets yet (or read through any of the many threads here to start getting an idea of what qualities would make for a good pairing), so I'm hoping someone will chime in on this aspect.
name of the game is to adjust current contribution of parts used to desired ratio
if there is a reason to do that, then change it
if there is no reason, don't
can't see that major factor is mentioned at all
if there is a reason to do that, then change it
if there is no reason, don't
can't see that major factor is mentioned at all
Major factor being what, transconductance @ bias current?
It's largely about distortion cancellation, correct? (in addition to reducing output impedance & increasing output stage efficiency, of course...)
It's largely about distortion cancellation, correct? (in addition to reducing output impedance & increasing output stage efficiency, of course...)
in my Book it is about wanted THD Spectra, and (in my Book again) that's not in usual meaning of distortion cancellation , but shaping of same
also, when speaking of THD Spectra, I'm observing the same in entire power range, not just at 1W
some constructions can be all Honney and Nut at 1W, while all hell loose going up in power
some are more polite
anyway, just wanting to say that you need first firm definition of project/investigation goal.......
let's say that pure DEF is one made of parts allowing no source resistors (gate control is other page, and pretty much trivial), thus allowing full nature of parts, combined
if that nature combined we want to change, there are the ways .......
also, when speaking of THD Spectra, I'm observing the same in entire power range, not just at 1W
some constructions can be all Honney and Nut at 1W, while all hell loose going up in power
some are more polite
anyway, just wanting to say that you need first firm definition of project/investigation goal.......
let's say that pure DEF is one made of parts allowing no source resistors (gate control is other page, and pretty much trivial), thus allowing full nature of parts, combined
if that nature combined we want to change, there are the ways .......
Certainly, the distortion character is mainly what we wish to influence for the better - the usual avoidance of high-order harmonics. And, to the extent we can reduce the low-order HD, this is also good. Perhaps of greatest importance to me is the reduction of IMD - though this is something which more generally requires a reduction of THD, since they are inter-related.
Better damping factor and more efficiency are gravy, if we can get those benefits too! But not the main point here.
Better damping factor and more efficiency are gravy, if we can get those benefits too! But not the main point here.