Actually seeing your and dvv s posts I think youd go the route I expect you to go, no " I rather use a output power transistor as vas as it sounds better than a video transistor" stuff, I think you know what I mean.
In the case of dvv I meant the part where he states that two To92 mpsa92s have lower distortion than one bf721. This is true only at low frequencies.
How does this Jfet driver compare with standard Bjt follower ?? Ive used mosfets here but as expected due to their higher parasitic capacitances high frequency distortion is usually worse off.
In the case of cascoding I agree it does nothing, well I dont consider .001 and less figure improvements as worthwhile so its implementation is purely for limiting the vas voltage.
In the case of dvv I meant the part where he states that two To92 mpsa92s have lower distortion than one bf721. This is true only at low frequencies.
How does this Jfet driver compare with standard Bjt follower ?? Ive used mosfets here but as expected due to their higher parasitic capacitances high frequency distortion is usually worse off.
In the case of cascoding I agree it does nothing, well I dont consider .001 and less figure improvements as worthwhile so its implementation is purely for limiting the vas voltage.
Cascoding seems more efficent than degeneration in respect
of the VAS toplogy , at least in simulations...
Attachments
A buffer connected before or after the VAS ?..
It depend , but in case of post buffering , that is after the VAS ,
this latter will highly benefit from the higher loading impedance ,
hence the cascoded stage will have higher gain , wich will
reduce distorsions as well.
It depend , but in case of post buffering , that is after the VAS ,
this latter will highly benefit from the higher loading impedance ,
hence the cascoded stage will have higher gain , wich will
reduce distorsions as well.
Attachments
15 -20 amps .... !!! When playing, not unusual to pull 15 amps , POP that fuse a few times ..
Careful T, those values border on Big bwoy PSU size ............
In the graph above the VAS has no pre buffer ,
yet cascoding reduce notably THD.
That said , adding a pre buffer will reduce THD even further.
Hi,
Without knowing what you simulate (eg. Circuits) I would not know.
Why not sim the Buffer/VAS combo I have shown open loop and then add a cascode...
Ciao T
Without knowing what you simulate (eg. Circuits) I would not know.
Why not sim the Buffer/VAS combo I have shown open loop and then add a cascode...
Ciao T
Hi,
I don't know what you mean, actually. I'd rather use a Tube as VAS, but as said, for my particular case this was not on the cards. BTW, if it DID sound better I WOULD use an output transistor as VAS.
Well, it depends.
Why not look at the datasheets for the Fet's I selected and have a look.
The circuit for AC essentially boils down to a standard complementary push pull follower (I realised I need to shift one connection of one cap for that), so Ciss is bootstrapped into the VAS input load (which is quite high for the extended audio range), hence Crss dominates as input impedance, which is comparable to BJT's.
I make it 30pF in total for the whole follower @ 1MHz which is around the f3 for the input stage / Vas interface. I think this is quite adequate.
An Emitter follower can have lower output impedance, maybe a touch less capacitance (or maybe not), but that is not the only figure of merit and not necessarily what I am looking for. Most of the time it will have a more non-linear and lower input impedance...
In the end I would agree in principle that a cascoded J-Fet buffered, Mosfet cascoded bipolar VAS could do better than what I have shown, but compared to other sources of distortion in the Amp, would it be material enough to worry about?
I'd like to have all harmonics from the VAS (and IPS) around 10dB below those of the output stage, that's all...
Ciao T
I don't know what you mean, actually. I'd rather use a Tube as VAS, but as said, for my particular case this was not on the cards. BTW, if it DID sound better I WOULD use an output transistor as VAS.
Well, it depends.
Why not look at the datasheets for the Fet's I selected and have a look.
The circuit for AC essentially boils down to a standard complementary push pull follower (I realised I need to shift one connection of one cap for that), so Ciss is bootstrapped into the VAS input load (which is quite high for the extended audio range), hence Crss dominates as input impedance, which is comparable to BJT's.
I make it 30pF in total for the whole follower @ 1MHz which is around the f3 for the input stage / Vas interface. I think this is quite adequate.
An Emitter follower can have lower output impedance, maybe a touch less capacitance (or maybe not), but that is not the only figure of merit and not necessarily what I am looking for. Most of the time it will have a more non-linear and lower input impedance...
In the end I would agree in principle that a cascoded J-Fet buffered, Mosfet cascoded bipolar VAS could do better than what I have shown, but compared to other sources of distortion in the Amp, would it be material enough to worry about?
I'd like to have all harmonics from the VAS (and IPS) around 10dB below those of the output stage, that's all...
Ciao T
Hi,
Which one will reduce high order HD components more, given a high source impedance, cascoding or buffering (if we accept we may not do both). And how about standard cascode vs. Hawkesford cascode?
Ciao T
Which one will reduce high order HD components more, given a high source impedance, cascoding or buffering (if we accept we may not do both). And how about standard cascode vs. Hawkesford cascode?
Ciao T
Basically , at about 1Khz , nothing that would have any other
effect than self satisfaction , but at higher frequencies where
amps generaly struggle to be in par with CD standard , it will
help keep HD contents close below the -90dB line...
Hi,
I do not know what "Big bwoy PSU size" is.
I only use what I consider necessary.
In this case it is a 1200VA Transformer and > 200,000uF in total for a 180W Stereo Amp.
In my pet project I am limited to what can fit the existing hardware.
The original designer was not obliging on giving me much space for the electrolytics and I normally emphasise quality over quantity. So it's going to be only 144,000uF in total for a 150W Stereo Amp. But the designer was kind enough to fit a 1200VA torroid on an oversized core, with separate windings for the two channels and another one for the frontend.
Ciao T
Careful T, those values border on Big bwoy PSU size ............
I do not know what "Big bwoy PSU size" is.
I only use what I consider necessary.
In this case it is a 1200VA Transformer and > 200,000uF in total for a 180W Stereo Amp.
In my pet project I am limited to what can fit the existing hardware.
The original designer was not obliging on giving me much space for the electrolytics and I normally emphasise quality over quantity. So it's going to be only 144,000uF in total for a 150W Stereo Amp. But the designer was kind enough to fit a 1200VA torroid on an oversized core, with separate windings for the two channels and another one for the frontend.
Ciao T
Also germane and recommended: Dimitri Danyuk's October 2008 AES paper, "On the Optimization of Enhanced Cascode", where he uses a Boxall (a.k.a. Baxandall) pair in a cascode (Dimitri refers to the complementary stage as an "optimized cascode").
Brad
Hi,
With respect, having HD content at -90dB in an Amp also has no other effect than self satisfaction. Despite repeated requests by me you have not offered anything that demonstrated either that such low levels of HD reliable produce "good sound" or indeed that such low levels of HD can be delivered to the listner using any available speaker...
Ciao T
With respect, having HD content at -90dB in an Amp also has no other effect than self satisfaction. Despite repeated requests by me you have not offered anything that demonstrated either that such low levels of HD reliable produce "good sound" or indeed that such low levels of HD can be delivered to the listner using any available speaker...
Ciao T
That s a complementary differential topology with symetrical VAS
wich is the usual complementary common emitter , either with or
without cascodes and bufferings.
As for your schematic , i ll give t a try , although i have long
abandonned single differential input topologies that seems to me
inadequate for large signal and current amplification.
It is said in this forum that Hawksford cascode has better behaviour ,
so i did make some simulations and was quite surprised that it didnt
bring anything with the schematics i usualy play with.
As for the best VAS topology , it is obvious that a single transistor
is not a good option.
What will increase its gain and linearity will be an output buffer ,
cascoding and eventualy an input buffer.
Implementing all theses mods will yield the best performances
albeit with a drawback since bufferings and even cascoding will
increase phase shifts , hence more tight compensation is needed.
Last edited:
We have discussed of this earlier after i made some listening
tests where i noticed that HD contents are audible with a 1KHZ
sine at levels of -60dB.
Agree that -90dB HD contents are hardly audible but as i said ,
i used the CD standard as baseline , wich seems to me quite logical.
As for sounding better , who knows , but i can assure you that
it will not sound worse , for sure....
Hi,
You mean it is a complete amplifier, including looped feedback, following your philosophy of maximising looped feedback. It is my experience that such designs do not produce sound quality I like.
I am more interested in the behaviour of only the VAS Stage without looped feedback AND with the levels of degeneration that I am likely to apply (> 30dB)
Output Buffer in my case is not necessary, as the load is only a pair of the TO-220 ("small") Lateral Fets in Class A. It is hard to find any BJT Buffer that under practical conditions will do better (I tried).
I do NOT want to increase gain and thus the amount of feedback. Those are your design goals and they run completely counter to my own ones.
I instead want to increase open loop linearity (and bandwidth) at a given (resistive) load and at a given level of gain. I have already decided what level of HD (NOT THD - THD is meaningless) is acceptable and I only want this to be dominated by the output stage, so I only want the IPS and VAS distortion to be lower than the OPS one (at all harmonics).
The output stage will be biased for minimal high order components. This requires "overbias" compared to what is called "optimum bias" (but should more precisely be called minimum THD bias - which is not an optimum from an audibility viewpoint). Looped Feedback will be used only to bring the resulting low order components below -60dB and no further.
Knowing the behaviour of the output stage in terms of HD distribution and levels as well as the level of HD I wish to permit determines the amount of Feedback (selected at 14dB) and thus the amount of open loop gain (40dB) I require.
I have no use for more gain, only for more inherent linearity and that only to around 10dB below the HD levels of the output stage, though I'm not going to loose sleep if I have lower HD.
Ciao T
You mean it is a complete amplifier, including looped feedback, following your philosophy of maximising looped feedback. It is my experience that such designs do not produce sound quality I like.
I am more interested in the behaviour of only the VAS Stage without looped feedback AND with the levels of degeneration that I am likely to apply (> 30dB)
Output Buffer in my case is not necessary, as the load is only a pair of the TO-220 ("small") Lateral Fets in Class A. It is hard to find any BJT Buffer that under practical conditions will do better (I tried).
I do NOT want to increase gain and thus the amount of feedback. Those are your design goals and they run completely counter to my own ones.
I instead want to increase open loop linearity (and bandwidth) at a given (resistive) load and at a given level of gain. I have already decided what level of HD (NOT THD - THD is meaningless) is acceptable and I only want this to be dominated by the output stage, so I only want the IPS and VAS distortion to be lower than the OPS one (at all harmonics).
The output stage will be biased for minimal high order components. This requires "overbias" compared to what is called "optimum bias" (but should more precisely be called minimum THD bias - which is not an optimum from an audibility viewpoint). Looped Feedback will be used only to bring the resulting low order components below -60dB and no further.
Knowing the behaviour of the output stage in terms of HD distribution and levels as well as the level of HD I wish to permit determines the amount of Feedback (selected at 14dB) and thus the amount of open loop gain (40dB) I require.
I have no use for more gain, only for more inherent linearity and that only to around 10dB below the HD levels of the output stage, though I'm not going to loose sleep if I have lower HD.
Ciao T
- Status
- Not open for further replies.