Hi Hugh, as soon as I saw the thread title I guessed it might be yourself responding.
Just to go off the original OP's thread slightly, did you see this thread of mine,
http://www.diyaudio.com/forums/soli...ry-design-sounded-great-time.html#post2165381
I missed the VAS PNP stage off, but the main question was that it was a quasi design but with the darlington arrangement as shown on only one side of the LTP.
Wondered what your reaction might be to that... I always remembered it sounded really good at the time, and never found out why it had been designed like that.
Just to go off the original OP's thread slightly, did you see this thread of mine,
http://www.diyaudio.com/forums/soli...ry-design-sounded-great-time.html#post2165381
I missed the VAS PNP stage off, but the main question was that it was a quasi design but with the darlington arrangement as shown on only one side of the LTP.
Wondered what your reaction might be to that... I always remembered it sounded really good at the time, and never found out why it had been designed like that.
I prefer the sound of quasi.
But if you are going to design or build a quasi then care needs to be taken with bias and DC offset. I found I needed a bit more bias current on the quasi which can cause thermal problems if you run too much bias current.
Member
Joined 2009
Paid Member
The top driver is an emitter follower so has a gain of one.
The bottom driver is an inverter with a gain of 1 unless you add a Baxandall diode, in which case it has a much higehr gain and also drives the mosfet gate harder just like the top driver.
This extra gain of course disturbs the balance and then most often yuou find you need a 220pf from c-b on the lower driver.
I don't really know what the idea was, and have never found any reference to this old design anywhere on the web. The design was AC coupled, so any DC drift didn't matter... I don't know.
I've done it both ways, and I believe that Q-C sounds better, even if it measures worse. Others won't agree. Another case of YMMV.
I confirm quasicomplementary sounds better in MOSFET and BJTs
have test it with NMOS
and listening 12 years with my own ears in installed quasicomplementary PA Amplifier in BIG CLUB for MID and HIGHS (speakers was JBL 2445)
Model BGW 250C for Midrange and Model 100 for Tweeters
After replace 1 months with todays complementary amp ....quasicomplementary PA Amplifier was in service...... the sound was horrible
Mooly,
It is interesting that lots of people say Quasi is best sound, Full Comp measures best, yet no one (yet) draws the long bow and concludes that thd is not correlated well with sound quality. If only we could get this debate going.....
Bring on Earl Geddes, he might stir things up!
Hugh
It is interesting that lots of people say Quasi is best sound, Full Comp measures best, yet no one (yet) draws the long bow and concludes that thd is not correlated well with sound quality. If only we could get this debate going.....
Bring on Earl Geddes, he might stir things up!
Hugh
Hugh... it does indeed seem the case from the subjectivst point of view, that quasi stages have a certain "something" sonically, and let's be honest, it's how it all sounds ultimately that matters most.
I would really be interested in your "reaction" though, to the LTP as shown here (I've added the VAS stage too).
Any thoughts as to what the darlington arrangement does to the THD spectrum etc.
thanks
I would really be interested in your "reaction" though, to the LTP as shown here (I've added the VAS stage too).
Any thoughts as to what the darlington arrangement does to the THD spectrum etc.
thanks
Attachments
Hi Mooly,
Interesting approach! I really don't feel that a soft start is necessary; I'm more concerned about how the additional pn junctions to rail formed by the darl soft start would affect the drive for the VAS; the base of same would be referenced via the darlington to the rail and this would likely bring more non-linearities to the VAS operation. Neither can I see an advantage to the fb device having darl configuration, as the impedance of the fb network is generally low anyway, dominated by the shunt resistor (680R in your schemat). A cfp maybe, but not a darlington.
The VAS current source which you show as 2k2 is really best in constant, or near constant, current. Either a CCS or a bootstrap is best here.
Lastly, a useful quasi refinement is a medium sized diode on the emitter of the pnp phase splitter, parallelled with a 100R resistor and a 100nF cap, all to simulate the sizable junction of a complementary output. It cleans up the crossover event very nicely, take a look on LTSpice, works well, the brainchild of Peter Baxandall in the sixties.
I like the cap coupled output - affords wonderful protection, no relay required!
I hope this doesn't seem too negative. As in all cases, for a definitive answer, build it and listen. Hell, it might sound wonderful, stranger things have happened.....
Cheers,
Hugh
Interesting approach! I really don't feel that a soft start is necessary; I'm more concerned about how the additional pn junctions to rail formed by the darl soft start would affect the drive for the VAS; the base of same would be referenced via the darlington to the rail and this would likely bring more non-linearities to the VAS operation. Neither can I see an advantage to the fb device having darl configuration, as the impedance of the fb network is generally low anyway, dominated by the shunt resistor (680R in your schemat). A cfp maybe, but not a darlington.
The VAS current source which you show as 2k2 is really best in constant, or near constant, current. Either a CCS or a bootstrap is best here.
Lastly, a useful quasi refinement is a medium sized diode on the emitter of the pnp phase splitter, parallelled with a 100R resistor and a 100nF cap, all to simulate the sizable junction of a complementary output. It cleans up the crossover event very nicely, take a look on LTSpice, works well, the brainchild of Peter Baxandall in the sixties.
I like the cap coupled output - affords wonderful protection, no relay required!
I hope this doesn't seem too negative. As in all cases, for a definitive answer, build it and listen. Hell, it might sound wonderful, stranger things have happened.....
Cheers,
Hugh
There is a lot of discussion here about the topology in what I would call the 'DC' condition, but no reference to the dynamic operating conditions. Do you not think that aspects like input LTP overload, bandwidth, slew rate (these three preceeding factors having a critical impact on TIMD performance), gain and phase margin are actually the really important parameters that define amplifier sound?
Last edited:
Hi Hugh,
Thanks very much for your thoughts, there not at all negative, I'm just reminiscing really... this was the first amp I ever built from scratch... and it always sounded really musical to me... it was only later as skills developed that I realised how unusual it was, and how good it was compared to others.
" Hell, it might sound wonderful, stranger things have happened....."
Exactly, it was wonderful, very different sonically to other offerings of the period, the Texan and so on.
I realise there are many things that could be improved on it, but that darlington fb configuration, weird weird weird, that was the real puzzler.
Maybe that slow start and the non linearities you suspect... could that be it's secret
It was just a design that as a youngster caught my imagination... and I was thrilled when I got it all to work. I'm sure there was a 1nf (yep that big) across the C-E of the VAS.
Thanks for sharing your views anyway.
Thanks very much for your thoughts, there not at all negative, I'm just reminiscing really... this was the first amp I ever built from scratch... and it always sounded really musical to me... it was only later as skills developed that I realised how unusual it was, and how good it was compared to others.
" Hell, it might sound wonderful, stranger things have happened....."
Exactly, it was wonderful, very different sonically to other offerings of the period, the Texan and so on.
I realise there are many things that could be improved on it, but that darlington fb configuration, weird weird weird, that was the real puzzler.
Maybe that slow start and the non linearities you suspect... could that be it's secret
It was just a design that as a youngster caught my imagination... and I was thrilled when I got it all to work. I'm sure there was a 1nf (yep that big) across the C-E of the VAS.
Thanks for sharing your views anyway.
QUOTE=Mooly.
Maybe that slow start and the non linearities you suspect... could that be it's secret
hello.
i think this is not only a slow start.....it is a filter (a capacitance multiplier) too .........to clean up the power supply voltage.this kind of technic has positive effects on the sound.other examples in this place are a res-z diode network or volt regs.so this could really be part of the (sound)secret.
greetings
Maybe that slow start and the non linearities you suspect... could that be it's secret
hello.
i think this is not only a slow start.....it is a filter (a capacitance multiplier) too .........to clean up the power supply voltage.this kind of technic has positive effects on the sound.other examples in this place are a res-z diode network or volt regs.so this could really be part of the (sound)secret.
greetings
What he's calling a "soft start" is really an active decoupler (sometimes called a "capacitance multiplier") I've used them for some projects: at the front end of a 120db(v) gain audio stage that was a subsystem of a TRF longwave receiver; as the low current negative rail for a vacuum tube amp. Being that it's basically a DC emitter follower, and represents a very low AC impedance (which is the whole point: isolating the sensitive early stages from the DC rail) it's sonically benign. I question whether it's really necessary to stabilize that design, though.
It looks like that thing unbalances the LTP -- not good.