Q3/6 are the cascodes..biased by the leds at 4V below the rails.
I might lose a little headroom , but they have many benefits.
1.As aksa and others were discussing in the CM thread, cascodes
in the VAS diminish the "early effect".
2. The amp/VAS is much faster than a standard "blameless"
without the cascode.
3.The open loop gain is better than the frugal1/blameless,
again this is the cascode, it also clips "gracefully".
OS
I might lose a little headroom , but they have many benefits.
1.As aksa and others were discussing in the CM thread, cascodes
in the VAS diminish the "early effect".
2. The amp/VAS is much faster than a standard "blameless"
without the cascode.
3.The open loop gain is better than the frugal1/blameless,
again this is the cascode, it also clips "gracefully".
OS
Previously, C7 and C9 was connected to LED's, and now they are connected to rails (as they should be, I think). What happens - Im just curious? 
And more important: Q5 will dissipate much more than Q7, so there will be some imbalance. It should be cascoded, or both should be thermaly tied somehow.
With the cascode VAS Isn't faster, because that is mainly determined by Cdom capacitor. That capacitor is also much larger than Ccb (base-collector) with these transistors. I don't know how much it affect Early effect (these transistors have high output resistance anyway), but I think, that dropping the cascode for simplicity reason is woth trying.
And more important: Q5 will dissipate much more than Q7, so there will be some imbalance. It should be cascoded, or both should be thermaly tied somehow.
With the cascode VAS Isn't faster, because that is mainly determined by Cdom capacitor. That capacitor is also much larger than Ccb (base-collector) with these transistors. I don't know how much it affect Early effect (these transistors have high output resistance anyway), but I think, that dropping the cascode for simplicity reason is woth trying.
Then I would have a triple like in the original APT topology,
but also lose another volt of rail capability.The OP stage
seems to have enough gain as to not load the vas.
C7/9, tied to the rails seem to increase OLG and moderate
saturation at clipping.. I might baker clamp the VAS. (still
saturates at LF a little)
Q5/7 , Q8/9,Q17/18 will be "face to face" T-coupled..
Amp balances out good , (electrically) under 1mv offset
It sure seems faster, at least in simulation , as far as dropping
the cascodes, another $1 does not matter Fa2 had 6 VAS
trannies and the board was not to hard to layout.
No cascode..
UGP is 100Khz below first plot ....slower.
Overall gain is 62 vs 66 with cascode.. phase remans the same..
The real question is whether this small increase in UGP and
global gain is worth 8 components. Also how will the cascode
"sound" ??Some say they sound awful ,but that is a personal
opinion..
OS
UGP is 100Khz below first plot ....slower.
Overall gain is 62 vs 66 with cascode.. phase remans the same..
The real question is whether this small increase in UGP and
global gain is worth 8 components. Also how will the cascode
"sound" ??Some say they sound awful ,but that is a personal
opinion..
OS
Attachments
Hi Os,
Have you spotted the problem with the Holman amp yet ?. You should do this as it affects the sound in a negative way IMHO.
Have a look at the gross imbalance in the LTP Im pretty sure youll have. I think you might prefer the sound of your blameless if you dont overcome this first. You will also get incredible distortion figures once youve mastered the topology, I did this about 12 odd years ago. No simms back then. Distortion figures around -120db are possible with this topology although also possible with the blameless. I always take my figures at 20khz although they will deteriote at high powers. You were worried about ltp imbalances with the fruga1 and 2, do the same here. There are 3 ways to correct this, possible more. Andrew is thinking along one of the possible solutions, I think, if he spotted the problem that is.
BTW Im of the opinion that cascodes do not sound good in general, that is with bjts, with fets its different because you should deal with the gate capacitance, their low voltage capabilty and other problematic properties, but with BJTs with below 2 pf cob capacitance its of debateable use. Their use can make sound thin,very detailed but you lose bass definition and warmth and the way they are biased is of vital importance. These things you cannot predict with simms. All a matter of balance and matching, there is no perfect amp and then these have to be matched to sources and speakers.
I would suggest you breadboard the holman amp first and experiment with it a little before building it. Maybe youll be happy with the sound as it is, maybe dissapointed compaired blameless or ltp vas.
Have you spotted the problem with the Holman amp yet ?. You should do this as it affects the sound in a negative way IMHO.
Have a look at the gross imbalance in the LTP Im pretty sure youll have. I think you might prefer the sound of your blameless if you dont overcome this first. You will also get incredible distortion figures once youve mastered the topology, I did this about 12 odd years ago. No simms back then. Distortion figures around -120db are possible with this topology although also possible with the blameless. I always take my figures at 20khz although they will deteriote at high powers. You were worried about ltp imbalances with the fruga1 and 2, do the same here. There are 3 ways to correct this, possible more. Andrew is thinking along one of the possible solutions, I think, if he spotted the problem that is.
BTW Im of the opinion that cascodes do not sound good in general, that is with bjts, with fets its different because you should deal with the gate capacitance, their low voltage capabilty and other problematic properties, but with BJTs with below 2 pf cob capacitance its of debateable use. Their use can make sound thin,very detailed but you lose bass definition and warmth and the way they are biased is of vital importance. These things you cannot predict with simms. All a matter of balance and matching, there is no perfect amp and then these have to be matched to sources and speakers.
I would suggest you breadboard the holman amp first and experiment with it a little before building it. Maybe youll be happy with the sound as it is, maybe dissapointed compaired blameless or ltp vas.
So which is it..???
I have noticed a "bit " of imbalance, much less than the blameless
but more than FA2 . At full power about a 1/100th of a milliamp
LTP but none at the base of the 2 active VAS trannies.
As far as the sound goes , a french audiophile aready has enlightened me on the use of cascodes...
http://peufeu.free.fr/audio/memory/memory-5-vas.html
He did not use the holman way, but he did use the "Magic resistor"
and described it's sound using a more unbalancedtopology ,the "blameless". ON the board I have included
the resistors and worst case could jumper the cascodes..
omit them..
OS
The comment of the balancing was incorrect, had a second look now at your and holman schematics, good simple and reliable way, I did it differently but same idea of push pull vas on single LTP and then some more. I didnt know of the existence of tom holman back then. Build it I think youll be happy with results. Its faster than typical blameless and there is some advantages with speed if you like faster paced music like rock.
Opinions will differ as do our likes, makes life interesting, read what Nelson Pass had to say when trying to cascode ouput stages.
Somewhere out there one will always find someone else that will share a common opinion. If every ones opinion was that cascodes were better one wouldnt find a single amp without them. I dont read much in someone s subjective opinion, one mans gold is the other mans dirt and then there is the marketing factor which will come into play with some opinions that are given. Someone else will interpret his use of muffled and fluid differently, I have never come across a cfp sounding muffled or making a circuit sound muffled because of its use in replacing a single transistor before, although I find a difference, bigger one with a fet bjt combination, but not muffled.
Reading and believing too much in objective opinions will put you into a
and lead to 
and finally 
, in the end you have to try it yourself and listen to what you like.
The best amp is the one you yourself enjoy listening to the most, irrespective of specs or others opinions. Some say this amp sounds nothing special, I find it better than typical blameless using same quality parts, the topology works however executed, strangely enough it either doesnt raise much interest around here or all have secretly tested and evaluated.
I am aware of peufeu,s work, I will try the magic resistor sometime in a circuit when I get a chance, have been working on no global feeback type circuit to listen what all the hype is about concerning this. So far mixed results.
Opinions will differ as do our likes, makes life interesting, read what Nelson Pass had to say when trying to cascode ouput stages.
Somewhere out there one will always find someone else that will share a common opinion. If every ones opinion was that cascodes were better one wouldnt find a single amp without them. I dont read much in someone s subjective opinion, one mans gold is the other mans dirt and then there is the marketing factor which will come into play with some opinions that are given. Someone else will interpret his use of muffled and fluid differently, I have never come across a cfp sounding muffled or making a circuit sound muffled because of its use in replacing a single transistor before, although I find a difference, bigger one with a fet bjt combination, but not muffled.Reading and believing too much in objective opinions will put you into a
and lead to and finally , in the end you have to try it yourself and listen to what you like.The best amp is the one you yourself enjoy listening to the most, irrespective of specs or others opinions. Some say this amp sounds nothing special, I find it better than typical blameless using same quality parts, the topology works however executed, strangely enough it either doesnt raise much interest around here or all have secretly tested and evaluated.
I am aware of peufeu,s work, I will try the magic resistor sometime in a circuit when I get a chance, have been working on no global feeback type circuit to listen what all the hype is about concerning this. So far mixed results.
Thank ostripper, very interesting readingostripper said:As far as the sound goes , a french audiophile aready has enlightened me on the use of cascodes...
http://peufeu.free.fr/audio/memory/memory-5-vas.html
He did not use the holman way, but he did use the "Magic resistor"
OS
Thats the funny thing, without the cascodes I can only
lower Cdom to 47pf before ringing on SW.
At that value unity gain is only @ 700k.
With the cascodes it stays stable to 33pf and does a 50k SW perfectly. The linearity of the vas is improved, especially
at HF. At this level unity gain is 1.15mhz.
In a third test I added the "magic resistor"
(47R)and used a 27pf for Cdom by lowering OLG a little.
So the effect of the cascode seems to be quite beneficial
electrically , as far as sound , I guess I'll have to build it.
In conclusion, as I have simmed all the simple
(30 transistors and under)DIY amps ,this
one beats them all except for the krill and roenders amp
(mjl's abomination) the latter is also cascoded and
shows the same gain plots as FA3..
Update.. I saw no real improvement with CFP inputs ..so
out they go..(2 fewer trannies)
22 more hours here..but same to you..
OS
As promised to Danielwritesback (I keep my promises), here is
the FA1 COMPLETE DIY package.
Anything I do here from now on will be complete, WTF..
if you can only discuss it and not build it..it's not DIY,
just blah,blah,blah..
It contains LTspice .asc, models used (ksa/m/njl),PCB and
parts layout (pretty board),BOM.. All triple checked for
accuracy. http://71.203.202.83/pdf1/Electronics/Projects/Audio_amp/Frugalamp/
Go to my "WWW" to get individual "works" and updates..
(or any of my other TB of stuff..
Direct link to frugal folder.. file is FA1_full.zip..
a preview below..
OS
Happy- happy New Year to all
the FA1 COMPLETE DIY package.
Anything I do here from now on will be complete, WTF..
if you can only discuss it and not build it..it's not DIY,
just blah,blah,blah..
It contains LTspice .asc, models used (ksa/m/njl),PCB and
parts layout (pretty board),BOM.. All triple checked for
accuracy. http://71.203.202.83/pdf1/Electronics/Projects/Audio_amp/Frugalamp/
Go to my "WWW" to get individual "works" and updates..
(or any of my other TB of stuff..
Direct link to frugal folder.. file is FA1_full.zip..
a preview below..
OS
Happy- happy New Year to all
Attachments
Hi Os
Some thoughts for abominating this holman if I may, and what is possible. To show theres no hard feelings about my previous comment and that it was given in good faith and honesty and learned by experience, and as proof that some guys that blah blah a lot have built 100s and sold many more(I speak for myself here) can contribute a fair amount, I will give you a nice tip for the new year.
As your design has cascodes and you are going to use them, and leaving the circuit as simple as possible and that maybe you have already designed a pcb that you dont want to change, I recommend a two tranny substitution, this way the cascodes will be even more usefull, actually you cant do without them. Replace the vas trannies with mega high gain darlingtons, the highest you can find. Some darlingtons have hfe 20000 typical, the more gain the better. Zetex should have something useful thats easy to obtain. Darlingtons have high cob and most are low voltage and may seem useless here but a cascode addresses these issues. Now think what a darlington is going to do in the vas.
Note I havent tried this with this particular circuit but I think it should be of reasonable usefullness. You could construct the darlington with 2 trannies and fine tune it but its more complication. Although Im younger than most here I design more old school than simulator, but you could verify this within seconds with your sim.
Now watch all the cascoded darlington circuits spring up all over the place,
Happy new year to everyone
Some thoughts for abominating this holman if I may, and what is possible. To show theres no hard feelings about my previous comment and that it was given in good faith and honesty and learned by experience, and as proof that some guys that blah blah a lot have built 100s and sold many more(I speak for myself here) can contribute a fair amount, I will give you a nice tip for the new year.
As your design has cascodes and you are going to use them, and leaving the circuit as simple as possible and that maybe you have already designed a pcb that you dont want to change, I recommend a two tranny substitution, this way the cascodes will be even more usefull, actually you cant do without them. Replace the vas trannies with mega high gain darlingtons, the highest you can find. Some darlingtons have hfe 20000 typical, the more gain the better. Zetex should have something useful thats easy to obtain. Darlingtons have high cob and most are low voltage and may seem useless here but a cascode addresses these issues. Now think what a darlington is going to do in the vas.
Note I havent tried this with this particular circuit but I think it should be of reasonable usefullness. You could construct the darlington with 2 trannies and fine tune it but its more complication. Although Im younger than most here I design more old school than simulator, but you could verify this within seconds with your sim.
Now watch all the cascoded darlington circuits spring up all over the place,
Happy new year to everyone
Fa1 is only the "blameless amp" schoolbook for all "nuubs"
But yours is the useful "blah blah"!! I was talking about the esoteric
40 transistor schemas that have no pcb backup or anything
just a discussion/exercise in design...
That is brilliant.. you are right.. I already have subbed many
different devices for the "active" vas components and
the cascodes seem to moderate their differences, so
a darlington might give me better balance by reducing
load on LTP/CM..
I know , I see some of my reason "pop up" elsewhere..
OS
What does your sim say ???
The action of a darlington used like this in a vas is a little more complicated, you should think more what the effect of extra gain does in a vas. But Im a bad teacher for these things, andyc or glen could explain better, they much more on top of the theory than myself.
Excellent performance can be obtained by simple circuits, I have been able to succeed with some topologies without the 40 tranny error correction way. You must understand I cant place all my ideas here. But dont discard the blameless topology, my 12 yr version is very close in speed compaired to the fc100, below 120db thd, all thd after 3rd below 150db, transient analysis very good, a true blameless, well with respect to THD. Figures could be better but I make use of some fets and a mosfet which affects this but I prefer the sound with them. My blameless doesnt have cascodes, could try though, in the ltp. With just two more tannies in the vas a blameless can get wingssssss.
Please nobody email me about this, I will not tell, its so simple youll kick yourself for not having thought it out yourself. Be creative.
The action of a darlington used like this in a vas is a little more complicated, you should think more what the effect of extra gain does in a vas. But Im a bad teacher for these things, andyc or glen could explain better, they much more on top of the theory than myself.
Excellent performance can be obtained by simple circuits, I have been able to succeed with some topologies without the 40 tranny error correction way. You must understand I cant place all my ideas here. But dont discard the blameless topology, my 12 yr version is very close in speed compaired to the fc100, below 120db thd, all thd after 3rd below 150db, transient analysis very good, a true blameless, well with respect to THD. Figures could be better but I make use of some fets and a mosfet which affects this but I prefer the sound with them. My blameless doesnt have cascodes, could try though, in the ltp. With just two more tannies in the vas a blameless can get wingssssss.
Please nobody email me about this, I will not tell, its so simple youll kick yourself for not having thought it out yourself. Be creative.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.