Lumba Ogir said:
Thank you.
Dis you read all posts in that thread? There are no so many ...
http://www.diyaudio.com/forums/showthread.php?postid=1146544#post1146544
Lumba,
You and I appear to hold similar views; however, don't underestimate either Pete or Mihai - I have watched with astonishment how quickly Pete has picked this stuff up through his simulations, far outstripping me..... and I have corresponded with Mihai and know his mathematical insights to be profound and accurate.
Loading down the VAS to 'swamp' the output stage drive variations is not such a bad idea, and it is a clever way to minimise the lag compensation even if it does throw away some of the OLG.
I'm not contributing here, but I have designed some very good sounding amps, and I salute Peter's efforts. Mind you, he should come to me for a decent pcb layout, but that's another matter.
Hugh
You and I appear to hold similar views; however, don't underestimate either Pete or Mihai - I have watched with astonishment how quickly Pete has picked this stuff up through his simulations, far outstripping me..... and I have corresponded with Mihai and know his mathematical insights to be profound and accurate.
Loading down the VAS to 'swamp' the output stage drive variations is not such a bad idea, and it is a clever way to minimise the lag compensation even if it does throw away some of the OLG.
I'm not contributing here, but I have designed some very good sounding amps, and I salute Peter's efforts. Mind you, he should come to me for a decent pcb layout, but that's another matter.
Hugh
Yes , it might be, but a reversable one
so on the real thing, I will try with and
without.
I have no idea how they would sound in real life,but
on LTspice 4 they totally "choke" amp off >2mhz and change
the unity gain plot.. so I assume the real world would reflect this.
ahh , one I didn't know about ....will read ,thanks
OS
BTW.. Since you guys have experience with this topology,do
you consider it unstable. ? I see no evidence to support this
at least in simulation.
And boy do I have enough OLG... almost 9 Db more than FA1
Thanks for comment, totally opposite "you know who" as he
talks down to me but only drives me further onward.(bootstrap thread)
he finally got to me.. I will not allow "threadstalking"here anymore and will boldly oppose it..
I would welcome that, hugh.. but I have no printer . All of doug selfs general layout rules including
NFB takeoff, grounding, short traces between stages,separated
OP and input stages,etc. (both
yours and mine break rules with rails on opposite sides
) I could still make minor changes now (as i have not etched them yet) short of a major overhaul..
BTW - All those jumpers in OPS at top of board form 18ga solid
wire "rails" and save on a lot of foil (sony/sanyo HT receiver "trick")allows for 4-way use of board (sandwich.90 and 180,or as driver board)
An externally hosted image should be here but it was not working when we last tested it.
OS
ostripper said:
Sometimes something is right in front of me, and I still don't see it...
Thanks.
ostripper said:
BTW, I had no problem modeling the symasym after giving it
matched Cdoms (same as mine, elliots has no Cdom)
just to see how/why it worked.
OS
I tried the second Miller cap (100pF) on my sim, but it didn't completely stabilize it.
Does LT have a real time o-scope? If so, probe the output from the LTP into the VAS. Here's what I see for the stock Symasym (on top) and the most well behaved amp that I have modeled so far - the RMI by roender. This is with both amps handling a 1k squarewave.
Attachments
Hello ostripper.
I would not change very much from this schematic
from December 1
http://i302.photobucket.com/albums/nn112/ostripper/frugal/FA2bestsim.jpg
Have you thought about using a separate Transformer/supply for the Output Stage. = for the drivers+power transistors?
You know, simulations uses ideal voltage sources.
In real life there are no such things
VAS and input stage should benefit a lot from there own 'clean supply'.
Preferably like ~5 Volt higher than output
-----------------------------------------------
I am not sure about the 'base stopper' resistors.
Should they be at BASE of the output transistors? Or like now 47 Ohms at drivers?
-----------------------------------------------
About the 68k R11/R12 resistors.
I think you should experiment with values. In the real amplifier.
Changing these in values can make a lot of difference!
There is a sweet spot to find ....
-----------------------------------------------
About Q6, the cascode in VAS.
Here is an old idea I posted to quasi in January 2006.
Even if you are using a current mirror, it may be worth a try to connect the cascoding transistor Q6 to follow the voltage of the Vbe multiplier.
(Of course the vas transistors have to take the Vce in question)
Let's see what you find out, ostripper
Lineup Regards
http://www.diyaudio.com/forums/attachment.php?s=&postid=810128&stamp=1136798176
Lineup Idea post:
http://www.diyaudio.com/forums/showthread.php?postid=810128#post810128
I would not change very much from this schematic
from December 1
http://i302.photobucket.com/albums/nn112/ostripper/frugal/FA2bestsim.jpg
Have you thought about using a separate Transformer/supply for the Output Stage. = for the drivers+power transistors?
You know, simulations uses ideal voltage sources.
In real life there are no such things
VAS and input stage should benefit a lot from there own 'clean supply'.
Preferably like ~5 Volt higher than output
-----------------------------------------------
I am not sure about the 'base stopper' resistors.
Should they be at BASE of the output transistors? Or like now 47 Ohms at drivers?
-----------------------------------------------
About the 68k R11/R12 resistors.
I think you should experiment with values. In the real amplifier.
Changing these in values can make a lot of difference!
There is a sweet spot to find ....
-----------------------------------------------
About Q6, the cascode in VAS.
Here is an old idea I posted to quasi in January 2006.
Even if you are using a current mirror, it may be worth a try to connect the cascoding transistor Q6 to follow the voltage of the Vbe multiplier.
(Of course the vas transistors have to take the Vce in question)
Let's see what you find out, ostripper
Lineup
Regardslineup said:
Workhorse said:
Lineup Idea schematic:MikeB said:
http://www.diyaudio.com/forums/attachment.php?s=&postid=810128&stamp=1136798176
Lineup Idea post:
http://www.diyaudio.com/forums/showthread.php?postid=810128#post810128
Sqauredance (nice)
Interesting, as I never gave the original symasym a dirty
sqaurewave before moving on to mine ..just an idea
to improve upon.
Here is that LTP to VAS pulse from FA2, looks like roenders..
To get real ,I did the next shot at 20Khz/.25v input for this output..
[ "gross compensation" at VAS output seems to round corners
of square more ,looking like my bootstrap simulation]
Slightly rounded rising edge, with 22pf cdom it just started to overshoot, but at full 70vp-p only .5us (100+ v slew)
balanced throughout, real
low Cdom (33p,with 22p I crashed sim) , no other compensation. this amp seems to stay stable and gets faster (slew)with
just a unity gain shift (here at 1mhz)..
[dark blue -output /green-yellow..VAS/L.blue- purple LTP]
One last sim before I etch it,
Both halves of the LTP Re...very symmetrical , all frequencies
all output levels ,current levels are locked as well 1.8Ma LTP
8.5Ma VAS , I think these are the most linear regions in
my device choices (device Hfe specs confirm this).
besides the widlar buffer and MJL's 2SA/C devices.
is important. Also, I "spike" my voltage sources with 5vAC
(2 frequencies to make them out of phase )
when doing simulations.
OS
Interesting, as I never gave the original symasym a dirty
sqaurewave before moving on to mine ..just an idea
to improve upon.
Here is that LTP to VAS pulse from FA2, looks like roenders..
An externally hosted image should be here but it was not working when we last tested it.
To get real ,I did the next shot at 20Khz/.25v input for this output..
An externally hosted image should be here but it was not working when we last tested it.
[ "gross compensation" at VAS output seems to round corners
of square more ,looking like my bootstrap simulation]
Slightly rounded rising edge, with 22pf cdom it just started to overshoot, but at full 70vp-p only .5us (100+ v slew)
An externally hosted image should be here but it was not working when we last tested it.
balanced throughout, real
low Cdom (33p,with 22p I crashed sim) , no other compensation. this amp seems to stay stable and gets faster (slew)with
just a unity gain shift (here at 1mhz)..
An externally hosted image should be here but it was not working when we last tested it.
[dark blue -output /green-yellow..VAS/L.blue- purple LTP]
One last sim before I etch it,
An externally hosted image should be here but it was not working when we last tested it.
Both halves of the LTP Re...very symmetrical , all frequencies
all output levels ,current levels are locked as well 1.8Ma LTP
8.5Ma VAS , I think these are the most linear regions in
my device choices (device Hfe specs confirm this).
about Q6 , its addition is what really made this VAS "tight",
besides the widlar buffer and MJL's 2SA/C devices.
thats why a good CCS/w those extra caps (low PSRR)
is important. Also, I "spike" my voltage sources with 5vAC
(2 frequencies to make them out of phase )
when doing simulations.
OS
Hi OS,
I complement you on a fine job here. I have taken the time to sim this and WOW it looks good.
It is (according to my sim) head and shoulders over the Symasym - performance and stability. THD is certainly low enough, with the 2nd stronger than the 3rd, with the higher order out of the picture (though to give roender credit, the 3rd and especially the 5th order harmonics jumped up with your "gross compensation" - might want to chop that).
I'm tempted to do a layout for this for myself. Having spent the time to do roender's RMI for my own use (reception to my "tampering" with his design and layout has been, shall we say, "cold" ), and seeing performance from yours on a par with his much more complicated amp, it's probably worth my time.
Excellent work.
I complement you on a fine job here. I have taken the time to sim this and WOW it looks good.
It is (according to my sim) head and shoulders over the Symasym - performance and stability. THD is certainly low enough, with the 2nd stronger than the 3rd, with the higher order out of the picture (though to give roender credit, the 3rd and especially the 5th order harmonics jumped up with your "gross compensation" - might want to chop that).
I'm tempted to do a layout for this for myself. Having spent the time to do roender's RMI for my own use (reception to my "tampering" with his design and layout has been, shall we say, "cold"
), and seeing performance from yours on a par with his much more complicated amp, it's probably worth my time.Excellent work.
The one I use is 32% (on the label).
It will warm up on it's own when it starts to work.
I have done traces as thin as .1mm so undercut isn't a problem. Just don't leave it any longer than you need to.
It takes 5 minutes or so, depends upon how much solution you have and the size of the board/thickness of copper.
I read that you don't have a laser printer. If you want to send a layout file in pdf to me I'd gladly print out a sheet for you and send it off.
It will warm up on it's own when it starts to work.
I have done traces as thin as .1mm so undercut isn't a problem. Just don't leave it any longer than you need to.
It takes 5 minutes or so, depends upon how much solution you have and the size of the board/thickness of copper.
I read that you don't have a laser printer. If you want to send a layout file in pdf to me I'd gladly print out a sheet for you and send it off.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.