Symasym HighPower, it's working and it is stable!
I had to make some modification in order to make it work:
1. I dropped the 2sk246 as cascode jfet in LTP because all my jfets are Y grade with only 200mV Vgs at 2ma LTP tail current. I put in that position 2n5551 and i added 12 volt zener diode
loaded by 1mA CCS in common base of cascode devices. All BJT's used as cascode has base stoppers in order to prevent possible HF oscillations.
2. I had to increase C8 and C9 in VAS from 100p to 220p and to decrease C15 in feedback network to 6p8, to stop some random oscillation.
3. I put two antiparalel 1n4148 diodes at the output of LTP for preventing heavy VAS saturation on clipping
4. Iq is set at 140mA (70mA on each output peer)
Some characteristics:
- very powerful compared with original symasym, ease to increase the number of output BJT's and rail voltages to gain more power.
- stable in occasional 2ohm load
- very stable bias current
- with input in short circuit, only -0.1mV Vout and this DC voltage is very stable with temperature
- no turn on/off thump
- very silent, even with input not shorted to ground
- if any fuse blow, the output will drop at near zero DC ( suitable for crowbar DC protection)
How it sound?
It is very pleasant surprise to be more dynamic than symasym and the soundstage is huge. It sound very uncompressed, with more clarity and volume in mid range and blade sharp but not
annoying trebles.
Bass is very very tight but not so deep, my first impression is that it is a little under symasym at this chapter, an "audiophile bass", well suitable for jazz and ambiental music.
In my opinion, it is a good amplifier and I will invite you my friends to build it.
ps. I will put later some pics and pcb EAGLE files
pps. to moderators ... maybe it's a good idea to make a separate thread?
I had to make some modification in order to make it work:
1. I dropped the 2sk246 as cascode jfet in LTP because all my jfets are Y grade with only 200mV Vgs at 2ma LTP tail current. I put in that position 2n5551 and i added 12 volt zener diode
loaded by 1mA CCS in common base of cascode devices. All BJT's used as cascode has base stoppers in order to prevent possible HF oscillations.
2. I had to increase C8 and C9 in VAS from 100p to 220p and to decrease C15 in feedback network to 6p8, to stop some random oscillation.
3. I put two antiparalel 1n4148 diodes at the output of LTP for preventing heavy VAS saturation on clipping
4. Iq is set at 140mA (70mA on each output peer)
Some characteristics:
- very powerful compared with original symasym, ease to increase the number of output BJT's and rail voltages to gain more power.
- stable in occasional 2ohm load
- very stable bias current
- with input in short circuit, only -0.1mV Vout and this DC voltage is very stable with temperature
- no turn on/off thump
- very silent, even with input not shorted to ground
- if any fuse blow, the output will drop at near zero DC ( suitable for crowbar DC protection)
How it sound?
It is very pleasant surprise to be more dynamic than symasym and the soundstage is huge. It sound very uncompressed, with more clarity and volume in mid range and blade sharp but not
annoying trebles.
Bass is very very tight but not so deep, my first impression is that it is a little under symasym at this chapter, an "audiophile bass", well suitable for jazz and ambiental music.
In my opinion, it is a good amplifier and I will invite you my friends to build it.
ps. I will put later some pics and pcb EAGLE files
pps. to moderators ... maybe it's a good idea to make a separate thread?
Hi Roender, nice job! I like it. 
Would you mind making a layout for normal MJLs? I have so many left...
Have you tried/evaluated the changes to each stage?
I'm quite sure that the 0.1mv DC is because of the modified current mirror. Do you have any experience about the effect of the cascodes?
I am happy to read that you were able to further improve the dynamics, i guess it's because of further improved psrr?
Good to see that you use a central power-gnd star. What color is the LED? IR?
Some will call that amp a complexity beast...
BTW, thanks for having done my job...
Mike
Would you mind making a layout for normal MJLs? I have so many left...
Have you tried/evaluated the changes to each stage?
I'm quite sure that the 0.1mv DC is because of the modified current mirror. Do you have any experience about the effect of the cascodes?
I am happy to read that you were able to further improve the dynamics, i guess it's because of further improved psrr?
Good to see that you use a central power-gnd star. What color is the LED? IR?
Some will call that amp a complexity beast...
BTW, thanks for having done my job...
Mike
MiiB said:
MiiB, my comments...
You really shouldn't use a current mirror in 1st stage, it defeats symasym's balanced operation. Symasym should have enough openloopgain without. A 3-bjt wilson current mirror in 2nd stage should be sufficient.
Feeding a mosfet outputstage without drivers from a low current to92 vas is perfect for ... subwooferamps, not fullrange amp.
But in this case i would use the much cheaper IRFPs.
But... just my thoughts...
roender said:
Gimme more! I like pictures...
Scoped squarewaveresponse will be interesting!
Mike
Hi
I just cut the two 100k resisotors from VAS output to ground. After 2 hours of listening I can say that this is a major sonically improvement. Sound more relaxed and fluid/natural
It is a very strange thing that this mod do not apply to original symasym (47k load resistors)
I will simulate the new situation as soon as possible.
I just cut the two 100k resisotors from VAS output to ground. After 2 hours of listening I can say that this is a major sonically improvement. Sound more relaxed and fluid/natural
It is a very strange thing that this mod do not apply to original symasym (47k load resistors)
I will simulate the new situation as soon as possible.
Hi, Roender,
What is the effect if you split R18(68ohm) into 2 resistors, each 136ohm, one for emitor of Q8 and one for emitor of Q9 (splitting the differential VAS)?
I see this change in one amp, the earlier model uses differential VAS, but in the later model, they split the VAS (not forming differential VAS anymore). What is the effect?
What is the effect if you split R18(68ohm) into 2 resistors, each 136ohm, one for emitor of Q8 and one for emitor of Q9 (splitting the differential VAS)?
I see this change in one amp, the earlier model uses differential VAS, but in the later model, they split the VAS (not forming differential VAS anymore). What is the effect?
AAK said:
In original symasym there is no difference with or without 47k's because it has huge masking 2nd harmonic which cover all the nasty things introduced by this resistors.
In my version, the output stage have very high input impedance and any supplementary VAS load is well shown in simulation.
I will upload some pics (FFT-THD20) from simulator to pruve this
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.