OK , but you will have something that has been beat to death on this forum.
All great amps ,
http://www.diyaudio.com/forums/solid-state/154717-my-new-symasym-pcb-design-rev_1-3-a.html
OR...
http://www.diyaudio.com/forums/solid-state/97144-symasym-roender-style.html
Even in the original , mike B. proposed and elaborated on the FET/cascode mod.
In fact , read this thread...
http://www.diyaudio.com/forums/solid-state/171244-help-wanted-diy-audio-forum-guys-china-hong-kong-taiwan.html
This amp is reported to sound wonderful, laterals and all.
So basically the goldmund HAS been cloned , sold , and "beat to death" laterals and all.
My adaptation ,a beta enhanced VAS w/ current mirror has not been as thoroughly explored at DIYA. Nico and ampslab have first "dibs" on this , so we know it works. Why not "vanilla" , add your tube preamp or DSP and have a transparent reproduction of the soft clipping attributes enjoyed by previous technologies. Most of the Goldmund's magic is in it's deployment ... separate power supplies , anti-resonant construction , choice devices, to-3 laterals (we don't have them anymore) , and of course the $30USD 2n5565.
Already an expensive build even if ported to source-able, available parts , to capture the original goldmund might be a hard to reach goal for many DIY'ers.
With all this in mind , why not explore as DEEPLY as possible what may make this amp sound good, (best ??? - very subjective) objectively.
We know it needs a better power supply than your basic DIY build (more boards/money$$) , I feel the laterals are most of the "magic" (I KNOW that). Why not give it a superior input/voltage stage ???
OS
All great amps ,
http://www.diyaudio.com/forums/solid-state/154717-my-new-symasym-pcb-design-rev_1-3-a.html
OR...
http://www.diyaudio.com/forums/solid-state/97144-symasym-roender-style.html
Even in the original , mike B. proposed and elaborated on the FET/cascode mod.
In fact , read this thread...
http://www.diyaudio.com/forums/solid-state/171244-help-wanted-diy-audio-forum-guys-china-hong-kong-taiwan.html
This amp is reported to sound wonderful, laterals and all.
So basically the goldmund HAS been cloned , sold , and "beat to death" laterals and all.
My adaptation ,a beta enhanced VAS w/ current mirror has not been as thoroughly explored at DIYA. Nico and ampslab have first "dibs" on this , so we know it works. Why not "vanilla" , add your tube preamp or DSP and have a transparent reproduction of the soft clipping attributes enjoyed by previous technologies. Most of the Goldmund's magic is in it's deployment ... separate power supplies , anti-resonant construction , choice devices, to-3 laterals (we don't have them anymore) , and of course the $30USD 2n5565.
Already an expensive build even if ported to source-able, available parts , to capture the original goldmund might be a hard to reach goal for many DIY'ers.
With all this in mind , why not explore as DEEPLY as possible what may make this amp sound good, (best ??? - very subjective) objectively.
We know it needs a better power supply than your basic DIY build (more boards/money$$) , I feel the laterals are most of the "magic" (I KNOW that). Why not give it a superior input/voltage stage ???
OS
A lateral MOSFET will conduct less as the temperature rises. (negative coefficient) That is why they are so easy to implement. (No Vbe - less parts). it's BJT driver will do the opposite .. keeping equilibrium (stable at all temps)
That worked well in that Pioneer M-90 clone .. huh? THD20 did rise there , too. In that case , there it was just a second "tandem" differential with stability problems , not the VAS itself.
OS
Last edited:
An experience I am sure other may share. At a site where a hundred or so African workers were digging trenches, they sang while they toiled.
These were not professional artists, they have no musical training nor practiced the song before, only simple laborers. The lyrics was maybe five or so words but they sang in absolute harmony and perfect rhythm while digging.
Hearing this mde the hair on my neck rise, gave me goosebumps and a slug in my throat. I have not in all my audio years, ever heard any recording chain, regardless of complexity, power, weight or cost to reproduce this sound or emotional experience - not even close.
But there is al lot of fun in trying
Here, here! I totally agree... There are a lot of moments like that that I have experienced, but it doesn't necessarily have to be "perfect rhythm and harmony" to make that impact. It's a combination of the setting, the company (people you're sharing the experience with), the "story" behind the story, the smell of the air, the landscape back drops, the intensity and emotion that the "performer" is putting into the music that you're witnessing in real life... they all add up to something that absolutely can't be replicated. Our bodies and minds are incredibly complex and astounding to me. Carpe diem.
Salas, I could. Could you remind me of the scheme with a schematic or example?
The problem I see is that the frontend sees a mainly capacitive load. Both types of compensation, lead compensation and miller compensation (and including TMC I believe), only respond to voltage change. If there is no voltage change, as is true for a capacitive load at HF, the compensation does not respond and the full OLG is dumped into the capacitive load. The type of compensation I just employed responds to current changes, not voltage changes, and so is more effective against capacitive loads.
- keantoken
The problem I see is that the frontend sees a mainly capacitive load. Both types of compensation, lead compensation and miller compensation (and including TMC I believe), only respond to voltage change. If there is no voltage change, as is true for a capacitive load at HF, the compensation does not respond and the full OLG is dumped into the capacitive load. The type of compensation I just employed responds to current changes, not voltage changes, and so is more effective against capacitive loads.
- keantoken
The guy recognised the superiority of the transparent amp, he just would not care as much about information as for approaching the hall experience. By knowing him being an opera goer my understanding is that transparency to recordings is bringing them to stark relief, that overloads him. When the tube amp could just mask the edges -along with hiding some info- and bring the tonal finesse to the fore. You don't get so much info in halls, mostly tone you get. Recordings and acoustics are the main manipulation.
You could hear the miking and artificial reverb splashing like when wearing AKG 1000s through the open back planar speakers with the hi tec amp in studio Jazz and modern. As a whole some look for a total system that always reminds music, others endure the recording's nature. Me I prefer the truth. Alex Jones bullhorn outing kind of an amp. LOL.
I know some are disappointed in my decision to work with the original design. Calm down! Sheesh! Once we tackle this, other directions seem interesting.
There has to be other things we can do, concerning layout and implementation rather than a "blind" schematic. When you have an oscillating FET amp, what do you do first? Surely not dive right in and remodel the amplifier's compensation! Do you use ferrite beads? Where?
- keantoken
There has to be other things we can do, concerning layout and implementation rather than a "blind" schematic. When you have an oscillating FET amp, what do you do first? Surely not dive right in and remodel the amplifier's compensation! Do you use ferrite beads? Where?
- keantoken
Around post #615 ... http://www.diyaudio.com/forums/solid-state/169590-mongrel-supersym-ii-62.html
OS
Okay, TMC could work, but I don't know how to figure out the values, except by the babylonian approach.
OS, on this version both lead compensation and Miller compensation were increased by the same amount. For the lead compensation, it should probably be the same. After a point, increasing it won't help, and may in fact make things worse. Phase shift definitely decreases.
We should probably increase Miller compensation, and find the "sweet spot" for lead compensation.
Also, after discovering the original rails are less voltage, we can probably use the original transistors too.
OS, the compensation idea works the same either way. Point is that the RC senses the current output of the following stages. And I don't think it was your VAS that was unstable, but the output stage throwing the frontend off.
- keantoken
OS, on this version both lead compensation and Miller compensation were increased by the same amount. For the lead compensation, it should probably be the same. After a point, increasing it won't help, and may in fact make things worse. Phase shift definitely decreases.
We should probably increase Miller compensation, and find the "sweet spot" for lead compensation.
Also, after discovering the original rails are less voltage, we can probably use the original transistors too.
OS, the compensation idea works the same either way. Point is that the RC senses the current output of the following stages. And I don't think it was your VAS that was unstable, but the output stage throwing the frontend off.
- keantoken
post311.
What does Q9 give us? Is there temp compensation or higher PSRR or ?
The 4pair paralleling of the drivers is going to make for set up difficulties.
The output bias will depend on individual Vbe of driver + Vgs of FET + tolerance of 330r.
A common driver pair will remove the Vbe variation.
What does Q9 give us? Is there temp compensation or higher PSRR or ?
The 4pair paralleling of the drivers is going to make for set up difficulties.
The output bias will depend on individual Vbe of driver + Vgs of FET + tolerance of 330r.
A common driver pair will remove the Vbe variation.
Last edited:
at what SPL?
At average listening SPL, speakers can be down around 0.01% to 0.1% H2 & H3.
H4 and higher almost unmeasurable.
It gives us a very boring CCS. Since the Vbe tempcos cancel, the Zener tempco is the only worry. PSRR and Zout are much worse compared to the 2-Q version I use. Only advantage may be that it's an extremely inert CCS, with no feedback loops which may give trouble.
- keantoken
or simply remove Q9 to leave a Vref applying voltage over the fixed resistor+Vbe.
I run my personal amp in bridge mode on a regulated +-45V psu.
Non-bridged you can use +-65V providing about 200 watt into 8 ohm.