Maybe another Goldmund killer
Hi all, Sorry Os for the intrusion. Here is my take for the Goldmund killer:
Note that the modest components choosed are what I have on hand here in Chile.
output power :60W (8 OHM load with a SPEAKER model)
output impedance: 0.001 OHM
Voltage Gain: 21 (3vpp in, 60,8Vpp out)
Slew Rate: ~ 200V/usec (rising & falling edge at full power)
THD's at full power
Frec. THD(%) Amplitude Group delay (nanosec)
400KHZ 0.02638 29.48 5.7
200KHZ 0.01185 30.89 5.4
100KHZ 0.00559 31.29 5.1
40KHZ 0.00211 31.39 5.0
20KHZ 0.00096 31.41 5.0
10KHZ 0.00056 31.42 5.0
4KHZ 0.00045 31.42 5.0
Attached the circuit and a fourier analysis at 10KHZ 13,57 Volts out, see the nice fade out of high harmonics content. I am sorry but I have only a .cir (text) file as I use a text mode simulator (ngsipce). DC servo not shown.
Cheers.
Hi all, Sorry Os for the intrusion. Here is my take for the Goldmund killer:
Note that the modest components choosed are what I have on hand here in Chile.
output power :60W (8 OHM load with a SPEAKER model)
output impedance: 0.001 OHM
Voltage Gain: 21 (3vpp in, 60,8Vpp out)
Slew Rate: ~ 200V/usec (rising & falling edge at full power)
THD's at full power
Frec. THD(%) Amplitude Group delay (nanosec)
400KHZ 0.02638 29.48 5.7
200KHZ 0.01185 30.89 5.4
100KHZ 0.00559 31.29 5.1
40KHZ 0.00211 31.39 5.0
20KHZ 0.00096 31.41 5.0
10KHZ 0.00056 31.42 5.0
4KHZ 0.00045 31.42 5.0
Attached the circuit and a fourier analysis at 10KHZ 13,57 Volts out, see the nice fade out of high harmonics content. I am sorry but I have only a .cir (text) file as I use a text mode simulator (ngsipce). DC servo not shown.
Cheers.
Thank you guys for the comments, this encourage me to take the next step, bring to life this little monster. The cascoded CLTP pairs, CVAS and the CB (complementary buffer) do the trick in simple and direct way, ... symmetry has it's goods. I totally agree with AKSA that the most important feature for human 'music quality' (lurking in his web site), is indeed how high order harmonics combine and fade out.
Cheers
Arturo
Cheers
Arturo
Much less than it seems ... 20 signal transistors transistor ..., 2 dependent current sources for 2 cascoded LTP, 1 complementary diff. VAS 2 complementary buffers with it's current sources and the output power follower, and that's all, no mystery. D. Self in his excellent book argues that a few more signal transistors and PCB area only represents a small percentage of the total cost, so it worth for a DIY project if it helps to achieve a blameless amp.
Cheers.
PS. I will not post here as it is Os thread, BTW was very inspiring his work, maybe we share the same perfection mania, my apologies Os.
Cheers
Cheers.
PS. I will not post here as it is Os thread, BTW was very inspiring his work, maybe we share the same perfection mania, my apologies Os.
Cheers
Complicated circuit ??.... That is simplicity city !!!
This or something similar will be the LX1.0 (designing and building now) hope I don't run out of devices first. I like your circuit, Artu. Perhaps a joint venture on the LX ??
I can go from your schema to a voltage board in hours.... do you mind ??? My only changes will be RED led CCS's , and what do Q12/14 do ??
OS
Last edited:
AX1.1 all done ... works great (with TMC)
I went the route of D. self... a "real blameless". Sounds nearly identical to the "blame ES" clone , just a little less bass "slam" at high volumes. I took a DPDT switch , using 1 pole to short out the big TMC cap and the other pole to open the FB resistor (Rtmc). (pix 1-2 - AX1.1 closeups)
TMC/no TMC sounded about the same at normal volumes , at moderate to high volumes , TMC had more clarity, depth ... the image deepened.. So , I don't know why the other "Camp" had issues with it . Actually , I do know why ... the ES has much less open loop gain than my AX/ BX.. (pix 3)
The BX , overall is the winner ... I went from 330-680R and 100/270 to 430p ,
I settled on 470R / 100 - 430p for all 4 amps. TMC was noticeable at high volume , and if I reduced the output stage bias. In NO WAY did the extra cap and FB resistor degrade the sound at ANY volume or in any of the combinations that I tried. Doug Self has essentially my AX on his desk and says he is excited ... WHO YOU GONNA BELIEVE !!!
On to my next creation , The LX (leech clone) ... Artu , I really like your design besides the IC servo
.... no cheeps here , all BJT yours beat mine (close to original leech ) by a factor of 7 (.0007 to .0001% - pix 4/5)
Mine is in the last attachment (pix6) ... let combine designs... I will build it (voltage stage).
OS
I went the route of D. self... a "real blameless". Sounds nearly identical to the "blame ES" clone , just a little less bass "slam" at high volumes. I took a DPDT switch , using 1 pole to short out the big TMC cap and the other pole to open the FB resistor (Rtmc). (pix 1-2 - AX1.1 closeups)
TMC/no TMC sounded about the same at normal volumes , at moderate to high volumes , TMC had more clarity, depth ... the image deepened.. So , I don't know why the other "Camp" had issues with it . Actually , I do know why ... the ES has much less open loop gain than my AX/ BX.. (pix 3)
The BX , overall is the winner ... I went from 330-680R and 100/270 to 430p ,
I settled on 470R / 100 - 430p for all 4 amps. TMC was noticeable at high volume , and if I reduced the output stage bias. In NO WAY did the extra cap and FB resistor degrade the sound at ANY volume or in any of the combinations that I tried. Doug Self has essentially my AX on his desk and says he is excited
... WHO YOU GONNA BELIEVE !!!On to my next creation , The LX (leech clone) ... Artu , I really like your design besides the IC servo
.... no cheeps here , all BJT yours beat mine (close to original leech ) by a factor of 7 (.0007 to .0001% - pix 4/5)Mine is in the last attachment (pix6) ... let combine designs... I will build it (voltage stage).
OS
Attachments
Hi ostripper,
I've been swapping back and forth between a "non-blameless" BAKSA and BAKSA TMC for the last couple of days and the results are too close to call given all tests have been done in mono and in the garage. My "impressions" are the TMC has a slightly better top end and maybe less sibilance. I was therefore very surprised to hear Carlos's results where the difference was very obvious and negative.
I believe that TMC is definitely worth pursuing and I will be testing in stereo on my main system were differences may be more obvious.
Points to note:
- It takes over 1 minute to change from TMC amp to non-TMC.
- The standard amp uses Mica caps and the TMC uses polystyrene caps.
I used a switch on the "test pig" above. I let the rails collapse , flipped the switch (right on the board real close to cdom) .. so maybe 3-4 seconds. My AX/BX's (after setting CCS adjust for exactly 6mA VAS) , can be entirely swapped in 30 seconds to the same OPS board. I think this speed was the only way to be able to audibly hear any differences in compensation.
I actually use a 100p poly@160 / 430p mica@ 300v on the AX's , I have noticed no major differences between them and the BX , which used mica's exclusively. Another myth - a PF is a PF !!
I simmed carlo's original "ES" , his operating points/degeneration combined with his devices gives roughly 85db OLG. My blameless based designs are up to over 120db. This is the bigger factor in the successful implementation of TMC. Over most of the audio band ,TMC's compensation is nearly the same as 2 caps in series (just like a single miller cap) ... hence, "transitional".
OS
This is the opposite :
Over the audio band , TMC include the OS in the miller loop.
It s only at high frequencies that it behave as a classical
miller compensation.
Increasing the VAS OLG decrease the frequency at wich the inclusion
of the OS in the miller loop start to have an effect...
Complicated circuit ??.... That is simplicity city !!!
OS
Compared to my circuits its really complicated!!
I've been thinking about trying dual LTP to drive both high and low side but im not sure if there are enough advantages and improvments in sound to be worth it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.