Yes. I have installed bracket for a large parallel capacitor for the PSU - which would bring it close to 1,000uF in total.
Then I read on this forum a lot of experts warning against such action.
I'm no guru, so I decided not to proceed with increased PSU capacitance.
I think I may have badly miscalculated the 21mA on the second 6SN7 because I overlooked the fact that it is two double triodes, served by the same shared cathode resistor.
In any event, moving cathode resistor R6 to 12K and load resistors R9/R10 to 22K seems to have done the trick. There is no longer any nasty hardness and dryness to the sound.
Given that cathode resistor R6 was factory set at 18K and load resistors were 30K - originally optimized for 6SL7/6SN7 with 100K of GNFB - I do wonder what I have achieved electronically.
Is it simply a case of running the second 6SN7 tubes a bit hotter, to gain greater linearity?
As far as plate voltages go, the second 6SN7 is now running with slightly higher voltage (now ~130V) than the original factory setting, but the first 6SN7 now operate at less than half the plate voltage (now ~50V) compared to the 6SL7 tube it replaced - which was ~115V.
Would it now be prudent to reduce the R2 and R3 cathode resistor on the first 6SN7 to bring the plate voltage back to ~115V? Or is this unnecessary?
i am no guru myself, but i like to overbuild things like my transformers and chokes and lots of caps in psu, i do not impose my ideas on others, i post only to inform that options exist, that they are free to diy as they like, just do not get electrocuted in the process...
recently i discovered CBB60 caps now available up to 100ufd@450vac they are good for kt88 amps, i will make an amp using them in future...https://datasheetspdf.com/pdf-file/647431/ETC/CBB60/1
in any common cathode amplifiers, the plate load resistors are there to determine voltage gains, remember that the Rp of the 6sn7(around 7K) the next stage grid leak resistors(maybe 220K) and the plate load resistors are all in parallel as far as ac goes, this being said, the voltage gain can never exceed half the mu of the tube, x11 in the case of the 6sn7 that has a mu of 20..
it is the plate voltage that determined output voltage swings, either you increase the B+ supply, or lower the plate load resistors, you choose...
Last edited:
you can try, do it in small incremental steps...the kt88 grid is biased at around 40 volts negative wrt cathodes, so a 115vdc plate on the second 6sn7 seemed plenty, but if i can i will add more to that plate voltage...
Last edited:
yes, those resistors are gm reduction resistors to an already moderate gm tube, i wonder what they were doing there in the first place...
if i were to build this amp, i would choose a Russian 6H23 in the first hole and Russian 6H30 in the second hole, 6sn7 to me seems whimpy, though linear, but overly hyped imho...
Last edited:
R6 is back to 18K.
R9 and R10 are now at 22K7.
The second 6SN7 now has ~260V on the plate.
It's sounding pretty good with no smoke or red-plating.
Here's the present numbers.
Still running in TRIODE ONLY with zero GNFB.
I have spent the last few years, trying a number of different operating points for our beloved BEWITCH 6550 push-pull, Class AB1 integrated amplifier - ever seeking to improve the (subjective) sound quality.
My conclusion is that it is simply NOT possible to configure the BEWITCH 6550 for optimum sound for both 6550 and KT88 output tubes.
It seems the two different - albeit very similar - tubes, require different operating points, in order to sound their very best.
Understanding this, I have settled on 6550 configuration, knowing that when my current (curvy) Russian 6550 tubes fail, some reconfiguration will be required to bring the new Shuguang KT88-98 tubes on-song, in this amp.
Through this process of experimentation, I have also tried:
Changing preamp tubes. 6SN7 v 6SL7 in both first and second holes, applying various amounts of cathode resistance and load resistance.
I have tried varying the GNFB and tried zero GNFB - with various tube combinations.
Trying different rectifier tubes from the factory (5Z4P), experimenting with (NOS) USSR 5U4C and (NOS) WESTINGHOUSE CANADA 5U4G; each measuring quite differently in terms of voltage output, thereby subtly impacting the sound.
In the end, I must concede that the designers of this BEWITCH 6550 knew precisely what they were doing.
That said, my final configuration serves to improve reliability, improve some of the parts quality and tailor the sound to both my own loudspeakers and my personal taste.
Our trusty LEBEN tube amp has served as a valuable source of comparison through this journey - not unlike taking a whiff of coffee beans between trying samples of different perfume - in an attempt to return to a meaningful baseline. It works!
We have always run a simple stepped attenuator upstream of this amp, so our BEWITCH 6550 is now no longer an integrated amplifier - but simplified to power amp status, with a single input.
Changes include:
PLUG, WIRE & JUMPERS:
Installed pure OFC copper INPUT RCAs.
Installed pure OFC copper INPUT WIRING and JUMPERS.
Removed factory selector switch and volume pot from circuit.
Upgraded to pure OFC copper FURUTECH IEC AC plug.
Installed FERRITE RING AC filter.
Removed TRIODE/UL. switch from circuit. (Now wired for Triode only)
NON-MAGNETIC RESISTORS:
All resistors replaced with:
4W VISHAY Wirewound. (PSU)
12W MILLS Wirewound. (GNFB, Cathode)
2W TDK Metal Film. (Other)
NON-MAGNETIC CAPACITORS:
Upgraded .47uF coupling capacitors using CORNELL-DUBILIER C940 Polypropylene.
Installed 2K2 bypass capacitors using CLARITYCAPS ESA Polypropylene.
Installed all .01uF bypass capacitors using CORNELL-DUBILIER C940 Polypropylene.
Installed 4.7nF capacitors using VISHAY Ceramic. (AC input filter)
Installed 470pF capacitors using generic Polystyrene. (IN4007 diode bypass)
Installed 9.5pF capacitors using generic Silver Mica. (GNFB bypass)
EL CAPACITORS:
Installed 4700uF bypass capacitors using NICHICON Aluminium Electrolytics.
RECTIFIER TUBES:
Upgraded to (NOS) WESTINGHOUSE 5UFG Black Plate tubes. (Circa 1940s)
AMPLIFIER FEET:
Upgraded to 04 x FINITE ELEMETE CERABALLS.
We are delighted with the outcome! This is a superb amplifier.
I provide the final schematics below, to help anyone else on a similar journey.
I'd like to thank the many experienced diyAudio members who faithfully supported my learning process. Appreciated. Special mention must go to @TonyTecson, who has been a constant and tireless source of knowledge, competence and inspiration.
Here's the final outcome - with measured values attached.
SONDEKNZ BEWITCH 6550 UPGRADED SCHEMATICS: August 2023.
My conclusion is that it is simply NOT possible to configure the BEWITCH 6550 for optimum sound for both 6550 and KT88 output tubes.
It seems the two different - albeit very similar - tubes, require different operating points, in order to sound their very best.
Understanding this, I have settled on 6550 configuration, knowing that when my current (curvy) Russian 6550 tubes fail, some reconfiguration will be required to bring the new Shuguang KT88-98 tubes on-song, in this amp.
Through this process of experimentation, I have also tried:
Changing preamp tubes. 6SN7 v 6SL7 in both first and second holes, applying various amounts of cathode resistance and load resistance.
I have tried varying the GNFB and tried zero GNFB - with various tube combinations.
Trying different rectifier tubes from the factory (5Z4P), experimenting with (NOS) USSR 5U4C and (NOS) WESTINGHOUSE CANADA 5U4G; each measuring quite differently in terms of voltage output, thereby subtly impacting the sound.
In the end, I must concede that the designers of this BEWITCH 6550 knew precisely what they were doing.
That said, my final configuration serves to improve reliability, improve some of the parts quality and tailor the sound to both my own loudspeakers and my personal taste.
Our trusty LEBEN tube amp has served as a valuable source of comparison through this journey - not unlike taking a whiff of coffee beans between trying samples of different perfume - in an attempt to return to a meaningful baseline. It works!
We have always run a simple stepped attenuator upstream of this amp, so our BEWITCH 6550 is now no longer an integrated amplifier - but simplified to power amp status, with a single input.
Changes include:
PLUG, WIRE & JUMPERS:
Installed pure OFC copper INPUT RCAs.
Installed pure OFC copper INPUT WIRING and JUMPERS.
Removed factory selector switch and volume pot from circuit.
Upgraded to pure OFC copper FURUTECH IEC AC plug.
Installed FERRITE RING AC filter.
Removed TRIODE/UL. switch from circuit. (Now wired for Triode only)
NON-MAGNETIC RESISTORS:
All resistors replaced with:
4W VISHAY Wirewound. (PSU)
12W MILLS Wirewound. (GNFB, Cathode)
2W TDK Metal Film. (Other)
NON-MAGNETIC CAPACITORS:
Upgraded .47uF coupling capacitors using CORNELL-DUBILIER C940 Polypropylene.
Installed 2K2 bypass capacitors using CLARITYCAPS ESA Polypropylene.
Installed all .01uF bypass capacitors using CORNELL-DUBILIER C940 Polypropylene.
Installed 4.7nF capacitors using VISHAY Ceramic. (AC input filter)
Installed 470pF capacitors using generic Polystyrene. (IN4007 diode bypass)
Installed 9.5pF capacitors using generic Silver Mica. (GNFB bypass)
EL CAPACITORS:
Installed 4700uF bypass capacitors using NICHICON Aluminium Electrolytics.
RECTIFIER TUBES:
Upgraded to (NOS) WESTINGHOUSE 5UFG Black Plate tubes. (Circa 1940s)
AMPLIFIER FEET:
Upgraded to 04 x FINITE ELEMETE CERABALLS.
We are delighted with the outcome! This is a superb amplifier.
I provide the final schematics below, to help anyone else on a similar journey.
I'd like to thank the many experienced diyAudio members who faithfully supported my learning process. Appreciated. Special mention must go to @TonyTecson, who has been a constant and tireless source of knowledge, competence and inspiration.
Here's the final outcome - with measured values attached.
SONDEKNZ BEWITCH 6550 UPGRADED SCHEMATICS: August 2023.
Last edited:
Thanks, you aren't quite done yet methinks, but good job indeed....
the last frontier imho would be to try out different OPT's say from EI's to C-Cores, and also beefier PTX, there are still differences to be discovered...even heavier PTX for psu stiffness...
up to today, i have not seen a side by side comparison between the EI and the C Core in the same amp...
anyway, amplifiers are designed to a certain price point, and i believe the ultimate test would be to come up with the best sounding amplifier to a fixed price...
how can a budget amplifier be made to sound like a "price no object amplifier"...now that would be a challenge and an adventore.....
the last frontier imho would be to try out different OPT's say from EI's to C-Cores, and also beefier PTX, there are still differences to be discovered...even heavier PTX for psu stiffness...
up to today, i have not seen a side by side comparison between the EI and the C Core in the same amp...
anyway, amplifiers are designed to a certain price point, and i believe the ultimate test would be to come up with the best sounding amplifier to a fixed price...
how can a budget amplifier be made to sound like a "price no object amplifier"...now that would be a challenge and an adventore.....
Thanks for the inspiration, Tony.
I have to confess that I have yet to learn much about OPTs and PTXs, so this might be a project for further down the track.
What I can say is that we are so delighted with the sound, it is hard to imagine we need anything more.
Famous last words, of course.
@TonyTecson
BTW Tony, I still have not forgotten our discussion about potentially running the BEWITCH 6550 using EL34 output tubes with zero GNFB - and (perhaps) getting better bass results from the present OPT, due to the reduced output power.
I have been studying the KT88 v EL34 TRIODE OPERATION datasheets and from my newbie perspective, it would appear that I could probably just drop a quad of EL34s into the present BEWITCH configuration - with no other changes - without too many concerns.
Obviously, things would change again, once the EL34s were in and GNFB was then removed.
Would you agree with my assessment?
If not, what have I missed?
Am I heading for disaster?
BTW Tony, I still have not forgotten our discussion about potentially running the BEWITCH 6550 using EL34 output tubes with zero GNFB - and (perhaps) getting better bass results from the present OPT, due to the reduced output power.
I have been studying the KT88 v EL34 TRIODE OPERATION datasheets and from my newbie perspective, it would appear that I could probably just drop a quad of EL34s into the present BEWITCH configuration - with no other changes - without too many concerns.
Obviously, things would change again, once the EL34s were in and GNFB was then removed.
Would you agree with my assessment?
If not, what have I missed?
Am I heading for disaster?
Last edited:
Thanks Chris.
I actually tried 100K Grid Leak/Stop resistors for V1a and the sound quality took a backward step.
(Not sure whether to describe them as Grid STOP or Grid LEAK, so I have attached a photo of the signal input, with said resistors in place...)
Everything sounded a bit rolled-off and overly smooth - so I removed them.
Do I need to spend big on low-noise resistors for this position?
I'm not sure what I risk, by leaving them out completely, to be honest...
Last edited:
Above photo shows a volume control not on the schematic. If that's immediately between the input jacks and the circuit as drawn you don't need another grid leak - the pot provides one. But I assure you, you can't hear or measure a 100K resistor in that location. A grid stop is a smaller resistor, maybe about 1K Ohm in series with the signal, mounted right at the valve socket. Suppresses RFI and helps stability. Can't hurt. The output valves should have grid stops also.
All good fortune,
Chris
All good fortune,
Chris
@Chris Hornbeck
Thanks Chris.
The ALPS volume pot is indeed completely out of circuit.
Not sure why I felt those 100K Grid Leak Resistors smoothed things out too much.
Here’s the full schematic with values,,,
What is the risk of leaving those Grid Leak Resistors out please?
Thanks Chris.
The ALPS volume pot is indeed completely out of circuit.
Not sure why I felt those 100K Grid Leak Resistors smoothed things out too much.
Here’s the full schematic with values,,,
What is the risk of leaving those Grid Leak Resistors out please?
An example of "grid leaks" in your amplifier is R11, R12 at the output stage. These set the output valves' grids at 0VDC, and without them the output valves would have no negative bias and would run away and self destruct. V1a isn't in danger without a grid leak, because its anode current is limited by its anode resistor R5, but it won't be properly biased without its grid at 0VDC.
Presumably whatever is driving this amplifier is providing a 0VDC "ground reference" so you're fine when amp is connected to source. But when disconnected, you can't test your amp for DC voltages etc. by itself. Why invite grief when the solution is so simple? (and universal - all amps have this). If worried about loading a very high impedance source, you can make the input grid leaks 1M Ohm and that'll work fine.
All good fortune,
Chris
Presumably whatever is driving this amplifier is providing a 0VDC "ground reference" so you're fine when amp is connected to source. But when disconnected, you can't test your amp for DC voltages etc. by itself. Why invite grief when the solution is so simple? (and universal - all amps have this). If worried about loading a very high impedance source, you can make the input grid leaks 1M Ohm and that'll work fine.
All good fortune,
Chris