Building a PP KT120 amplifier, 65 watts Class A, fixed bias (-43V), 450V B+ at 120mA idle each side. Earlier prototype monoblock has been working flawlessly since September, the other monoblock has undergone numerous revisions, mainly power supply tweaking, delay start and circuit protection. Prototypes are on pcb with the HT hard wired to the socket pins to avoid thermal/stress failure. Bias adjustment is by two trimmers: one sets bias voltage and the other balances with LED's sensing off a 10-ohm cathode resistor, with the cathode voltage fed to the bases of differential NPN's that have the LED's pulled through the collector, whose emitters are held steady through a common current sink. The ccs and the collector are off a half wave rectifier from the second secondary of a xfmr; the other secondary feeding the 80V bias circuit. There is a B+ time delay of one minute and the bias voltage is on the grid well before the B+ turns on.
Noticed one KT120 tube burned out on the newer monoblock, but the fuse intact. Replaced the tube, started amp and noticed one LED brighter than the other. After a few minutes, the replacement tube started glowing very brightly and the other tube's bias LED went dark... and immediately shut down power. After two tubes in the same socket, suspected bias voltage loss.
Removed the power tubes and started the amplifier with a variac to keep the B+ to 450V. Measured both pin 5 voltages with the probes in the sockets and adjusted the bias to -50V. The voltage held steady for over an hour. Plate and screen voltages remained constant at 450V during that time also.
Installed the tubes (swapped sockets to rule out tube). Placed a DMM probe on the cathode of the suspected socket and another DMM probe to the grid bias. Started the amplifier. Bias held steady at -50v and the cathode voltages each measured 0.90V (90mA). Corresponds with the plate curves. The tubes were balanced and operating for about three to four minutes when the tube cathode voltage in the problem socket climbed to over 2.00V and the bias dropped to -18V before I shut it down. Couldn't tell which came first, both voltage changes appeared to happen at the same time.
The bias was measured at the downstream lead of the coupling capacitor. Even if the grid pin at the pcb was faulty or lost contact from heat, shouldn't the voltage stay constant at the capacitor? The only other thing I can think of are the trimmer resistors -- they have been pulled from previous prototypes and reused, maybe five or six times.
Can anyone offer other things to check? This is my first build and learned quite a bit. But I'm stumped here. Appreciate any assistance. Thanks.
Noticed one KT120 tube burned out on the newer monoblock, but the fuse intact. Replaced the tube, started amp and noticed one LED brighter than the other. After a few minutes, the replacement tube started glowing very brightly and the other tube's bias LED went dark... and immediately shut down power. After two tubes in the same socket, suspected bias voltage loss.
Removed the power tubes and started the amplifier with a variac to keep the B+ to 450V. Measured both pin 5 voltages with the probes in the sockets and adjusted the bias to -50V. The voltage held steady for over an hour. Plate and screen voltages remained constant at 450V during that time also.
Installed the tubes (swapped sockets to rule out tube). Placed a DMM probe on the cathode of the suspected socket and another DMM probe to the grid bias. Started the amplifier. Bias held steady at -50v and the cathode voltages each measured 0.90V (90mA). Corresponds with the plate curves. The tubes were balanced and operating for about three to four minutes when the tube cathode voltage in the problem socket climbed to over 2.00V and the bias dropped to -18V before I shut it down. Couldn't tell which came first, both voltage changes appeared to happen at the same time.
The bias was measured at the downstream lead of the coupling capacitor. Even if the grid pin at the pcb was faulty or lost contact from heat, shouldn't the voltage stay constant at the capacitor? The only other thing I can think of are the trimmer resistors -- they have been pulled from previous prototypes and reused, maybe five or six times.
Can anyone offer other things to check? This is my first build and learned quite a bit. But I'm stumped here. Appreciate any assistance. Thanks.
w/o schematic we can only guess ...
KT120 demands control grid leak resistor to be less than 50 kohms in fixed bias;
much higher can start runaway, depending on tube specimen;
in which case grid current develops which loads down your bias setup causing it to drop which in turn causes more plate current causing more grid current a.s.o.
have in mind that the total series resistance counts ... including pot etc ....
Is the design original, or has it been proven?
Those are mighty serious valves for a first build! You're showing great courage.
Is the output transformer rated for Class A use? Is the screen voltage well-controlled; is it an 'ultra linear' design?
As Sorento mentioned, posting the schematic will probably be necessary to get significant help. Don't worry if it isn't 'pretty'. 😉
Regards
Those are mighty serious valves for a first build! You're showing great courage.
Is the output transformer rated for Class A use? Is the screen voltage well-controlled; is it an 'ultra linear' design?
As Sorento mentioned, posting the schematic will probably be necessary to get significant help. Don't worry if it isn't 'pretty'. 😉
Regards
The schematic for the bias and balance:
Bias measured at the test probe and cathode voltage at the test port common at the spdt switch
Bias measured at the test probe and cathode voltage at the test port common at the spdt switch
Rick:
The transformer is by Electra Print. I gave Jack the specs for Zaa 6,000 ohms, UL 40% Class A operation with 300mA max, 65+ watts and a 3-ohm secondary. My speakers (Thiel 3.6) have a steady 2.5-ohm impedance for most of the spectrum so the OPT sees a 5k inductive load. Used the load line calculator at vtadiy for confirmation.
The design is sort of original, I guess. I've got info from Turner Audio and I saw other schematics online. I wanted the least amount of gain stages so I used a 6SL7 in a totem pole to swing 50 Vpp from a 2 volt signal to drive the KT120's (and maybe later KT150's) . I wanted fixed bias but to use it without an inverter meant that it's balanced input only. The power transformer is an Edcor 360-360 450mA with a 10.9V @4A for the heaters. Using a 555 timer to delay the B+, and during the delay I use a multivibrator to flash the LED through a normally closed optocoupler. The 12.6V is routed to the 555 and the mosfet optocouplers. When the 555 times off it grounds the 12.6V and sends voltage to close the B+ optocoupler (CPC1979J) and to open the flash optocoupler, handing over the led's to the cathode voltage.
I've attached a schematic of the whole thing...
The transformer is by Electra Print. I gave Jack the specs for Zaa 6,000 ohms, UL 40% Class A operation with 300mA max, 65+ watts and a 3-ohm secondary. My speakers (Thiel 3.6) have a steady 2.5-ohm impedance for most of the spectrum so the OPT sees a 5k inductive load. Used the load line calculator at vtadiy for confirmation.
The design is sort of original, I guess. I've got info from Turner Audio and I saw other schematics online. I wanted the least amount of gain stages so I used a 6SL7 in a totem pole to swing 50 Vpp from a 2 volt signal to drive the KT120's (and maybe later KT150's) . I wanted fixed bias but to use it without an inverter meant that it's balanced input only. The power transformer is an Edcor 360-360 450mA with a 10.9V @4A for the heaters. Using a 555 timer to delay the B+, and during the delay I use a multivibrator to flash the LED through a normally closed optocoupler. The 12.6V is routed to the 555 and the mosfet optocouplers. When the 555 times off it grounds the 12.6V and sends voltage to close the B+ optocoupler (CPC1979J) and to open the flash optocoupler, handing over the led's to the cathode voltage.
I've attached a schematic of the whole thing...
To calculate the 200K, is it the 150 plus the 47 kohm resistors? If so, then my mistake was assuming they were in parallel with the stopper.
Thank you...
I assume the other amp running with no problem is just luck and will eventually suffer the same fate.
Thank you...
I assume the other amp running with no problem is just luck and will eventually suffer the same fate.
The grid stopper is the 2.2k which is in series with the 150k which is again in series with all the bias gear.
Yes, you have to fix them all.
Have in mind that when you lower the grid leak resistors that you will loose some bass response, unless you increase the coupling capacitors.
They are now 0.33uF which form a low pass filter with the 200 kohm grid leaks, which gives a -3dB point at 2.4 Hz.
This will rise to 9.6 Hz with 50 kohm.
Might still be OK because the OPT is probably still limiting bass response.
Otherwise you could try 1uF ...
Was trying to check if the drive circuit was OK to drive 47K. However some bits of the drive tube seem unconnected.
The driver seems to be a cathode follower , so should work with 50K
Just 150K without what is in series 47K in parallel with pots and so on , is too high for many much weaker output tubes .
Just 150K without what is in series 47K in parallel with pots and so on , is too high for many much weaker output tubes .
OK the voltage across the 150K should indicated grid current draw. The cathode follower can also run out of current too but more abruptly. Could not see the value of the resistor in the schematic maybe the windows pdf reader is not drawing it all.
Last edited:
Wrap the bias back to the grid of the cathode follower, move the cap, return cathode to the most negative supply you have. You need a few more volts of bias but the cathode follower grid resistor can be >500K easily.
Ampeg SVT (6550) 1971 is a VERY well-proven huge amplifier with power tube(s) bias injected at a cathode follower. Your monster amp is less heroic than this. You won't need the beefy 12BH7 as CFs, 12AU7 is fine. R16 R19 150k series grid resistor is for OVER-drive (yes, players slammed these 300++ Watt heads back in the day) and 20k-50 will be fine for good music.
https://ampeg.com/support/files/Schematics/S Series/SVT (1972, 6550 tube)/SVT 1971 6550 Schematics.pdf
Ampeg SVT (6550) 1971 is a VERY well-proven huge amplifier with power tube(s) bias injected at a cathode follower. Your monster amp is less heroic than this. You won't need the beefy 12BH7 as CFs, 12AU7 is fine. R16 R19 150k series grid resistor is for OVER-drive (yes, players slammed these 300++ Watt heads back in the day) and 20k-50 will be fine for good music.
https://ampeg.com/support/files/Schematics/S Series/SVT (1972, 6550 tube)/SVT 1971 6550 Schematics.pdf
The schematic shows a broken connection from the cathode to the coupling cap because I am hard wiring the signal path from that point. All other tracks are pcb.
I'm attaching the LTSpice schematic for the input (SRPP) stage with a 6SL7 (rp 44k, mu 70). I changed the bias circuit per suggestions above... using 47K leak, 220R stopper and 5k half of balance pot and 2.5K half of bias set pot, total 58K grid to cathode resistance. It appears the Zout of the SRPP is on the order of 20K, which is about half the grid circuit load.
Good work so far. But there's a point that still hasn't been addressed . .
Input imbalance.
You may not be ready to consider this, since you want to minimize stages . . and every designer I know IS CERTAIN that no unbalanced input will ever be applied.
But if it is, a toroidal output transformer (if present) may suffer permanent damage. It might just be me, but I always encourage designers to convert Balanced-to-SE-and-back if There's Any Possibility Whatsoever that a signal might be applied to one line, absent the other.
Cheers
Input imbalance.
You may not be ready to consider this, since you want to minimize stages . . and every designer I know IS CERTAIN that no unbalanced input will ever be applied.
But if it is, a toroidal output transformer (if present) may suffer permanent damage. It might just be me, but I always encourage designers to convert Balanced-to-SE-and-back if There's Any Possibility Whatsoever that a signal might be applied to one line, absent the other.
Cheers
Or just used with less-than-perfect equipment. I’ve got a few Behringer “Pro” pieces whose XLR outs show unequal amplitudes at the hot and cold. Balanced outs my ***. Transformer input coupling solves it (as well as the ground loop that “balanced” connections were intended to solve).
Balanced input stages should be designed with a high CMRR (CCS bias, etc) so as not to propagate problems though.
Balanced input stages should be designed with a high CMRR (CCS bias, etc) so as not to propagate problems though.
So I think since its AC coupling only the bias can cause a DC imbalance in the OPT. Is there such a thing as a balanced SRPP where the bottom part is a LTP.
Maybe I was thinking of one of the AB classes -- or was just generally confused. 😱 (likely)
Your 'balanced SRPP' question is above my pay grade -- maybe wg_ski can answer that one; I don't know.
Regards
Your 'balanced SRPP' question is above my pay grade -- maybe wg_ski can answer that one; I don't know.
Regards
Yeah, if it is AB you would get DC in the core if you ended up driving one side harder than the other. In principle this wouldn‘t happen in class A - BUT class A isn’t perfect. There will be some 2 HD, therefore some DC shift in the driven side, while the undriven side stays at quiescent. Not enough to upset an EI core, but a toroid maybe.
I dont see any reason you couldn’t connect a pair of SRPPs to a CCS and run it differential. They do transistor amps and ICs with long tails and active loads all the time. Might need some degeneration to enforce the match - which would give the ability to trim offsets and AC balance.
I dont see any reason you couldn’t connect a pair of SRPPs to a CCS and run it differential. They do transistor amps and ICs with long tails and active loads all the time. Might need some degeneration to enforce the match - which would give the ability to trim offsets and AC balance.