Tweet
Now that I've come to terms with the fact that my amp will never leave my bench, I've been wanting to experiment with changing the output section. I've been really unhappy with my EL84 set up in this amp. First, it's just plays really loud, and the high frequency component has been really difficult to tame.
However, after the results from subminiature amp project over the last year, I've become much more encouraged. In fact, I decided to tear my EL84 amp down to the chassis, and transformers and rebuild it using the same approach I took with the submini build.
Before:
After:
Haha! She was so pretty before, but it is so much quieter and stable in it's current ugly state. But I digress...
My first idea, was to install a pair of 5902 output tubes in addition to the EL84s low output option. My PT already had a second HT winding, and was gonna run them off that as a designated power supply. The challenges with that idea were how/where to mount the additional tubes, and they would require an additional output transformer. In addition to those, I was worried that I would loose some of the effects of excursion and power supply dynamics you get from overdriving a push pull amp by running the output tubes from a separate supply. But since I was unhappy with the original output section anyways, decided to look into what I needed to do to replace them with 6AQ5s
(For those who aren't familiar with 6AQ5s, they are 7-pin beam tetrode designed to operate as a 6V6 equivalent with a plate and screen voltage of 250V. They were more common to find in early gibson/supro/valco type amps)
The good thing was that the output transformer specs are shared closely between 6AQ5s and EL84s, so I could use the existing one. Obviously I would need to use a socket adapter or replace the 9-pin sockets with a pair of 7-pin.
The other challenge was that even though the output power was comparable, all the data sheets I could find specified that the max screen voltage and plate voltage for a 6AQ5 is 250V. My existing PT was spec'd to provide a B+ of 330V under an idle load. In reality, Heyboer ended up sending me a transformer that was about 4% off on both my HV secondary windings, so the B+ is between 317V-320V. Still, that's enough of a difference, where I'd seen modern amp companies design amps that cause output tube and output stage failures because of overvoltage conditions, particularly on the screens.
This gave me the opportunity to test an Idea I came up with for switching a while back. Rather than trying to drop the entire B+ by using any number of common techniques, I had some STP9NK90Z Mosfets with an RDS(on) resistance of <1Ω at @ 250mA which is the most my HT secondary is set to pull before the fuse action, so I decided to essentially lift the 6AQ5 cathode voltages to provide safe B+ levels. Because the Max voltages listed on Tube data sheets is in reference to the cathode, I could use a 32V zener diode to set the Mosfet Drain voltage to +32V, and then use the drain to be the low side of the 400Ω cathode bias resistors to bring the screen/cathode voltage to the safe 250V level. It's a supper simple design, works fantastic. I had to do some series/parallel connections to get the values I was looking for. But that's only because I'm impatient and wanted to do this quickly.
However, after the results from subminiature amp project over the last year, I've become much more encouraged. In fact, I decided to tear my EL84 amp down to the chassis, and transformers and rebuild it using the same approach I took with the submini build.
Before:
After:
Haha! She was so pretty before, but it is so much quieter and stable in it's current ugly state. But I digress...
My first idea, was to install a pair of 5902 output tubes in addition to the EL84s low output option. My PT already had a second HT winding, and was gonna run them off that as a designated power supply. The challenges with that idea were how/where to mount the additional tubes, and they would require an additional output transformer. In addition to those, I was worried that I would loose some of the effects of excursion and power supply dynamics you get from overdriving a push pull amp by running the output tubes from a separate supply. But since I was unhappy with the original output section anyways, decided to look into what I needed to do to replace them with 6AQ5s
(For those who aren't familiar with 6AQ5s, they are 7-pin beam tetrode designed to operate as a 6V6 equivalent with a plate and screen voltage of 250V. They were more common to find in early gibson/supro/valco type amps)
The good thing was that the output transformer specs are shared closely between 6AQ5s and EL84s, so I could use the existing one. Obviously I would need to use a socket adapter or replace the 9-pin sockets with a pair of 7-pin.
The other challenge was that even though the output power was comparable, all the data sheets I could find specified that the max screen voltage and plate voltage for a 6AQ5 is 250V. My existing PT was spec'd to provide a B+ of 330V under an idle load. In reality, Heyboer ended up sending me a transformer that was about 4% off on both my HV secondary windings, so the B+ is between 317V-320V. Still, that's enough of a difference, where I'd seen modern amp companies design amps that cause output tube and output stage failures because of overvoltage conditions, particularly on the screens.
This gave me the opportunity to test an Idea I came up with for switching a while back. Rather than trying to drop the entire B+ by using any number of common techniques, I had some STP9NK90Z Mosfets with an RDS(on) resistance of <1Ω at @ 250mA which is the most my HT secondary is set to pull before the fuse action, so I decided to essentially lift the 6AQ5 cathode voltages to provide safe B+ levels. Because the Max voltages listed on Tube data sheets is in reference to the cathode, I could use a 32V zener diode to set the Mosfet Drain voltage to +32V, and then use the drain to be the low side of the 400Ω cathode bias resistors to bring the screen/cathode voltage to the safe 250V level. It's a supper simple design, works fantastic. I had to do some series/parallel connections to get the values I was looking for. But that's only because I'm impatient and wanted to do this quickly.
Comment