Featured Vacuum Tube: The 807
Posted 15th January 2013 at 11:05 PM by Miles Prower
Featured VTs: The 807
807
The first of the beam formers is also one of the most enduring types: the 6L6. Developed by RCA in the mid-1930s, this type was originally intended for use as an audio final. It included other, then new, features besides the elimination of an actual, physical suppressor grid required to smooth out the screen grid "kinks". This included the now standard Octal base (up to eight pins possible, and with a keyed base for proper socket alignment) and a metal envelope. The latter was made in one of two ways: a glass envelope VT slipped into a metal shield can, or using the shield can as the envelope, with a glass base to bring out the connections. Other improvements was to give the control grid and screen grid the same pitch and wire diameter. By overlaying these two grids, the negative control grid serves to "shadow" the screen, thereby reducing the useless screen current for improved overall efficiency.
Though the metal envelope provided excellent shielding, these types ran very hot. Even the small signal metal envelope tubes get unusually warm. As a result, the Octal base was adapted to the glass envelope. Thus, the 6L6G type appeared. The first had the "ST" profile. The 6L6GT with a tubular glass envelope appeared by the early 1950s. The 6L6 (in its various iterations) proved to be quite successful, and is still in production today.
By the early 1940s, the 6L6 was adapted to the "ST" glass envelope, the then-standard five pin base, and with a top cap connection to the plate. This RF version became the 807. The main drawbacks to the 6L6 are that the highest voltage pin is right next to the lowest. For the usual audio operating conditions, this was of no consequence. It did, however, limit the available power when operating as a Class C, RF final. The top cap is also highly convenient, as this allows the output circuitry to be shielded from the input by the chassis itself. It also allows for plate voltages that would cause flash-overs between the plate and heater pins.
As for the characterizations of the type, the spec sheets cover all operations from Class A1 SE, through Class C. Audio final applications are explained for everything from Hi-Fi through PA and AM modulator duties. Having the AM plate modulator capable of using the same finals as the RF deck was a convenience. For PA and AM modulation, efficiency and power output are more important than sonic performance. The data for efficiency includes Class AB2 and Class B (actually a very deep Class AB2). Fidelity by Class A1 or Class AB1, push-pull. As with any pentode, the SE performance isn't so good as a triode final.
Even though this type boasts some outstanding THD figures (THD= 1.8% for push-pull operation) it has gotten a reputation as being a guitar amp final. Perhaps because that's where most are used these days? The 807 does have a tendency to make lots of nasty, high order harmonic distortion. This shows up when performing the "Twin-T" test. The residual after nulling out the fundamental is quite distorted, resembling a sawtooth wave at three times the frequency. This waveform must contain a lot of harmonics, H5 and above.
While running open loop, 807s do sound nasty: lots of listener fatigue. How nasty depends on the program material, but is always there. This particular deficiency was noted by the type's inventor: O. H. Schade. He recommended local NFB to tame that harmonic nastiness. This is frequently accomplished by "Ultralinear" operation. This isn't possible since the 807 has a very low screen voltage limit. Unlike many audio pentodes, the screen can't be run at the same DC voltage as the plate unless you're going to sacrifice a lot of output. That leaves either cathode or parallel NFB. Schade recommended feeding 10% of the AC plate voltage back to the control grid, anode follower style.
When this is implemented, the 807's sonic performance improves. All that remains is to take off some of the "edge" with some additional gNFB (~7.0db does nicely if the rest of the open loop design is well implemented). The nastiness can be tamed, and the type capable of excellent sonic performance. With cathode follower grid drivers, the 807 can easily produce some 30W of audio output, though specced at 26.5W. If you're willing to go Class AB2, you can get 80W or more, which requires increasing the plate voltage to 600Vdc, and is not recommended with 6L6s. This wasn't done very often back in "the day", due to the problems inherent with providing suitable grid current from a source with the lowest possible output impedance. That usually meant interstage transformer coupling, and that's not compatible with Hi-Fi since it becomes very difficult to include gNFB, due to the pecular phase performance of any xfmr with ferromagnetic cores. These days, a MOSFET source follower can easily drive the grids positive with very low output impedances.
Other Considerations
This type definitely likes to make RF. The spec sheet recommends the use of plate stoppers, either 47R or 100R, C-comp resistors, or a suppressor coil. The latter can be ten spaced turns with an ID= 7/16 inch to provide about 1.0uH of inductance. This is low enough to be of no consequence at even the highest audio frequencies. The coil can be "de-Q'd" by paralleling it with a 100R/2W C-comp resistor (or four 470R/0.5W C-comps in parallel. Mount the resistor(s) inside the coil. The plate stopper needs to be mounted right at the top cap connector, as is the case with any stopper resistors or coils.
The 807 also likes to make "snivets" when operated in Class AB. When one final goes into plate current cutoff, this triggers a damped Barkhausen oscillation. The frequency being determined by the leakage inductance and stray capacitance of the OPT (usually 60KHz+). The fix here is to include screen suppressors. 1K5/0.5W, C-comp resistors should be soldered to the screen pins with the least lead length. These stoppers will prevent the snivets. (The other possible cure would be a slight positive bias on the beam formers, but that's not possible since these are internally connected to the cathode, and not brought out to a pin.)
If the design work is done correctly, and attention paid to details, there is no reason not to use 807s for Hi-Fi work.
One other application that is seen quite frequently is the use of the 807 in pseudotriode mode as a series pass voltage regulator.
Other 6L6-oids: 1625 and 6BG6
6BG6 TV HD Final
The original type proved so useful that it spawned other spin-off types, of which I've discussed one. The 1625 is the 12.6V heater version of the 807. It has the same shape and size, but uses an uncommon seven pin base, presumably to keep people from sticking the wrong VT in the wrong hole.
The 807 also proved useful as a horizontal deflection PA for B & W TV sets (before screens got too big, that is). It was desirable to have an Octal version, and so the 6BG6 is the Octal "807". Internally, and electronically, the types are the same. You could consider the 6BG6 to be an Octal 807, or a 6L6 with a top cap. As with the 807, the top cap keeps the high voltage well away from any low voltage points to prevent flash-over. The 1625, 807, and 6BG6 will all work in any circuit designed for the 6L6.
One thing that needs clarification is the oft stated notion that the 1624 is a "DH" 807. It isn't, as the 1624 was designed for a very specific application: mobile transmitters. Being a DH type, the filament could be turned off during receive and/or monitoring to save battery power. It's an RF type from the get-go, and was intended for Class AB, B, or C. At Vgk= 0, the plate current is quite low, under 100mA, and it won't produce any more than ~10W as a Class AB1, push-pull amp. You can get a bit more power at half the voltage from a pair of 6V6s or 6AQ5s. The 1624 must be operated as a Class AB2 amp in order to get the most out of it. In that case, it gives just a bit less than what a pair of push-pull 807s will provide. If there is a similarity, it's that the 1624 also boasts of some excellent THD performance if driven from a sufficiently Lo-Z grid driver.
807
The first of the beam formers is also one of the most enduring types: the 6L6. Developed by RCA in the mid-1930s, this type was originally intended for use as an audio final. It included other, then new, features besides the elimination of an actual, physical suppressor grid required to smooth out the screen grid "kinks". This included the now standard Octal base (up to eight pins possible, and with a keyed base for proper socket alignment) and a metal envelope. The latter was made in one of two ways: a glass envelope VT slipped into a metal shield can, or using the shield can as the envelope, with a glass base to bring out the connections. Other improvements was to give the control grid and screen grid the same pitch and wire diameter. By overlaying these two grids, the negative control grid serves to "shadow" the screen, thereby reducing the useless screen current for improved overall efficiency.
Though the metal envelope provided excellent shielding, these types ran very hot. Even the small signal metal envelope tubes get unusually warm. As a result, the Octal base was adapted to the glass envelope. Thus, the 6L6G type appeared. The first had the "ST" profile. The 6L6GT with a tubular glass envelope appeared by the early 1950s. The 6L6 (in its various iterations) proved to be quite successful, and is still in production today.
By the early 1940s, the 6L6 was adapted to the "ST" glass envelope, the then-standard five pin base, and with a top cap connection to the plate. This RF version became the 807. The main drawbacks to the 6L6 are that the highest voltage pin is right next to the lowest. For the usual audio operating conditions, this was of no consequence. It did, however, limit the available power when operating as a Class C, RF final. The top cap is also highly convenient, as this allows the output circuitry to be shielded from the input by the chassis itself. It also allows for plate voltages that would cause flash-overs between the plate and heater pins.
As for the characterizations of the type, the spec sheets cover all operations from Class A1 SE, through Class C. Audio final applications are explained for everything from Hi-Fi through PA and AM modulator duties. Having the AM plate modulator capable of using the same finals as the RF deck was a convenience. For PA and AM modulation, efficiency and power output are more important than sonic performance. The data for efficiency includes Class AB2 and Class B (actually a very deep Class AB2). Fidelity by Class A1 or Class AB1, push-pull. As with any pentode, the SE performance isn't so good as a triode final.
Even though this type boasts some outstanding THD figures (THD= 1.8% for push-pull operation) it has gotten a reputation as being a guitar amp final. Perhaps because that's where most are used these days? The 807 does have a tendency to make lots of nasty, high order harmonic distortion. This shows up when performing the "Twin-T" test. The residual after nulling out the fundamental is quite distorted, resembling a sawtooth wave at three times the frequency. This waveform must contain a lot of harmonics, H5 and above.
While running open loop, 807s do sound nasty: lots of listener fatigue. How nasty depends on the program material, but is always there. This particular deficiency was noted by the type's inventor: O. H. Schade. He recommended local NFB to tame that harmonic nastiness. This is frequently accomplished by "Ultralinear" operation. This isn't possible since the 807 has a very low screen voltage limit. Unlike many audio pentodes, the screen can't be run at the same DC voltage as the plate unless you're going to sacrifice a lot of output. That leaves either cathode or parallel NFB. Schade recommended feeding 10% of the AC plate voltage back to the control grid, anode follower style.
When this is implemented, the 807's sonic performance improves. All that remains is to take off some of the "edge" with some additional gNFB (~7.0db does nicely if the rest of the open loop design is well implemented). The nastiness can be tamed, and the type capable of excellent sonic performance. With cathode follower grid drivers, the 807 can easily produce some 30W of audio output, though specced at 26.5W. If you're willing to go Class AB2, you can get 80W or more, which requires increasing the plate voltage to 600Vdc, and is not recommended with 6L6s. This wasn't done very often back in "the day", due to the problems inherent with providing suitable grid current from a source with the lowest possible output impedance. That usually meant interstage transformer coupling, and that's not compatible with Hi-Fi since it becomes very difficult to include gNFB, due to the pecular phase performance of any xfmr with ferromagnetic cores. These days, a MOSFET source follower can easily drive the grids positive with very low output impedances.
Other Considerations
This type definitely likes to make RF. The spec sheet recommends the use of plate stoppers, either 47R or 100R, C-comp resistors, or a suppressor coil. The latter can be ten spaced turns with an ID= 7/16 inch to provide about 1.0uH of inductance. This is low enough to be of no consequence at even the highest audio frequencies. The coil can be "de-Q'd" by paralleling it with a 100R/2W C-comp resistor (or four 470R/0.5W C-comps in parallel. Mount the resistor(s) inside the coil. The plate stopper needs to be mounted right at the top cap connector, as is the case with any stopper resistors or coils.
The 807 also likes to make "snivets" when operated in Class AB. When one final goes into plate current cutoff, this triggers a damped Barkhausen oscillation. The frequency being determined by the leakage inductance and stray capacitance of the OPT (usually 60KHz+). The fix here is to include screen suppressors. 1K5/0.5W, C-comp resistors should be soldered to the screen pins with the least lead length. These stoppers will prevent the snivets. (The other possible cure would be a slight positive bias on the beam formers, but that's not possible since these are internally connected to the cathode, and not brought out to a pin.)
If the design work is done correctly, and attention paid to details, there is no reason not to use 807s for Hi-Fi work.
One other application that is seen quite frequently is the use of the 807 in pseudotriode mode as a series pass voltage regulator.
Other 6L6-oids: 1625 and 6BG6
6BG6 TV HD Final
The original type proved so useful that it spawned other spin-off types, of which I've discussed one. The 1625 is the 12.6V heater version of the 807. It has the same shape and size, but uses an uncommon seven pin base, presumably to keep people from sticking the wrong VT in the wrong hole.
The 807 also proved useful as a horizontal deflection PA for B & W TV sets (before screens got too big, that is). It was desirable to have an Octal version, and so the 6BG6 is the Octal "807". Internally, and electronically, the types are the same. You could consider the 6BG6 to be an Octal 807, or a 6L6 with a top cap. As with the 807, the top cap keeps the high voltage well away from any low voltage points to prevent flash-over. The 1625, 807, and 6BG6 will all work in any circuit designed for the 6L6.
One thing that needs clarification is the oft stated notion that the 1624 is a "DH" 807. It isn't, as the 1624 was designed for a very specific application: mobile transmitters. Being a DH type, the filament could be turned off during receive and/or monitoring to save battery power. It's an RF type from the get-go, and was intended for Class AB, B, or C. At Vgk= 0, the plate current is quite low, under 100mA, and it won't produce any more than ~10W as a Class AB1, push-pull amp. You can get a bit more power at half the voltage from a pair of 6V6s or 6AQ5s. The 1624 must be operated as a Class AB2 amp in order to get the most out of it. In that case, it gives just a bit less than what a pair of push-pull 807s will provide. If there is a similarity, it's that the 1624 also boasts of some excellent THD performance if driven from a sufficiently Lo-Z grid driver.
Total Comments 0
