It's been noted that that regular 1 wire channel link 'Torres' mod for 2 channel BF/SF Fenders makes the normal channel bassier than the vibrato channel.
I wondered how that could be, given that the coupling caps, 0.047 and 0.022uF, driving into the 3M3+ load of the reverb mixer, has a sub sonic low frequency -3dB breakpoint.
I scoped it out on the bench and found that it was true - feed the same signal into those 2nd stage grids and when they're mixed together, measured at the 3rd stage mixer plate, at 80Hz the normal channel is boosted about 2.5dB, the vibrato channel cut 3dB (compared to mid level 400-1kHz).
I've reasoned that this is due to, rather than the 3M3 mixer being the load, the load for one channel is actually the plate resistance of the other channel, + the impedance of the 2 coupling caps. Effectively a capacitive potential divider with a resistive load, in the form of the other channel's plate resistance. That's why the channel with the bigger cap gets a low boost, while the channel with the smaller cap actual gets a low cut.
A frequency dependant mixer has been created, due to the 2 caps being different values.
If that's what you want then fine, but I've been thinking about how this artificial cut/boost can be avoided. 1 way would be to fit equal value capacitors, however, a simpler and cleaner solution is to directly link the plates of the 2 stages to be mixed, rather than mix them via coupling caps.
Therefore the alternative channel link mod is to remove the 0.047uF normal channel coupling cap, and to fit a short wire link between terminal 6 of V1 and V2, either on the tube socket lugs, or on the eyelet board.
See
http://i963.photobucket.com/albums/a...nelLinkMod.jpg
The frequency dependance of the channel mixing is removed, both channels are mixed evenly across the audio band.
I felt uneasy about dc linking the plates, however, I reasoned that the cathodes are dc coupled already, some designs use the 2 sections in V1 linked together, and power tube plates get linked (in parallel and parallel push/pull output tube arrangements) with no ill effect.
I've tested the mod now on a couple of amps and can report that it works great. I arranged a switch in order to flip between the 2 mixing methods (with standby engaged), and the direct plate link seems to sound warmer and more open, the 'Torres' method sounding constricted and less natural in comparisson.
Maybe it's because the coupling cap isn't getting loaded so heavily, the 0.022uF is now having a easy life driving the 3M3 mixer load, rather than the ~40k of the other plate.
I think that it may be preferable for both V1 and V2 to be similar 12AX7 types - if there's a mix of a 12AY7 and 12AX7, or 12AU7 and 12AX7, then the widely differng gains, plate resistances and current draws might cause operating issues. However, I can confirm that 12AX7 and 5751 are fine.
I'd be interested to hear what anyone else makes of this method. Apologies if I've re-invented the wheel and all the above is common knowledge, it's been new to me at least! Peter.
I wondered how that could be, given that the coupling caps, 0.047 and 0.022uF, driving into the 3M3+ load of the reverb mixer, has a sub sonic low frequency -3dB breakpoint.
I scoped it out on the bench and found that it was true - feed the same signal into those 2nd stage grids and when they're mixed together, measured at the 3rd stage mixer plate, at 80Hz the normal channel is boosted about 2.5dB, the vibrato channel cut 3dB (compared to mid level 400-1kHz).
I've reasoned that this is due to, rather than the 3M3 mixer being the load, the load for one channel is actually the plate resistance of the other channel, + the impedance of the 2 coupling caps. Effectively a capacitive potential divider with a resistive load, in the form of the other channel's plate resistance. That's why the channel with the bigger cap gets a low boost, while the channel with the smaller cap actual gets a low cut.
A frequency dependant mixer has been created, due to the 2 caps being different values.
If that's what you want then fine, but I've been thinking about how this artificial cut/boost can be avoided. 1 way would be to fit equal value capacitors, however, a simpler and cleaner solution is to directly link the plates of the 2 stages to be mixed, rather than mix them via coupling caps.
Therefore the alternative channel link mod is to remove the 0.047uF normal channel coupling cap, and to fit a short wire link between terminal 6 of V1 and V2, either on the tube socket lugs, or on the eyelet board.
See
http://i963.photobucket.com/albums/a...nelLinkMod.jpg
The frequency dependance of the channel mixing is removed, both channels are mixed evenly across the audio band.
I felt uneasy about dc linking the plates, however, I reasoned that the cathodes are dc coupled already, some designs use the 2 sections in V1 linked together, and power tube plates get linked (in parallel and parallel push/pull output tube arrangements) with no ill effect.
I've tested the mod now on a couple of amps and can report that it works great. I arranged a switch in order to flip between the 2 mixing methods (with standby engaged), and the direct plate link seems to sound warmer and more open, the 'Torres' method sounding constricted and less natural in comparisson.
Maybe it's because the coupling cap isn't getting loaded so heavily, the 0.022uF is now having a easy life driving the 3M3 mixer load, rather than the ~40k of the other plate.
I think that it may be preferable for both V1 and V2 to be similar 12AX7 types - if there's a mix of a 12AY7 and 12AX7, or 12AU7 and 12AX7, then the widely differng gains, plate resistances and current draws might cause operating issues. However, I can confirm that 12AX7 and 5751 are fine.
I'd be interested to hear what anyone else makes of this method. Apologies if I've re-invented the wheel and all the above is common knowledge, it's been new to me at least! Peter.
Comment