-
Feeling lucky? Check out the AK 24th Anniversary Giveaway thread and join in the fun!
Favorite 5687 driver stage for PP 6550's
- Thread starter lugnut1
- Start date
Register to hide this ad
The answer to your question is quite complex for the arbitrary case because it depends on the amp's front end topology--that being the voltage gain stage, phase inverter stage, and driver stage, the amount of feedback added, as well as the desired input sensitivity.
But if we make some assumptions, a lot of the discussion can be distilled down. Assume a "Mullard style" front end, meaning a common cathode gain stage coupled to a differential pair inverter that drives the output stage. In other words a topology that looks like this:
Typically for a 6550 PP output stage delivering say 60 watts, you would want about 38 dB open loop gain in the front end stages, with 14 dB global feedback, for approximately 2V input sensitivity. Back of napkin math says with 14 dB feedback and with V1 being a 6DJ8 and V2 being a 5687, you'd have about 26 dB forward gain, which would get you maybe 1.8V input sensitivity. It will work. Note though, you may need to AC couple V1 to V2, depending on power supply voltages available.
But if we make some assumptions, a lot of the discussion can be distilled down. Assume a "Mullard style" front end, meaning a common cathode gain stage coupled to a differential pair inverter that drives the output stage. In other words a topology that looks like this:
Typically for a 6550 PP output stage delivering say 60 watts, you would want about 38 dB open loop gain in the front end stages, with 14 dB global feedback, for approximately 2V input sensitivity. Back of napkin math says with 14 dB feedback and with V1 being a 6DJ8 and V2 being a 5687, you'd have about 26 dB forward gain, which would get you maybe 1.8V input sensitivity. It will work. Note though, you may need to AC couple V1 to V2, depending on power supply voltages available.
Depends on the topology. A Mullard style frontend isn’t ideal for a 12AT7 in either phase inverter or voltage gain positions because it will offer too much gain. You want input sensitivity in the 1.5 to 2.5V range, that is if you are going to drive the amp with an outboard preamp and if you are going to add say 12 to 14 dB feedback.
A 12AT7 in combo with a 6FQ7 or 6DJ8 would make more sense if you wanted to add say 20 dB feedback. 20 dB feedback is difficult to apply though. Most low end output transformers will struggle with that much. About the most I’d shoot for is 14 dB, unless you’ve got the Peerless or Acrosound or other top end output transformer.
A 12AT7 in combo with a 6FQ7 or 6DJ8 would make more sense if you wanted to add say 20 dB feedback. 20 dB feedback is difficult to apply though. Most low end output transformers will struggle with that much. About the most I’d shoot for is 14 dB, unless you’ve got the Peerless or Acrosound or other top end output transformer.
Sure, then under these conditions:
- 60 watts output
- 4.3K UL OPT
- Mullard-like frontend as shown above
- 12AT7 voltage amp, 5687 inverter/driver
- 6 dB feedback: 0.35V input sensitivity
- 8 dB feedback, 0.45V input sensitivity
- 10 dB feedback: 0.55V input sensitivity
- 12 dB feedback: 0.7V input sensitivity
- 5687 voltage amp, 12AT7 inverter/driver
- 6 dB feedback: 0.38V input sensitivity.
- 8 dB feedback: 0.45V input sensitivity
- 10 dB feedback: 0.6V input sensitivity
- 12 dB feedback: 0.75V input sensitivity
- 6DJ8 voltage amp, 5687 inverter/driver
- 6 dB feedback: 0.7V input sensitivity.
- 8 dB feedback, but with cathode bypass on 6DJ8 stage: 0.67V input sensitivity
Last edited:
Why not just follow the original design Chicago commissioned for use with their BO-14?
tubebooks.org/file_downloads/chicago_100W_amp.pdf
Dave
tubebooks.org/file_downloads/chicago_100W_amp.pdf
Dave
kward, thanks for the numbers. Looks like 12at7/5687 will work for me.
Dave, I have the article and I have used the 6sn7 many times with great results but would like to try something different. I also prefer to NOT use a preamp, hence a little more gain in the VA with a 12at7. I may end up trying both at some point just to hear what Olson had in mind.
Dave, I have the article and I have used the 6sn7 many times with great results but would like to try something different. I also prefer to NOT use a preamp, hence a little more gain in the VA with a 12at7. I may end up trying both at some point just to hear what Olson had in mind.
Gordon, Yes, another fine option. So many tubes and a lot of satisfying time for a 70yr old retired tubeguy. I have a lot of respect for the BO-14 output trannys and I just hope I can do them justice. This hobby has served me well for a very long time and I have encountered some wonderful people along the way. I still have not met anyone who has actually heard these BO-14's do their stuff. I wonder if my mental image will be met or exceeded by the "finished" amp. I am optimistic!
Dave, I have looked at the schematic for the triode version. It shows a.02 coupling cap to the grid (pin 5) of the 6550. This doesn't seem correct as the cutoff would be around 37HZ. What do you think?
It’s discussed in some detail in the text as to why those values were chosen—to aid in avoiding low frequency instability. But look at the max power vs. frequency plot. In triode mode, max power response is achieved down to < 20 Hz. While true the open loop response of that filter has knee at 37 Hz, that filter is inside the global feedback loop, which acts to extend flat response down to the stated value.
Last edited:
Just remember as you play with the front end design, that the pentode configured output stage requires some 100 volts peak of drive voltage per tube due to the NFB provided by the cathode feedback connection. And, it must be available across the full audio bandwidth into a 50 kΩ load as well. That's what necessitated the use of the 5687 driver tube in the first place.
Dave
Dave
PakProtector
AK member
The BO14 poses some issue with respect to drive. The CFB winding is about 10%, so voltage drive for the 6550 is going to increase vs the usual grounded cathode. IMO, to heck wid the high voltage B+ and AB1 operation. Go for B+ around 350 and bias that pup into Class A. You will need 100mA of idle current, give or take. Keep in mind whilst drawing the load line that the load will *NOT* be the advertised 5k, as you are adding 10% to the cathodes, or 20% to the total load...6k The difference in output power will be negligible to your ears. Also, with friendly speakers presenting a mostly resistive load, you won't need the global loop of NFB.
On a 350V, Class A bias point, the grid 1 voltage will be around -30V. You will get the cathodes moving 10% of the plate voltage, so add another 30V. That is a zero to peak...or +/- 60V will be needed. go AB1 from a 450V OP and bias will be -55, and you'll add 40 to it...or +/-95V...so there is the math behind Dave's number...
As far as 5687's go, the Russian 6H6Pi are slightly better IME, and as a whole as well matched section to section as the impeccable GE 5-Star 5687's.
I am looking forward to seeing how it turns out; I also have a pair of those outputs...
cheers,
Douglas
On a 350V, Class A bias point, the grid 1 voltage will be around -30V. You will get the cathodes moving 10% of the plate voltage, so add another 30V. That is a zero to peak...or +/- 60V will be needed. go AB1 from a 450V OP and bias will be -55, and you'll add 40 to it...or +/-95V...so there is the math behind Dave's number...
As far as 5687's go, the Russian 6H6Pi are slightly better IME, and as a whole as well matched section to section as the impeccable GE 5-Star 5687's.
I am looking forward to seeing how it turns out; I also have a pair of those outputs...
cheers,
Douglas
