Ok, I've been lurking in the background for a while and I am trying to get the opinions and facts about what I think are the two most popular driver stages for a push pull tube audio amp.
1 the split load "concertina" or
2 the differential stage, i.e. 2 matched triodes with a common constant current source cathode.
I am an objectivist I don't believe in what you imagine you hear, or really hear, unless you can demonstrate it to another person without telling them to hear it. In other words, I believe you can hear something, but I know it has a cause and can be demonstrated to another person. The most effective way I know to prove something is with measurements. If you think measurements are ineffective, invent a new one and submit it to peer review and we will name a new audio effect after you. Everything you really hear has a reason and I've yet to see a cause that wasn't RILL or C based. If it isn't, I will shake your hand in Switzerland.
I see a lot of opinion on the two dominant driver stages, but little fact. What I'd like is a FLAME LESS discussion of the merits of concertina VS differential drivers. The parameters of interest are distortion, affect of feedback on the stage, harmonic structure of the distortion, ability to drive Miller capacitance, slew rate, best tubes for each service, bandwidth and noise. Any other relevant criteria are welcome too. Again, I don't want objectivist calling subjectivist stupid and subjectivist calling objectivist tone deaf. And please no "Magic Pixy Dust"
1 the split load "concertina" or
2 the differential stage, i.e. 2 matched triodes with a common constant current source cathode.
I am an objectivist I don't believe in what you imagine you hear, or really hear, unless you can demonstrate it to another person without telling them to hear it. In other words, I believe you can hear something, but I know it has a cause and can be demonstrated to another person. The most effective way I know to prove something is with measurements. If you think measurements are ineffective, invent a new one and submit it to peer review and we will name a new audio effect after you. Everything you really hear has a reason and I've yet to see a cause that wasn't RILL or C based. If it isn't, I will shake your hand in Switzerland.
I see a lot of opinion on the two dominant driver stages, but little fact. What I'd like is a FLAME LESS discussion of the merits of concertina VS differential drivers. The parameters of interest are distortion, affect of feedback on the stage, harmonic structure of the distortion, ability to drive Miller capacitance, slew rate, best tubes for each service, bandwidth and noise. Any other relevant criteria are welcome too. Again, I don't want objectivist calling subjectivist stupid and subjectivist calling objectivist tone deaf. And please no "Magic Pixy Dust"
Both the differential and "concertina" phase splitters should be built around high gm types. It's a matter of avoiding slew limiting.
When executed properly, both splitter types perform well. The differential splitter has the advantage of providing gain. We took advantage of the gain a differential splitter provides in "El Cheapo" to create a 2 stage "integrated" amp. The "El Cheapo" saga can be found here.
When executed properly, both splitter types perform well. The differential splitter has the advantage of providing gain. We took advantage of the gain a differential splitter provides in "El Cheapo" to create a 2 stage "integrated" amp. The "El Cheapo" saga can be found here.
TUBESMAN, you should start by asking a more rational question, then you might receive rational and succinct answers. What you describe as 'drivers' are not drivers at all, they are phase splitters. In fact, the concertina makes a poor driver unless the output tubes are very undemanding, such as EL84s. That is one reason the Dynaco front end has been redesigned countless times.
The dif also gives much greater swing and does not sound so ugly if overdriven on transients. The concertina is unpleasant if overdriven, which is one reason it should be located before a driver with gain, rather than trying to use it directly as a driver.
As to the types of tubes that are most suitable, a low-mu or medium-mu triode (e.g. 6SN7) is generally considered best for a concertina splitter (lower OP impedance), while a high-mu double triode (e.g. 6SL7) works best in a dif splitter (better balance).
PS you ain't gonna get 25 watts or more out of triode-strapped EL34s. Maximum is probably 18 watts, even with fixed bias.
Say again? The dif has excellent PSRR, a notorious failing of the concertina. In fact, Jim Broskie went to some lengths to tame power supply noise from the concertina splitter with capacitor dividers between B+ and ground.
The dif also gives much greater swing and does not sound so ugly if overdriven on transients. The concertina is unpleasant if overdriven, which is one reason it should be located before a driver with gain, rather than trying to use it directly as a driver.
As to the types of tubes that are most suitable, a low-mu or medium-mu triode (e.g. 6SN7) is generally considered best for a concertina splitter (lower OP impedance), while a high-mu double triode (e.g. 6SL7) works best in a dif splitter (better balance).
PS you ain't gonna get 25 watts or more out of triode-strapped EL34s. Maximum is probably 18 watts, even with fixed bias.
I like nutshell answers. After reading many post on the subject of tube types and configurations I decided that I'd like to try a Dif amp with a CCS. I've run Tube Cad simulations and if they can be believed, the dif can be configured for symmetrical output swings easier than a concertina. But the dif has more noise. I figure about 3db more noise power. So I'd take 2 6DJ8's parallel the sections, put 1 in each leg of a dif amp, that should make the noise a wash. Use a CCS in the tail, like the one Morgan Jones described, and call it a day. I want to drive either EL34's, KT77's or something like that , preferably as triodes with fixed bias, to get around 25-35 Watts. A little less power is ok too. Is there an obvious flaw with my idea that I am blind too? Can you use (Vcc)*(Vcc)/2Po to get the a-a load impeadence? If not what other than a graphical method can be used?
Hi,
Indeed! And for truly symmetrical output, add a little common-mode local feedback.
Attached is a phase splitter with CMFB I've adapted for a 6V6 amp, based on Fred Nachbaur's (RIP) tube version at: http://dogstar.dantimax.dk/tubestuf/driver02.htm
Cheers!
** edit **
Right attachment this time
TUBESMAN said:
Indeed! And for truly symmetrical output, add a little common-mode local feedback.
Attached is a phase splitter with CMFB I've adapted for a 6V6 amp, based on Fred Nachbaur's (RIP) tube version at: http://dogstar.dantimax.dk/tubestuf/driver02.htm
Cheers!
** edit **
Right attachment this time
Attachments
Tubesman, nothing at all wrong with your idea, but run at a decent current, a single 6DJ8 section is already ridiculously quiet. Paralleling the sections doubles all the capacitances for not much return.
I'm puzzled about the symmetric swing comments. Neither cc diff amp nor concertina have any appreciable asymmetry as long as their loading is symmetric. If the loading is nonsymmetric, neither circuit will have a symmetric output.
I'm puzzled about the symmetric swing comments. Neither cc diff amp nor concertina have any appreciable asymmetry as long as their loading is symmetric. If the loading is nonsymmetric, neither circuit will have a symmetric output.
Power supply noise rejection is not the same as broadband noise from the device itself. A dif amp I've read, right here by one of you guys, has more noise than a single end stage. seems reason able for a bunch of reasons. The comment about the symetrical output is based on what tube cad shows as the maximum positive vs maximum negative output voltage from each type of circuit. With the level of P-P drive voltage I need it's likely its not a factor. I can live with 18 watts, but I'd like a little more, perhaps parallel pairs or triode connected KT88's
Geek, the circuit you posted looks like a hybrid version of something I was interested in from Glass Audio. They used a tube to read the balance error of the dif stage and feed it back to correct it. The way it looked, I think it could do two things, provide whatever level of balance you want for the next stage and make sure the two phase signals are exact mirrors of each other.
Hi Tubesman,
That is exactly how this works.
I put it through it's paces on the bench, using various tubes with matched and unmatched sections and the highest imbalance was 20mV with 60V P-P output swings per tube. Without the CMFB, the imbalance was 5V at that output level.
Another advantage of this circuit is you do not need to buy tubes with perfectly matched sections
Finally, the current setting is a lot more stable and far less touchy to adjust.
Cheers!
TUBESMAN said:
That is exactly how this works.
I put it through it's paces on the bench, using various tubes with matched and unmatched sections and the highest imbalance was 20mV with 60V P-P output swings per tube. Without the CMFB, the imbalance was 5V at that output level.
Another advantage of this circuit is you do not need to buy tubes with perfectly matched sections
Finally, the current setting is a lot more stable and far less touchy to adjust.
Cheers!
Sure, drop a 1 to 5K variable between the two 22K resistors and see what you can do
Didn't bring the breadboard to the computer that night, so no figures there.
But I can tell you that noise perceptibly dropped to around 1/4 compared to a resistor in the tail and dropped noticeably compared to without the CMFB.
TUBESMAN said:
Didn't bring the breadboard to the computer that night, so no figures there.
But I can tell you that noise perceptibly dropped to around 1/4 compared to a resistor in the tail and dropped noticeably compared to without the CMFB.
Geek said:Hi,
Indeed! And for truly symmetrical output, add a little common-mode local feedback.
Attached is a phase splitter with CMFB I've adapted for a 6V6 amp, based on Fred Nachbaur's (RIP) tube version at: http://dogstar.dantimax.dk/tubestuf/driver02.htm
Cheers!
** edit **
Right attachment this time
Geek, would this driver work for a Dynaco Mark III? I'm looking to replace the orignal driver board with something better.
Thanks.
rick
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.