I have learned heaps while being part of this forum and this is more or less my goodbye giving back present: this weekend I'll be moving into our new house but unfortunately the subdivision has not yet wired broadband so I'll be going back to dial up (yuk!) (With the hills around mobile boradband is not an option either)
It may be a year to 18 months before I get hooked up so I'll say thanks to all who have helped me move forwards during the time I've been part of the forum.
During the past week I've been learning LtSpice and modelling the different options for my 6L6GC SE amplifier. The power supply is not yet on paper but I'll be using a CCS feeding a OA2 and OB2 gas voltage regulator (more stable and less noise than zeners and the heat is above chassis) and use that voltage reference through a 10 turn put to "fine tune" the screen voltage. (The pot will be feeding a high voltage mosfet which will be acting for the current supplier to the screen)
I've mentioned elsewhere the existence of a self adjusting voltage regulator (with a mosfet too) so I am not tied to using a fixed B+ voltage. (Basically it becomes an electronic choke)
I did search long and hard for some decent 6l6GC SE amps but they are thin on the ground. In the end I found Alex Kitic's RH807 but I did not like the screen voltage dropping resistor which goes against recommendations made by KenRad for longevity of the tube. Some others remarked that the 12AT7 was not the best choice so in the end I wanted to find out what was true and what was old wifes tales.
I am attaching two sets of schematics - I am imposing a 30V pp limit on the 12AX7 output voltage swing since above that level distortion increases a lot (just my personal view on the matter, others are welcome to their views.)
The 12AX7 turned out to have a lower noisefloor (depending on design between 5 and 10 dB) than the 12AT7 plus the THD for the 12AX7 was about half to two thirds of the 12AT7. Confirmed that the 12AX7 is the lower distortion tube. There is also some interaction between the feedback resistor value and optimum performance. Came up with two OPT configurations: one for 2500 Ohm and the other for 5000 Ohm. The latter is the most flexible and can be adapted to different voltages with decent performance yet the 2500 did not exhibit that same flexibility. I ran out of time trying to find a 2500 solution with other voltages. It yielded the lowest distortion though.....
All transformers are James that I obtained directly from Taiwan.
I ran out of time to model properly an OPT so I made a "simulated" fix by using a choke and a resistor. Since it is not a part of the feedback loop I can see no issue with this. (having seen it elsewhere)
Enjoy and goodbye for now
AM
PS design is for personal use only and not for commercial use
It may be a year to 18 months before I get hooked up so I'll say thanks to all who have helped me move forwards during the time I've been part of the forum.
During the past week I've been learning LtSpice and modelling the different options for my 6L6GC SE amplifier. The power supply is not yet on paper but I'll be using a CCS feeding a OA2 and OB2 gas voltage regulator (more stable and less noise than zeners and the heat is above chassis) and use that voltage reference through a 10 turn put to "fine tune" the screen voltage. (The pot will be feeding a high voltage mosfet which will be acting for the current supplier to the screen)
I've mentioned elsewhere the existence of a self adjusting voltage regulator (with a mosfet too) so I am not tied to using a fixed B+ voltage. (Basically it becomes an electronic choke)
I did search long and hard for some decent 6l6GC SE amps but they are thin on the ground. In the end I found Alex Kitic's RH807 but I did not like the screen voltage dropping resistor which goes against recommendations made by KenRad for longevity of the tube. Some others remarked that the 12AT7 was not the best choice so in the end I wanted to find out what was true and what was old wifes tales.
I am attaching two sets of schematics - I am imposing a 30V pp limit on the 12AX7 output voltage swing since above that level distortion increases a lot (just my personal view on the matter, others are welcome to their views.)
The 12AX7 turned out to have a lower noisefloor (depending on design between 5 and 10 dB) than the 12AT7 plus the THD for the 12AX7 was about half to two thirds of the 12AT7. Confirmed that the 12AX7 is the lower distortion tube. There is also some interaction between the feedback resistor value and optimum performance. Came up with two OPT configurations: one for 2500 Ohm and the other for 5000 Ohm. The latter is the most flexible and can be adapted to different voltages with decent performance yet the 2500 did not exhibit that same flexibility. I ran out of time trying to find a 2500 solution with other voltages. It yielded the lowest distortion though.....
All transformers are James that I obtained directly from Taiwan.
I ran out of time to model properly an OPT so I made a "simulated" fix by using a choke and a resistor. Since it is not a part of the feedback loop I can see no issue with this. (having seen it elsewhere)
Enjoy and goodbye for now
AM
PS design is for personal use only and not for commercial use
Attachments
Last edited:
The main issue with Kitics designs is the use of a triode as the driver. This cannot deliver the best that this circuit can offer. It is so so simple to substitute a 6AU6 or similar for the triode that you really should consider it. A very simple transitor regulator for the screen is all that is needed. This allows you to pass at least 5mA of driver current and makes the driver relatively immune from the cripplingly low input impedance of the output stage (think about 1K load for your poor triode).
This feedback mechanism was never intended to work with triode drivers. Anything you do with the output stage screens is a waste of time if you don't address the driver problem. I know this from personal experience having built various implementation of screen circuits for the RH807 and found almost no sonic difference - so its not old wives tails
Shoog
This feedback mechanism was never intended to work with triode drivers. Anything you do with the output stage screens is a waste of time if you don't address the driver problem. I know this from personal experience having built various implementation of screen circuits for the RH807 and found almost no sonic difference - so its not old wives tails
Shoog
I am totally agree with this !
But so few understand the problems this days ... as it was not in the past .
Last edited:
Politely disagree - it has a major impact on the longevity of the output tube.
Pentode or triode - that is a personal choice but don't forget that the not bypassed cathode resistor ensures that the triode takes on a pentode like behaviour. Happen to have a stash of 12AX7 and am not going to cut more holes in the chassis, the layout will becoming awkward.
PS Be aware that not all 6L6GC are the same, some that are marketed as a 6L6GC (e.g. Ei) have no higher ratings than the 6L6 / 6L6G /6L6GB and have only a 19 Watt plate rating as opposed to 30 Watts for the real 6L6GC(-STR).
AM
It may have an effect on the life of the tube, but it will have very little if any on the sonics.
It is your choice what you do about the driver, but using a triode is an abuse of the Schade principle and you will not get the best out of your design. I have built about 10 schade amps so far (RH807's version been about three), and can tell you that a pentode is a better option. Still the amp will sound fine with a triode - just not as good as it could be.
Shoog
It is your choice what you do about the driver, but using a triode is an abuse of the Schade principle and you will not get the best out of your design. I have built about 10 schade amps so far (RH807's version been about three), and can tell you that a pentode is a better option. Still the amp will sound fine with a triode - just not as good as it could be.
Shoog
I happen to like the triode sound and the lower background noise level of a triode. THD figures are good enough for me.
However the sound was not the the only objective - it is all a balancing act between different priorities.
My primary objective was to build an amp that would still be going long after I've gone and hence I like tubes that are in the top 10 of most sold tubes. The 12AX7 is #1 and IIRC the 6L6GC is # 7. So if any they are having the best chance of surviving for quite some time.
Peace and I am signing off now, have to pack stuff up before the move.
All the best
AM
Yes, and its the fact that that Rp is invarient with output level.
To answer the point about liking the sound of a triode over a pentode, I do to, but this is not a triode amp - this is a Schade amp and it just doesn't work like a normal triode amp. You are notlistening to the sound of a triode driver, you are listening to a feedback arrangement which combines the front and back ends.
I suggest going back to first principles and reading the Tubcad article which explains why you need a pentode.
Shoog
I do not care on iota what chip on your shoulder is for triodes, nor do I care for what Tubecad writes. The engineers in the 1940's forgot more than todays self proclaimed "experts" will ever know.
Schade himself mentioned that this would work with triodes and the cult following that the RH84 has built up is a testament to that. I just happened to have a quad of Tungsol 6L6GC-STR that I would rather use running far below their max ratings than have an EL84 going flat out.
Similarly I happened to have ECC83's lying around so I am using that.
Your harping on about a pentode makes me regret having started this thread.
I am done with this and I'll have second thoughts about ever coming back to DIYaudio thanks to your truly.
No AmadeusMozart, u must not quit diyaudio. We need counterweights to all the EEs around here and their 'good engineering'
(Actually I'm an EE too, but completely amateur in this field).I try to tell myself to have a wide open door when entering such discussion sites. I know I have what it takes to build competent amps, but I still find new ways and new ideas and even keep learning new stuff here. The day I become an expert (as if that will ever happen), I will have nothing more to learn, and therefore nothing more to give either, and then I will stop coming to this forum.
It's a hard balance, trying to answer in a short and concice manner, and still get all your intentions through. If some tiny detail is left out, some 'expert' is right there to let you know, and the feeling is they are standing up there looking down on you with a 'I know more' grin. But it's hard to try and share your knowledge in a to-the-point way, and at the same time not be patronising. I'm sure Shoog doesn't mean any harm, and it is ok to discuss topologies even though it can be a bit like childplay sometimes.
Thats a shame that you have taken that attitude.
I have built that amp and others of the same design principle. I have no chip on my shoulder about triodes - I love triodes and love to use them where they work best.
Learn from people who have done this before - or not, that's your choice. Broskie is one of the best and Kitic is a very pale imitation. It was meant as friendly advise, that's all.
Build what you like I'm sure it will sound good anyway.
I am no expert and am learning all the time- by listening to those who have done things before me.
Shoog
I have built that amp and others of the same design principle. I have no chip on my shoulder about triodes - I love triodes and love to use them where they work best.
Learn from people who have done this before - or not, that's your choice. Broskie is one of the best and Kitic is a very pale imitation. It was meant as friendly advise, that's all.
Build what you like I'm sure it will sound good anyway.
I am no expert and am learning all the time- by listening to those who have done things before me.
Shoog
Last edited:
Schade driver loadlines
I think the notion that triode drivers are unsuitable for this service is a gross oversimplification of the question.
How about looking at some idealized loadlines for the driver with low effective load resistance of the feedback network (assuming current mode feedback)?
This will illustrate the issue some have with using a triode.
First slide is the 12AX7 in your circuit. You can see why the swing is limited to 30V P-P but actually it doesn't look too nonlinear up to about 50% FS. This is what you get, pretty much. Not shown is how the effective load resistance decreases as the gm of the 6L6 decreases, which actually curves the load line flatter at lower (toward the left) plate voltages. Using an unbypassed cathode resistor further linearizes the V-I curve and increases the plate resistance, making it a better V-I converter.
next is a 12AT7 and you can see why this one is unlikely to work well due to linearity and interaction of it's own low plate resistance with the V-I function.
Third is a pentode, showing why this is a better voltage-to-current converter because of the flat plate current lines. As the effective load resistance changes through the gm curve of the output tube, the output current remains a more linear function of the pentode grid voltage.
Last, an illustration of why it's not always evil for triodes to work into low load resistance. This is a 6922 as the bottom device in a cascode.
To me it looks like some triodes are suitable in this position and AM's measurements seem to indicate satisfactory performance.
Where's the beef?
Michael
I think the notion that triode drivers are unsuitable for this service is a gross oversimplification of the question.
How about looking at some idealized loadlines for the driver with low effective load resistance of the feedback network (assuming current mode feedback)?
This will illustrate the issue some have with using a triode.
First slide is the 12AX7 in your circuit. You can see why the swing is limited to 30V P-P but actually it doesn't look too nonlinear up to about 50% FS. This is what you get, pretty much. Not shown is how the effective load resistance decreases as the gm of the 6L6 decreases, which actually curves the load line flatter at lower (toward the left) plate voltages. Using an unbypassed cathode resistor further linearizes the V-I curve and increases the plate resistance, making it a better V-I converter.
next is a 12AT7 and you can see why this one is unlikely to work well due to linearity and interaction of it's own low plate resistance with the V-I function.
Third is a pentode, showing why this is a better voltage-to-current converter because of the flat plate current lines. As the effective load resistance changes through the gm curve of the output tube, the output current remains a more linear function of the pentode grid voltage.
Last, an illustration of why it's not always evil for triodes to work into low load resistance. This is a 6922 as the bottom device in a cascode.
To me it looks like some triodes are suitable in this position and AM's measurements seem to indicate satisfactory performance.
Where's the beef?
Michael
Attachments
Last edited:
Here's an illustration of the effect of the output tube gm curve on the driver load line. As the driver plate voltage decreases, it has to swing more voltage (dotted red line) to the output tube grid to maintain the output tube plate swing. This is the Schade current mode feedback at work.
Attachments
No beef Michael. AM seems a little touchy when contradicted - it appears that to provide a conflicting pov is a matter of personal affront to him. In that he is far from alone. I note however that he is performing that most stressful of human activities - moving house. My sympathies lie with him currently!
Shoog has given advice that is generalised but fairly widely accepted as a rule of thumb. He certainly didn't give it as emphatically as I have seen it put by others.
Thanks for the diagrams - they illustrate at least two points: 1 some but not all triodes can be used effectively in this topology and 2 pentodes have an inbuilt design characteristic that makes them a more likely candidate for this topology.
The curves also gave me a better insight into what is actually going on....
Best of luck to AM in the move. I detest moving. It is too bad that there are apparently no small signal pentodes in current production but I suppose the 6AU6 is plentiful enough for many years. Of course then one is working with two itty-bitty sockets instead of one simply tiny socket.
Hmm, I wonder about 6AU6 LTP PI with Schade for PP.
It is too bad that there are apparently no small signal pentodes in current production but I suppose the 6AU6 is plentiful enough for many years. Of course then one is working with two itty-bitty sockets instead of one simply tiny socket.Hmm, I wonder about 6AU6 LTP PI with Schade for PP.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.