As tubetvr just said, that resistors would make the tube impedance (as seen from the cathode follower) around 10kohm.
Why not use a simple 10 kohm instead? Or even 22kohm,... and you save one half tube.
Like I said, it wasn't perfect. I stopped building these sorts of circuits once I built my first version of Sy's Heretical preamplifier. I stopped using triodes as loads and went resistor or CCS from then on out.
If the signal inverts at the plate, how is it positive feedback? I'm not doubting you, just want some explanation. My personal versions never used feedback to the cathode (I just simply ran an unbypassed cathode resistor) but I've seen plenty of similar circuits use the same scheme without mention of it being positive feedback, always said to be negative...
If the signal inverts at the plate, how is it positive feedback? I'm not doubting you, just want some explanation. My personal versions never used feedback to the cathode (I just simply ran an unbypassed cathode resistor) but I've seen plenty of similar circuits use the same scheme without mention of it being positive feedback, always said to be negative...
Last edited:
This is what the screen grid of a pentode is for...that 13M claimed is at DC. Start increasing the frequency and that fine number is going to drop like a rock. A 6AU6 makes a very fine CCS when its plate is loaded.
Use some of the same tricks to increase the plate resistance.
cheers,
Douglas
Use some of the same tricks to increase the plate resistance.
cheers,
Douglas
In order to have negative feedback you have to subtract a % of the output from the input. In a single stage common cathode amplifier you can consider the cathode and grid as inputs. Signals applied at the grid will appear 180 degrees out of phase with signal at the plate, signals applied at the cathode are in phase with the signal at the plate.
Feed back applied at the cathode requires an additional phase inversion, in a single stage you would apply the feedback at the grid in order to get the inversion required..
Someone else will probably explain this way better than I did.. 😀
some seem prefer an anode - follower sonically - the input Z on those will be lower than with CF. A negative rail for the cathode R might be an option vs using a triode on the bottom. As others say, high gm, low rp for CF. IIRC, some of Bruce Moore's stuff used triodes. A pentode would be an option.
Fixed bias makes the input and output impedances more independent of source and load impedances, which is normally what you want from a buffer. You can still get very high input impedance -just make the biasing resistors bigger (see page 261 of my book).
Can you show how input and output impedance would more independant of load using fixed bias compared to self bias?
BTW with fixed bias the cathode resistor for the upper tube is not necessary.
BTW with fixed bias the cathode resistor for the upper tube is not necessary.
From my book:
"Although self-biasing is commonly used in traditional circuits, it suffers from some non-obvious problems. Firstly, the input resistance, although potentially very large, is dependent on valve characteristics and loading, rather than being nicely constant under all conditions. Secondly and more subtly, the grid leak provides a positive feedback path from cathode to grid which subtracts from the otherwise 100% negative feedback existing inside the valve itself. The total amount of negative feedback is therefore not unity but depends on the source impedance of the previous stage; the greater the source impedance, the smaller the feedback fraction. This means all the characteristics of the stage –in particular the output impedance– are dependent on the source impedance,"
If you do the derivations or a SIM you can see this happening -the output impedance increases as the source impedance increases. With infinite source impedance the output impedance would become equal to ra||Rl, which is a lot higher than the 1/gm we're all used to thinking of!
Yes, true.
Merlin, I'm reading your book, and it definitely is great (so is your website, by the way). I just miss that you don't go up to the output stage in your book (particulary, ultralinear stage design, but the title is clear about that)
Going back to my original post.
As expected, I replaced the triode active load with a simple bipolar CCS connected to -9V supply, and the same CF can now drive 20k load with a 100 pk-pk signal, where the one with the triode CCS struggled at 100k.
Going back to my original post.
As expected, I replaced the triode active load with a simple bipolar CCS connected to -9V supply, and the same CF can now drive 20k load with a 100 pk-pk signal, where the one with the triode CCS struggled at 100k.
Output stages need a whole book -and it's a lot of work. I was rather hoping Tubelab would do it, but he's dragging his feet. 😉
I simulated five cathode followers, and compared them.
All based on 12AX7 (which we all know, is not the best choice).
The best performer (distortion and gain) is "D", using BJT CCS.
The next best one is "C", very close to the one I originally posted.
Then come "E" and "B".
My original circuit is half way between E/B and D.
It can drive 36 Vrms signal into 100 kohm load, such as some SE KT88 output stage. Also 7.5 Vrms into a hard 20 kohm load.
It's not as bad as it seemed. Is it?
Just it is not fixed bias, but none of these are, anyway.
Circuit "E" active load is around 180k impedance. A little bit higher than those 118k of "B".
The circuit A is just another thing. It can only work for input signals less than Vgk (2.8V)
So, my conclusion is that, for "all-valve" circuit, using 12AX7, the "C" topology is not a bad choice.
What do you think??
All based on 12AX7 (which we all know, is not the best choice).
The best performer (distortion and gain) is "D", using BJT CCS.
The next best one is "C", very close to the one I originally posted.
Then come "E" and "B".
My original circuit is half way between E/B and D.
It can drive 36 Vrms signal into 100 kohm load, such as some SE KT88 output stage. Also 7.5 Vrms into a hard 20 kohm load.
It's not as bad as it seemed. Is it?
Just it is not fixed bias, but none of these are, anyway.
Circuit "E" active load is around 180k impedance. A little bit higher than those 118k of "B".
The circuit A is just another thing. It can only work for input signals less than Vgk (2.8V)
So, my conclusion is that, for "all-valve" circuit, using 12AX7, the "C" topology is not a bad choice.
What do you think??
Attachments
I think that due to limitations in sourcing current, the 12AX7 is a poor choice of triode for a cathode follower. I would use at least a 12AT7 or something with even better gm, like the 6DJ8...
Btw, circuit "D" is pretty much the heretical preamp by Sy, and my usual recommend way to go for a cathode follower- it works great with the 6DJ8 in particular, but also works well with the 12AU7, 12AT7, 6N1P, and other types that can source adequate current.
Btw, circuit "D" is pretty much the heretical preamp by Sy, and my usual recommend way to go for a cathode follower- it works great with the 6DJ8 in particular, but also works well with the 12AU7, 12AT7, 6N1P, and other types that can source adequate current.
D does not need a negative rail. 2N3055 (whatever that is this year) is about the least-likely transistor to use here; which does make it a good test.
All will work good for small work. Some will fall flat pulling heavy loads. There is significant difference in cost (not so much for DIY where fun exceeds the cost of the bottles). The choice will depend on goals and economics.
All will work good for small work. Some will fall flat pulling heavy loads. There is significant difference in cost (not so much for DIY where fun exceeds the cost of the bottles). The choice will depend on goals and economics.
The cascode as mentioned there is described as 'similar' to a pentode...and in the horizontal plate curve section that is so...but their claim that the u of a cascode is based on the product of the mu factors of the triodes used to build the cascode is incomplete at best. The cascode gain should be calculated with the gm of the lower device times the plate load....just like a pentode...🙂
cheers,
Douglas
cheers,
Douglas
I remember Allen Wright's preamp cookbook which was all aboit differential preamps....and CCS. I believe he was a strong suporter of cascoded designs with a triode and a mosfet, so thatthe actual signal triode was fed by another triode and below a jfet or mosfet...has anyone listened to such a construction with a tube vs. pure silicon ?
- Status
- Not open for further replies.