I've read many times that triodes are not good as CSS.
I've seen design examples with CSS triodes with AC resistance in the order of 100 kohm, which, obviously, could be replaced with a simple resistor.
But I've just designed a cathode follower with triode active load (both 12AX7/ECC83) with a AC resistance of 13 Mohm, which I honestly think it isn't a low value. It is a higher value than with most bipolar transistor designs I've seen in books and websites.
It increases a little bit the gain, and reduces a little bit the distortion, compared with using a simple 120 kohm resistor as load.
What am I not getting right?
I've seen design examples with CSS triodes with AC resistance in the order of 100 kohm, which, obviously, could be replaced with a simple resistor.
But I've just designed a cathode follower with triode active load (both 12AX7/ECC83) with a AC resistance of 13 Mohm, which I honestly think it isn't a low value. It is a higher value than with most bipolar transistor designs I've seen in books and websites.
It increases a little bit the gain, and reduces a little bit the distortion, compared with using a simple 120 kohm resistor as load.
What am I not getting right?
Attachments
The triodes in a 12AX7/ECC83 make poor cathode followers, as the gm is very low. The O/P impedance of a CF is approx. 1/gm. Another objection to the 'X7 triode as a CF is low plate current. You want to be able to charge and discharge interconnect cable capacitance.
Try your idea with the 12AT7/ECC81, which has reasonably high gm and can handle a few mA. of plate current, unlike the 'X7 triode at approx. 900 μA.
Try your idea with the 12AT7/ECC81, which has reasonably high gm and can handle a few mA. of plate current, unlike the 'X7 triode at approx. 900 μA.
Please add 1 Kohm carbon comp. grid stoppers. The triodes in a 6922 are seriously prone to oscillate, in a parasitic manner.
Like Eli said, the 12AX7 is not good for a CF. They work but they are not the most optimal solution for best performance. Even Mcintosh finally realized that a 12AT7 was more practical as a CF in their amps. The 12AX7 is a great voltage amp but it can't fight its way out of a wet paper bag. It never made sense why they used it as a driver for when they could have lowered global NFB and used a 12AT7 or even a tri-pent.
Who says that?
Mullard or Philips had a long white-paper on differential amps, with particular attention to the tail impedance. Pentodes look good until you account for the screen feed. In AC coupled work you can do OK, but for DC amplifiers you need several more parts and typically an active device to work the screen. They favored triode tails for most work.
150K times mu=100 makes 15Meg, sure.
But how much impedance do you need, with a "100K" load? More than 100K, of course. But 15Meg? 13Meg? 1Meg? may all be about the same with 100K loading. And 1Meg could be got with a mere 10K cathode resistor, dropping (for 0.5mA flow) only 5V, not wasting so much of your 250V supply.
A complete spec of what you need done. If this is a lab buffer, load cable may be short and/or frequency response may be limited. However in Audio we often reach for a 10 foot (3m) cable, 300pFd. At 20KHz this is about 28K load, not 100K. Since Z(out) is a few K at most, it will drive, but run out of current (0.5mA) at 14V peak, and THD may be rising above a few Volts at the top of the audio band. Which is indeed rare, so I'm sure it "works".
But until you define what the amp has to DO, we are just bench-racing a pig in a poke.
Regarding the circuit shown by Kodabmx:
Why do you use that arrangement for biasing the upper tube? It is way easier to return the grid resistor to the lower anode and get much higher input resistance.
Why do you take output from the lower tube anode? why not from the upper cathode?
I have been using a triode as active load for a cathode follower with very good result but I use a circuit more similar to the one posted by Elerion, (even though I don't use 12AX7 nor 6DJ8).
Why do you use that arrangement for biasing the upper tube? It is way easier to return the grid resistor to the lower anode and get much higher input resistance.
Why do you take output from the lower tube anode? why not from the upper cathode?
I have been using a triode as active load for a cathode follower with very good result but I use a circuit more similar to the one posted by Elerion, (even though I don't use 12AX7 nor 6DJ8).
That's just how I biased it. 333k is plenty for me. It's basically an Aikido Cathode Follower without the Aikido.
Taking the output from the anode linearises it at a cost of a slightly higher Zout.
From Broskie: "Loading a triode with the same triode—under the same cathode-to-plate voltage and idle current and with the same cathode resistor—works well to flatten the transfer curve out of that triode."
Taking the output from the anode linearises it at a cost of a slightly higher Zout.
From Broskie: "Loading a triode with the same triode—under the same cathode-to-plate voltage and idle current and with the same cathode resistor—works well to flatten the transfer curve out of that triode."
What you have quoted doesn't imply that it will be more linear if output is taken from the anode, it only says that loading a triode with another triode will give better linearity which I don't argue against. Can you link to the text by Broskie so we can see the whole text?
Anyway linearity doesn't get better if the triodes are the same, the higher the eqivalent load resistor is the better linearity you get so it can be better to use another triode, (e.g. one with higher u) as load.
Anyway linearity doesn't get better if the triodes are the same, the higher the eqivalent load resistor is the better linearity you get so it can be better to use another triode, (e.g. one with higher u) as load.
I had no target specs. I just wanted to play around to try to understand some theory I've been reading about.
Several authors of tube amplifier books (like Blencowe, I.Popovich,...), state that CCS is better done with transistors (or at least, that triodes are just not good at it), which I'm sure it is right.
But I've read these affirmations, along some design examples with triodes as CSS, where these triodes CCS have an AC impedance of around 100kohm, which obviously, isn't very interesting.
Right. My circuit can drive a 100V p-p signal bot only to 110kohm external load, then it gets out of current. Or, of course, a lower voltage swing if lowering this external load impedance. So, it seems not a very useful circuit, because a common cathode could do that, with gain.
Several authors of tube amplifier books (like Blencowe, I.Popovich,...), state that CCS is better done with transistors (or at least, that triodes are just not good at it), which I'm sure it is right.
But I've read these affirmations, along some design examples with triodes as CSS, where these triodes CCS have an AC impedance of around 100kohm, which obviously, isn't very interesting.
Right. My circuit can drive a 100V p-p signal bot only to 110kohm external load, then it gets out of current. Or, of course, a lower voltage swing if lowering this external load impedance. So, it seems not a very useful circuit, because a common cathode could do that, with gain.
As I said it's based on the ACF. Aikido Cathode Follower & More Triadtron
and Aikido Cathode Follower 2 & Impedance Multiplier Circuits and the PS-6
and Aikido Cathode Follower 2 & Impedance Multiplier Circuits and the PS-6
If you need gain, you can direct-couple the input to the cathode follower to the anode of a preceding grounded cathode voltage amplifier stage, keeping a protection diode between cathode and grid. This will work well as long as your previous stage has the plate sitting around 1/2 of the supply voltage. Use a tube appropriate for the gain you need, and run something like a 6DJ8 for the cathode follower and associated load triode. Need less gain? run some feedback from the cathode follower output (after the output capacitor) to the cathode of the input stage, and don't bypass the cathode resistor.
I have built up a few of these, and for low gain applications (5~18) something like a 6SN7/6CG7 for the input, and a 6DJ8 for the output will work very well. Wrap a bit of feedback around it and you will have a very nice little low-THD preamp. Great for driving zero-gain mosfet follower/pass type amplifiers as well with a bit more gain with some easy parts selection.
Poorly doodled schematic-
bbcode img
I have built up a few of these, and for low gain applications (5~18) something like a 6SN7/6CG7 for the input, and a 6DJ8 for the output will work very well. Wrap a bit of feedback around it and you will have a very nice little low-THD preamp. Great for driving zero-gain mosfet follower/pass type amplifiers as well with a bit more gain with some easy parts selection.
Poorly doodled schematic-
bbcode img
Last edited:
What resistor value did you use for the load-triode cathode resistor to ground?
It depends on what sort of voltage you run. From memory I used something like a 270~330R with a 225~250 supply voltage. It can be made larger or smaller to reach the desired idle current that works for you with different tubes or supply voltage. An LED of the appropriate voltage will work too, but will lower the impedance of the pseudo-CCS we are trying to accomplish here, but not enough to really matter. The cathode resistor on the upper triode can be anything from 47~220 ohms if you like, or can be safely (usually) omitted entirely. It basically functions as sort of a stopper resistor for use with longer interconnects.
Circuits like this sometimes can have DC at the output while warming up and the capacitor is charging- a shorting relay or some sort of dethumper circuit may be necessary in some situations or when used with sensitive gear.
I intentionally left out measurements, parts values, etc to show the basic idea. It's a nice adaptable circuit than can be made to work with many different tubes by simply choosing the parts to fit what you have on hand. It ain't perfect, but it sounds damn fine
My personal favorite "overkill" version that I've put together ran 375v regulated supply, 6SN7 input and outputs, and featured a startup mute delay circuit. Power supply was shunt regulated with VR tubes (2x 0D3, 1x 0A3 in series, raw supply voltage before regulation was 430 volts) and it used an industrial control transformer for the power transformer, with a separate filament transformer. It ran just enough feedback to get a nice clean transparent sound, I think something like 1/10 divider ratio (I can't do dB calc in my head well) and it was dead silent without signal.
Circuits like this sometimes can have DC at the output while warming up and the capacitor is charging- a shorting relay or some sort of dethumper circuit may be necessary in some situations or when used with sensitive gear.
I intentionally left out measurements, parts values, etc to show the basic idea. It's a nice adaptable circuit than can be made to work with many different tubes by simply choosing the parts to fit what you have on hand. It ain't perfect, but it sounds damn fine
My personal favorite "overkill" version that I've put together ran 375v regulated supply, 6SN7 input and outputs, and featured a startup mute delay circuit. Power supply was shunt regulated with VR tubes (2x 0D3, 1x 0A3 in series, raw supply voltage before regulation was 430 volts) and it used an industrial control transformer for the power transformer, with a separate filament transformer. It ran just enough feedback to get a nice clean transparent sound, I think something like 1/10 divider ratio (I can't do dB calc in my head well) and it was dead silent without signal.
You get biggest advantage when going from a simple cathode follower loaded only by its own cathode resistor to the same follower loaded by something that is significantly higher.
When you increase the load to higher and higher values the improvement is less and less, this was described already in Valley and Wallman , "Vacuum tube amplifiers, chapter 11" Using a triode as load with a very small cathode resistor is however a waste, with a cathode resistor of only 150 ohm a 6DJ8 is equivalent to a resistor of about 5k, using a cathode resistor 10 times higher would increase that eqivalent resistance up to almost 50k and linearity would be much better. Even a cathode resistor of 1.5k would only waste about 22V which can be easily compensated for.
When you increase the load to higher and higher values the improvement is less and less, this was described already in Valley and Wallman , "Vacuum tube amplifiers, chapter 11" Using a triode as load with a very small cathode resistor is however a waste, with a cathode resistor of only 150 ohm a 6DJ8 is equivalent to a resistor of about 5k, using a cathode resistor 10 times higher would increase that eqivalent resistance up to almost 50k and linearity would be much better. Even a cathode resistor of 1.5k would only waste about 22V which can be easily compensated for.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.