SY,
I have just put together a cascode stage, valve on top, jfet underneath. It has enough gain to match two of my previous stages.
Resolution, WOW.
Allen Wright spoke of high transconductance and single electrons. I was interested in his write up but skeptical. Now, I know he knows his subject. Allen wright is in no doubt a genius, and you, SY, know a good amp when you hear one.
Dynamics, the word effortless springs to mind. I know that this is a characteristic of the cascodes vertical load line for the bottom device. I seem to have tried all the other topologies, and all the time it was a cascode I wanted.
Can anyone recommend devices to use? I am using a 12au7 and 2n5484 as I had these available. I tried a 12at7 but no thanks. I don't have a 6922 at this point. I may set up for octal if I hear it is worth it.
I have just put together a cascode stage, valve on top, jfet underneath. It has enough gain to match two of my previous stages.
Resolution, WOW.
Allen Wright spoke of high transconductance and single electrons. I was interested in his write up but skeptical. Now, I know he knows his subject. Allen wright is in no doubt a genius, and you, SY, know a good amp when you hear one.
Dynamics, the word effortless springs to mind. I know that this is a characteristic of the cascodes vertical load line for the bottom device. I seem to have tried all the other topologies, and all the time it was a cascode I wanted.
Can anyone recommend devices to use? I am using a 12au7 and 2n5484 as I had these available. I tried a 12at7 but no thanks. I don't have a 6922 at this point. I may set up for octal if I hear it is worth it.
Huh? Would someone please explain THAT one to me? I won't argue the subjective impression, but when someone tries to explain a desirable sonic attribute by saying that it's due to a loading condition that increases non-linearity, well... Please help me to understand why this condition would reveal dynamics more accurately than a high Z plate load. If the quiescent operating point is set where the cascode plate curves are more crowded, and you were to hit the stage with a dynamic peak, one side of the waveform would be "expanded" while the other side would be further compressed. That might sound like increased dynamics, but I think it's more like signal processing. To his credit I think, Allen tends to use his cascodes in differential configurations, and this will tend to cancel the even distortion mechanism. Is this phono amp cascode single-ended or differential?
I haven't tried Allen's particular combination of tube and FET, but with a slightly degenerated NTE458 on the bottom and an ECC88-type on top, my 25 year old old phono stage still managed to clock in at about 0.1% THD midband at 5mV in, totally dominated by second HD. The small signal swings help. I think I can do rather better now and I'd like to knock that distortion down a decimal place or two.
You might try a 6SL7 on top with an NTE458 on bottom, running at 2mA. The NTE has good transconductance at this lower current, is cheap, and easy to get.
You might try a 6SL7 on top with an NTE458 on bottom, running at 2mA. The NTE has good transconductance at this lower current, is cheap, and easy to get.
Brian Beck said:
http://www.passlabs.com/downloads/articles/cascode.pdf
This is a good read.
Thanks SY.
Whilst Nelson's article about cascodes is interesting, it is largely concerned with the problems of Class AB vs Class A transistor output stages and Early effect. Triodes are an entirely different kettle of fish and do not exhibit Early effect. As has been pointed out by Brian Beck, a triode with a vertical loadline suffers greatly increased second harmonic distortion. This distortion can be tamed either by operating at small signal levels (which is why cascodes can be useful as the first stage of an RIAA stage), or by cancelling the second harmonic in a differential pair.
I may try Mr. Wright's design some day, if I can find a reasonable way of purchasing matched FETs.
I didn’t mean to rain on the cascode parade. If it sounds good to you, that’s great. I was just jumping on the assertion that “effortless” dynamics could be explained by the vertical load line. Cascodes have the potential for certain other advantages, such as reduced Miller capacitance (and therefore high BW) and high gain, although the latter drives one to use higher upper plate loads, which increases the lower tube’s Miller capacitance, so these advantages tend to go in opposite directions. The parallels to the pentode keep coming back, except that cascodes can be quiet due to lack of screen partition noise.
SY, as you suggest, the 0.1% THD at 5mV is not a terrible situation, especially being primarily second HD, but I would think at that exceedingly low level, a conventional grounded cathode stage with high plate load would do much better.
SY, as you suggest, the 0.1% THD at 5mV is not a terrible situation, especially being primarily second HD, but I would think at that exceedingly low level, a conventional grounded cathode stage with high plate load would do much better.
Kevin Carter uses this as well for his K&K audio phono stage and for the Art Audio phono stage (which he designed). (The first stage for a moving magnet cartridge consists of a very high transconductance JFET / triode (6N1P) cascode amplifier.)
So to answer your question on what tube. The 6n1p seem suitable. (Very little microphonics as well by the way) I never found out which JFET he uses.
Brian,
Before I built this cascode I was running a conventional grounded cathode stage. Since I changed, I took my cd player from its position and put it away. An impulsive move - I just don't want to look at it right now. I guess I just know that if I do I'll wan't to completely rebuild it and I'm too busy right now. Hope this speaks for itself.
Before I built this cascode I was running a conventional grounded cathode stage. Since I changed, I took my cd player from its position and put it away. An impulsive move - I just don't want to look at it right now. I guess I just know that if I do I'll wan't to completely rebuild it and I'm too busy right now. Hope this speaks for itself.
No problem Brian.
As NP put it, the jfet is shielded from voltage variation. This is what makes the load line vertical. He says and I quote:
NP: Because voltage-induced nonlinearities take the form of "compressive" intermodulation, it was not surprising to discover the sonic effects of utilising cascode operation throughout a power amplifying system corresponded to an impression of a dynamic range capability considerably beyond what the rated power would suggest. This effect is pronounced at high transient levels and imparts a sense of effortlessness in the reproduction of demanding material.
As NP put it, the jfet is shielded from voltage variation. This is what makes the load line vertical. He says and I quote:
NP: Because voltage-induced nonlinearities take the form of "compressive" intermodulation, it was not surprising to discover the sonic effects of utilising cascode operation throughout a power amplifying system corresponded to an impression of a dynamic range capability considerably beyond what the rated power would suggest. This effect is pronounced at high transient levels and imparts a sense of effortlessness in the reproduction of demanding material.
lndm said:
Yes, but that's talking about a power amplifier, and you're at the opposite end of the spectrum. The main distortion from FETs comes from changing the current through them, not the voltage across them. And you don't even change much voltage. I'm not disputing that your phono stage may sound nice, but that NP paper has nothing to do with it.
OK, maybe you have found a limit to my current understanding. Until a week ago, I ignored the cascode circuit. Maybe you could tell me how it works?
Edit: I would especially like to understand what and why certain attributes are desired in each of the two active positions, and also why many stages don't use Ra > 3 x Rp x mu ?
Edit: I would especially like to understand what and why certain attributes are desired in each of the two active positions, and also why many stages don't use Ra > 3 x Rp x mu ?
lndm said:
The cascode is an ideal RF circuit. It works by placing two devices in series so that the upper device provides most of the gain and loads the lower device with the upper device's load divided by its voltage gain. This means that the lower device sees a very low resistance load (near vertical loadline) and that reduces its voltage gain. From an RF point of view, this is a good thing because it reduces Miller capacitance from the output of the lower device to its input. From an audio point of view, a low impedance load invariably increases distortion (have a look at output characteristics of FETs, pentodes, or triodes). However, if you use a cascode carefully, that isn't a problem; you need to keep the output signal small, and if you can use it in balanced form, then you can cancel the (mainly) 2nd harmonic distortion.
Allen Wright seems to speak the praises of the single ended version, I am tending to follow his topology.
For which device is transconductance more important (or is it both).
Could a pentode prove to be ideal for some reason?
The RDH4 suggests a cascode similar to 6sn7's wants an Ra of half a Meg, how do I calculate the best Ra for dissimilar devices - I want to maximise gain.
Is low Rp good for either of the devices?
For which device is transconductance more important (or is it both).
Could a pentode prove to be ideal for some reason?
The RDH4 suggests a cascode similar to 6sn7's wants an Ra of half a Meg, how do I calculate the best Ra for dissimilar devices - I want to maximise gain.
Is low Rp good for either of the devices?
Transconductance is important for the lower device because high transconductance reduces noise. For a phono stage, low noise is the primary consideration. It may be that you can juggle DC current to get an optimum compromise between noise and distortion. I hate to go against RDH4, but a 500k load resistor would be a disaster. Start by designing the lower stage for low noise, then juggle that with the upper stage for good linearity. Expect to need quite a bit of HT current.
Low Ra means low gain. Almost as low as SE. The RDH suggests a 33k Ra with a 6sn7 cascode would be like SE 6sn7 with Ra=1k65. The low on resistance of the fet must therefore contribute to the ability to use a low Ra but I'm not yet sure how.
BTW, noise performance is important but I do put high expectations on other aspects of performance.
BTW, noise performance is important but I do put high expectations on other aspects of performance.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.