I did not want to spoil this thread http://www.diyaudio.com/forums/showthread.php?threadid=42259 with vacuum tubes.
Based on the Zeus amplifier by Susan Parker I drew schematics of a pentode version. Any thoughts about it?
Based on the Zeus amplifier by Susan Parker I drew schematics of a pentode version. Any thoughts about it?
An externally hosted image should be here but it was not working when we last tested it.
Grid 2 connection and input impedance
I thought that improving input impedance could be a good idea. In this case Grid 1 and Grid 2 change their electric potential in phase with Cathode and Grid 3. Thus the electric field around Grid 1 and Grid 2 should be constant regardless of the AC signal. Of course that is not exactly the case, but I suppose that the proposed connection would improve input impedance compared to a simple connection from Anode to Grid 2.
Yes I know that using a bootstrap capacitor from the Cathode to Grid 2 would be a simple way to do this, but then there would be a) feedback and b) a capacitor.
Kent Smith said:
I thought that improving input impedance could be a good idea. In this case Grid 1 and Grid 2 change their electric potential in phase with Cathode and Grid 3. Thus the electric field around Grid 1 and Grid 2 should be constant regardless of the AC signal. Of course that is not exactly the case, but I suppose that the proposed connection would improve input impedance compared to a simple connection from Anode to Grid 2.
Yes I know that using a bootstrap capacitor from the Cathode to Grid 2 would be a simple way to do this, but then there would be a) feedback and b) a capacitor.
Nikolas Ojala said:
Yeah, it isn't necessary. The main problem that the Zeus was designed to answer is the very large, and highly nonlinear, Cgs of the power MOSFET. The corresponding capacitance of a pentode is neither very large, nor is it nonlinear. Therefore, nothing needs to be done about it.
Secondly, a cathode follower topology isn't going to help you in the slightest to reduce the load resistance of that sucker. If the PP loadline says you need, let's say, 5.0K P-2-P in common cathode, then that's what you need in common plate as well. The VT doesn't care whether its load is connected from plate to AC ground, or whether it's connected from the cathode to AC ground. Either way, that OPT is going to see the same voltage across it. With the common cathode connection, you have voltage gain to help you get there. In a cathode follower topology, all that output voltage needs to come from the front end. If that's 20W into 5.0K, that's 447Vp either way, or 223Vp from each VT. The distortion you lose from the cathode follower output may very well be replaced by that of a front end that needs to swing the whole 447Vp-p, and it will in all probability require a much higher voltage power supply than the finals do.
"Yes I know that using a bootstrap capacitor from the Cathode to Grid 2 would be a simple way to do this, but then there would be a) feedback and b) a capacitor."
A cathode follower is already doing 100% feedback. You haven't eliminated it at all.
I'd say FUGEDABOUDIT: the Zeus is a solid state design to solve solid state problems that simply don't exist where hollow state is concerned.
What the idea was and still is
I think Susan Parker explained that the reasons were some other.Miles Prower said:
Look at the input transformers: They reduce the input impedance.
Excuse me Sir, but is that a straw man?
Now you see the reason for using a transformer. But that leads to the question of input impedance.
Feedback loops were eliminated. A loop contains some parasitic inductance.
There are some problems that are common to both worlds, such as distortion and noise.
Re: What the idea was and still is
Yes, they do. A Lo-Z is a very difficult type of load for a VT to drive. Also, 120 : 1 isn't what you'd call a standard ratio for an interstage xfmr.
If you don't believe me, perhaps you'll take Mr. Broskie's word for it.
What works quite well with a low voltage, high current, low impedance device like a power MOSFET isn't necessarily as workable with a high voltage, low current, high impedance device. There are far easier ways to implement a pentode, push-pull VT amp.
Nikolas Ojala said:
Yes, they do. A Lo-Z is a very difficult type of load for a VT to drive. Also, 120 : 1 isn't what you'd call a standard ratio for an interstage xfmr.
If you don't believe me, perhaps you'll take Mr. Broskie's word for it.
What works quite well with a low voltage, high current, low impedance device like a power MOSFET isn't necessarily as workable with a high voltage, low current, high impedance device. There are far easier ways to implement a pentode, push-pull VT amp.
Re: Re: Pentode Zeus
A connection from anode to G2 would make it a triode follower with a (potentially vastly) increased miller capacitance, so it would not be a pentode follower any more.
The phalacy in your idea of improving the input impedance is that you seem to forget that G2 is positive and draws signifficant current, the AC part of which will have to be sourced from the input - or in other words, G2 has a rather low impedance. This impedance will be further lowered by a factor of 14400 in your design, hardly benign for the previous stage. To make things even worse, the impedance of G2 is not constant with the input signal. In a pentode, the G2 current rises when the plate voltage becomes lower than the G2 voltage. The relationship is not linear and in a push-pull pentode output, either common cathode or anode, the sum of two G2 currents is NOT constant, in fact it is full of even order products of the plate current.
Actually, there is inherent feedback in every follower, regardless of Susan Parker touting that Zeus has no feedback. This is simply not true. A follower, in this case, uses the transconductance of the active element to drive current proportional with the DIFFERENCE OF THE INPUT AND OUTPUT TERMINALS, note the emphasis, and if you look, you will find it almost literally in the definition of 'negative feedback loop' - just replace gm (which is a form of amplification) with the more general 'amplification'.
Bootstrapping from the anode is a form of feedback. The reason being, although one might have the knee-jerk reaction of thinking it is positive feedback, it is actually a very low negative feedback mechanism. The reason for this lies in imperfect bootstrapping - the bootstrap cap is of fiite size, the follower does not have a perfect gain of 1 (because gm is finite too), and the G2 current does not stay constant with Vkp. As a consequence, increasing cathode voltage sees a decrease of the G2 voltage with respect to cathode, where it should actually remain unchanged. At small signal conditions and a well chosen pentode, this might not be a real problem. i repeat: might.
If you really want to make a pentode follower, and prevent the influence of G2 voltage and current to the output as much as possible, you might want to look into having two separate and floating G2 supplies. These would be connected between the pentode's G2 and K. because G2 current is returned to K, what you would have is the pure pentode curves, assuming the G2 power supply is low impedance and stable. G1 would be perfectly shielded and present a very linear capacitance. The follower output would get some of it's gm 'stolen' to drive the capacitance of G2 to plate. In essence, you get a vacuum tube 'MOSFET' and a true equivalent to the Zeus circuit with VTs. What you would still have as a VERY serious problem, is the input transformer. 1:120 from a relatively high impedance input may n ot even be posible with any sort of usable bandwidth - the input capacitance of the pentodes will be a completely negligible part of the total, which would be dominated by the interwinding and stray capacitances of the transformer itself.
Nikolas Ojala said:
A connection from anode to G2 would make it a triode follower with a (potentially vastly) increased miller capacitance, so it would not be a pentode follower any more.
The phalacy in your idea of improving the input impedance is that you seem to forget that G2 is positive and draws signifficant current, the AC part of which will have to be sourced from the input - or in other words, G2 has a rather low impedance. This impedance will be further lowered by a factor of 14400 in your design, hardly benign for the previous stage. To make things even worse, the impedance of G2 is not constant with the input signal. In a pentode, the G2 current rises when the plate voltage becomes lower than the G2 voltage. The relationship is not linear and in a push-pull pentode output, either common cathode or anode, the sum of two G2 currents is NOT constant, in fact it is full of even order products of the plate current.
Actually, there is inherent feedback in every follower, regardless of Susan Parker touting that Zeus has no feedback. This is simply not true. A follower, in this case, uses the transconductance of the active element to drive current proportional with the DIFFERENCE OF THE INPUT AND OUTPUT TERMINALS, note the emphasis, and if you look, you will find it almost literally in the definition of 'negative feedback loop' - just replace gm (which is a form of amplification) with the more general 'amplification'.
Bootstrapping from the anode is a form of feedback. The reason being, although one might have the knee-jerk reaction of thinking it is positive feedback, it is actually a very low negative feedback mechanism. The reason for this lies in imperfect bootstrapping - the bootstrap cap is of fiite size, the follower does not have a perfect gain of 1 (because gm is finite too), and the G2 current does not stay constant with Vkp. As a consequence, increasing cathode voltage sees a decrease of the G2 voltage with respect to cathode, where it should actually remain unchanged. At small signal conditions and a well chosen pentode, this might not be a real problem. i repeat: might.
If you really want to make a pentode follower, and prevent the influence of G2 voltage and current to the output as much as possible, you might want to look into having two separate and floating G2 supplies. These would be connected between the pentode's G2 and K. because G2 current is returned to K, what you would have is the pure pentode curves, assuming the G2 power supply is low impedance and stable. G1 would be perfectly shielded and present a very linear capacitance. The follower output would get some of it's gm 'stolen' to drive the capacitance of G2 to plate. In essence, you get a vacuum tube 'MOSFET' and a true equivalent to the Zeus circuit with VTs. What you would still have as a VERY serious problem, is the input transformer. 1:120 from a relatively high impedance input may n ot even be posible with any sort of usable bandwidth - the input capacitance of the pentodes will be a completely negligible part of the total, which would be dominated by the interwinding and stray capacitances of the transformer itself.
Re: Straw man
OK, you know better than me, you know better than John Broskie; better than anyone else on this whole forum. Go ahead and do it your way.
There: you win the e-penis waving contest. I concede. Are you happy now?
When it doesn't work, don't come back here and whine about it.
Nikolas Ojala said:
OK, you know better than me, you know better than John Broskie; better than anyone else on this whole forum. Go ahead and do it your way.
There: you win the e-penis waving contest. I concede. Are you happy now?
When it doesn't work, don't come back here and whine about it.
Another version
Here is a second try. There is a ECC85 voltage amplifier and a push-pull cathode follower with EL95 pentodes in this case. I'm not saying that EL95 would be an ideal pentode for that purpose, but I just have some of them.
Here is a second try. There is a ECC85 voltage amplifier and a push-pull cathode follower with EL95 pentodes in this case. I'm not saying that EL95 would be an ideal pentode for that purpose, but I just have some of them.
An externally hosted image should be here but it was not working when we last tested it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.