The Miller effect does not apply to voltage followers. The transfer capacitance of the FET does matter, but being low Rp/high gm the 'T7 is well suited to drive it. Look here, where sokke had problems using the HIGH Rp EF86. Sokke's Thread A tip of the hat to cogsncogs.
There's a nice exposition available on MOSFET use. Read MOSFET "Follies". MOSFET "Follies"
There's a nice exposition available on MOSFET use. Read MOSFET "Follies". MOSFET "Follies"
Actualy Douglas I aim to be close to class B for cool running.
Hacknet, I'm sure gate capacitance would be an issue with my original choice of anode resistors.
However there are changes afoot. Miller shouldn't be an issue with a 'follower.Yes I can see that now..
Ah, hadn't considered the difference. Class B is a whole new experience!
Will change to that now I realise the difference... although I LIKE grid current
Well, with enough current to drive it, I might as well. This circiut should even work with gassy valves... Well, at least the bias drift will be less.
Thanks Eli
Drawing revision coming later today.
Hi Chris,
I mainly go solid state, but I learned on tube amps and have always felt that there should be a better way to do the driver stages so I explore them every 5 or 10 years. I got back to tubes this time in thinking about building a guitar amp with my son. Also looking at rebuilding a few old stereo tube amps.
The tube prices came from Parts Express but I've not bought any.
Pete
I mainly go solid state, but I learned on tube amps and have always felt that there should be a better way to do the driver stages so I explore them every 5 or 10 years. I got back to tubes this time in thinking about building a guitar amp with my son. Also looking at rebuilding a few old stereo tube amps.
The tube prices came from Parts Express but I've not bought any.
Pete
anatech said:
Hi Pete,
I have a similar comfort level with solid state. Trying to get them sounding smooth is a challenge I enjoy. Solid state can be used to advantage in tube circuits as well.
With tube gear, I have been lucky to pick up some chassis' over the years. I experiment with them. The Eico I rebuilt took me completely by surprise. I don't want to mess with it any more.
Funny you mentioned a guitar amp. I have to build or rebuild one for my daughter. Trout has done a lot of work there.
Anyhow, give New Sensor a try at some point. I have to import them, which is a pain. Costs me a lot more than it does you. The EH line is just a good, reliable brand of tube. You may like them. The signal tubes are very quiet too.
-Chris
I have a similar comfort level with solid state. Trying to get them sounding smooth is a challenge I enjoy. Solid state can be used to advantage in tube circuits as well.
With tube gear, I have been lucky to pick up some chassis' over the years. I experiment with them. The Eico I rebuilt took me completely by surprise. I don't want to mess with it any more.
Funny you mentioned a guitar amp. I have to build or rebuild one for my daughter. Trout has done a lot of work there.
Anyhow, give New Sensor a try at some point. I have to import them, which is a pain. Costs me a lot more than it does you. The EH line is just a good, reliable brand of tube. You may like them. The signal tubes are very quiet too.
-Chris
John,
I've looked at the latest signal drawing revision.
C1 is too small. The 56 nF. value was chosen to protect the questionable O/P "iron" in El Cheapo. With decent O/P "iron", the high pass pole should be at about 17 Hz.
There are compromises in every design. It seems FET buffer drain current and 7591 grid leak resistor value are intertwined. A lower Id equates to less waste heat. Is an Id of approx. 15 mA. and a 47 KOhm 7591 grid leak resistor acceptable to you? If so, R7 and R8 are 24.9 KOhm Mills MRA-12s in series with 1.5 KOhm/1 W. Carbon parts. The Carbon parts are connected to the FETs to isolate them thermally from the NIWWs' heat.
BTW, 680 nF. coupling caps are "perfect" for 47 KOhm grid leak resistors, but they are a tad small for 33 KOhm GLRs.
I've looked at the latest signal drawing revision.
C1 is too small. The 56 nF. value was chosen to protect the questionable O/P "iron" in El Cheapo. With decent O/P "iron", the high pass pole should be at about 17 Hz.
There are compromises in every design. It seems FET buffer drain current and 7591 grid leak resistor value are intertwined. A lower Id equates to less waste heat. Is an Id of approx. 15 mA. and a 47 KOhm 7591 grid leak resistor acceptable to you? If so, R7 and R8 are 24.9 KOhm Mills MRA-12s in series with 1.5 KOhm/1 W. Carbon parts. The Carbon parts are connected to the FETs to isolate them thermally from the NIWWs' heat.
BTW, 680 nF. coupling caps are "perfect" for 47 KOhm grid leak resistors, but they are a tad small for 33 KOhm GLRs.
John,
I'm beginning to see how much heat generation is an issue with you.
You could revert to the cap. coupled gate/DC coupled source topology that you originally put in the CAD drawing. You'd have the annoyance of constructing a bipolar medium voltage PSU for the FETs, but I^2R losses in resistors will come down. I suppose that B2 operation for occasional peaks might be OK. 0.1 mA. of g1 current very occasionally is probably OK. I don't think a "steady diet" of g1 current is good for 7591 longevity.
A 2nd option is to forego FET buffers completely and use 68 KOhm 7591 grid leak resistors. The approach has some open questions. Will there be enough open loop gain? Is 68 KOhms too large for fixed bias 7591s? Someone more astute than I am will have to answer the questions.
I'm beginning to see how much heat generation is an issue with you.
You could revert to the cap. coupled gate/DC coupled source topology that you originally put in the CAD drawing. You'd have the annoyance of constructing a bipolar medium voltage PSU for the FETs, but I^2R losses in resistors will come down. I suppose that B2 operation for occasional peaks might be OK. 0.1 mA. of g1 current very occasionally is probably OK. I don't think a "steady diet" of g1 current is good for 7591 longevity.
A 2nd option is to forego FET buffers completely and use 68 KOhm 7591 grid leak resistors. The approach has some open questions. Will there be enough open loop gain? Is 68 KOhms too large for fixed bias 7591s? Someone more astute than I am will have to answer the questions.
Helllllloooooo, not enough open loop gain? for the faux-pentode cascode, it is gm*plate-load. Take a 5687 and run it to 10 mA/V gm, and put 20k plate loads on it. ten to -2 times ten to 4 is 100, or specifically in this case 200. Drive only one grid, and single plate gain is still going to be 100x. Drop this by a third from the grid circuit load....and you still have a single phase gain of ~65!
Look at Merlin v2, on ART's Group Build Projects folder for the hybrid cascode:
http://audioroundtable.com/GroupBuild/Projects/
cheers,
Douglas
Look at Merlin v2, on ART's Group Build Projects folder for the hybrid cascode:
http://audioroundtable.com/GroupBuild/Projects/
cheers,
Douglas
Good Grief! Splitter/driver gain of 65 is WAY too much. Gain of 12 would be enough if the 7591s were triode wired and no NFB of any kind was used. The 7591 is as easy to drive as a 6V6.
Also, an important point made by Jim McShane is that the non-linearity of the 'T7 works to advantage in a PP amp. Even order HD generated in the "finals" is cancelled. When that spectrum is combined with that of the SE 'T7, a PLEASANT sounding "waterfall" pattern (2nd>3rd>4th, etc.) for HD is the result.
IM distortion is a bigger villain than harmonic distortion. The local and loop NFB in the circuit are pretty effective at taming IMD. Stable g2 B+ also contributes to better distortion behavior in pentode mode.
Speaking of stable g2 B+, 8X glow diode bottles occupy a lot of "real estate". Perhaps adjustable SS series regulators would be a better choice. They go on the underside of the chassis and the adjustability could come in handy in maximizing overall performance. 255 V. g2 B+ is a GUESS on my part.
Also, an important point made by Jim McShane is that the non-linearity of the 'T7 works to advantage in a PP amp. Even order HD generated in the "finals" is cancelled. When that spectrum is combined with that of the SE 'T7, a PLEASANT sounding "waterfall" pattern (2nd>3rd>4th, etc.) for HD is the result.
IM distortion is a bigger villain than harmonic distortion. The local and loop NFB in the circuit are pretty effective at taming IMD. Stable g2 B+ also contributes to better distortion behavior in pentode mode.
Speaking of stable g2 B+, 8X glow diode bottles occupy a lot of "real estate". Perhaps adjustable SS series regulators would be a better choice. They go on the underside of the chassis and the adjustability could come in handy in maximizing overall performance. 255 V. g2 B+ is a GUESS on my part.
Eli,
Douglas,
Actually, before Eli suggested this concept I was deliberating over using a DRV134 as a phase splitter (within the feedback loop) for my own design - but that would have been truly hybrid.
My mind is still open, things are not set in stone until the parts are purchased, and even then there's some flexibility.
Well the reason for liking class B is more the "coolness" than the power supply cost. With the CCS's, buffers and regulators, the SS components generate almost as much as the valves. This amp has to be enclosed. I've worked out a convection system with the possibility of a low speed (quiet) fan for really hot days.
OK I've done that, but with more sensible values this time, I think. I could add clamping diodes to reduce the chance of grid current, but I think I'll chance it for now.
Well I had been thinking about those. Although they'd be really pretty all in a row, the repeatability of stab voltage might be an issue, especially for fixed bias valves, and I hadn't realised that the 225V was a guess! Yes I'll "make" a programmable zener instead.
Douglas,
Hey hey, common gate. Nice idea, but brings me into the territory of another stage. I rather burned my bridges when I named the thread "2 stage".
Actually, before Eli suggested this concept I was deliberating over using a DRV134 as a phase splitter (within the feedback loop) for my own design - but that would have been truly hybrid.
My mind is still open, things are not set in stone until the parts are purchased, and even then there's some flexibility.
dhaen said:Hey hey, common gate. Nice idea, but brings me into the territory of another stage. I rather burned my bridges when I named the thread "2 stage".
My mind is still open, things are not set in stone until the parts are purchased, and even then there's some flexibility. [/B]
That is not a multi-stage amp. Two exactly. The common gate you refered to is the upper sections of a cascode diff amp phase splitter. Instead of the voltage requirement of another valve, I used an even higher output Z MOSFET. They are in PP so the gate capacitance cancells itself for the most part. At the operating point, the capacitance is fairly constant...
With the E-Linear circuit, the higher the output Z of the driver stage, the more NFB you get with a given tap choice. If you want more or less, you use a different tap on the special OPTx. I have a 20-40% set of options.
You set gain like a pentode stage, pick the valve for desired gm, and load value for a good looking load line, and you're off.
Critical thing is to ref. the upper gate/grid to the cathode node. It sounds sterile and lifeless if you ref to ground.
cheers,
Douglas
Thanks Eli,
OK I'll do some research.
What I meant about the cathode feedback is that presumably since I'll be pumping -ve signal voltage into the cathodes, the Ra will be lowered. Does that mean that in pracice I could choose a slightly lower a-a impedance at the detriment of damping factor?
OK I'll do some research.
What I meant about the cathode feedback is that presumably since I'll be pumping -ve signal voltage into the cathodes, the Ra will be lowered. Does that mean that in pracice I could choose a slightly lower a-a impedance at the detriment of damping factor?
All,
I initially approached the Class "B" PP 7591 idea from an O/P power perspective. John has a LOW waste heat requirement and is working up the PSU and everything else from that perspective.
In a well ventilated, "normal", situation my thoughts for the B+ PSU are simplistic. A Hammond 715 trafo drives a pair of 6AU4 damper diodes. A choke I/P filter is employed. Icrit for a 10 H. inductor is approx. 41 mA. Even if a bleeder resistor proves necessary, there should be plenty of current available. The Hammond 715 is rated for 300 mA. DC working 24 X 7. The 715 is capable of providing more current on an occasional basis. That's EXACTLY the situation with Class "B". Also, the 6AU4 is designed for high pulse current. The combination looks good to me. Follow the initial LC section with paired (1/channel) LC sections. Channel separation will be good and ripple low.
I initially approached the Class "B" PP 7591 idea from an O/P power perspective. John has a LOW waste heat requirement and is working up the PSU and everything else from that perspective.
In a well ventilated, "normal", situation my thoughts for the B+ PSU are simplistic. A Hammond 715 trafo drives a pair of 6AU4 damper diodes. A choke I/P filter is employed. Icrit for a 10 H. inductor is approx. 41 mA. Even if a bleeder resistor proves necessary, there should be plenty of current available. The Hammond 715 is rated for 300 mA. DC working 24 X 7. The 715 is capable of providing more current on an occasional basis. That's EXACTLY the situation with Class "B". Also, the 6AU4 is designed for high pulse current. The combination looks good to me. Follow the initial LC section with paired (1/channel) LC sections. Channel separation will be good and ripple low.
Hammond 715 trafo:
I wish this reasonably priced transformer was available here.
Hammonds are available in the UK at Bluebell Audio , but the range is very limited.
I'll draw up Eli's power supply circuit as described.
I wish this reasonably priced transformer was available here.
Hammonds are available in the UK at Bluebell Audio , but the range is very limited.
I'll draw up Eli's power supply circuit as described.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.