Thread revive!
Since Alex has taken his toys and gone home, I don't think he should mind my using this thread for my own purposes.
I'm gathering parts to assemble Gary Pimm's PP47 amp that uses "partial feedback". I prefer "plate to grid" feedback, but whatever.
One thing that occurred to me while reading the TubeCad article on this topology is that the 6AU6 driver tube may not be the best option as driver here. An EF86 (or PF86 if you're a cheapskate like me) would have higher Rp and be happier running at 1.5ma of current. At first I thought gain might be an issue, but then I realised it isn't voltage swing that's required, but current swing.
I noticed a post by Gary saying the 6AU6 was almost ideally suited with it's highish Gm and high Rp, but I'm wondering why high Gm is a consideration for the driver tube. 5ma/V would seem excessive given the current swing required to drive the output stage. Maybe I'm missing something?
Now the question is how to modify the circuit for the EF86. It seems to me that with the CCS in the tail of the driver stage, that the EF86 would practically "drop in". All that is required is a screen voltage adjustment until the idle plate voltage sits at the desired point. I'm not sure I'm thinking about this correctly, though, as I'm not familiar with a circuit like this where the driver stage hangs from the output stage.
What do you guys think?
Since Alex has taken his toys and gone home, I don't think he should mind my using this thread for my own purposes.
I'm gathering parts to assemble Gary Pimm's PP47 amp that uses "partial feedback". I prefer "plate to grid" feedback, but whatever.
One thing that occurred to me while reading the TubeCad article on this topology is that the 6AU6 driver tube may not be the best option as driver here. An EF86 (or PF86 if you're a cheapskate like me) would have higher Rp and be happier running at 1.5ma of current. At first I thought gain might be an issue, but then I realised it isn't voltage swing that's required, but current swing.
I noticed a post by Gary saying the 6AU6 was almost ideally suited with it's highish Gm and high Rp, but I'm wondering why high Gm is a consideration for the driver tube. 5ma/V would seem excessive given the current swing required to drive the output stage. Maybe I'm missing something?
Now the question is how to modify the circuit for the EF86. It seems to me that with the CCS in the tail of the driver stage, that the EF86 would practically "drop in". All that is required is a screen voltage adjustment until the idle plate voltage sits at the desired point. I'm not sure I'm thinking about this correctly, though, as I'm not familiar with a circuit like this where the driver stage hangs from the output stage.
What do you guys think?
Some links:
This is the version of Gary's amp I'll most likely start with:
http://home.pacifier.com/~gpimm/47pp_8.gif
Here is the Tubecad article on partial feedback amplifiers:
http://www.tubecad.com/march2001/index.html
This is the version of Gary's amp I'll most likely start with:
http://home.pacifier.com/~gpimm/47pp_8.gif
Here is the Tubecad article on partial feedback amplifiers:
http://www.tubecad.com/march2001/index.html
jeff mai said:
That is interesting. We have just "discovered" triode strapped EF86 as driver for SE EL84. I was chatting with Bill Perkins about this amp and he suggested battery biasing the suppressor grid (IIRC 3.5V negative wrt ground) instead of just strapping it to the cathode... then he suggest 6AU6 would be a good substitute for EF86 -- which tickled me pink, since i've got almost no EF86 and a box of 6AU6 including some real pretty TFKs....
dave
Hi Jeff,
The EF86 is on my list of input tubes to try also. I've got a set of the Russian equivalents to try.
The real trick to figure out is what the transconductance at the operating point used in the amp really is. The transconductance of the 6AU6 is much lower at the low current operating point of the input stage. I was leaning towards higher transconductance as the current amplifiers, both the 47 and the Tabor just barley have enough gain to work with 2V rms sources. If the transconductance drops any lower you may need another gain stage, something I desparetly want to avoid.
Also up for testing is running the amplifier without an input transformer letting the differential input handle the phase splitting duties. This is brought about from 2 days of listening and testing both the Tabor and Karna amplifiers.
At the begining of the testing the Karna was showing a touch better low level detail at low volume. The Tabor was doing better at high volumes.
The obscuring of low level detail that was happening in the Tabor got me to trying things. The thing that helped the most was removing the cathode bypass cap on the output stage and running the amplifier in full differential mode. The down side to full differential mode on the output is loss of power (5 watts VS. 9 watts) but with my new 95 dB Cascadia speaker project I don't think I'll miss the 4 watts. The Karna was still just a touch clearner but they were really close.
Next up was transformer changes on the Karna amplifiers. We have 2 new sets of transformers to tryout. For the outputs we now have a pair of O/Netics level 3 output transformers with copper Litz wire wound secondaries. The second pair of transformers are Tribute input/phase splitters. These are 1 to 1+1 transformers wound on nanocrystaline cobalt toriods. The input transformers got wired into the input transformer select/phase/mute switch that was already in the amps. This switch allow us to switch between 2 different input transformers, invert the phase if wanted and have a mute position for swapping cables and such. The output transfomers replace a pair of O/Netics level 1
The combo of Tribute input and O/Netic level 3 output transformers is really something to hear. The Lundahl LL1635 and LL1671 interstage transformers are doing a great job of passing the signal on through the amp.
Getting back on track here... The Tribute input transformer is noticibly more transfparent than the Sowter 3575. This gets me to thinking about running the Tabor amp without an input transformer at all and letting the input stage do the work might sound better. The last time I tried this test and decided that I liked the transformer better was a rather long time ago working on the 47. With all the changes and improvements made since then the results of this test may be different. One more thing to try this weekend!
I also have a desire to try to add an interstage between the input stage and the output stage. In this instance the termination of the interstage would be "backwards" compared to a traditional triode amplifier- the termination resistance would be the low input impedance of the output stage and the high impedance side would be the output of the input stage. By having the interstage present brings up a couple of improvements/changes.
1. The power supply is more simple. Won't need the stacked supplies anymore.
2. With the input stage connected to the interstage transformer the input stage operating current is not tied to the current in the feedback resistors. We can crank up the current to get the transconductance higher and the input stage into a more linear part of the transfer curve.
The interstage idea is still just a thought experiment at this time.
Gary
The EF86 is on my list of input tubes to try also. I've got a set of the Russian equivalents to try.
The real trick to figure out is what the transconductance at the operating point used in the amp really is. The transconductance of the 6AU6 is much lower at the low current operating point of the input stage. I was leaning towards higher transconductance as the current amplifiers, both the 47 and the Tabor just barley have enough gain to work with 2V rms sources. If the transconductance drops any lower you may need another gain stage, something I desparetly want to avoid.
Also up for testing is running the amplifier without an input transformer letting the differential input handle the phase splitting duties. This is brought about from 2 days of listening and testing both the Tabor and Karna amplifiers.
At the begining of the testing the Karna was showing a touch better low level detail at low volume. The Tabor was doing better at high volumes.
The obscuring of low level detail that was happening in the Tabor got me to trying things. The thing that helped the most was removing the cathode bypass cap on the output stage and running the amplifier in full differential mode. The down side to full differential mode on the output is loss of power (5 watts VS. 9 watts) but with my new 95 dB Cascadia speaker project I don't think I'll miss the 4 watts. The Karna was still just a touch clearner but they were really close.
Next up was transformer changes on the Karna amplifiers. We have 2 new sets of transformers to tryout. For the outputs we now have a pair of O/Netics level 3 output transformers with copper Litz wire wound secondaries. The second pair of transformers are Tribute input/phase splitters. These are 1 to 1+1 transformers wound on nanocrystaline cobalt toriods. The input transformers got wired into the input transformer select/phase/mute switch that was already in the amps. This switch allow us to switch between 2 different input transformers, invert the phase if wanted and have a mute position for swapping cables and such. The output transfomers replace a pair of O/Netics level 1
The combo of Tribute input and O/Netic level 3 output transformers is really something to hear. The Lundahl LL1635 and LL1671 interstage transformers are doing a great job of passing the signal on through the amp.
Getting back on track here... The Tribute input transformer is noticibly more transfparent than the Sowter 3575. This gets me to thinking about running the Tabor amp without an input transformer at all and letting the input stage do the work might sound better. The last time I tried this test and decided that I liked the transformer better was a rather long time ago working on the 47. With all the changes and improvements made since then the results of this test may be different. One more thing to try this weekend!
I also have a desire to try to add an interstage between the input stage and the output stage. In this instance the termination of the interstage would be "backwards" compared to a traditional triode amplifier- the termination resistance would be the low input impedance of the output stage and the high impedance side would be the output of the input stage. By having the interstage present brings up a couple of improvements/changes.
1. The power supply is more simple. Won't need the stacked supplies anymore.
2. With the input stage connected to the interstage transformer the input stage operating current is not tied to the current in the feedback resistors. We can crank up the current to get the transconductance higher and the input stage into a more linear part of the transfer curve.
The interstage idea is still just a thought experiment at this time.
Gary
Gary P said:
Yes, this makes sense to me. With this in mind, do high Gm pentodes actually have much higher Gm than the EF86 at only 1.5ma?
Interesting idea with the transformer coupling of the two stages. It would definitely improve the operating conditions for the driver stage. I'm interested to hear how this experiment develops.
I'm still curious with regards to how the driver stage finds its operating point when hanging off the output stage in this direct coupled arrangement. Since my previous post I've realized the drop across Rfb must set the voltage difference between output plate and grid, so the only thing that would appear free to move is the driver stage cathodes. So dropping in the EF86 (aside from obvious things like socketing, heaters, etc.) would involve adjusting the screen supply voltage until you had the desired bias on the cathodes. Am I thinking clearly here?
Thanks for the reply Gary!
BTW, you might be interested to know that I intend to use the PP47 (once I've learned a bit more about it) to drive PP 300Bs through an interstage. Lynn mentioned this idea on AA a year or so ago, though I never heard that you guys tried it. I think it should work very well, and if it doesn't I can always slot in 45s for the 47s in a Karna-like arrangement.
Jeff
jeff mai said:
When I started the thread, I didn't call it "Alex's thoughts on partial feedback amplifiers" or something like that...
On the subject of driver valves, I was going to use a 6BX6 in this position ('cause they're cheap - were apparently used by the bucketload in old TVs), plus they look cool. ra is quite high as well, though they would appreciate a little more current (so the feedback resistor will probably have to be a little smaller).
The construction is pretty cool - there are two anodes surrounded by a mesh screen, and you can see the glowing cathode and the insides
Attachments
Brett said:
The gain is lower than the pentode connection used in the Mullard 3-3 circuit I built for a first project. I also plan on using (cue ominous organ sounds) mild NFB in the circuit.
Edit: Arghhh, you are of course right Brett. I pulled the wrong cicuit from the pile. Sorry, here's the correct one.
Attachments
Yep, you are thinking clearly. This amplifier design is a bit different with respect to what controls the operating points.
You can think of B+ as the reference point. The grid voltage of the output stage is controlled by the current drawn through the feedback resistors. This current is set by the CCS on the input stage.
Basically, the input stage CCS sets the output stage operating voltage. The output stage CCS (or cathode resistor) sets the output stage operating current.
On the input stage the plate voltage is set by the current through the feedback resistors. The grid voltage is 0 volts. The cathode voltage is influenced by the screen voltage. Raise the screen voltage and the grid bias voltage will increase. This effect applied to the input screens differentially is used to balance the amplifier.
Just got through removing the input transformers in the Tabor amplifier. Was a good move! Now it looks like the Tabor has just a touch more texture and resolution than the Karnas. Still have to listen some more to get a feel for the total change. As a summary, removing the output stage cathode bypass cap and the input transformer has resulted in a big improvement in overall sound quality.
Keep us posted on your 47 escapades. I think that it will make a killer driver stage.
Gary
audiousername said:
They do look cool - they'd be good subs for the original 6AU6.
I'm going to build the circuit with 6AU6s first, but I couldn't help wondering whether the EF86 might have advantages in this topology after reading the Tubecad article.
Is this a current project for you or is it in the queue?
Hi Brett,
The Cascadia speakers are a 2.5 way that uses 2 Eminence Beta 8's and the Hi-Vi RT8 tweeter. The box size is 4.25 cubic feet and is tuned around 40 hz. Output is ~95 dB/watt
The system is setup using an EBS alignment that has ~3dB droop before system resonance. The lower Beta 8 driver only does bass to make up for the drooping response and some of the baffle step.
The upper Beta 8 driver has a .4mh series inductor to compensate for the rising response of the Beta 8 above 1Khz. The upper Beta 8 is allowed to roll off naturally.
The RT8 has a 2 pole crossover consisting of a .24mh choke and a 2uf GE paper in oil cap. There is a 35 ohm resistor in parallel with the RT8.
The system is quite flat and very revealing. Lots of slam from the Beta 8's. Goes down to the low to middle 40's on the bottom and 25K on the top. The Beta 8 is a real good driver that is overlooked by most folks. The RT8 is a great hi rez tweeter. In the near future I'll get to see how the RT8 holds up to the Arum Cantus G1 ribbon when I setup a crossover to match the G1's to the Beta 8's for a friend.
This sysem can kick you hard then turn around and play harpsicord and other delicate instruments with real finnesse.
Gary
Hi Gary,
I hope you'll be able to find time to update your site to include the Tx coupled version too.
I ran a PP47 based on your design before the stacked supplies for about a year before I moved for a work contract. It's currently in bits awaiting a cabinet. It's really a nice amp, and was quite a bit better to my ears than the Auroras I built before. Might be those nice Mullards 47's I have...
Been thinking the relatively high output Z might work to advantage in some conical midhorns with compression drivers.
I hope you'll be able to find time to update your site to include the Tx coupled version too.
I ran a PP47 based on your design before the stacked supplies for about a year before I moved for a work contract. It's currently in bits awaiting a cabinet. It's really a nice amp, and was quite a bit better to my ears than the Auroras I built before. Might be those nice Mullards 47's I have...
Been thinking the relatively high output Z might work to advantage in some conical midhorns with compression drivers.
NOT A CHILD
Hi, guys!
I see that you have revived the thread (I got an automated notice on that, actually) so I was curious what was it like. Well, I would say it's just what it was before I took the liberty of taking my toys home.
Who is actually childish here? The schoolboys, or the schoolmaster? Pardon this offense, but it's just right for the "lingo" used on these pages.
No, Jason, you did not... but actually all the links mentioned in your opening post are linked to something I did or said in the previous developments. Except for TubeCad links -- and I do not want to comment J.B. more than I already did.
Unfortunately, the sound-quality of the tube used seems to have a huge effect on the sound of the circuit, however good or bad. Px marked tubes are always poor sounding as compared to regular Ex versions... not to mention that there are plenty of NOS USA and similia versions lying around. I do not believe the cheap...skatedness one is surrounded on most places: it seems everyone's thriving on social security?! I almost feel pity -- and to think that an IT guy in Serbia earns less than what he would have as unemployed in the States... I have to carry my burden -- the legacy of Mr. Milosevic -- with a smile, using cheapo tubes like 6B4G or RCA 6550 ST shape or Brimar 807 or... while you guys seem to have heavier burdens on your shoulders!
Good on you, Dave... ! But all this while, you have not yet converted any of your "mules" to RH84 and discovered whether it was worth the fuss?!
What a kind answer! Just like Brett...
Well, guys, I've devoted you a little time, trying to lift up a mirror and show you what is going wrong with your attitude, or your thread. I leave you on your own now, hoping that you will find something better to write about. Actually, you are already writing something else than partial feedback -- read the last 5-6 pages and you'll see that there is amasingly little, or none of partial feedback issues in this thread that was supposed to be dedicated to various approaches to partial feedback, and eventually stemming some new designs as a collective "brainstorming" and "working effort"!
That's it, thank you for your patience and effort in reading this post about partial feedback.
Alex
Hi, guys!
I see that you have revived the thread (I got an automated notice on that, actually) so I was curious what was it like. Well, I would say it's just what it was before I took the liberty of taking my toys home.
Who is actually childish here? The schoolboys, or the schoolmaster? Pardon this offense, but it's just right for the "lingo" used on these pages.
No, Jason, you did not... but actually all the links mentioned in your opening post are linked to something I did or said in the previous developments. Except for TubeCad links -- and I do not want to comment J.B. more than I already did.
Unfortunately, the sound-quality of the tube used seems to have a huge effect on the sound of the circuit, however good or bad. Px marked tubes are always poor sounding as compared to regular Ex versions... not to mention that there are plenty of NOS USA and similia versions lying around. I do not believe the cheap...skatedness one is surrounded on most places: it seems everyone's thriving on social security?! I almost feel pity -- and to think that an IT guy in Serbia earns less than what he would have as unemployed in the States... I have to carry my burden -- the legacy of Mr. Milosevic -- with a smile, using cheapo tubes like 6B4G or RCA 6550 ST shape or Brimar 807 or... while you guys seem to have heavier burdens on your shoulders!
Good on you, Dave... ! But all this while, you have not yet converted any of your "mules" to RH84 and discovered whether it was worth the fuss?!
What a kind answer! Just like Brett...
Well, guys, I've devoted you a little time, trying to lift up a mirror and show you what is going wrong with your attitude, or your thread. I leave you on your own now, hoping that you will find something better to write about. Actually, you are already writing something else than partial feedback -- read the last 5-6 pages and you'll see that there is amasingly little, or none of partial feedback issues in this thread that was supposed to be dedicated to various approaches to partial feedback, and eventually stemming some new designs as a collective "brainstorming" and "working effort"!
That's it, thank you for your patience and effort in reading this post about partial feedback.
Alex
Pentodes as Triodes
In reference to the EF86 as triode circuit.
It is an interesting concept using pentode as triodes. There are a number of tubes that are close to the plate impedance and Uhmos of a 6267/Ef86. 6SK7, 6SJ7, 5879, 6AU6, 6BA6 etc..
I have canvassed the books matching up pentodes and triodes from different time periods. Also 57, 58, 77 and 78 and others. I have found a number of pentodes like 6KT6 that is very like a 5879/EF86 pentode. This pentode is one-tenth the impedance of a EF86 and would give a lot lower output impedance in triode mode.
I have used five pin tubes like 37, 76, and 39/44 in Mike preamps and line stages and they are very warm and stable.
6AG7 is an older metal pentode that is a high-power amp but not a output that would work great in triode mode.
6EH and J 7 pentodes are built just like EF86 and are lower impedance and higher power. I have used 6EJ7 as Phono preamp 1st. stage and it sounds great..
You guys may have discovered a whole new area of triodes.
I always have a small amount of resistance between the Kathode cap/resistor and the Kathode. The tube likes a little neg feedback at the Kathode. It lowers the noise current and sweetens the tube up. I use values from 10 to 250 ohms. I found this a really usefull trick building Git-Amps to control Hissssss in High-Gain Git-Amps...See Pic of baby Deluxe head..
The different plates structures could yeild some interesting results in triode mode. I hooked the suppressor to the kathode in some circuits and got better transit response and warmer dynamics out of the pentode used.
It is interesting how you can discover tubes even though you have looked at the books for ages. I can't find 6BX6 except in a sub book. It subs to; 6BW7, 6BY7, 6EL7 and 6EC7.
Other untried duals are the "dissimilar double triodes" like 6FM7, 6EW7, 6FD7, 6EA7, 6EM7, and 6FJ7 that have really low plate impedance 10W second triodes. I have used these in headphoneamp/line stages with the big triode idling around 6.7W as Kathode follower. I would like to try two as a push-pull triode amp at 3K primary.
The 10W triodes in these tubes sound great and would make good drivers. Even 6CS7 or 6CY7 have 6.5W high-volt second triodes. I use the first triode for gain, and the second as Kathode follower.
ANyway, sounds like cool fun. Wish I had more time to build gear for myself. There are a lot of untried tubes out there.
In reference to the EF86 as triode circuit.
It is an interesting concept using pentode as triodes. There are a number of tubes that are close to the plate impedance and Uhmos of a 6267/Ef86. 6SK7, 6SJ7, 5879, 6AU6, 6BA6 etc..
I have canvassed the books matching up pentodes and triodes from different time periods. Also 57, 58, 77 and 78 and others. I have found a number of pentodes like 6KT6 that is very like a 5879/EF86 pentode. This pentode is one-tenth the impedance of a EF86 and would give a lot lower output impedance in triode mode.
I have used five pin tubes like 37, 76, and 39/44 in Mike preamps and line stages and they are very warm and stable.
6AG7 is an older metal pentode that is a high-power amp but not a output that would work great in triode mode.
6EH and J 7 pentodes are built just like EF86 and are lower impedance and higher power. I have used 6EJ7 as Phono preamp 1st. stage and it sounds great..
You guys may have discovered a whole new area of triodes.
I always have a small amount of resistance between the Kathode cap/resistor and the Kathode. The tube likes a little neg feedback at the Kathode. It lowers the noise current and sweetens the tube up. I use values from 10 to 250 ohms. I found this a really usefull trick building Git-Amps to control Hissssss in High-Gain Git-Amps...See Pic of baby Deluxe head..
The different plates structures could yeild some interesting results in triode mode. I hooked the suppressor to the kathode in some circuits and got better transit response and warmer dynamics out of the pentode used.
It is interesting how you can discover tubes even though you have looked at the books for ages. I can't find 6BX6 except in a sub book. It subs to; 6BW7, 6BY7, 6EL7 and 6EC7.
Other untried duals are the "dissimilar double triodes" like 6FM7, 6EW7, 6FD7, 6EA7, 6EM7, and 6FJ7 that have really low plate impedance 10W second triodes. I have used these in headphoneamp/line stages with the big triode idling around 6.7W as Kathode follower. I would like to try two as a push-pull triode amp at 3K primary.
The 10W triodes in these tubes sound great and would make good drivers. Even 6CS7 or 6CY7 have 6.5W high-volt second triodes. I use the first triode for gain, and the second as Kathode follower.
ANyway, sounds like cool fun. Wish I had more time to build gear for myself. There are a lot of untried tubes out there.
Attachments
Thanks for the reply Gary.
Looking at the datasheet I have for the 6AU6, the Gm rises significantly as you reduce the screen voltage while plate current is held constant. For convenience of comparing I used 2ma of plate current. Gm rises from 1.8ma/V @ 150Vg2 to 2.5ma/V @ 50Vg2. Unfortunately the grid bias voltage drops from -3.5V to -0.7V at the same time, so you lose input headroom.
A cool tube that I used to drive a 300B once was a 6AC7. It might be a good tube to try here. Where the 6AU6 appears to have Gm of around 1.6ma/V @ 1.5ma with 100Vg2, the 6AC7 has Gm of 2.5ma/V at the same operating conditions and even biases in the same spot - around -2.4V The Rp is similar. The downside is that it's a metal tube. GT versions were made but are rare. I wonder if there was a miniature equivalent made?
The EF86 is harder to judge (I can't find a datasheet plotting grid bias volts vs Gm at different screen voltages) but from the slope of the grid bias vs plate current curves it looks like the Gm will be somewhat less than 1.5ma/V at the same 100V g2 and 1.5ma plate current. That makes it only marginally worse than the 6AU6 in this respect.
Looking at the datasheet I have for the 6AU6, the Gm rises significantly as you reduce the screen voltage while plate current is held constant. For convenience of comparing I used 2ma of plate current. Gm rises from 1.8ma/V @ 150Vg2 to 2.5ma/V @ 50Vg2. Unfortunately the grid bias voltage drops from -3.5V to -0.7V at the same time, so you lose input headroom.
A cool tube that I used to drive a 300B once was a 6AC7. It might be a good tube to try here. Where the 6AU6 appears to have Gm of around 1.6ma/V @ 1.5ma with 100Vg2, the 6AC7 has Gm of 2.5ma/V at the same operating conditions and even biases in the same spot - around -2.4V The Rp is similar. The downside is that it's a metal tube. GT versions were made but are rare. I wonder if there was a miniature equivalent made?
The EF86 is harder to judge (I can't find a datasheet plotting grid bias volts vs Gm at different screen voltages) but from the slope of the grid bias vs plate current curves it looks like the Gm will be somewhat less than 1.5ma/V at the same 100V g2 and 1.5ma plate current. That makes it only marginally worse than the 6AU6 in this respect.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.