Below is the latest version of my 6B4G PP amplifier. I have decided on the 6N6Ps for the first stage as they sound more natural than the 6C45pi (in this case - I have a use for the 6C45s driving 845s or 813s where they are better than the 6N6Ps...)
I have been working through the Royal Navy Wireless Telegraphy Handbook 1938 edition and there is a lot of interesting stuff in there. Prompted by TL comments to examine the WE designs more closely and by Lynn Olsen's comments on AA, I paid particular attention to signal return paths in each stage and to the use of neutralizing capacitors to help overcome the input capacitance limitations of triodes.
Reading the RNWTH reinforced my feelings that valves do not care what circuit topology they are deployed in but purely respond to the ‘instantaneous’ voltages applied to their terminals i.e. they work in grounded grid, grounded cathode and grounded anode mode simultaneously and it is up to us to implement a circuit that maintains the correct external conditions on the valve to achieve the results we require. Hence if we want a valve to work in grounded cathode mode – we must ensure that signal input and output currents are referenced to the cathode and, for signal, the cathode is at ground. Hence the input return must be coupled to the cathode and the output return must also be coupled to the cathode - with the cathode coupled to the amplifier ground. Failure to do this allows the psu and other stages to interact with the stage under consideration. Note: Ideally this coupling is at all ac frequencies (not just audio) i.e. any frequency other than dc (0Hz). This lead me to add the capacitors between center tap of input and output transformers and the cathode. This cleaned up the sound leaving it more relaxed and more dynamic.
Adding the neutralizing capacitors to the output stage opened up the treble, removing a bit of hardness that was there. Doing the same on the input stage had no effect that I could hear…
Hope this is useful…
Ciao
James
I should add that the fixed bias can be varied for each side of the PP independently to allow the balance to be tweaked. This isn't shown in the drawing as it complicates working out the signal grounding scheme...
I have been working through the Royal Navy Wireless Telegraphy Handbook 1938 edition and there is a lot of interesting stuff in there. Prompted by TL comments to examine the WE designs more closely and by Lynn Olsen's comments on AA, I paid particular attention to signal return paths in each stage and to the use of neutralizing capacitors to help overcome the input capacitance limitations of triodes.
Reading the RNWTH reinforced my feelings that valves do not care what circuit topology they are deployed in but purely respond to the ‘instantaneous’ voltages applied to their terminals i.e. they work in grounded grid, grounded cathode and grounded anode mode simultaneously and it is up to us to implement a circuit that maintains the correct external conditions on the valve to achieve the results we require. Hence if we want a valve to work in grounded cathode mode – we must ensure that signal input and output currents are referenced to the cathode and, for signal, the cathode is at ground. Hence the input return must be coupled to the cathode and the output return must also be coupled to the cathode - with the cathode coupled to the amplifier ground. Failure to do this allows the psu and other stages to interact with the stage under consideration. Note: Ideally this coupling is at all ac frequencies (not just audio) i.e. any frequency other than dc (0Hz). This lead me to add the capacitors between center tap of input and output transformers and the cathode. This cleaned up the sound leaving it more relaxed and more dynamic.
Adding the neutralizing capacitors to the output stage opened up the treble, removing a bit of hardness that was there. Doing the same on the input stage had no effect that I could hear…
Hope this is useful…
Ciao
James
I should add that the fixed bias can be varied for each side of the PP independently to allow the balance to be tweaked. This isn't shown in the drawing as it complicates working out the signal grounding scheme...
Attachments
Hi James,
Interesting post and design. I've been looking at the 6C45 as possible drivers, oddly enough for 813's and GM70's, but it was the drive requirements that put me off. I don't recall seeing neutralising capacitors like this used much before, and I've never tried it. I might just have to now and see how it sounds.
I've also read Lynn's comments on AA with interest, but have yet to apply the 'WE' style bypasses in a finished circuit. The plan is to do it in the 813/GM70 boatanchor, if I can ever get the output stage DC filaments/cathode bias to work to my satisfaction; ac isn't quiet enough.
Your third paragraph I agree with.
Cheers
Interesting post and design. I've been looking at the 6C45 as possible drivers, oddly enough for 813's and GM70's, but it was the drive requirements that put me off. I don't recall seeing neutralising capacitors like this used much before, and I've never tried it. I might just have to now and see how it sounds.
I've also read Lynn's comments on AA with interest, but have yet to apply the 'WE' style bypasses in a finished circuit. The plan is to do it in the 813/GM70 boatanchor, if I can ever get the output stage DC filaments/cathode bias to work to my satisfaction; ac isn't quiet enough.
Your third paragraph I agree with.
Cheers
Hi Brett,
Thank you for your comments.
TL pointed me in the direction of the neutralising capacitors on a WE schematic. It seemed a neat idea and whilst reviewing my RF notes from 25years ago I found a reference to the RNWTH for theory. That it works at audio as well as RF is neat but not really unexpected.
I'll post my 6C45PP IT 845SE soon. It's working really well and uses +/- 400V supplies for the output stage to avoid 800V to ground in the circuit... The IT allows me to separate the grounding schemes from driver to output stage - another RF feature... I want to try 813s in triode mode next as I think they might beat the 845s...
I was fortunate to acquire some 10-14V 8amp. dc regulated (and RF filtered) supplies for the 845 heaters...
ciao
James
Thank you for your comments.
TL pointed me in the direction of the neutralising capacitors on a WE schematic. It seemed a neat idea and whilst reviewing my RF notes from 25years ago I found a reference to the RNWTH for theory. That it works at audio as well as RF is neat but not really unexpected.
I'll post my 6C45PP IT 845SE soon. It's working really well and uses +/- 400V supplies for the output stage to avoid 800V to ground in the circuit... The IT allows me to separate the grounding schemes from driver to output stage - another RF feature... I want to try 813s in triode mode next as I think they might beat the 845s...
I was fortunate to acquire some 10-14V 8amp. dc regulated (and RF filtered) supplies for the 845 heaters...
ciao
James
James D. said:
Look forward to seeing it. For my driverstage I was also looking at PP, CCS loaded 6C45's running at 200V/35mA driving the IT in parafeed, as well as a simple standard IT config.
I think you may be right about the 813's, but I'm biassed...
Also look at the GM70 triode and the GK71 pentode. Cheap as dirt from Gintaras or Anthony Welsh.
Nice. Are you using fixed or cathode bias?
Currently I'm leaning towards a variant on the John Camille +/- 5V design used in his 211 amp published in Sound Practices a while back.
Cheers
Hi Brett,
Just posted the sketches for the signal path and bipolar psu.
Op points on the 6C45s is as yours... 200V, 35mA and aimed for -2.25V on the grid - fixed bias individually adjustable...
On the 845 I'm running 700V at 90mA and -85 Volts cathode bias. 11K OPT.
I may well change the 845 to fixed bias once I build some confidence with it!!!
The GM70 doesn't look as linear as 813 or 845? I don't know the GK71 at all - more research
I've scratched my SP CD and can't get it to read at present...
ciao
James
Just posted the sketches for the signal path and bipolar psu.
Op points on the 6C45s is as yours... 200V, 35mA and aimed for -2.25V on the grid - fixed bias individually adjustable...
On the 845 I'm running 700V at 90mA and -85 Volts cathode bias. 11K OPT.
I may well change the 845 to fixed bias once I build some confidence with it!!!
The GM70 doesn't look as linear as 813 or 845? I don't know the GK71 at all - more research
I've scratched my SP CD and can't get it to read at present...
ciao
James
Re: ...apples with pears...
James,
I am using AC heating. For SE, 20 volt AC heaters would be a bugger.
Your power supply is intriguing. I have not given it too much thought so far, but cannot quite see the advantage of 2 chokes in the zero volt line.
At present, I'm converting my power supply to use hollow-state rectifiers. Have just ordered some EY81 and PY500's. He he
Why? Why not...
Cheers,
James D. said:snip...
Why aren't you using GM70 instead of 845 in your amp
Do you like my bipolar psu for the 845?
ciao
James.
James,
I am using AC heating. For SE, 20 volt AC heaters would be a bugger.
Your power supply is intriguing. I have not given it too much thought so far, but cannot quite see the advantage of 2 chokes in the zero volt line.
At present, I'm converting my power supply to use hollow-state rectifiers. Have just ordered some EY81 and PY500's. He he
Why? Why not...
Cheers,
John said...
To be honest neither can I, It felt right from a symmetry point of view and is modeled on 3 phase power filters... Actually I haven't got any in the 0V line at present - I have to buy some more before I can try it...
I managed to acquire some 83uF and 166uF 1000V polyp. delta config. capacitors so the configuration was sort of forced as it seemed a shame to short one capacitor of the delta out to parallel up the other two... It is working very well on the bread board.
I thought you would go for PY500 after your comments in the euro diodes thread.
ciao
James
To be honest neither can I, It felt right from a symmetry point of view and is modeled on 3 phase power filters... Actually I haven't got any in the 0V line at present - I have to buy some more before I can try it...
I managed to acquire some 83uF and 166uF 1000V polyp. delta config. capacitors so the configuration was sort of forced as it seemed a shame to short one capacitor of the delta out to parallel up the other two... It is working very well on the bread board.
I thought you would go for PY500 after your comments in the euro diodes thread.
ciao
James
James,
Yes it has a nice symmetry about it. And if you have the components, why not?
I forgot to add that another thing against the GM-70 is the socket.
Most people seem to convert (drill out) septar sockets, which seems a little unreliable to me. Only later, did I find a US company that can supply the real thing.
Nice design. Why is the 22uF cap that size? Should it be 0.22uF ?
Cheers,
Yes it has a nice symmetry about it. And if you have the components, why not?
I forgot to add that another thing against the GM-70 is the socket.
Most people seem to convert (drill out) septar sockets, which seems a little unreliable to me. Only later, did I find a US company that can supply the real thing.
Nice design. Why is the 22uF cap that size? Should it be 0.22uF ?
Cheers,
Guys,
It does indeed.
There's a Yahoo group for it with a lot of great members. The group started out dedicated to the GM70 but has evolved into the Boatanchor Amp Club open to all with an interest in big heavy amps. There are some building 212 and GM100 monsters.
Here's a GM70 datasheet
I'm still hunting for some copper anode versions.
How much hum do you have at the output? Even in PP I couldn't make mine quiet enough, but if you're doing OK in AC, maybe I'll try again.
Would you mind sharing who can do the sockets?
I'm using a big metal cap clamp to hold mine in place, and some brass crimp-on connectors pushed on to the pins. I figured it's worth the hassle because these aren't tubes you can roll.
James D. said:
It does indeed.
There's a Yahoo group for it with a lot of great members. The group started out dedicated to the GM70 but has evolved into the Boatanchor Amp Club open to all with an interest in big heavy amps. There are some building 212 and GM100 monsters.
Here's a GM70 datasheet
I'm still hunting for some copper anode versions.
dhaen said:
How much hum do you have at the output? Even in PP I couldn't make mine quiet enough, but if you're doing OK in AC, maybe I'll try again.
dhaen said:
Would you mind sharing who can do the sockets?
I'm using a big metal cap clamp to hold mine in place, and some brass crimp-on connectors pushed on to the pins. I figured it's worth the hassle because these aren't tubes you can roll.
Brett,
Hum:
Last time I checked it was the magic 5mV. My speakers are quite inefficient, though I can't put a figure to them.
GM-70 socket:
This is the post I remember from AA:
http://db.audioasylum.com/cgi/m.pl?forum=tubediy&n=19072&highlight=gm70+socket&r=&session=
It actually doesn't look promising because of the minimum order.
There has been a lot of sicussion on AA & DIY which is searchable.
Cheers,
Hum:
Last time I checked it was the magic 5mV. My speakers are quite inefficient, though I can't put a figure to them.
GM-70 socket:
This is the post I remember from AA:
http://db.audioasylum.com/cgi/m.pl?forum=tubediy&n=19072&highlight=gm70+socket&r=&session=
It actually doesn't look promising because of the minimum order.
There has been a lot of sicussion on AA & DIY which is searchable.
Cheers,
Hi John,
5mV is a fraction loud on very high eff speakers, but I still might revisit AC, and see if I can better the 2mV I had.
I've spoken with Doug and other Boatanchor members and the Brenner's socket is apparently a modified septal, and not all that great. Then there's the $US100 min order. My solution might be the best, and certainly the cheapest.
Toodlepip
5mV is a fraction loud on very high eff speakers, but I still might revisit AC, and see if I can better the 2mV I had.
I've spoken with Doug and other Boatanchor members and the Brenner's socket is apparently a modified septal, and not all that great. Then there's the $US100 min order. My solution might be the best, and certainly the cheapest.
Toodlepip
Sorry all guru's are busy at present, but will I do?
Yves,
Neutraliising caps are fitted to couteract or cancel the Miller capacitance of triodes.
I believe the term comes from RF electronics, where the technique is widely used.
The idea is to feed an "equal and opposite" signal to the grid, thereby "neutalising" the efect of the Miller capacitance.
Miller capacitance is due to the proximity of the grid to the anode (plate). It is worse in high mu valves.
Cheers,
Yves,
Neutraliising caps are fitted to couteract or cancel the Miller capacitance of triodes.
I believe the term comes from RF electronics, where the technique is widely used.
The idea is to feed an "equal and opposite" signal to the grid, thereby "neutalising" the efect of the Miller capacitance.
Miller capacitance is due to the proximity of the grid to the anode (plate). It is worse in high mu valves.
Cheers,
...it's a positive improvement...
Hi Yves,
As they are connected anode to grid between a push-pull pair they are providing positive feedback and frequency dependent feedback at that to 'help' drive the input capacitance of the triode. This would be a modern description.
Traditionally it is designed to increase the high frequency current drive into the triodes input capacitance providing extra current drive at higher frequencies when the driving stage may be starting to run out of steam. Hence the intention is to 'neutralise' the high input capacitance of the output triode.
It is an old technique (1920s first use) more commonly found in RF circuits before pentodes took over. However Western Electric, amoungst others, used it in PP audio circuits.
Lynn Olsens mentions it in his Rosetta Stone article. A Rosetta Stone for Audio . He found that it caused too much overshoot on squarewaves but I have found it helps in my circuit. It is very component variable and I am using quite different components from Lynn. As always try it and see YMWV... (Your Milage Will Vary).
ciao
James
Hi Yves,
As they are connected anode to grid between a push-pull pair they are providing positive feedback and frequency dependent feedback at that to 'help' drive the input capacitance of the triode. This would be a modern description.
Traditionally it is designed to increase the high frequency current drive into the triodes input capacitance providing extra current drive at higher frequencies when the driving stage may be starting to run out of steam. Hence the intention is to 'neutralise' the high input capacitance of the output triode.
It is an old technique (1920s first use) more commonly found in RF circuits before pentodes took over. However Western Electric, amoungst others, used it in PP audio circuits.
Lynn Olsens mentions it in his Rosetta Stone article. A Rosetta Stone for Audio . He found that it caused too much overshoot on squarewaves but I have found it helps in my circuit. It is very component variable and I am using quite different components from Lynn. As always try it and see YMWV... (Your Milage Will Vary).
ciao
James
...practice makes...
Hi John,
The 22uF capacitor connected to the negative grid supply was arrived at experimentally... Origionally I had tghe 2uF and 22uF the other way around but tried swopping them and prefered the sound that way around. It doesn't make sense to me and my lab. is in pieces so I can't measure what is really going on...
Any suggestions welcome - 'cos I'm confused!
ciao
James
Hi John,
The 22uF capacitor connected to the negative grid supply was arrived at experimentally... Origionally I had tghe 2uF and 22uF the other way around but tried swopping them and prefered the sound that way around. It doesn't make sense to me and my lab. is in pieces so I can't measure what is really going on...
Any suggestions welcome - 'cos I'm confused!
ciao
James
very clever...
...the neutralising cap technique, and those who
chipped in too Sure, you qualify, John
Regarding the 2uF and 22uF, does it have anything to
do with the resonance of the capacitor with the inductances
of the transformers? If there is a particular frequency band
that sounds particularly good with the swap, maybe it
is an indication of this effect.
Yv
...the neutralising cap technique, and those who
chipped in too
Sure, you qualify, JohnRegarding the 2uF and 22uF, does it have anything to
do with the resonance of the capacitor with the inductances
of the transformers? If there is a particular frequency band
that sounds particularly good with the swap, maybe it
is an indication of this effect.
Yv
Hi Yves,
I drew out the signal input circuit loop to look for resonances, see below. Is this what you had in mind?
I can't immediatly see where the 22uF Cbypass would help reduce a resonance effect. Have you any idea what is going on here? Any suggestions would be appreciated.
ciao
James
I drew out the signal input circuit loop to look for resonances, see below. Is this what you had in mind?
I can't immediatly see where the 22uF Cbypass would help reduce a resonance effect. Have you any idea what is going on here? Any suggestions would be appreciated.
ciao
James
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.