I'm in need of a phase splitter with about 2X to 3X gain, so I've been goofing with loadlines and SPICEing ideas. I came up with this, using one of my favorite twin triodes, a 5687. I picked that one for its low ra, since I expect to drive powered monitor speakers with this.
Note R15, C11 and R15, C12. Those are NFB loops around U1 and U2 (the 5687).
Also note that there is no explicit grid leak resistor for the LTP. R2 and R3 take the place of that, combining with R15 and R16 to create a 'virtual ground' at the grids of the 5687 triodes.
The volume control at the input appears to work as expected (in simulation). I think it should work in reality because R3 lifts the grid of U2 above ground, so the volume control is working in a balanced configuration.
Whenever I come up with something that I haven't seen others use before, I get suspicious. There must be a serious downside that I can't see due to my inexperience, lack of education in physics, etc.
So, before I try to prototype this thing, what say you, the more knowledgeable folks around here? Will this work fed from unbalanced sources like a CD player or similar?
Note R15, C11 and R15, C12. Those are NFB loops around U1 and U2 (the 5687).
Also note that there is no explicit grid leak resistor for the LTP. R2 and R3 take the place of that, combining with R15 and R16 to create a 'virtual ground' at the grids of the 5687 triodes.
The volume control at the input appears to work as expected (in simulation). I think it should work in reality because R3 lifts the grid of U2 above ground, so the volume control is working in a balanced configuration.
Whenever I come up with something that I haven't seen others use before, I get suspicious. There must be a serious downside that I can't see due to my inexperience, lack of education in physics, etc.
So, before I try to prototype this thing, what say you, the more knowledgeable folks around here? Will this work fed from unbalanced sources like a CD player or similar?
Last edited:
-10V is probably too low for the current generator. I had to use some -30 to -40V in a similar setup.
You can spare C11 and C12 if you put degenerating resistors in the cathodes of U1 and U2. 100R, 220R, something like that.
You can spare C11 and C12 if you put degenerating resistors in the cathodes of U1 and U2. 100R, 220R, something like that.
I figure the LM317 CCS with 14V dropped across it should be OK. But maybe I could put another few volts across it. This circuit is likely to see no more than 2V rms signal at its inputs (absolute maximum), so it won't need to swing more than about 5V rms at its plates.
Degeneration would increase Zout, which is something I need to avoid here. But that's a good idea if this circuit were to drive a high impedance load of 100k ohms or more.
Edit to add:
Experimenting with degeneration from unbypassed cathode resistors...
Degeneration would increase Zout, which is something I need to avoid here. But that's a good idea if this circuit were to drive a high impedance load of 100k ohms or more.
Edit to add:
Experimenting with degeneration from unbypassed cathode resistors...
- To reduce the gain to 2.5x I had to use a 680R in each triode's cathode. Calculated Zout rose to 6k ohms.
- With the NFB arrangement Zout is less than 750 ohms. That's a very big difference.
Last edited:
If the voltage of the cathodes does not drop below .5V you may manage with a current mirror rather than a negative rail.
Hi Rongon...
If It was Me 🙂 i would have used A Cathodyne / SourceOdyne In front of your circuit,
( And ReUsed C11, C12, As Coupling from the cathodyne Phase-Splitter Stage, )
And if you need more gain then what you allready have, you could use other tubes with higher Mu
( Higher Ra ) Or Reduce The Amount of feedback you have around U1 & U2, as you don't need Low Z out, As Long as you buffer (witch is usual engeneering Practice anyway)
The stage with some kind of follower, or other Low Z output to UnLoad Your Tube Gain Stage..
And Or As Buffers to the Outside world, hook on a pair Of Cathode Followers, source followers, Mu-Followers, Gyrators, Or Alike, as it's usual not the best configuration To have tubes/Gain-Stages connected directly to the outside world, with capacative cables or other unknowns..... 🙂 🙂
If It was Me 🙂 i would have used A Cathodyne / SourceOdyne In front of your circuit,
( And ReUsed C11, C12, As Coupling from the cathodyne Phase-Splitter Stage, )
And if you need more gain then what you allready have, you could use other tubes with higher Mu
( Higher Ra ) Or Reduce The Amount of feedback you have around U1 & U2, as you don't need Low Z out, As Long as you buffer (witch is usual engeneering Practice anyway)
The stage with some kind of follower, or other Low Z output to UnLoad Your Tube Gain Stage..
And Or As Buffers to the Outside world, hook on a pair Of Cathode Followers, source followers, Mu-Followers, Gyrators, Or Alike, as it's usual not the best configuration To have tubes/Gain-Stages connected directly to the outside world, with capacative cables or other unknowns..... 🙂 🙂
Thanks.
I actually don't need any more gain. Just phase splitting with about 2x gain.
With the NFB in place, the output impedance looks to be 750 ohms. The 5687 triodes sink 15mA each. What kind of load would upset that?
I could use an LTP made from a 6SN7, put feedback grid-to-plate to get the gain down around 2x, and put source followers on the 6SN7 plates as buffers. The source followers would sink the current and provide the low Zout for driving the load.
I'd still need the NFB, though. A 6SN7 has way too much gain for my needs. Actually, what triode has a mu of 2? The closest would be something like a DHT, 2A3 or 45, maybe. Now things are getting really complicated. All that plate current and the expensive filament supply, worries about uneven current draw along the filament... That would have to be another thread (and it is: https://www.diyaudio.com/community/threads/dht-otl-linestage-tram-2.156854/ and https://www.diyaudio.com/community/threads/2p29l-preamp.336620/).
I actually don't need any more gain. Just phase splitting with about 2x gain.
With the NFB in place, the output impedance looks to be 750 ohms. The 5687 triodes sink 15mA each. What kind of load would upset that?
I could use an LTP made from a 6SN7, put feedback grid-to-plate to get the gain down around 2x, and put source followers on the 6SN7 plates as buffers. The source followers would sink the current and provide the low Zout for driving the load.
I'd still need the NFB, though. A 6SN7 has way too much gain for my needs. Actually, what triode has a mu of 2? The closest would be something like a DHT, 2A3 or 45, maybe. Now things are getting really complicated. All that plate current and the expensive filament supply, worries about uneven current draw along the filament... That would have to be another thread (and it is: https://www.diyaudio.com/community/threads/dht-otl-linestage-tram-2.156854/ and https://www.diyaudio.com/community/threads/2p29l-preamp.336620/).
If I use MOSFET source followers added to the original idea, I get what amounts to a distortion-less line stage, thanks to the NFB.
If I reduce the current set by the LM317 in the 6SN7 cathodes, 3rd harmonic creeps up. But the distortion stays incredibly low. I should run this simulation with different 6SN7 models to see if there's a huge variance in predicted outcomes. (The model I'm using is Adrian Immler's i4 version created from a batch of 6SN7WGTs).
I think the only advantage this circuit would have over an op-amp would be the glowing 6SN7 bottles on top of the chassis. I also have a vintage art- deco-styled choke I could throw on there, just for looks. Maybe add a 5Y3GT rectifier tube, so there's another hot bottle on display.
Believe it or not, the simulation is predicting 0.002% THD at 1V rms output. I know not to trust SPICE generated THD predictions, but still...
If I reduce the current set by the LM317 in the 6SN7 cathodes, 3rd harmonic creeps up. But the distortion stays incredibly low. I should run this simulation with different 6SN7 models to see if there's a huge variance in predicted outcomes. (The model I'm using is Adrian Immler's i4 version created from a batch of 6SN7WGTs).
I think the only advantage this circuit would have over an op-amp would be the glowing 6SN7 bottles on top of the chassis. I also have a vintage art- deco-styled choke I could throw on there, just for looks. Maybe add a 5Y3GT rectifier tube, so there's another hot bottle on display.
Believe it or not, the simulation is predicting 0.002% THD at 1V rms output. I know not to trust SPICE generated THD predictions, but still...
Hei Rongon... 🙂
There you have it, Much Better Circuit to Couple to external Loads, With a coveted Low Z Out, To Better Drive External Loads with Un-Known Paracitical L & C's.....
And with a cathodyne in Front ( Witch in my experience is the best and best balanced Phase Splitter there is )
It would be perfect IMHO, and much more room to tweek Gain Z out And Other parameters...
There you have it, Much Better Circuit to Couple to external Loads, With a coveted Low Z Out, To Better Drive External Loads with Un-Known Paracitical L & C's.....
And with a cathodyne in Front ( Witch in my experience is the best and best balanced Phase Splitter there is )
It would be perfect IMHO, and much more room to tweek Gain Z out And Other parameters...
And actually, for driving a line to a differential input, you don't even need a "balanced" output. All the differential input needs to work its magic is equal impedances on both polarities. You can do this with signal at a defined output resistance feeding pin2 and an identical resistor from pin3 to signal ground. Very common in prosumer gear, where it's called "impedance balanced".
All good fortune,
Chris
All good fortune,
Chris
By a gain of 2, I meant that my 1V unbalanced input needs to become 2V for each phase. I guess that's a gain of 4 then.
Hi Rongon,
With a cathodyne in front you could probably betterd that Low distortion by 12 to 2o Db, because of allmost IMHO, Perfect signal balance and therfor in better distortion cancelation,
But sadly less tube single ended caracter
Because cancelationt of only even harmonics H2, H3, H4, H5, H6, H7, H8, H9, H1o,
Hi Rongon.. 🙂
I Myself have a soundSystem ( Mostly Self Buildt )) Comprized of hybrid preamp, after a heavyle modded bheringer DCX Active filter with thre analog & ONE SP/DiF Input, + the analog-devices Sharc Prosessor on bourd, and a host of diverce tube, hybride tube-mose & hybrid mos-Tube,,, PreAmps,, Buffers,, PowerAmp GainStages........
On Your Post Rongon 🙂
The Best You Can Use is ConstantCurrentSource CCS fed shunt power suplly,, its the only best way to go rongon...
I Myself have a soundSystem ( Mostly Self Buildt )) Comprized of hybrid preamp, after a heavyle modded bheringer DCX Active filter with thre analog & ONE SP/DiF Input, + the analog-devices Sharc Prosessor on bourd, and a host of diverce tube, hybride tube-mose & hybrid mos-Tube,,, PreAmps,, Buffers,, PowerAmp GainStages........
On Your Post Rongon 🙂
The Best You Can Use is ConstantCurrentSource CCS fed shunt power suplly,, its the only best way to go rongon...
For simplicity sake, I'd just use a simple gain stage direct coupled to a source follower, just like half of your differential circuit, and cap coupled to a 4:1 (10K:600R) output transformer. You can tweak the feedback resistor to optimize gain. Now you have either balanced or single-ended output. And you save half of the differential circuit and parts.
Yeeeah 🙂
Thats The Beuty Theres so many topologys, and endless configuration possabillitys, So it's realy uP To preferences and topology / design Of what you make,,, Still it facinates me that there is only three fundamentel electronic principals that underlys everything we try to do in the Hi-Fi World and those are : :
L= inductance whether known or Paracitic
C= Capactitance also wheter known Or Paracitic
And Dielectric Constant In Insulating Materials
And to a degree R= Z / Ohz
Thats The Beuty Theres so many topologys, and endless configuration possabillitys, So it's realy uP To preferences and topology / design Of what you make,,, Still it facinates me that there is only three fundamentel electronic principals that underlys everything we try to do in the Hi-Fi World and those are : :
L= inductance whether known or Paracitic
C= Capactitance also wheter known Or Paracitic
And Dielectric Constant In Insulating Materials
And to a degree R= Z / Ohz
