Hi Anatoly,
I think you refer to the use of R6 and R11. They are used to get a symmetrical entry point to Class B operation when one of the output stage mosfet cuts off during part of a cycle. I think this one requires non trivial measurement and listening to find a value with good sound. In simulation, the use of source degeneration on one side affect generated distortion spectra, an incorrect value will most likely affect sonic impression negatively.
On thermal issue, since both of the ZVN and ZVP have positive tempco, the negative tempco of the laterals should help reduce bias variation against temperature.
Another key point on this design is that the very low Vgs(off) of the laterals, together with square law transfer function, allows the output mosfets avoid cutoff up to about 20V peak, 8 ohm load at moderate bias of 800ma.
I think you refer to the use of R6 and R11. They are used to get a symmetrical entry point to Class B operation when one of the output stage mosfet cuts off during part of a cycle. I think this one requires non trivial measurement and listening to find a value with good sound. In simulation, the use of source degeneration on one side affect generated distortion spectra, an incorrect value will most likely affect sonic impression negatively.
On thermal issue, since both of the ZVN and ZVP have positive tempco, the negative tempco of the laterals should help reduce bias variation against temperature.
Another key point on this design is that the very low Vgs(off) of the laterals, together with square law transfer function, allows the output mosfets avoid cutoff up to about 20V peak, 8 ohm load at moderate bias of 800ma.
HI,
Why?
Do we have a bad linearity when i will run 4p1l with 325mA /4,2V in linestage?
No, but is better with parallel filaments, explanation from Anatoliy at the start of this thread
Either way linerarity of 4P1L is so good that isn't the challenge. Challenge is microphony if you're looking to use it as line stage. There are several recommendations to tackle that here so won't repeat
Interesting, after 400hours of continuous use the microphony is tamed significantly
Either way linerarity of 4P1L is so good that isn't the challenge. Challenge is microphony if you're looking to use it as line stage. There are several recommendations to tackle that here so won't repeat
Interesting, after 400hours of continuous use the microphony is tamed significantly
I have been pondering on this for a while. I have not found good references either. John Curl once mentioned that there was an attempt to give some kind of weighting in calculating THD value as an attempt to include the fact that some higher (especially odd) harmonics sound pretty nasty even at a very low level (-100dB?) while H2 is hardly noticeable at pretty high level (-60dB?) but was largely ignored. Still now, many of us including myself, lack understanding on the sonic effects of the trade off we are making.
What I use as guideline on my hybrid design is:
- trading for lower than 20dB H3 compared to H2,
- by accepting H2 at -60 to -80dB,
- pushing for lower than -160dB for H5 and above,
- with downward cascading spectra whenever possible
at 6.3V rms.
This of course is possible in simulation (with unfortunately not trustworthy result) but very difficult to measure. I would expect a lot worse distortion figures in real life. I also found some interesting tricks allowing cancellation of H2 and H3 based on choice of gyrator part, Rmu, bias current and other form of loading on the sim that I use to tweak output distortion spectra as similar as possible to the distortion spectra of CCS loaded valve stage.
Another point rarely discussed is the effect on the phase of harmonics on sonics. On the sims, I found that one of the difference between CCS loaded trioded 4P1L to the 26 is the 180 degree phase difference of H3 which may affect sonic. Salas had hiss and "SS sound" on CCS loaded trioded 6V6 while others are fine with their CCS loaded triode. However, RC and gyrator loaded trioded 4P1L and 4P1L with output taken from Screen has similar H3 phase to the 26. Trouble is how to measure.
Still another intriguing property generally unexplored is the variation of mu against signal level related to filament property as pointed by Steve Bench. It shows that the 26 and 864 or other oxide filament direct heated valves could potentially double function as an expander, somewhat partially compensate recording compression. I noted some difference on spatial cues between the TT and oxide filaments through listening but have not performed any sim nor measurement. But this one is easier to measure.
Do you have measuring set which is capable -at least- -180dB ????
Simulation is good thing, but it is only mathematics, as good as math models good.
As Wavebourn said: "The Devil is not so terrible as his math model is! "
Last edited:
Hi Bela,
Not at the moment, but since design of buffer involves high input signal, I'm toying with the design of some form of wider notch filter to "magnify" H5 and higher and multiple measurement average trick to reduce noise. As I stated, very difficult. Even with the very best available tools. But hopefully not impossible. I'll measure with what I have and worry about verifying the -160dB sim result after the design pass listening confirmation.
I understand that simulation is just a limited tool. For me, it is one of the necessary steps used to optimize a design. The sim points to some properties that highlight some differences of tube characteristics compared to SS. A lot of people sneered that tube amps produce higher THD and those who prefer tube as people who enjoy distortion. Quite true, but for direct heated tubes at listening level, the high distortion consisted of only H2 - H4 with H5 and above below the noise grass. I am sure that other factors also contribute to the good sound, but in this hybrid design, I'd like to focus my attention to the properties listed on my previous post.
I simulated many buffer topologies, some with sub ppm (about -140dB) level THD and found that when a tube voltage amplifying stage is inserted in front, output H5 and above distortion rose to -120dB or worse with various odd amount of phase shifts that could be noticeable at H3, H4 and H5. I wonder if these are the cause why hybrids are not so popular. Biggest problems I experience are taming the wild level of H5 and keeping phase shift of H3 in line.
I try to reduce the unavoidable trial and error tweaks on the build by understanding the expected behavior of parts in a topology a bit better by simming.
Would the grounding connections be configured different with filament bias (Coleman regs.)?
upgrades to my line stage
I picked up one of these second hand. It was made by Radu (abszero). I love it. But of course wondering how it could improve it. I have some basic kit building skills but all that means is I can follow instructions. Any suggestions on where to start? I know I can ask and/or send it back to the but I'd like to attempt myself and not pester Radu with a bunch of questions. thanks!
I picked up one of these second hand. It was made by Radu (abszero). I love it. But of course wondering how it could improve it. I have some basic kit building skills but all that means is I can follow instructions.
Any suggestions on where to start? I know I can ask and/or send it back to the but I'd like to attempt myself and not pester Radu with a bunch of questions. thanks!I'm thinking of building the Siberian G3. I'll first try to make it sound good as a headphone amp for some 32R headphones. If that doesn't work, I'll use it as a linestage in my main system. My first question: I have an ll1660 interstage transformer gapped for 18mA. Can I parallel the primaries on this to make it gapped for 36mA? Many thanks!
yes, you can, both alt T and alt V on the datasheet show how to connect (either side) of the windings in parallel, for double current handling (36mA) in your case.
you can combine alt V with alt N: alt V connects both C winding in parallel, these go to the tube anode. alt N connects all other windings (2x A , 2xB) in parallel. This way you will get a 2,25:1 transformer.
But do you want to use this to drive your 32R cans? That won't work well, I am afraid. 2,25:1 gives an impedance ratio of about 5:1. With 32R on the output, you get 150 ohms on the priomary, no good. Even in a 4,5:1 configuration you would get only about 600 ohms on the primary (4.5^2*32R).
I've been listening to the 2P29L for a week now, I have to say I like the sound. Not as much as 01a or 10Y, I think thoriated tungsten filaments have something special.
Anyhow, microphonic noise is minimal and gain stage is great. This is a fantastic driver
http://www.bartola.co.uk/valves/2017/05/20/2p29l-dht-preamp/
Cheers
Ale
Anyhow, microphonic noise is minimal and gain stage is great. This is a fantastic driver
http://www.bartola.co.uk/valves/2017/05/20/2p29l-dht-preamp/
Cheers
Ale
Hi Alex, I would like to know your opinion about 6e5p preamp compare with:
4p1L and 10Y. I like a lot the sound of 6e5p but is true that I don't listen the 4p1l or 10y.
There's no simple answer to that am afraid. The 6e5p has too much gain for my system so I don't use it currently. It all depends on the end to end system. I remember it sounded great when I used it before but I prefer the 6e6p.
I currently settled with a 01a/VT-25 driver stage into a 4P1L PSE. That is my best implementation so far.
I keep testing variants though!
I currently settled with a 01a/VT-25 driver stage into a 4P1L PSE. That is my best implementation so far.
I keep testing variants though!
Is there some tried and true setups for grid stoppers and grid leaks in a 4p1l triode connected preamp? I have a 100K volumpot and a 1K carbon on the grid and that's about it. It works well, but sometimes there is a very low scratching noise in one of the channels and I suspect the carbon resistor. Before changing it, it would be nice to know if there is a more optimal configuration of the resistors on the 4p1l tube.
The source of the noise mostly due to the oxidized contacts of tube and socket.
Try to pull down/push up tubes for several times.
Try to cleaning tube pins (for example DeoxIT).
Swap tubes. If the noise remains in the same channel, try to cleaning tube socket inner parts.