To James or to Baxandall?

[Bob Richards]

Going the Baxandall route (James inside of a negative feedback loop), you get to use linear pots (less likely to have interchannel tracking issues and LOG curve variations from one pot to another), and the curves are symmetrical (if that matters and it may not), the pot gives you a flat response when centered, but because there's negatory feedback, the circuit may generate higher order harmonic distortion products. My mu follower tube circuits have the perfect product rolloff. I'd hate to lose that.

Going the James route (passive Baxandall - no negative feedback), the curves have some unusual curve shape change as the pots are turned, there's some interaction between the bass and treble control, turnover frequency shifts a little bit as the controls are turned, and the circuit as a whole seems likely to be noisier. Finding good LOG taper pots has proved to be a royal pain in the rumpster. None are 10% or 30% as per the choices in the Duncan James simulator program, which could have been a very big time saver.

I looked at a bunch of preamps and recievers from the "Golden Era" of tube stuff; McIntosh, Marantz, Scott, Heathkit, etc.etc.etc., to see what they were using. It appeared to be about half and half. Earlier stuff was more likely James, but that may be the stuff everyone thinks sounds so good... Any thoughts?

 

[Eli Duttman]

Tone controls are a "can of worms". Very few recordings benefit from their presence. If TCs are installed, they should be completely defeatable, as in switched out of the signal path. The uploaded Max Robinson implementation of Baxendal style is "unity" gain and well suited to switch in/switch out. IMO, 4X 12AX7s for a stereo setup is OTT. The bottle count can be reduced and the signal path simplified by replacing the cap. coupled cathode followers with DC coupled ZVN0545A source followers.

 

[DF96]

A good Baxandall circuit should not give problems with high order distortion because it has plenty of feedback. Use a high gain low distortion valve and don't try to push too much signal through it, just enough to overcome noise.

I have never tried it myself, but a mu follower or a common cathode bootstrapped by a cathode follower (they are almost the same circuit) might work quite well as you can use lower value pots than usual so get less thermal noise. An ECC83 bootstrapped by an ECC81 is one possibility - you would have nearly 40dB of feedback around a linear valve.

 

[artosalo]

Here is one example. Omit the first triode and you have unity-gain circuit that can feed as low as 33 kohms load.

An externally hosted image should be here but it was not working when we last tested it.

 

[jan.didden]

[snip] but because there's negatory feedback, the circuit may generate higher order harmonic distortion products. [snip]

Only if you have not enough feedback. Move past 30-40dB and all your harmonics are lower than without.

jan didden

 

[Bob Richards]

Thanks much for the above comments. I think I'll build a Baxandall, with the negative feedback, using a 6SL7 set at 2mA, with a Gary Pimm mu follower MOSFET current source plate load (I'm already using the Pimm circuit in my poweramp and it works perfect), and then look at it on a spectrum analyzer. If the distortion spectrum rolls off fast at all settings of the tone controls, it will be the best tone circuit I know of. The more recent (post 1960) highly regarded McIntosh preamps have the Baxandall w. feedback topology. How bad can it be? There will be a bypass switch, and I want to be able to compare bypass with tone enabled and set exactly flat. Getting the James circuit to give you a flat curve is difficult. BTW, I've been told by a very reliable source that the 6SL7 is significantly more linear than the 12AX7. Not quite as much gain as a 12AX7, but with a current source load I'll have the full mu of the tube (70).

 

[DF96]

You won't get the full mu of the valve as the tone network will load the valve. A mu follower type load will help, though it will only approximate a CCS if unloaded. To get full mu you need a CCS load, then a cathode follower to drive the tone network.

 

[Bob Richards]

You won't get the full mu of the valve as the tone network will load the valve. A mu follower type load will help, though it will only approximate a CCS if unloaded. To get full mu you need a CCS load, then a cathode follower to drive the tone network.

The Pimm I source has a "mu" output that can be used optionally. It's a low Z output. So you get the benefits of a current source plate load (tube gain approaching mu spec), and have a low Z output that can drive 20Kohms no sweat (based on the distortion spectrum shape and amount of distortion). It may be able to drive considerably less than 20K, but I ran out of time and didn't measure any lower than that. Maybe there's more to it. Guess I'll find out.