6V6 line preamp

[wrenchone]

I don't know if there is any interest in this one, as it uses both nasty sand and evil feedback, but check it out if you're curious.

https://www.diyaudio.com/community/threads/the-evil-sandman-preamp.155459/

This design uses the 6/12AQ5, the little sibling of the 6V6 with similar characteristics. I have since refined the design to use 6P1P, as they're lots cheaper...

 

[rongon]

Well, that's a lot of information there!
Have we decided whether or not LTspice simulations are reasonably predictive of real-world circuit behavior, at least in regard to this line stage circuit?

 

[wrenchone]

If you're talking about the "Sandman" design, I actually built the thing, lit it up, and tested it on the bench - no simulation - the same is true for its 6P1P version. I simulate my solid state designs all the time before I build them up and test them, but I've been too lazy to import the tube models into Spice for simulation there.

 

[rongon]

Oh, sorry. I got out of sync with the thread.
I was referring to the comparison of simulations with the built Salas 6V6-triode preamp.

 

[Salas]

Other's and my LTspice simulations verify the original design's tests reasonably close as we saw. -55 to -58dB 2ndH.

 

[rongon]

Interesting.
By increasing the B+ to 450V, increasing the Ra to 15k, and then adding plate-grid NFB, I was able to get the 6V6 circuit down to below 0.01% at 1V output, with about 4x gain -- in a simulation, of course.
I hope I can get around to whacking one of these together fairly soon. I have an ugly old chassis I can reuse, which has a suitable power transformer mounted on it. I could use a good line stage with 4x gain from a toob.

 

[Salas]

I once plate loaded it with a Mosfet Cascode, enhancment mode totem with LEDs for Vgs. I could measure around 20dB gain and surely less THD but the sound went south and I reverted it. Kinda tight and trebly plus bit hissy. I expect better results in your case because keeping the plate load a resistor.

 

[Salas]

Keep in mind that some tubes will have bit better or worse THD and hiss. You could see the RCA was different to the Russian tubes in my cathode follower FFTs (#4054 #4055).

 

[rongon]

Re: MOSFET CCS: Yes, I can definitely see that happening. Especially the 'hissy'. I'm thinking I'd use a big wirewound resistor instead of a CCS for the plate load. Also, I find that many times I prefer the sound I get using the old-school Rk//Ck combination over an LED for cathode bias. I know there's no good reason for that, but that's what I like. I most recently ran into that using a triode-wired 12HL7 for a 'spud' headphone amp (just the triode into an OPT, that's it). I could hear a more 'hissy' sort of thing happening with paralleled green LEDs in place.

I'll check out the tube comparisons, thanks. I have a large selection of vintage 6V6s, although most are pulls. I also have several new 6P1P, including a few 6P1P-EV. Unfortunately I don't have any 6AQ5 or others from that family.

 

[kodabmx]

I use MO for plate load...
6P1P will just drop right in to a 6V6 circuit in most cases. Too bad about the bastard pinout.

 

[rongon]

I have used metal oxide resistors for plate loads. I don't like them there. Big wirewounds sound a bit smoother to me, less 'sizzle' to the high frequencies, which I like. I have no idea why that should be, but that's how I hear it.

 

[wrenchone]

Maybe the MO resistors vary with applied voltage.

 

[kodabmx]

Or the wirewound ones are inductive and the MO usually aren't?

 

[wrenchone]

Metal oxide resistors are made usually with deposited tin oxide, sometimes with antimony oxide added to vary the oxide film resistivity. They have better thermal stability than carbon comp, but worse than metal film. In tube circuits, there may be a change in resistance vs. applied voltage - this is certainly the case with the high voltage variants of these resistors,, though to be fair, if they are used in HV voltage dividers, the values will be in hundreds of Mohm to Gohm, which will impact stability due to the formulation needed to get that sort of resistance. The voltage coefficient is a function of voltage gradient, so longer resistors are better in that respect. as would also be the case for multiple resistors in series to attain a given value. It may be instructive to do a shootout with a plate load consisting of several series MO resistors as compared to a single wirewound.

 

[Koonw]

I settle for these settings, it's about as much details one can hope to see, with Bohman Windows.

.tran 0 {simtime} {dlytime} {timestep}
.options plotwinsize=0
.options method=gear
.options numdgt=7
.param numcyc=150
.param dlycyc=1
.param FFT=18
.param simtime=numcyc/Freq+dlytime
.param dlytime=dlycyc/Freq
.param timestep=(simtime-dlytime)/2**FFT
.four {Freq}  V(out)
.param Freq=1k

As compression is disable, simulation is slow, use ".save V(out)" command to speed up simulation. Ensure that all compressions are turned off.

 

[rongon]

I tried both sets of directives.
The end result is very similar.
The output voltage is ever so slightly different, by about 1mV peak.
The THD is very close to the same for both; 0.126% at 1V rms out, using the Ayumi 6V6 model.
Unless I'm misunderstanding something, it looks like both serve the purpose well enough.

 

[Salas]

That makes sense

 

[Salas]

I had simulated this design once. I found it in my files. It adopts NFB. Showing 6dB & 9dB gain options (x2,~x3) but uses more moderate B+ and plate load values than what you plan. Maybe useful to you.

 

[Salas]

*Rser 5k is a classic series type volume pot's worst case output impedance. Occurs at half voltage divider setting for 20K nominal value pot. RL is for simulating the preamp's THD when driving 20K load like a solid state power amp. Its not an actual output capacitor DC discharge termination resistor.

 

[jean-paul]

Mmmm that standard 20 kOhm potentiometer can not be found anymore I just found out