Why does my pre-amp sound like poo?

[m1garandusa]

I've built a very basic stereo pre-amp, which is just a gain stage, followed by a cathode follower. Both stages use a 6SN7. It has weak bass, and although I haven't measured it with a spectrum analyzer, I'd swear there is a dip in the upper bass register. Frankly it just doesn't sound that great. Please take a look at the schematic and let me know if I've committed a design flaw that will just suck the life out of a pre-amp.

The gain stage is as basic as it gets (common cathode). It is biased so that it's passing 10 mA (which seems to be a popular point for the 6SN7). It is DC-coupled to a cathode follower. The potential divider between the 2 stages should only attenuate the AC signal, leaving the DC voltage the same as at the anode of the gain stage. The diode between the grid and cathode of the cathode follower is only there to protect the tube from arcing during power-up when the grid is at B+ and the cathode is at ground, and should effectively drop out of the circuit once the cathode starts conducting. This stage settles in when the grid is about 4V lower than the cathode and produces an idle current of ~10mA.

The potential divider is there because I don't really need any gain in my system, but used a gain stage on purpose to add some 2nd harmonic magic. Output voltage (AC) after the divider is ~1/4 of the input voltage.

As you can see, the power supply is fully passive CLC and CRC topology.

Here is my background with building pre-amps. About a year ago, I built an e-bay knock-off of the Jadis J200, using 12AX7s. It had beautiful velvety mid-range, crystal-clear highs, and tight bass, if not particularly strong. However I abandoned that design as it had horrible ground-loop noise that I and a pro amp building friend could never track down. We spent about a month trying to track the inject point.

For my next attempt, I decided to use an octal triode, and found a mu-follower design on the web for the 6SN7. I built it, point-to-point and it worked without a hitch (no ground-loop). It produced full, thunderous bass, so I know the 6SN7 is capable of producing low-end "umph." But the pre-amp had no magic. It was dry and clinical, and didn't get me emotionally involved in the music.

This final attempt was based on the assumption that a basic triode common-cathode gain-stage would be less linear than a mu-follower, and perhaps might give me some warm, inviting music. But it just sounds blah, and there is no low-end. Have I unintentionally created some kind of high-pass filter? Any thoughts would be appreciated.

[nyazzip]

"...the pre-amp had no magic. It was dry and clinical, and didn't get me emotionally involved in the music"....

"...it just sounds blah, and there is no low-end..."

have you looked in to therapy for Seasonal Affective Disorder? tis the season

[m1garandusa]

I have one of those light boxes for SAD, so I'm good to go. In all seriousness, I do fully believe that mood affects the perceived quality of sound.

[Geek]

You have some good answers here:
http://diyaudioprojects.com/Forum/viewt ... f=9&t=4577

[EWBrown]

I'd bet that the 300K resistor from VA plate to CF grid is causing the freq response problem, along with the 1 uF cap and 100K resistor to ground,
which would insert some weird equalization curves.. Also, there should be a 47 K to 100K resistor on the output side of the 1 uF cap to ground,
to eliminate any DC voltage getting into the power amp. The diode connected between the CF grid and cathode may cause biasing issues.


This one by my friend Gary Kaufman works very well, it also uses a 6SN7, and B+ voltages not too different from yours.

http://www.the-planet.org/linestage2.html

If the gain is excessive, just delete the 330 uF / 35V VA cathode bypass cap.

I also tried one very similar to the "baby blue" 407A line stage, using WE396As, it sounds very nice, but has far more gain than necessary,
or desirable in most cases. It would be perfect for driving an EL34 or 6L6GC in an SE amp design.

I have another approach using the 6SN7 or 6CG7, with 150VDC regulated B+ and different value plate and cathode resistors, which sounds excellent,
it's a "mutant" of Doc Bottlehead's "Foreplay II" linestage.

Also uses DC voltage for the filaments, just to eliminate the last vestiges of any AC hum. I'll append the details later, it's late and my eyes are starting to cross...


/ed B

[m1garandusa]

Ed,
Thanks for the advice. I knew I wanted a 4-to-1 reduction in gain, but that leaves open an infinite number of resistor pairs to produce that ratio. I went with the highest values I could find in a carbon-film resistor, which was in the hundreds-of-kila-ohms range, which is how I came up with 300K and 100K. Would a different range of resistors produce a different EQ effect? I think I know the answer to that, but the one thing I'm horrible with is looking at a circuit with active devices and knowing which parts are going to introduce a high-pass or low-pass effect.

The diode should be invisable (open) to the circuit once the cathode goes positive relative to the grid. I've seen it used in a number of cathode follower circuits to prevent arcing at turn-on when the grid goes straight to B+, and the cold cathode is still at ground.

One thing I just noticed was that the 1uF orange-drop cap at the bottom of the voltage divider is actually .1uF (in fact they don't even make orange-drops that big). I'll try switching to a 1uF cap and see what effect that has.

BTW, regarding a grounding resistor after the coupling cap: Would a DC offset at the input of the power amp affect the sound? The first stage of the power amp has a 47K resistor to ground, so my gut reaction would be that it would take care of any DC offset, just the same as putting one after the coupling cap in the pre-amp. Then again, it's a Dynaco III, with a two-prong power plug, so who knows what potential "ground" is at the Dynaco vs. ground in the pre-amp

[Balle Clorin]

The LT spice simulation shows a a dramatic bass and midrange cut due to the lack of resistor/load after the output capacitor. More bass cut w1th lower loads, and you operting point gives 1% THD.

[Jerryz1963]

I might have missed it, but nobody mentioned two factors. The 22 K resistor. I experimented with the 6SN7 a lot. There is a reason you see 47K in 99% of the 6SN7 schematics. THe other factor that you could change in conjunction with the 22 K resistor is increasing B+. The 6SN7 likes a lot of voltage.

The 300 K resistor and cap is probably playing a big role. IMHO, you are better off capacitor coupling. I experimented with leaving out the capacitor ala Williamson and frankly couldn't tell the diff. I think it just unnecessarily complicates things.

It makes no diff what device you use as a buffer. There are high voltage MOSFETS that are very affordable that will fill the bill nicely for the output stage and do a BETTER job than a tube. Look at '"MOSFET FOLLIES." Buffers don't add anything to the tonal qualities. They just pass the signal along.