Need Help with First Tube Build - WE91 300B Parafeed Derivative

[EricS]

Thanks, guys!  It's been quite a bit of physical work in addition to wiring and troubleshooting.  Only have a few hours on them, but they sound great!  Midrange is something special and there seems to be more bass than with my solid state stuff.  Need to start enjoying them now...

[johnsonad]

Those look fantastic Eric, nice work!  I'm glad you did this project.  I remember PJ posted on AA about it years ago.  You're the first person I've read of that finished them, enjoy!

[EricS]

Aaron: I know Al Sawicki, also an AA member, built this design several years ago.  He's also been offering some build advice for me as my project has progressed.  So that's two that I am aware of. 

[Karl5150]

The inlay in the bases and the pinstripes play well together, visually. Great work, enjoy.
Karl

[2wo]

I have been following your build from the beginning. Very handsome....John

[EricS]

Thanks for sharing your kind thoughts, Karl and John.

Here are a few under the hood shots of the finished amp. 

Paul, are the specs for the grounding resistor bundled with anti-parallel diodes something you can share, or is this information proprietary?  I'm curious to explore other grounding configurations.  Not sure why I'm seeing increased noise in the 20-40Hz range with the thermistor in place, though I am unable to hear it in my super-sensitive headphones.  Maybe it's not worth playing around with...

[Paul Birkeland]

The antiphase diode/resistor/cap thing certainly isn't proprietary.  Ideally you'd want a 10-20 ohm 5W resistor, a pair of diodes across the resistor anti-phase wired (non-schottky diodes that can do 20+A peak current), and a little 0.1uF ceramic cap to finish it off.

[Doc B.]

Nice looking!

[Jamier]

They look fantastic! I guess the parts collecting was an adventure in itself.

Jamie

[EricS]

Thanks, Doc and Jamie!  It was a pretty solid 2-3 years to track down all of the necessary parts.  I got lucky and scored a pair of TFA-2004 Mexico Jr. M4 pin stripes direct from Mike.  The output transformers came from the great folks on the forum here.

I spent some time experimenting this morning with different grounding schemes (thermistors vs. diodes, resistors, caps) and none of the variations made much difference.  I did note that when using diodes, the value of the resistor made a small difference of 2-3dB here or there, but we talking about something like -94dB vs -96dB.  I could see it in the FFT, but certainly couldn't hear the difference with my headphones.  For the record, I tried 4R, 10R and 20R.  The 10R provided the lowest noise floor (at least for my environment), but only by a very small difference to the others.  Overall, there is no appreciable difference between diodes+resistor and a thermistor to isolate the ground.

I'm not sure what to make of the FFT data in the 10-40Hz range, though - it just seems to flop around like the end of a whip.  I don't know if my AC mains ground is just more "dirty" than others, or there is some house-borne contamination, or this is just part of what happens with this amp.  I  measured one of my solid state amps the same way and I'm seeing about -90dB in the 10-40Hz range as opposed to -55dB to -65dB with the tube amp.

[Paul Birkeland]

I haven't seen that kind of response myself, what happens if you run a frequency response sweep instead of the FFT?

[EricS]

I wonder if this is an artifact of new computer hardware I'm using.  I have an old computer with FFT software and a sound card that I'll try for a comparison.  I should also be able to do a frequency sweep with REW.

This 10Hz-40Hz behavior is similar on both amps.  For each of these measurements, the input RCA is shorted and the output is connected to an 8R resistor, then connected to my FocusRite analog to digital box.  All metal parts in the amp are grounded to the chassis except for the caps. 

[EricS]

Here is a 5Hz to 20kHz frequency sweep of the 300B amp that I made with REW.  Using the 1kHz dB level as a reference (about 61dB), it looks like the -3dB points (58dB) are about 21Hz and  about 16.5 kHz.  This is rather what I expected it would look like...

An FFT made with my old computer set up looked pretty much like the ones I've already posted: high noise floor in the 10Hz-40Hz region.

[EricS]

Now I'm thoroughly confused   

I made an FFT with my old computer that looked similar to my previous graphs - output peaks in the 10-40Hz region.

So, using the integrated sound on my new computer (just built it last week), I'm getting what looks like much better FFT performance.  Maybe the FocusRite (ebay purchase) that I was using previously was abused by it's first owner...  Or the built in sound in my new computer is just not as sensitive...  I'm really not sure what is going on here.  I think I need to make a few more comparisons.

Anyhow, here are three new FFT plots using the built in audio of my new computer:

1) Noise floor of internal sound card, no audio connections
2) Noise floor of 300B amp, power on, input shorted
3) Noise profile of 300B amp, 1kHz test tone

At first I thought the odd harmonics in the third FFT graph are characteristic of the pentode driver tube, but after exploring other amps, it looks like this is an artifact of the built-in sound on my computer.

Do these graphs look better?

[Paul Joppa]

The low frequency behavior looks very believable - the gentle rise (about 10dB/decade below ~400Hz) indicates i/f noise, which most electronic devices have.