triode comparsion circuit, anyone build anything like this?

[terminalgs]

I've been wanting to build something like this for a while.   Everytime I get a tube that tests okay on a tube tester but fails the auditory test in an amp,  I think how useful this might be.     Now I have about 100 various NOS and used twin triodes  that I've like to go through and pick out the best sounding ones,   I'm compelled to put something like this together.

http://i.imgur.com/DMAkV.png



the "8BD" octal preamp tube would be for 6SN7's 6SL7s 6SQ7s,  anything with the pinout "8BD".   The one next to it would be for "9A" pinouts:  ECC81, ECC83, 6092, 5814, 5751, 5962, 6211,  etc....  the last one would be the one that be easy to re-solder to any non-"9A" pinout  (5755, for example).

the idea is you would only use 1 of the top 3 tubes.   All three are wired in parallel.  technically, you wouldn't hurt anything  by installing 2 or 3, but it might over draw the filament (maybe a 6v fuse to protect the filament PT?).

I'd feed it a signal into the input (guitar on a loop, test signal,  beatles mono CDs, whatever) and you'd select between the 'left triode' or the 'right triode'  for the test tube.   For the reference tube, you place a similar "known good tube",,,  the toggle in the middle would switch output between the test triode (left or right)  and the reference triode.

the Mu selector would select a voltage divider pair of resistors to drop the signal to +12dB for a given Mu (100,70,60,50,40,20,15) and then the att. pot would cut from there.   (I'd need to calibrate those resistors for some consistent B+,  probably max B+.

the filter jack is a stereo jack that you plug in separate filter. almost like an effect loop, but the idea is I could easily plug in a jack that has a 15pf to ground as a noise/hiss cancellation measure.  (on the fence for this, it could be a toggle switch too).

the two transformers each have their own power plugs.  the filaments must be at 6.3v all the time, so PT always needs 120v.  The 240" B+ transformer could go to the wall -or- to a variac.  That way B+ could be anything between 0v and 300v, which I can monitor via the banana clip plugs ("B+" "left plate" "right plate").   The idea here is to also be able to use the system to see if a tube sounds better @ 120v on the plate or 180v on the plate.., etc.  (the problem I have with this --see schematic--  the CF output as voltage drops might lose fidelity ,  but if I'm A/B'ing to the reference triode,  I suppose the CF deficiency would be equal for both A and B.

The output goes into a good amp,  or if the input was a CD player (mono I guess), the output could be a stereo. 

Ultimately, the idea is to use my ears to evaluate the fidelity of various tubes before putting them in amps.

http://i.imgur.com/GHhYC.png


http://i.imgur.com/Juauy.png


what I'm I overlooking?  forgetting?  not considering?

[tubeswell]

For myself, I would probably just plug the tube in question into an amp and see if I liked how it sounded.

But experiments are fun. And if you find you don't use it after you've built it, then you can always recycle the parts.

[terminalgs]

For myself, I would probably just plug the tube in question into an amp and see if I liked how it sounded.

thanks for the reply.

that's what I've been doing so far,  and of course it is useful.  I think what has led me to this idea is hitting some walls with that method.   for my amps (AC15, a couple  Magnatones (M14 and M10A), a BF Bandmaster, and Gemini I,  installing tubes in V1 is time consuming,  especially if you want to A/B a couple of tubes (swap in, swap again..,etc) Also it's not a great way to test the fidelity of smaller Mu tubes,  and it's difficult to gauge usefulness for non-guitar amp gain applications. Plus it doesn't help me with non 9A basing tubes,,  (I have a bunch of 5755s, 5867s,  and 50+ 6SN7, 6SL7, and 6SQ7s to go through).

But experiments are fun. And if you find you don't use it after you've built it, then you can always recycle the parts.

yep.  plus I'm using independent transformers.  both bought as surplus box lots.  with sockets, caps,  and materials  I think I'd be in $30 or so.

[Willabe]

I think it's pretty cool.

Sounds like you have most of the parts laying around and you do have a lot of tubes to test.........   


           Brad         

[LooseChange]

Besides playing my Desktop Studio Champ, the next best thing I use it for is to test Preamp and Power tubes. Pop the tubes in, turn up the volume, listen, tap on it, play through it and listen again. Build yourself a desktop model and have an cool amp too.
By the way, great illustrations.

[PRR]

For fixed plate load resistor, the cathode-bias resistor should generally track with Mu. If fashion suggests 12AX7 at 100K + 1.5K, then 12AU7 should bias well/similar with 100K + 7.5K (better, 47K + 3.3K).

The large unbypassed cathode resistor swamps a lot of the naked-tube curvature. THD will be less than half.

> guitar on a loop, test signal,  beatles mono CDs, whatever

Tube "color flavor" is a LOT about the Signal Level (actually signal level as a fraction of power supply voltage). Soft guitar may be 0.02V, hot CD is 2.0V.... BIG difference. A bypassed 12AX7 at any sane B+ will overload on CD. 12AU7 any reasonable B+ with soft guitar will be loafing, and be "undistorted" (unless the tube is sick).

Your "Mu" attenuator does not do what its label suggests. Full-down is infinity not -6dB. And the 2-pole switch could just as well be a single-pole potentiometer connected switch. While you could re-rig this to do what it says, I suggest just putting in a volume pot with marks for loss of 15:1 and 50:1 (AU7 and AX7).

Node impedance at the "filter" node is undefined. Taking some possible approximations, 20K to 150K depending on tube and attenuator settings. So the example "15pFd" could roll-off at 74KHz or 555KHz. Aside from that uncertainty, it is highly likely there will be more than 15pFd of stray capacitance in all that switching (or in a real amp).

Cathode bias for cathode follower is awful thin: 100K bias resistor. You may be looking at it upside-down, tho even so 10K is a large bias resistor. As a general guide, steal a plate-load stage (such as 100K+1.5K) then put the 100K under the 1.5K and gridleak.

"calibrate" note seems to suggest gain varies with B+. It does, but very-very slightly.

All of this could be worked-up and elaborated into a full PhD Thesis Project. Find, control, and document all variables.

Said it before, say it a gain. Build a Champ with several sockets. If you don't fret about power tubes (or only use high-power outputs which work different from Champ), then just build the preamp and some kind of fault-free power-amp and modeling speaker.

[terminalgs]

For fixed plate load resistor, the cathode-bias resistor should generally track with Mu. If fashion suggests 12AX7 at 100K + 1.5K, then 12AU7 should bias well/similar with 100K + 7.5K (better, 47K + 3.3K).

The large unbypassed cathode resistor swamps a lot of the naked-tube curvature. THD will be less than half.

> guitar on a loop, test signal,  beatles mono CDs, whatever

Tube "color flavor" is a LOT about the Signal Level (actually signal level as a fraction of power supply voltage). Soft guitar may be 0.02V, hot CD is 2.0V.... BIG difference. A bypassed 12AX7 at any sane B+ will overload on CD. 12AU7 any reasonable B+ with soft guitar will be loafing, and be "undistorted" (unless the tube is sick).

Your "Mu" attenuator does not do what its label suggests. Full-down is infinity not -6dB. And the 2-pole switch could just as well be a single-pole potentiometer connected switch. While you could re-rig this to do what it says, I suggest just putting in a volume pot with marks for loss of 15:1 and 50:1 (AU7 and AX7).

Node impedance at the "filter" node is undefined. Taking some possible approximations, 20K to 150K depending on tube and attenuator settings. So the example "15pFd" could roll-off at 74KHz or 555KHz. Aside from that uncertainty, it is highly likely there will be more than 15pFd of stray capacitance in all that switching (or in a real amp).

fantastic input!!

Cathode bias for cathode follower is awful thin: 100K bias resistor. You may be looking at it upside-down, tho even so 10K is a large bias resistor. As a general guide, steal a plate-load stage (such as 100K+1.5K) then put the 100K under the 1.5K and gridleak.

ah, yes.  I reversed them, and actually the 10k should 1.5K like you pointed out.  When I first started using CFs for outputs for tube pedal boxes, I found an effects loop tube 'send' stage on a Marshall schematic and the 1M gridleak, 1.5K cathode R, and a 100K under it all was exactly what Marshall did... 

"calibrate" note seems to suggest gain varies with B+. It does, but very-very slightly.

All of this could be worked-up and elaborated into a full PhD Thesis Project. Find, control, and document all variables.

I wonder if there is any grant money out there for vacuum tube research? :-)

I hope to be thorough in my testing.   in absence of a DMM with a USB interface  (boy, that would be handy for this..),  I'll keep my laptop sitting nearby and record and label data (and matching tubes) with care...

Said it before, say it a gain. Build a Champ with several sockets. If you don't fret about power tubes (or only use high-power outputs which work different from Champ), then just build the preamp and some kind of fault-free power-amp and modeling speaker.

yep.  sounding more and more like the way to go.   I like  LooseChange's desktop Champ.  It might be the next place to go...

after processing the great feedback on this thread, here's what I have so far:

- Mu  voltage divider switching removed.
- aux filter moved to between coupling cap and volume pot (don't know if it'll be of use,  other than a way to sneak an o-scope probe into the chassis.
- added a front input level pot.   at full tilt, its the 1M grid leak for the input triodes, can be used to tame inputs (in hopes it'll help with PRR's comment for CD input).
-added switched cathode bypass caps (I've got a rotary 3PDT).
-added complex cathode resistor circuit.

this last one might be a way to also select cathode R values for various Mu tubes (re: PRR's comment about this issue).  the switch is a triple gained break-before-make rotary (I *think* it has the current capacity to handle:  its DuMont NOS).

I should get 100k, 1.45k, 2.15k, and 7.5k (If i can figure 8K resistors!!).  a side benefit of this might be the switch mode of 100K cathode resistor,,  a way to model a cathodyne PI ??? (also way I switch the cathode bypass caps).

here's what I have so far:  continued thoughts / concerns / issues??

http://i.imgur.com/I6htR.png (link is better res).



PRR, thanks so much for your input!!!

[terminalgs]

Besides playing my Desktop Studio Champ, the next best thing I use it for is to test Preamp and Power tubes. Pop the tubes in, turn up the volume, listen, tap on it, play through it and listen again. Build yourself a desktop model and have an cool amp too.

cool looking amp.  what PT and OT did you use?

By the way, great illustrations.

thanks!   google sketchup is a killer app.  lots of youtube 'howto's as well...  in sketchup, its 100% 3D  (move it around, get under it, etc..).  I id that in about
2hrs...   which might seem like a waste,  BUT it was 2hrs of me not impulsively using a soldering iron... or impulsively taking sheet metal to the drill press... 

[terminalgs]

Here is what I have so far...  It works well.  the banana clip sockets at plate and cathode as well as at B+  let me do quick voltage measurements.  so far I've been using my Wavetek 110 to generate  various frequencies and the output goes to a Valve junior.,,  I can A/B the two sections of the tube and compare them to the reference section in the reference tube.

so far, so good.