[dsavitsk]

Hi

I have been looking at the schematic for the Torpedo (quick link : http://www.diyforums.org/Torpedo/schem/TORPEDOschem.gif)  and there are a few things I can't figure out. First is the bias current : is it only 2 or 3 mA ?  Second is the biasing method : does it serve a purpose to have the same current flowing through D12, D13 and D10 (other than saving a few resistors) ? Actually, why the use of a transistor + led under the cathode ? Wouldn't a string of leds or diodes do the same ?

A data I can't find anywhere as well is the B+ voltage. I guess it is in the 200V (and it doesn't matter because of the ccs) but since the bias current seems different than in the other l'espressivo designs ...

thanks

By bias current, you mean the current through the tube, or the current through D12/D13/D10? The former is ~16mA and is set by R4, the latter is 2-3mA and is set more or less by R6 and R7.

We use the same current as a method of recycling -- ~250V * 3mA is 0.75W. X2 it is the difference between throwing away 1.5W or 3W, which is enough to put significant extra stress on the power transformer, which is heat we don't want to worry about and noise we want to avoid.

The difference between this biasing method and a string of LEDs is that this method is lower impedance and considerably more linear. Keeping this Z low keeps rp lower which helps drive the transformers. And while an LED is an improvement over an RC circuit, this is much better then either.

Yes, B+ is not critical - IIRC it is around 250V, but Tomb probably knows better.

 

[tomb]

By bias current, you mean the current through the tube, or the current through D12/D13/D10? The former is ~16mA and is set by R4, the latter is 2-3mA and is set more or less by R6 and R7.

We use the same current as a method of recycling -- ~250V * 3mA is 0.75W. X2 it is the difference between throwing away 1.5W or 3W, which is enough to put significant extra stress on the power transformer, which is heat we don't want to worry about and noise we want to avoid.

The difference between this biasing method and a string of LEDs is that this method is lower impedance and considerably more linear. Keeping this Z low keeps rp lower which helps drive the transformers. And while an LED is an improvement over an RC circuit, this is much better then either.

Yes, B+ is not critical - IIRC it is around 250V, but Tomb probably knows better.

 
To be honest, I haven't measured it while the amp is operating.  Maybe that's something I need to do when I get up enough courage.

  I think you said once, though, that the voltage could go to as high as 275V before everything is loaded up, which is why the caps are rated to 300V minimum.

 

[bidoux]

Thank you both for your answers.
 
By biasing current I meant the current going through the tube, but I admit that the question was ambiguous. I am still stuck with calculating it ; if I say Vbe of Q3/Q4 is around 0.7V and forward voltage of D12/D13 around 1.8V I get I_bias=(1.8-1.1)/402 since R4/R5 is 402 Ohms. Can't find my mistake here ...
 
I get your explanation about saving some power on the transformer, I hadn't thought of it and it is as clear as it can be. Actually I was asking since there would be at least two ways of converting the circuit to balanced. Getting inspiration from here (http://www.nutshellhifi.com/preamp-pix.gif) there would be two ways of providing current to the LED. I guess it makes a lot of sense of chosing the one that keeps the electricity bill down

(I may have expressed myself very poorly in the last paragraph, it does need a schematic)

 

[dsavitsk]

I am still stuck with calculating it ; if I say Vbe of Q3/Q4 is around 0.7V and forward voltage of D12/D13 around 1.8V I get I_bias=(1.8-1.1)/402 since R4/R5 is 402 Ohms. Can't find my mistake here ...

Calculation is correct. What you missed is that we replaced D12 and D14 with a 6.8V Zener. The impedance of the circuit is approximately Hfe1 * Hfe2 * Rbias. By switching to a Zener, we can increase Rbias which increases PSRR considerably, and it makes everything work better. However, we did find that 6.8V was the sweet spot as larger Zeners were too noisy defeating the whole point. With more PCB space, you could put an arbitrarily long string of LEDs, or add a capacitor bypass, but in the limited space we had, this was a good solution.

 

[dsavitsk]

Actually I was asking since there would be at least two ways of converting the circuit to balanced. Getting inspiration from here (http://www.nutshellhifi.com/preamp-pix.gif) there would be two ways of providing current to the LED.

Note that with this bias in that balanced configuration, you could push the circuit into class B. You also include the transistor in the signal current loop. Not necessarily bad, but not necessarily what you are after.

If you want to force it to stay in class A, and eliminate the transistor from the signal path, you would use a CCS there. I've done both, and both work well in the right place.

 

[bidoux]

Cool

My preference obviously goes to the string of LEDs, what's not to like about much more lighting ?
 
I didn't pay attention when I picked the schematic from nutshellhifi. I didn't mean doing someting ultrapathstyle, rather having the primary of the output transformer going from one cathode plate(corrected) to another with the capacitor in the middle (can't find the drawing on the internet edit more or less that : http://www.raleighaudio.com/figure_18.htm).  I am guessing doing this doesn't include the transistor setting the cathode voltage in the signal loop. Does it still change the class of operation from A to AB ?
 
 
And out of curiosity, do you actually mean replacing the resistor which goes from ground to both cathodes (same drawing : http://www.nutshellhifi.com/preamp-pix.gif) with a current source ? Doing that, how does one makes sure the cathode sits at the right voltage ? Seems like it is going to charge the capacitor which also discharges through the transformer, what is the final state of such a configuration ?

 

[dsavitsk]

rather having the primary of the output transformer going from one cathode to another with the capacitor in the middle (can't find the drawing on the internet edit more or less that : http://www.raleighaudio.com/figure_18.htm).  I am guessing doing this doesn't include the transistor setting the cathode voltage in the signal loop. Does it still change the class of operation from A to AB ?

You mean from plate to plate?

On the K&K design, the cap is really there because they are using amorphous core transformers that cannot handle any DC. In fact, if you use too large of a cap, the current just to charge the cap will saturate the transformer which takes some time to recover. With a regular P-P transformer, it is probably not an issue.

That design is always class A, and you probably would not want a CCS in the tail (these are all long tailed pair variations, b/t/w/) as the CCSes would conflict. So you want a resistor there. But you do want it as large as possible (just that a CCS is too large -- though there is some evidence that this may not always be true).

And out of curiosity, do you actually mean replacing the resistor which goes from ground to both cathodes (same drawing : http://www.nutshellhifi.com/preamp-pix.gif) with a current source ? Doing that, how does one makes sure the cathode sits at the right voltage ? Seems like it is going to charge the capacitor which also discharges through the transformer, what is the final state of such a configuration ?

Yes, replacing that resistor with a CCS makes it a differential amp. The upside is that performance is better, the downside is that you are limited to class A. The CCS will drop the necessary voltage -- that's how a CCS works. Note that you have to balance the tubes you use with the CCS you use -- some CCSes require more operating voltage than others, so you may need something with very low dropout. That cap is there to shunt distortion away.

 

[bidoux]

Thanks dsavitsk, helps a lot.

 

[tomb]

  Thanks dsavitsk, helps a lot.

Yeah - glad I didn't have to answer all of that stuff.

 

[tomb]

Hallo,

when will the torpedo kits be available again on beezar?
Or did I miss them?

Florian

More transformers have arrived - the Torpedo kits are back in stock!

 

[tomb]

With some helpful suggestions from Jim Cross at vacuumtubesinc.com, we recently discovered the finest tube available for the Torpedo: the E90CC/5920.  It pushes the amp into an entirely new level of sound quality and performance (IMHO).  A friend at work who has a Torpedo says it sounds like a $1,000 amplifier (not my quote).
 
Unfortunately, the tubes are somewhat rare and there is a current non-audio demand for them: http://www.tnmoc.org/news/notes-museum/valves-attic .  These tubes were originally manufactured for early computers and calculators.  Some machines might use up to 10,000 tubes.  Most of them used 1,000 or more.  Burnouts were also common, one reason they went to solid-state and integrated circuits.  Unfortunately, the National Computer Museum (the link above) and others are using these and similar tubes by the thousands.  Dsavitsk remarked to me: "What a waste of tubes!"  I have to agree.
 
I'm not normally one to be against preserving fine old equipment - especially ones based in tubes - but these old computers seem a bit ridiculous and are an actual threat to the NOS tube market.  If you have contact with other tube dealers besides Beezar, my personal opinion would be to encourage them not to donate tubes to this museum.
 
Back to the tube: the data sheet even says it's not recommended for audio (probably a reason the computer museums feel justified in eating them up).  Yet, after using many pairs of these for months in my Torpedo, I'm convinced the only reason for this was to cover the manufacturer against microphonics.  There is more of a tendency toward microphonics than with most tubes, but not by much in my experience.  Microphonics can vary from a slight, quick echo if you tap directly on the tube or amp, or a loud ringing that can last for several minutes.  The latter is tolerable, the former is not.  So, please keep this in mind when purchasing these tubes (from anyone).  The performance gains are worth that much.
 
A side benefit is that these tubes are so tall that you don't have to remove the Torpedo case lid to swap them out

:
 

 

[H22]

the finest tube available for the Torpedo

So,.......... What is so great about them? What is it they do that bumps the torpedo to a whole new level?
Thanks, I am still scraping together funds for one of these amps, hopefully soon....
Thanks TomB!

 

[Punnisher]

I'll have to grab a pair of these. They are somewhat expensive, but definitely not out of reach.
 
TomB, do you have any recommendations for brand/quantity/origin? There are a fair amount on Ebay. I just want to ensure I get a quality pair that sounds the same between the two.

 

[tomb]

  I'll have to grab a pair of these. They are somewhat expensive, but definitely not out of reach.
 
TomB, do you have any recommendations for brand/quantity/origin? There are a fair amount on Ebay. I just want to ensure I get a quality pair that sounds the same between the two.

My friend at work actually has more diverse experience with them right now than I do.  I've tried Amperex, Phillips and Mullard.  The Mullards might be the most neutral, while the Phillips seem to have a bit of a bass bloom - it's quite nice at lower volumes.  The Amperex were the first I tried and I can't remember whether there were any differences - just that the detail was outstanding compared to a 6J6.  The Telefunkens may have the best quality construction, but I haven't tried a pair yet.  The European-origin E90CC's seem to have tall plates with little tabs that rise out of the top mica spacer, whereas the American tubes have plates that are the height of 6J6 tube plates, but with a getter that's raised on a stilt (necessary to get to the top). I'm not sure yet whether either one means it sounds better, except that all of them are a tremendous improvement over a 6J6.
 
Some of these are known for having a "pinched waist" where the glass was drawn taller and the diameter shrinks around the center section.  Others have tube tops where the glass is folded on the top with seams (!).  Both are pretty strange attributes.  In my experience with other tubes, however, excellent construction is no guarantee that a tube sounds better.  I've heard RCA's whose construction looked like the mfrs were drunk when they assembled them and they turned out to be among the best in sound quality.

 
 
EDIT: I may be over-generalizing when referring to European vs American manufacture.  It appears that many European tubes also had short plates, although I haven't seen many long plates on American brands.  There again, there may have been a lot of cross-supplying going on across the pond.

 

[tomb]

the finest tube available for the Torpedo

So,.......... What is so great about them? What is it they do that bumps the torpedo to a whole new level?
Thanks, I am still scraping together funds for one of these amps, hopefully soon....
Thanks TomB!

Quite simply, they are a superior tube - with detail that you'll not hear with a 6J6.  I kept listening for a month or two before I felt like it was safe to talk about these tubes.  My friend at work really helped me in that regard.  He confirmed that everything I was hearing was in fact, true.  I've been fooled before when some opamp or buffer was oscillating and it sounded like the "swishes" were more prominent.  This is something else - detail with great, authoritative bass (the Mullards may have a bit less bass*), and a midrange to die for.  All this without the harshness in complicated passages as might be heard with the 6J6.
 
Don't get me wrong - the 6J6 is still a great tube when looking at the Torpedo as a mid-fi amp (a very good one).  It's just that the E90CC/5920 pushes the amp into a higher class of transparency and detail.
 
* Depends on the headphone.  With Grado HF-2 (PS500's) and Sennheiser HD600's, the Mullards sound great.  The Phillips seem to have that extra bass push that really sounds neat with AKG K701's.  You'll notice the leanness again in the K701's with the Mullard E90CC's.