Noisefloor question....

[Dr. Toobz]

My Stereomour is currently running a pair of 45 tubes, but I have also wired the amp to use 2A3 tubes at times. In both cases, I can cancel out much of the 60Hz hum (though less so with NOS 2A3's), but am left with what sounds like tube rush or hiss on my tweeters. It's not that loud, but I'm curious about it. Is it another byproduct of AC heaters? My SEX amp is dead quiet on the same speakers - no hiss, noise, or hum to be heard, even with an ear to the speaker. It's definitely noticeable as you get within a few feet of each speaker, though it's pretty much a non-issue in the whole scheme of things. I'm just a little puzzled by it, given that I was jaded by the super dead-quiet SEX amp for so long.

I've swapped out the 12AT7 for an NOS 3-mica Sylvania, but the hiss is about the same. Are 12AT7's just a "hissy" tube, given the somewhat high gain?

 

[Paul Birkeland]

The hiss would most likely be the 12AT7.  How does it vary with the volume control?

 

[Dr. Toobz]

Changing the volume pot doesn't result in any change to the hiss. With the 45's and NOS 12AT7 in there now, I can only hear a bit of hiss on the tweeter horn within close proximity to the speakers. The stock EH 12AT7 was actually a bit worse. I took out my SEX amp, which is scheduled for a rebuild soon, and it's like the speakers are turned off. Maybe the SEX amp is just that quiet?

It's really not very noticeable, just more of a theoretical question.....

 

[Paul Joppa]

There is no reason to expect more white/pink noise from the Stereomour than from the SEX amp; it's only the hum that should be greater, due to the AC heating.

The gain of the 45 is very close to that of the 2A3, so if there is noise upstream of the power tube they would produce the same audible noise. Since they don't, it is logical to suspect that the noise occurs in the power tube.

You might check the AC voltage at the filament pins; if it is low on the 2A3s (below 2.25vRMS) that would be suggestive - especially if it is not too low with 45s. Obviously this requires a meter that is accurate at 2.5v of AC - many are not. Just a caution.

Otherwise, I think swapping 2A3s would be the more definitive test - though more difficult if you don't have a bunch of spares on hand!

 

[Dr. Toobz]

Actually, the 2A3 and 45 were similar in terms of hiss, with the 2A3 showing more hum. When I pulled the 12AT7 from the amp and ran the power tubes only, the hiss goes away completely. So that's why I assumed it might have something to do with the 12AT7. The hiss is uniform in both channels and very much like random pink noise in that's it's constant and doesn't change in pitch or intensity.

Could the regulators and/or CCS cause some low-level pink noise? Or might I perhaps have lucked out two times in a row with hissy 12AT7's?

 

[Paul Joppa]

Sorry, I misunderstood the posts.

It is most likely to be the 431 used as the cathode bias voltage source for the 12AT7. You can parallel a low-voltage capacitor of around 10000uF to test this theory. (FWIW, I calculate the noise from the specifications at 120dB below max power. We don't yet have the ability to measure such low noise levels.)

The capacitor may reduce the sound quality, however. Here are two alternatives:

1) replace the 431 with a resistor of 500 to 690 ohms (metal film preferred). This may reduce the high treble a dB or two unless it is bypassed - damaging the sonics again. I would not drop below 499 ohms. The 45 will not have this problem; its Miller capacitance is half that of the 2A3. This is similar to the SEX amp circuit.

2) replace the 431 with an HLMP-6000 LED. I'm not entirely comfortable with such a low bias voltage, due to the risk of risk grid current on transient peaks - though we did use it in the Paramour II with satisfactory results.

As you can see, I chose the 431 to avoid the Miller capacitance rolloff and the grid current possibility, figuring the noise would be well below most sources unless the speakers were exceptionally efficient. But you can make - or at least experiment with - the other choices and find the compromise that suits your situation best.

 

[Dr. Toobz]

Great info, Paul! I always learn something useful from your posts.

I may try out the cathode resistor option, especially since I am running the 45's and therefore, shouldn't have to worry about bypassing the cathodes for fear of losing treble detail. If anything, I could stand to lose a bit of treble, given the revealing nature of the 45's and the somewhat thin sound of the Klipsch speakers on this amp. Part of this may have to do with bass; I'm going to bump up the output coupling caps from the stock 3.3uF to some 6.8uF 630V Solens I have laying around when I get a chance. When testing out the 45's with headphones, the bass was rather light as well.

Otherwise, the extra hiss is covered by the music and may only be a problem of my fastidiousness versus anything having to do with actual sound quality. If I were a betting man, I'd say that the three drivers in the Klipsch speakers aren't terribly balanced, and the tweeters are probably over 100dBW sensitive. The woofer likely brings the average to 99dB. In short, it's the speakers, not the amp. For what it's worth, I may have liked these more on the SEX amp due to the warmer, "tubier" sound of that amp with the Speco line transformers, which may have worked around these flaws. Now that I'm running a DHT amp (and 45's at that), the emperor has no clothes!

 

[Dr. Toobz]

I calculate .72W dissipation on the 12AT7 (200V B+ @ 3.6mA, correct?), so I am assuming a 1W cathode resistor would suffice?

Also, let's say that I want to "correct" my Klipsch speaker's wiry sound by adding a bit of second harmonic distortion back into the picture. I could do this by ditching the C4S boards and going with a plate resistor on the 12AT7 as a load. Question is, how do I calculate this? I'm not good with the load lines ....

 

[Paul Birkeland]

The cathode resistor dissipation is different (lower most often) than the plate dissipation.

We have 2.5V of bias under the cathode, and 3.6mA, Ohm's Law says Power=Volts*Amps, so we get 2.5V*0.0036A=0.009Watts. 

From experience, a large safety margin is needed, so the cathode resistor should be rated to three times this power, or 0.027W.  (So a 1/40th Watt or greater part is OK)

For the plate load resistor, we know we have 388V of raw B+, and 200V on the plate of the 12AT7.  This is a difference of 188V.  With a target current of 3.6mA through the tube, we have voltage and current, so we can solve for resistance.

V/I=R, or 188V/0.0036A=52,222 Ohms (47K-56K will probably work)
We also have V*A=P, or 188V*0.0036A=0.7W, so a 3W resistor would be nice here.

The gain of the Steremour will drop with this change, the distortion will go up (as you like), and potentially the noise floor as well.  I don't think anybody has tried this, so if you do let us know what your observations are.

-PB

 

[Dr. Toobz]

Thanks! That makes a bit more sense.

The most sensible approach towards lessening the slight tweeter hiss, and perhaps adding back some harmonic goodness at the same time, may be to stick with an unbypassed cathode resistor in place of each regulator and leave the C4S as-is. If I don't like that approach, I may try a plate resistor, though I've noticed that the best way of voicing a system seems to be to leave the amp as accurate as possible, in favor of changes closer to the source. This may mean a return of the Quickie to my main system! Or, perhaps it's really time to ditch the Klipsch speakers in favor of something that doesn't require so much messing around with amplification.

 

[Paul Birkeland]

That's certainly a good approach, there is a fair amount to be lost by eliminating the C4S. 

One can also use a plate load resistor from plate to ground, while at the same time increasing the C4S current to account for this extra draw.  This will give you some of the additional distortion that you may want, while at the same time maintaining the power supply isolation of the C4S.  This would also let you try a whole range of different plate load resistors without changing the actual plate voltage appreciably. 

-PB

 

[Dr. Toobz]

Well, I'm happy to report that the hiss problem is gone, following the excision of the 431 shunt regulators. I put 680 ohm resistors in their place, since that's the only value I had in my junkbox between 499 and 700 ohms. Now the amp is dead silent, with no hiss or hum to be heard, even right up against the speaker (using NOS 45 tubes). The inky black silence of the SEX amp has been replicated, and on AC heaters to boot......

A quick listen doesn't suggest that the modification did much to change sound quality. Bass maybe isn't as tight, but with the Klipsches, that's not a bad thing.

 

[Dr. Toobz]

Sorry to resurrect my old thread.....

I have noticed lately that there is s definite loss of bass with the cathode resistor approach, as expected. It wouldn't be so bad with most speakers, except that my Klipsch H3's are bass-shy to begin with and drop off like a rock below about 55Hz. The 431's are a no-go, due to hiss, so my next idea would be to use an LED to get close to the 2.5V bias cited earlier. A green LED typically has a voltage drop of 2.2V, which would reduce the risk of positive grid current on transient peaks one might run into with the HLMP-6000, which I think drops just 1.56V. I may try this approach vs. a bypass capacitor. Despite the green LED not being ideal, I'd be willing to bet that the sonic penalty would be more noticeable with the capacitor than from any flaws with the LED.

As an aside, is there a drop-in replacement for the 431 that might exhibit less tweeter noise?

 

[Paul Birkeland]

If there was a quieter TL431 that we knew of, it'd be in the kit.

You could try putting two HLMP-6000's in series to get 3V of bias.  Otherwise, with the current levels in the Stereomour driver are low enough that to get 2.5V, you may want to look for a 3V LED that may drop to more like 2.5V at low current levels (a few mA).

 

[Dr. Toobz]

The only green LED's I have on hand appear to be no good. (No sound - and I checked to make sure the cathode/anode pins were oriented correctly).

I then tried the earlier idea of using a single HLMP-6000, since I don't have four on hand. The sound is slightly improved over the resistors - maybe a bit more dynamic and clean - though I'm hard pressed to tell if this is simply a matter of placebo effect. There's certainly not a huge change in bass output. 

I pulled the 431's around the same time that I went to 45's as output tubes. I think the next idea will be to keep the cathode resistors and go back to a 2A3 setup, as it may be that the 45's are just too bass shy and lean for my Klipsch speakers (which I can't afford to replace at this point). What I am attributing to the change in cathode biasing technique may be incorrect.

 

[Doc B]

Yes 45s are generally more lean in the bass compared to 2A3s and 300Bs.

 

[Paul Joppa]

Yes, there is no technical reason to think the cathode resistor replacing the LED bias will have any effect on the bass.

It is possible to use a single HLMP-6000 - the Paramour did so with no problems I am aware of.

A possibly more practical option would be something like the Fix, a passive bass boost of a few dB. At least you would know that you'd made a measurable difference in the bass level, and could then decide whether it is weak bass or some other quality of the bass that is bothering you. Over the decades, engineers (mea culpa) have put so much emphasis on frequency response that I think we sometimes ascribe other, more subtle effects to simple response differences.

 

[triode]

Hi PJ & others,
I just finished my Stereomour and got the same hiss problem. Mine are configured as 45, and the hiss still there both with 2A3 and 45 tubes. I also tried different 12AT7, still the same.

Would I better off with 2 series HLMP-6000 or just single one to replace the 431? any sonic degradation using series led vs single?
I recall Paramount (2A3 version) using just one led and seems no problem on mine.

FYI my speaker is Lowther front loaded horn with bullet tweeter. Bass is active so Stereomour does not need to work below 200Hz.

Thanks in advance

 

[Paul Joppa]

...
Would I better off with 2 series HLMP-6000 or just single one to replace the 431? any sonic degradation using series led vs single?...

As a reminder, from my earlier post (edited now to fix a typo):

1) replace the 431 with a resistor of 500 to 690 ohms (metal film preferred). This may reduce the high treble a dB or two unless it is bypassed - damaging the sonics again. I would not drop below 499 ohms. The 45 will not have this problem; its Miller capacitance is half that of the 2A3. This is similar to the SEX amp circuit.

2) replace the 431 with an HLMP-6000 LED. I'm not entirely comfortable with such a low bias voltage, due to the risk of risk grid current on transient peaks - though we did use it in the Paramour II with satisfactory results.

I'll add 3) replace the 431 with two HLMP-6000s in series. Two LEDs risks the opposite problem, running out of compliance at high voltages as the 12AT7 ages. They will be 3dB noisier that a single LED but that is probably not audible relative to the 12AT7's intrinsic noise.

The optimum voltage remains 2.5 volts, as I designed it.

As to the best choice, try them and see. It's neither expensive nor difficult! Personally I'd use the resistor with a 45 - simpler, more reliable, and closer to the optimum voltage. The LED(s) will be quieter, but I think the margin will be insignificant - a small fraction of a decibel. But that's just me, and I haven't done the experiment.

 

[triode]

Thanks PJ.
I'll try the resistor and see

Cheers