Tube DIY Asylum

Do It Yourself (DIY) paradise for tube and SET project builders.

Return to Tube DIY Asylum


Message Sort: Post Order or Asylum Reverse Threaded

ping deathtube 667 re: starved filaments

70.36.214.224

Posted on April 10, 2014 at 13:02:14
GSH
First, thanks for your continuing work and postings.
Since you have the ability to test this like most of us don't, would you please try lowering the filament voltage of whatever DHT circuits you do these THD tests on? For example either use a separate filament transformer and put a variac in front of it, or use low ohm pairs of resistors in series.

Run what ever sim you want at regular filament voltage and then try as suggested, perhaps at two lower levels. If it was as simple as dialing up the variac, then that shouldn't be too much trouble ?

Here's what Steve Bench produced along with this: http://diyaudioprojects.com/mirror/members.aol.com/sbench102/dht.html


I'm curious if you'll find similar with a complete circuit, rather than an isolated test using a plate resistor. Thanks in advance






 

Hide full thread outline!
    ...
RE: ping deathtube 667 re: starved filaments, posted on April 10, 2014 at 17:28:12
Tre'
Industry Professional

Posts: 17263
Location: So. Cal.
Joined: February 9, 2002



"So maybe we have explained the mystery; by starving the filament in DHT devices, the tube characteristics show a substantial region of not only constant mu, but constant gm and constant rp, thereby essentially removing all distortion causing mechanisms. "

I need some help seeing this.

Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"

 

RE: ping deathtube 667 re: starved filaments, posted on April 10, 2014 at 18:24:22
GSH
I agree these curves contradict was he's saying. The guy was no idiot however, and his other measurements show a dramatic lowering of harmonic levels, by 100's of percent, right? We don't know what test rig he had, which is why I asked deathtube 667 to try his rig on the question. Why not ? It will either show nothing or something, and the later may be worse or may be better. If it measures "better", then we can take a chance and ask DT
what he thinks of it's accuracy in reproduction, knowing then that we'll just have to try it ourselves to really find out. Seems productive.

Perhaps the original designers of DHT's over designed the electron cloud for "a reason", likely commercial in nature, maybe to make tubes last longer so the consumers of them would be satisfied with "decent" performance over a longer period? I don't know. But intuitively, I guess I smell something lurking behind the curtain here. Our use of these devises is very specific, and I believe most of us would trade shorter life for a cleaner sound, although Bench's life test seemed to show no
degradation.

 

RE: starved filaments, posted on April 10, 2014 at 19:32:17
If I get around to trying this idea, I will post the FFT findings here.

I noticed that there seems to be a lot of 60 & 120 cycle energy in my
SE DC 2A3 stereo circuit on the FFT. About -60dB from the fundamental.

It needs to be a bit better than that, so I got to redesign or go back to a one channel/mono circuit.

Always a problem with something.

 

RE: starved filaments, posted on April 10, 2014 at 20:37:23
GSH
The original DHT's were all designed to be lit with DC filaments at that time, batteries. AC operated tubes were an "economical" solution, and also more practical. Previous designs used higher voltage that certainly would be too high with AC (5-10V) 2.5VAC was "low enough" to "get away with".

When you can measure with no "music signal" ANY AC signal at the output, like ANY AC operated 2A3, then your amplifier is not just a music reproduction machine, it's a buzz bomb helper. Argue all you want. This is/was not the intention of AC filament design. It is a biproduct of an economical compromise, period. The modulation which has to, and will occur
with the music, is, because of the relative "lowness"of the frequencies,
the "warmness" it is attributed to. How this isn't obvious, I don't know.

So, clearly if you decided to try my request in a way as free as possible from interfering variables, then DC would be required, which may be a PITA understandably.

 

RE: ping deathtube 667 re: starved filaments, posted on April 10, 2014 at 21:47:47
dave slagle
Manufacturer

Posts: 5429
Location: NYC
Joined: April 27, 2001
Take a constant current load at 2ma and look at which has more uniformly spaced lines. this represents amplification.

now look at the slope of each line at 2ma and this is the Rp. Again which set of curves have the least amount of variation.

since Gm-Rp/mu, the more consistent Rp and mu remain the more consistent Gm will be.

Of course you do need to pay attention to where you operate things since you now have to avoid a "knee" at both low and high current but it can be easily seen that the starved tube is more linear.

dave

 

RE: starved filaments, posted on April 11, 2014 at 03:44:55
I use DC filaments on 2A3 finals. Non-regulated and I had previously used a rheostat bypassed with a power resistor for voltage control.

I'll need to try to set that up again to run the starved filament test.

When I ran the FFT bandwidth sweeps with white noise input and , I noticed that there was a big spike at 120 cycles that went up and down more than +/-5dB.

At 120 cycle spike was a big lower using pink noise.

Perhaps I don't have enough filtering for a stereo circuit or some other reasons.





 

Starved Tube Watchout, Higher Plate Resistance, posted on April 11, 2014 at 05:37:44
Maxamillion
Audiophile

Posts: 856
Location: New Jersey
Joined: May 26, 2006
Bench's work does show decreased distortion, but at the cost of increased plate resistance. If your circuit is sensitive to that, you may create more problems than you solve. As always, be sure the cure is not worse than the disease.

 

Thanks Dave, posted on April 11, 2014 at 08:36:37
Tre'
Industry Professional

Posts: 17263
Location: So. Cal.
Joined: February 9, 2002
I think I see this.

So starved filaments aside,

As I look at some triode plate curves, lower plate voltage and higher current (in many instances) gets us both linear gain and a more constant Rp which will equate to less distortion.

Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"

 

RE: ping deathtube 667 re: starved filaments, posted on April 11, 2014 at 13:12:29
Tre'
Industry Professional

Posts: 17263
Location: So. Cal.
Joined: February 9, 2002
I was starving the filaments on my 6b4gs and hated the way they sounded.

I wasn't starving them by much, maybe 5.8 volts. When I readjusted the supply to 6.3 they came to life.

Maybe I wasn't starving them enough?

Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"

 

RE: Thanks Dave, posted on April 11, 2014 at 14:00:20
Tre'
Industry Professional

Posts: 17263
Location: So. Cal.
Joined: February 9, 2002
I reduced the value of the cathode resistor on my CCS loaded 5687 driver.

That lowered the plate voltage and places my P-P swing in a more linear (Rp) place.

It sounds great but then it sounded great before.

I'll have to listen some more.

Thanks again.

Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"

 

The problem, posted on April 11, 2014 at 14:18:54
Paul Joppa
Industry Professional

Posts: 7294
Location: Seattle, WA
Joined: April 23, 2001
I see nobody has posted this yet.

At normal filament voltage there is a substantial cloud of electrons surrounding the cathode. This "space charge" serves three useful functions:

1) It protects the cathode from ion bombardment (cathode poisoning), extending the life of the tube.

2) By forcing a more uniform net flow of electrons from different parts of the cathode surface, it reduces the noise generated. (This is used in so-called noise diodes, which run a current through a starved cathode to generate white noise efficiently.)

3) By forcing a more uniform net flow of electrons from different parts of the cathode surface, it prevents the formation of hot spots on the cathode, extending the life of the tube.

Starving the cathode reduces the space charge nearly to zero, as show by the high-current limited emission.

 

RE: ping deathtube 667 re: starved filaments, posted on April 11, 2014 at 17:52:04
GSH
"hated the way they sounded", this doesn't "sound" like Tre'.

Although I don't doubt it, what would have been interesting is the measurements in both cases. You've reported before that after a known measurable improvement, you at first were not impressed or disappointed in the sound, but then after time realized it was more true to the recording
and ended up preferring the change. From AC to DC filaments I think was one.

I certainly don't know what amount if any, would be appropriate starvation, but I think the subject is interesting because of what at least one guy, Bench, measured. Usually, the "way" we (you and me, not everybody) listen, a measurable improvement is also a sonic one, even though some take time to get unaddicted to the previous "sweet distortion".

 

RE: ping deathtube 667 re: starved filaments, posted on April 11, 2014 at 19:31:47
Tre'
Industry Professional

Posts: 17263
Location: So. Cal.
Joined: February 9, 2002
"I certainly don't know what amount if any, would be appropriate starvation, but I think the subject is interesting because of what at least one guy, Bench, measured."

I agree. It's very interesting and worth investigating.


"Usually, the "way" we (you and me, not everybody) listen, a measurable improvement is also a sonic one, even though some take time to get unaddicted to the previous "sweet distortion". "

Well said.

Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"

 

Page processed in 0.032 seconds.