I agree that FFT bin width is definitely something to consider when performing measurements.
btw - right now I'm working on a build for some globe 46's. These old beauties will get DC heaters, but the reason is not to reduce hum or intermodulation distortion. 🙂
btw - right now I'm working on a build for some globe 46's. These old beauties will get DC heaters, but the reason is not to reduce hum or intermodulation distortion. 🙂
I really like 46, better than 45, I think. You might want to try 47, also, if you haven't already.
Win W5JAG
Win W5JAG
What are you designing here? SE or PP? External NFB or not?
As for "sonic differences", that depends on a lot of different factors about which you will find little consensus. The only thing that I can advise is that, if you use 300Bs, you need adequate grid drive. A lot of the complaints I see regarding the disappointing sonic performance of 300Bs traces back to inadequate grid drive.
SE and I'm not capable of designing. I'm a kit builder. Last one was a Tubelab build and needed to do quite a bit of searching to get that one done.
yes.......the reason for pentodes is the need for easy to drive amps....
triodes are indeed harder to dive in comparison.....
Yes, I believe his design incorporates power drive. I have an xtra board here. Just need to make a decision. I think some have had issues with the 2a3 builds. My 45 build went quite smooth after asking many questions and thread searching.
Yes, the tubelab SE can be built to run 300Bs and has power drive. It would certainly be a good introduction to the 300B. You should encounter no more difficulty with it than you did with the others. Resist the temptation to scrimp on output transformers.
The subject of harmonic distortion and intermod distortion keeps coming up.
But the effects of using AC on direct heated filaments are two completely different effects than the normal harmonic distortion and the normal intermod distortion tests that keeps coming up in this thread.
Most old 60Hz 6000Hz IM testers worked this way:
The large 60Hz signal and smaller 6000Hz signal (usually at a 4 to 1 ratio of voltage) were combined and applied to the amplifier input.
The amplifier output had a resistive load connected.
The amplifier output also went to a 6000Hz bandpass filter in the IM tester.
That 6000Hz bandpass filter was just wide enough to include the 5940Hz and 6060Hz
Intermod products (sidebands of 6000Hz).
The 6000Hz bandpass filter was applied to an AM Detector (Amplitude Modulation Detector).
The AM Detector output was applied to a low pass filter that only let 60Hz pass.
Unless the low pass filter on the AM Detector also let 120Hz pass without filter rolloff, then using such an intermod distortion analyzer will not work to test the effects of using AC on
a direct heated filament.
Non adjustable FFTs can have a problem with this test too.
Suppose the FFT covers 100Hz to 4kHz, and has 1000bins (FFT filters).
Disregarding from 0Hz to 100Hz (another FFT artifact, another subject)
The effect is 4000Hz/1000bins. Each bin is 400Hz wide.
That works great for testing the fundamental and first 4 harmonics of a 1kHz tone.
But it will not work to test 120Hz sidebands on a 1000Hz tone, because all 3 tones are in one and only 1 FFT filter bin. There is not enough resolution to separate the 3 tones.
This is like trying to use a very old 2000 x 1000 matrix CCD photo sensor to resolve the 100 lines per millimeter resolution of a Nikon true Macro lens that covers 36mm (3600 lines are not resolved by 2000 points.
I can see I have to get back to some AC powered direct heated filaments, and make some measurements and publish them here. But of course then I have to put DC on the same filaments and the same rest of the circuit, and publish the difference.
Now, where did that hour go?
But the effects of using AC on direct heated filaments are two completely different effects than the normal harmonic distortion and the normal intermod distortion tests that keeps coming up in this thread.
Most old 60Hz 6000Hz IM testers worked this way:
The large 60Hz signal and smaller 6000Hz signal (usually at a 4 to 1 ratio of voltage) were combined and applied to the amplifier input.
The amplifier output had a resistive load connected.
The amplifier output also went to a 6000Hz bandpass filter in the IM tester.
That 6000Hz bandpass filter was just wide enough to include the 5940Hz and 6060Hz
Intermod products (sidebands of 6000Hz).
The 6000Hz bandpass filter was applied to an AM Detector (Amplitude Modulation Detector).
The AM Detector output was applied to a low pass filter that only let 60Hz pass.
Unless the low pass filter on the AM Detector also let 120Hz pass without filter rolloff, then using such an intermod distortion analyzer will not work to test the effects of using AC on
a direct heated filament.
Non adjustable FFTs can have a problem with this test too.
Suppose the FFT covers 100Hz to 4kHz, and has 1000bins (FFT filters).
Disregarding from 0Hz to 100Hz (another FFT artifact, another subject)
The effect is 4000Hz/1000bins. Each bin is 400Hz wide.
That works great for testing the fundamental and first 4 harmonics of a 1kHz tone.
But it will not work to test 120Hz sidebands on a 1000Hz tone, because all 3 tones are in one and only 1 FFT filter bin. There is not enough resolution to separate the 3 tones.
This is like trying to use a very old 2000 x 1000 matrix CCD photo sensor to resolve the 100 lines per millimeter resolution of a Nikon true Macro lens that covers 36mm (3600 lines are not resolved by 2000 points.
I can see I have to get back to some AC powered direct heated filaments, and make some measurements and publish them here. But of course then I have to put DC on the same filaments and the same rest of the circuit, and publish the difference.
Now, where did that hour go?
still the $64 dollar question remains, how much of these are audible?
numbers please...
can anyone please point me to studies showing this?
numbers please...
can anyone please point me to studies showing this?
w5jag,
It looks like the filters on the IM tester might not have enough out of band attenuation.
The filter attenuation curves only show down to -20 dBc, there will certainly be more attenuation, but how much at the overlap of the filters is unknown.
The FFT analyzer you have with 20Hz resolution and 80dB dynamic range will probably work, depending on the other setting ranges it has.
Thanks for the pages you attached.
It looks like the filters on the IM tester might not have enough out of band attenuation.
The filter attenuation curves only show down to -20 dBc, there will certainly be more attenuation, but how much at the overlap of the filters is unknown.
The FFT analyzer you have with 20Hz resolution and 80dB dynamic range will probably work, depending on the other setting ranges it has.
Thanks for the pages you attached.
soulmerchant,
It was my pleasure to support physicists in using our spectrum analyzers (I did less support on scopes than I did on spectrum analyzers). Some were doing proton medical instruments, some were doing linear accelerators, MRI, etc.
It was my pleasure to support physicists in using our spectrum analyzers (I did less support on scopes than I did on spectrum analyzers). Some were doing proton medical instruments, some were doing linear accelerators, MRI, etc.
AJT,
As I said, if and when I get some time, I will do some AC and DC test comparisons.
Once the units have been built, and the measurements have been done, the listening
venue(s) is an easy thing.
I set up a double blind test once, that was to see if another factor could be audibly heard.
One listener was so frustrated that he threw the test switch across the room.
Fortunately nobody was hurt, except for the frustrated listener's feelings.
As I said, if and when I get some time, I will do some AC and DC test comparisons.
Once the units have been built, and the measurements have been done, the listening
venue(s) is an easy thing.
I set up a double blind test once, that was to see if another factor could be audibly heard.
One listener was so frustrated that he threw the test switch across the room.
Fortunately nobody was hurt, except for the frustrated listener's feelings.
you will be doing the community a great service when you do.....
guess we can not do much about the frustrated listener's feelings......😱
I visited Matt last week. Have known him since the early 80's when we were both engineering techs at Tektronix.
I think the above mentioned electrophonics is what is partially responsible for 300B sound. Some brands are worse about this than others.
Ron
Ron,
Good to hear from you on this thread.
We will have to talk at the local audio club sometime.
Now I have to figure out how to document electrophonic test results for this site.
Good to hear from you on this thread.
We will have to talk at the local audio club sometime.
Now I have to figure out how to document electrophonic test results for this site.
It's not just the 300B. It's all tubes. The 300B has become infamous for this because it is expensive and people expect miracles.🙂
In general vibrations affect in a negative way every piece of the stereo system but sadly people just ignore it or employ the wrong solution (typically bare damping) and go round in circles talking forever about the sound of....😀
Started out with Hammonds. Made the switch to Electra Prints. I would like to try MagneQuest next but the fella who sells these is a bit hard to connect with.
In theory I think it should work. In practice, not so sure. As I understand this old device, it's just an AM receiver. The high frequency tone is the "carrier" and the "passband" is the width of the high pass filter response. The "carrier" can fall anywhere in this "passband". The modulating frequency on the carrier can be any frequency within the low pass filter response. The "carrier" is detected, and the low pass filter makes sure that the ac vtvm only responds to the modulation products within the low pass filter response.
So, it seems to me to *try* to use this elderly device to measure the + - 60 Hz products on an AC heated filament, you just ditch the two tone set up like you would use for SMPTE IMD; and just put a single tone into the amp with the directly heated filament, say 3 kHz, and let the device detect and demodulate that single tone. The 60 Hz on the filament will modulate the tone. The meter is dumb, and doesn't know that the 60 Hz is from something other than a tone generator. The only question would be, is it strong enough to deflect a meter indication.
The practical problem would be if the vtvm is sensitive enough to display this. The other problem would be that since modulation is 1:1, how do you separate the + - 60 Hz modulation products, from the noise of the meter which, I guess, would be + - 120 Hz? Unless the 60 Hz modulation is strong enough to swamp the internal noise, I don't see a way to differentiate this. I'm assuming the meter jiggle/vibration artifacts are almost certainly internal noise artifacts of the meter.
Still might take a spectrum analyzer, but if it is beneath the meter's noise floor, is it really anything to be concerned about?
BTW, I totally accept that this 60 Hz ac modulation is unavoidable on a directly heated filament; it's outside the scope of my particular project to use DC on the filaments. If I could find a way to quantify the detriment that doesn't involve buying more equipment, I'm all in for that.
Thanks for taking a look at the limited info I have on the spectrum scope in that box, and making me think about this.
Win W5JAG