• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

2a3 vs 300b

No hum, but plenty of intermodulation distortion. Have a look sometime with an FFT, you'll be surprised. It's why tubelab, me, and others advocate for proper DC heating even though you can seemingly get away without it with 2.5V tubes.

It may not require something as sophisticated as an FFT. I've got an ac heated 2A3 on my bench right now, on a modified Tubelab SSE board.

With my analog distortion analyzer, I'm looking at 0.36% SMPTE IMD, and I can see an oscillation plus and minus about 0.36; it's about the width of the meter needle, but it is definitely there. It sure looks like the oscillation frequency is about 60 Hz.

Win W5JAG
 
I'm looking at 0.36% SMPTE IMD, and I can see an oscillation plus and minus about 0.36; it's about the width of the meter needle, but it is definitely there. It sure looks like the oscillation frequency is about 60 Hz.

what i would like to know is, is this audible? if not audible at what level is it audible?
and can you please compare with ac heating case?
 
AJT,

I am not talking about the IM distortion of 2 music tones with each other.
I am not talking about reducing that kind of IM.

What I am talking about is the IM distortion of two times the line frequency that appears on each and every musical tone that occurs when using AC filaments to power DHTs.
Although it is quite small on small amplitude tones, it is much larger when the grid to filament is near to 0V (i.e. 7, 5, or 3V difference). You do not have to be in clipping, and not have to be a 0V difference between either of the ends of the filament to the grid.

i.e. AC filament 60Hz, music tone 1kHz, gives IM tones at 1kHz +/- 120Hz;
music tone 440Hz, gives IM tones at 440Hz +/- 120Hz, etc.

fair enough, and my question is how much of this is audible?
 
the reason i resist the use of electronic regulators on filament supplies
was that i once was given a tube buffered gainclone that use filament
regulators, after a sometime, the regulator got fried and the amp
was sidelined, the technician cited the unavailability of the 7806 regulator,
when the unit was given to me, i simply replaced the regulator with a suitable sized
resistor and the amp was playing music again....
 
what i would like to know is, is this audible? if not audible at what level is it audible?
and can you please compare with ac heating case?

I doubt that I could hear it.

And it could be just an artifact of my mediocre equipment, a Heath AA-1. However, the meter is "new" - I built it from an unbuilt kit about this time last year.

Win W5JAG
 
Alex M,

Second harmonic distortion is when one direction of the sine wave is foreshortened versus the other direction of the sine wave (i.e. a short positive alternation versus the negative alternation).

A single triode SE output stage has this characteristic.
A single triode driver stage has this characteristic.

The short alternation occurs in one direction for each stage.
But the driver and output stage are in opposite phase.
Now, connect the driver stage to drive the output stage.
That can cause the short alternations to occur in both the positive alternation and also in the negative alternation. That cancels (all or part) of the second harmonic distortion.

Another way to do this is the push pull output stage. The push and pull are in opposite phases and the second harmonic is reduced or cancelled.

Second harmonic distortion cancellation were also mentioned by 50AE in his reference to Sakuma San.

But now we have foreshortened sine wave in Both directions. That is third harmonic distortion.

What once produced second harmonic distortion is combined with another source of second harmonic distortion, and now we have third harmonic distortion.

As maiko also stated second to third harmonic conversion can occur in a LTP.
There can be other examples too.
 
How do I say this in order to be understood?

AC powered DHt filaments does not create multi-tone IM, i.e it does not cause SMPTE IM.

It creates 2x line frequency to appear on each tone from the signal generator, or in the case of music 2x line frequency to appear one each music tone.

Can you hear it?
Only you can say.
 
AJT,

I used to use regulators for DC filaments on DHTs. Some oscillated, some had slow startups due to protective current limiting, some had not enough range for power line variations (all those could be easily re-engineered to work properly).

I only use CRC brute force filtering now to get DC filament power.
It is not right or wrong, it is just my preference.
 
AJT,

I used to use regulators for DC filaments on DHTs. Some oscillated, some had slow startups due to protective current limiting, some had not enough range for power line variations (all those could be easily re-engineered to work properly).

I only use CRC brute force filtering now to get DC filament power.
It is not right or wrong, it is just my preference.

+1, my choice too....😎
 
w5jag,

It has been years since I used the HP 3585 spectrum analyzer to look at the +/- 120Hz sidebands on the signal generator tones.

I generally use DC on DHT filaments now.
Part of that has been the 1 to 11mV ripple on the amp output that caused me to not use AC filaments. I am a stickler for hum.
I am willing to try AC one more time again, but if it has too many problems, it is 3 strikes and they're out!

I have too many other circuit ideas to try out first.
Yes, I would love the simplicity of AC filaments if it meets my other criteria.
And AC takes a lower amperage rating of the filament secondary to keep it cool, versus a secondary that has to supply a DC rectified and filtered circuit.
Here I go again, . . . wish me good results.
 
I just wired up a 46, triode, and it is still there

In looking at it more closely, I see two distinct artifacts - the first is the slower oscillation about a point, and the second is much more rapid - the needle has a sharp tip on it, but in looking at the IMD on this lowest scale, the sharp tip is indistinct - it's moving too fast to focus on.

I'm going to call it an artifact(s) of my mediocre equipment, otherwise it would really bug me.

Win W5JAG
 
w5jag,

I thought the 46 was a Direct Heated Tetrode, not a triode.
Sometimes I think I will have to try using Tetrodes too.

I might be more partial to some 4-65 tubes that I have.
Wow, scary thought, that is more ambitious, and requires much higher voltage.

I am not familiar with an analyzer that has a vibrating needle.

I really need to get a 2A3 amp with AC filaments up and running again.
If I can see the 120Hz sidebands on a 1kHz tone with my 8 bit FFT,
then I will start worrying about using AC filaments again.
But if I can not hear the difference, I may need to find an old 12 bit or 14 bit CD player.
 
> looking at the IMD on this lowest scale, the sharp tip is indistinct - it's moving too fast

Is this an old 60Hz+6KHz IMD meter?

They put 60+6K through the amp under test. If the amp is not perfect, the 6K is modulated by the 60Hz. Then in the meter a low-cut removes the 60Hz. If the amp were perfect the remaining 6K would be dead-constant. With a distorty amp the 6K will be wobbling at 60Hz. The meter driver usually filters this smooth. If filtering (either meter or main) is incomplete (or flawed build), the needle could wobble 60Hz.

It has been so very long since I used this type IMD that I can't recall if the needle vibrated.
 
Is this an old 60Hz+6KHz IMD meter?

Yes. Built it this time last year, verified integrity of components before assembly, so, old is ambiguous.

With a distorty amp the 6K will be wobbling at 60Hz. The meter driver usually filters this smooth. If filtering (either meter or main) is incomplete (or flawed build), the needle could wobble 60Hz. ....

Looking at it eye level, at a distance of eighteen inches or so, where it resides above my bench, these artifacts are not discernible. It has to be at it's highest sensitivity setting, and you have to be right up on the meter to discern it. At least, that's what it takes with my eyes. Not sure what compelled me to look at it that closely.

I'm relegated to mediocre junk test equipment because I'm cheap and refuse to devote scarce space in my radio room to a real computer. BUT in looking around here, I notice that I have a never unpacked, stand alone, digital whiz bang box for one of my radios, a Yaesu DMU-2000, been here maybe a decade, that, in looking at the manual, among its functions, has a digital scope and a limited FFT audio analyzer with an external input for both functions.

Looks like the audio FFT is spec'ed at 100 Hz to 4 kc, 80 dB dynamic range. Not all that useful outside of the comms context, but should be adequate to display residual ac heated sideband noise on a single tone, like we are discussing here. If residuals are below -80 dB, I'm not all that sure I would worry about it. I guess I'll have to rearrange things to unpack it and make room for it. Looks like it can also control my antenna rotor, so that frees up some space right there.

Win W5JAG

edit: I suppose this would also provide some limited harmonic distortion info, at least on a 1 kc tone?
 
Last edited:
One IM Tester:
The problem with using a 60Hz 6000Hz IM tester in this application may be that the filter to reject noise and get the 60Hz sidebands may be narrow enough to filter off some or all of the 120Hz sidebands. That will give a reading that is less than the actual amount of the 120Hz products caused by 60Hz AC on DH filaments.

The other tester:
Do try using the FFT.
The question is, what are the FFT filter bin widths? Can they be adjusted?
If they are not narrow enough, you will never see the 120Hz sidebands that are on the 1 kHz tone. Instead, you will see a single 1kHz tone (but that lumps the 120Hz sidebands in with it).

I worked with, engineered, used, and supported spectrum analyzers and FFTs for over 30 years.
Sometimes these are hard to properly apply, properly set up, and to properly interpret the measurement results.
 
soul merchant,

If you are using AC filaments, there is nothing you can do about the IM that is at 2 times line frequency, the IM is just there. You can not filter it out.
DC filaments totally eliminates this effect (using low enough ripple DC).
.

True, but when I did measurements using AC vs. DC (CLCL filter) there was not enough effect on THD for the 45, 46 and 2a3. Hum was lowered, but I don't put my ear right beside my speaker, and decided that my lundahl filament chokes could be saved for valves with thoriated tungsten which require higher voltages and can't be run with AC.

Overall, THD was better reduced by employing strategies in other parts of the circuit (especially input/driver).

In blind testing, I found that there was simply no gain to going DC on the heaters for these valves/tubes. For 5V, 6.3V and 7V heaters its a different story altogether though.

I won't be a quack and claim that AC heaters are better though - at least for the SE 2a3, 45 and 46 amplifiers that I have designed and built.

If you have ripple in the B+ supply, you can do something about that, just use more B+ filtering.
.

Extra filtering means higher impedance.... I prefer to cancel B+ noise in some of my designs/builds. This can be complex for PP amplifiers though.

Trying to cancel the AC filament wave, by using the B+ ripple wave has problems.
.

Yes, that's definitely a bad idea so I don't even try to do it. Once I cancel the B+ noise, its a real challenge to hear the AC heater noise. I need to put my ear right next to the speaker. Of course your designs are probably significantly different than mine, so its pointless to argue. 🙂

The shape and harmonic structure of those two are not the same.
Take a look with a 2 channel scope sometime.

Done many times. Even physicists can learn how to use a 'scope. 😉

IMHO, for 2a3, 45, 46, 300b, and many 'western' DHT's with thoriated tungsten filaments, I have found the best/easiest DC circuits use CLCL filtering. Regulated circuits can also be advantageous though...

Ian
 
Last edited:
One IM Tester:
The problem with using a 60Hz 6000Hz IM tester in this application may be that the filter to reject noise and get the 60Hz sidebands may be narrow enough to filter off some or all of the 120Hz sidebands

Just a single pole low pass filter that starts a gentle roll off at about 600 Hz. I've attached the manual page showing the response characteristics of the filters.

The other tester:
Do try using the FFT.
The question is, what are the FFT filter bin widths? Can they be adjusted?
If they are not narrow enough, you will never see the 120Hz sidebands that are on the 1 kHz tone. Instead, you will see a single 1kHz tone (but that lumps the 120Hz sidebands in with it).

Looks non adjustable. It would be definitely be using it for a purpose for which it was never intended, and may be totally unsuited, but I'm not using it at all now, haven't even bothered it to break the seal on the box and unpack it. Attached is a page from the manual - says it will resolve to 20 Hz. This is pretty much all I know about it ......

Win W5JAG
 

Attachments