Best place to connect with Mike might actually be on Facebook..
I'm using Electra-Print output transformers in my GM70 amps, they're quite good.
I've also used Monolith Magnetics custom transformers in a number of my recent projects and would rank them at the top of the product quality pyramid for the transformers I have worked with.
Quote:
Will do. Then I can figure out who you are.........
Ron,
492 & 496 test and cal. Back up to the aisle, nearest to Pat B.
Will do. Then I can figure out who you are.........
Ron,
492 & 496 test and cal. Back up to the aisle, nearest to Pat B.
By original I take it would mean the pre 1990s 300B? I am guessing very few people have heard those tubes 🙂
I am going to make another guess that the reason many prefer other tubes is in difficulty driving 300B. This thread is a good read. Of all 300B SET designs I've looked at Gary Pimm's is the one I'd like to hear the most.
Edit: added schematic
Attachments
Last edited:
People here are upset with 2nd hd, but i tell you, a 5% has negligible shape distortion. I would concentrate on the odds hd. 3-5-7
Why this gets notoriously repeated? It´s like somebody years ago just made opinion from photos, and people keep saying it again. I doubt anybody compared bias points of these tubes, with other 2a3´s
Agree. My SE 2a3 at full power has 2.5% thd, 98% of this 2nd order.
But if I use a poorer 2a3 sample, it can easily be more. There is no audible difference though.
Still, just for fun I am tempted to try dc filaments on it again. Its not hard to do, just a tight fit in the chassis.
Also, maybe take a closer look at the FFT bin size used in the spectrum analyser software. I'm not sure if im distortion is fully incorporated into THD assessment. At the end of the day, intermodulation is not a harmonic...
Last edited:
An FFT shows what frequencies are present, and the relative strength of each of them.
Examples:
A fundamental and the first x harmonics (harmonic distortion).
Two tones; and the difference frequency, and the sum frequency (2nd order Intermodulation distortion).
3rd order Intermodulation distortion:
One tone - 2 times a second tone.
The 2nd tone - 2 times the first tone.
FM carrier and modulation sidebands
AM carrier and the upper and lower sideband (by the way, this is the nature of AC on Direct Heated Filaments, a single tone, surrounded by upper and lower sidebands at 2x the line frequency). i.e. 1kHz, 880Hz, and 1120Hz.
Examples:
A fundamental and the first x harmonics (harmonic distortion).
Two tones; and the difference frequency, and the sum frequency (2nd order Intermodulation distortion).
3rd order Intermodulation distortion:
One tone - 2 times a second tone.
The 2nd tone - 2 times the first tone.
FM carrier and modulation sidebands
AM carrier and the upper and lower sideband (by the way, this is the nature of AC on Direct Heated Filaments, a single tone, surrounded by upper and lower sidebands at 2x the line frequency). i.e. 1kHz, 880Hz, and 1120Hz.
With an FFT of 8192 I just don't see these frequencies as significant in my spectrum. Sampled at 44.1 kHz, range 20Hz-20kHz THD=2.52% and THD+N=2.81% (that's full power - at reasonably loud listening levels it is around 1%)
Most of the noise is of course heater hum, which (at least for me) is completely inaudible. Maybe I am going deaf though? Still maybe its worth getting rid of the inaudible heater "noise" just to see this other effect.
Most of the noise is of course heater hum, which (at least for me) is completely inaudible. Maybe I am going deaf though? Still maybe its worth getting rid of the inaudible heater "noise" just to see this other effect.
Last edited:
FWIW, probably not much, I did try this, and the results were inconclusive.
I have not unpacked and set up the DMU-2000 to try the FFT. That would be work; I am lazy.
I did try it with the AA-1. The noise floor of the meter does seem sufficiently low for this purpose. The meter does seem to be sensitive enough to easily detect the 60 Hz ac modulation of a single tone. I did use 3 kHz for the test tone.
A 2A3 clearly showed more modulation than a 45, which seems plausible - 2A3 draws more current. 46 and 47 should fall in between these two on a current drawn basis, but didn't. Some directly heated tube types may be better unwanted modulators than others.
At these sensitivity levels, and with my setup, I can't rule out the influence of stray 60 Hz noise floating around. I'll need to make better test leads. And I guess I would have to set up and test each type with DC on the heaters as some sort of control. So more work is needed to see if this is usable or not.
But the level of this modulation - if my results are not fatally corrupted or otherwise just more junk science - was more than I expected. Whether someone can hear its contribution or not is a different question.
Win W5JAG
The level of intermodulation (IMD) products with 2A3 is lower than with 300Bs, but it is still enough to cause measurable and audible problems.
The trouble is that ac-heating modulates both V(grid-filament) AND V(anode - filament). A DHT amp with ac heating is usually set up by adjusting balance pot to achieve minimum hum. With the pot at its best position, the fundamental (50/60Hz) is minimized by locating a 'best fit' on the triode curves where the combination of the effective V(grid-filament) AND V(anode - filament) is symmetrical.
The mechanism that causes the hum at idle is the same one that causes filament-heating induced IMD. So the finding of the best-fit for curve symmetry usually reduces both hum and IMD.
The fly in the ointment for this approach is that the symmetry is only valid at idle - i.e. the static operating point. As soon as large signal swings are applied to the anode and grid voltage, the curve-shapes alter, and symmetry is lost.
The easy way to assess this effect is to try George's (Tubelab) test. Connect the amp to a PC soundcard and start up your favourite spectrum analyser. But instead of injecting continuous tones, connect a stringed instrument to the amp's input instead.
This way, you can see the levels of IMD while you play a single string. Often, the IMD is quite low for gentle playing, but suddenly rises above a certain input level - i.e. where the triode-curve symmetry breaks down more dramatically. And as you listen, you can hear the effect of the IMD quite clearly. It does not sound right, as you would expect for something that has no natural relationship with the music signal - and you can experiment further by swapping to dc, and see the IMD - and fuzzy sound - disappear. Using low-impedance filament heating causes different problems, but it suffices for examining the measurable effect of IMD, and the difference in sound without it.
The trouble is that ac-heating modulates both V(grid-filament) AND V(anode - filament). A DHT amp with ac heating is usually set up by adjusting balance pot to achieve minimum hum. With the pot at its best position, the fundamental (50/60Hz) is minimized by locating a 'best fit' on the triode curves where the combination of the effective V(grid-filament) AND V(anode - filament) is symmetrical.
The mechanism that causes the hum at idle is the same one that causes filament-heating induced IMD. So the finding of the best-fit for curve symmetry usually reduces both hum and IMD.
The fly in the ointment for this approach is that the symmetry is only valid at idle - i.e. the static operating point. As soon as large signal swings are applied to the anode and grid voltage, the curve-shapes alter, and symmetry is lost.
The easy way to assess this effect is to try George's (Tubelab) test. Connect the amp to a PC soundcard and start up your favourite spectrum analyser. But instead of injecting continuous tones, connect a stringed instrument to the amp's input instead.
This way, you can see the levels of IMD while you play a single string. Often, the IMD is quite low for gentle playing, but suddenly rises above a certain input level - i.e. where the triode-curve symmetry breaks down more dramatically. And as you listen, you can hear the effect of the IMD quite clearly. It does not sound right, as you would expect for something that has no natural relationship with the music signal - and you can experiment further by swapping to dc, and see the IMD - and fuzzy sound - disappear. Using low-impedance filament heating causes different problems, but it suffices for examining the measurable effect of IMD, and the difference in sound without it.
I'm reviving this thread as a result of trying out a 2a3 in a SE with 5K OPT. I used a Hammond 266M5 for AC heating and I'm getting very audible hum. Could be that I routed the filament wires close to the cathode bypass caps and OPT. Early days.
I usually use a 300b with O-Netics or NP amorphous or LL1682 OPTs. This OPT is a cheaper one, and I'm quite disappointed with the sound. Very clear and transparent but it doesn't have the 300b magic in terms of timbre and tone on acoustic instruments and voices.
Next step is to substitute a 6C4C and also try the 2a3 out with one of my better OPTs.
I usually use a 300b with O-Netics or NP amorphous or LL1682 OPTs. This OPT is a cheaper one, and I'm quite disappointed with the sound. Very clear and transparent but it doesn't have the 300b magic in terms of timbre and tone on acoustic instruments and voices.
Next step is to substitute a 6C4C and also try the 2a3 out with one of my better OPTs.
If you run the 2A3 with a cheap OPT how can you conclude that the problem is the 2A3?
Anyway, IMHO 5K is not great for the "real" 2A3/6A3/6B4G/6C4C. If you look at some real curves (that are generally more linear than those in datasheets) you still get the typical 2A3 trademark: it is very linear up to a certain plate voltage and then it degrades quicker than other DHTs. This means that too high load impedance does not bring much benefit. For this reason I have opted for 3.5K and this is what I consider a max (nominal) primary load for good performance.
2.5K-3.5K is the range in other words.
For me, personally, the only 300B I would buy are those with 2.5V filament: 2A3-40 or TJ300B-2.5.
There are other more exotic but they are too expensive for what they offer (little-to-nothing, I suspect)....
Anyway, IMHO 5K is not great for the "real" 2A3/6A3/6B4G/6C4C. If you look at some real curves (that are generally more linear than those in datasheets) you still get the typical 2A3 trademark: it is very linear up to a certain plate voltage and then it degrades quicker than other DHTs. This means that too high load impedance does not bring much benefit. For this reason I have opted for 3.5K and this is what I consider a max (nominal) primary load for good performance.
2.5K-3.5K is the range in other words.
For me, personally, the only 300B I would buy are those with 2.5V filament: 2A3-40 or TJ300B-2.5.
There are other more exotic but they are too expensive for what they offer (little-to-nothing, I suspect)....
Andy - If you heat a 2A3 with AC, you will not get the results you are used to.
It's a difficulty seen quite often: many folks believe that the low voltage of the filament means low hum, and good sound. But 100/120Hz (and upward series) of intermodulation spurs will still show up in the output spectrum, even if the hum can be nulled out. You'll also hear the sound changing, if you use different types of AC-heat transformer - with none of them sounding quite right.
This is how I recall the sound to be, It does not have the mid & lower body texture like the 300Bs. Rca 2a3 has better texrure as it's not as bright/light sounding.
Well, those curves in the JJ datasheet I suspect they are just the original RCA curves in a slightly different presentation (mostly background...). For me the 2A3-40 is a 300B with 2.5V filament which is not a bad thing at all. Actually I prefer it as is.
Many people think it's different from a 300B because using 2.5V filament the bias "might look different" but in reality they aren't IME. The plate is the same as the 300B and I think the grid too. In reality a real 2A3 is not much different from 300B in terms of gm, Rp and gain for the same test conditions and it's easy reshuffle things....
I have just bought a pair of NP Acoustic 3.5K SE transformers (NP3.5SE18F60, precisely). Soon will be able to do some comparisons. I could do 45 or 46 PSE, 2A3-40 or TJ300B-2.5 and 6C4C/2A3 double plate. For the initial part, where I just want to see what's coming out of the audio circuit I am thinking about a cheaper SMPS supply with a few supply voltages options but need to learn a bit on this subject.
Last edited:
Years ago I built a monos with 2A3-40, 4 each amp, very good tube.
I have also measured them with Sofia and in some parts of my pc I have the curves
I will look and then I will post, if found
Walter
I have also measured them with Sofia and in some parts of my pc I have the curves
I will look and then I will post, if found
Walter
I've moved on from the 2a3 outputs with AC heating and built a whole new amp using 6C4C outputs in DC with Rod Coleman's regs. I'm now hearing a better comparison with my usual 300b output stages, using the same input stage. I've just connected the output stage up and am very pleasantly surprised by the good sound. The two output stages aren't directly comparable since they use different OPTs, both nominally 5K. The 6C4C stage is brighter and more lively with the same detail as the 300b and good bass.
I've literally just started listening to it but I like what I'm hearing and I plan to use the 6C4C in another amp with one of my best liked OPTs to experiment a bit.
I've literally just started listening to it but I like what I'm hearing and I plan to use the 6C4C in another amp with one of my best liked OPTs to experiment a bit.
I have used the JJ 2A3-40 for over five years, probably closer to a decade and it always biases up exactly as a 2A3. I have been posting this for years but the unsupported notion that it is a 300B refuses to die.