Member
Joined 2009
Paid Member
From reading a few web pages I find a great deal of respect for the sound of the Fi 2A3 amplifier. It is reputedly based on an old design (aren't they all) that I saw once in 'practical amplifier diagrams' by Robin and Lipman 1947. It's a 6SF5 directly coupled to a 2A3, the driver tube is biassed at only 0.4 to 0.5mA and has a rather high internal resistance.
Many people who have been building amps for many years will say that this is simply no good, that the driver is too wimpy (too weak), go try a 6C45P or something. (I personally heard a 6C45P at relatively high current driving a 6C4C and it was a tad clinical, a triode wired 6E5P was nicer)
Yet there is no doubt about the reverence that the sound of the Fi 2A3 is held.
This reminds me of the AKSA amplifier by Hugh Dean. A simple and old design that has been tweaked up by somebody who knows their engineering but values sound above all. A topology that is now bypassed by most people on the search for something better. Yet the sound is simply lovely. And I believe Hugh has played with wimpy drivers (e.g. 6SL7) in his past.
Does anybody know why the wimpy driver approach works ?
Many people who have been building amps for many years will say that this is simply no good, that the driver is too wimpy (too weak), go try a 6C45P or something. (I personally heard a 6C45P at relatively high current driving a 6C4C and it was a tad clinical, a triode wired 6E5P was nicer)
Yet there is no doubt about the reverence that the sound of the Fi 2A3 is held.
This reminds me of the AKSA amplifier by Hugh Dean. A simple and old design that has been tweaked up by somebody who knows their engineering but values sound above all. A topology that is now bypassed by most people on the search for something better. Yet the sound is simply lovely. And I believe Hugh has played with wimpy drivers (e.g. 6SL7) in his past.
Does anybody know why the wimpy driver approach works ?
Last edited:
Cold feet? 
The amp is the sum of its parts. The perfect driver driving an imperfect output stage may be less than the "imperfect" driver, driving the same output stage. Per Mr. D., lots of older designs used driver tubes at lower currents than current designs. No sims. All they could do was try it.
Sheldon
The amp is the sum of its parts. The perfect driver driving an imperfect output stage may be less than the "imperfect" driver, driving the same output stage. Per Mr. D., lots of older designs used driver tubes at lower currents than current designs. No sims. All they could do was try it.Sheldon
The 6SF5 is hideously nonlinear, being that it was designed to serve as a high gain pre for AM or COMM receivers, and linearity wasn't a concern. There is always the possibility that its nonlinearities compensate for the nonlinearities of the 2A3 + OPT final. It's not something I'd want to stick in front of any DHT. I have some 6SF5s in the junk box, but if I ever use 'em it'll be as intended: receiver front end amp.
There's also the possibility of user preference.
"I personally heard a 6C45P at relatively high current driving a 6C4C and it was a tad clinical, a triode wired 6E5P was nicer".
"A tad clinical" looks to me like the driver was doing what it should: giving less distortion. If you want "nicer", you're basically wanting an FX box. Nothing wrong with that, as individual preference is, well, individual.
The 6F5/6SF5 doesn't look "hideously nonlinear" to me. It looks like it is quite linear for the input range used by Fi and many others. The issue is not non-linearity; the issue is whether it has the current necessary to drive the Miller capacitance of the 2A3.
http://www.mif.pg.gda.pl/homepages/frank/sheets/021/6/6F5.pdf
http://www.mif.pg.gda.pl/homepages/frank/sheets/021/6/6F5.pdf
I did some quick calculations and it appears that 2A3 needs only about .5mA in the driver to avoid slew rate limiting when driven to clipping at 20kHz. That's why is sounds okay.
The general rule of thumb in DIY is to use 5x what is needed, though because slew rate induced distortions sound really bad.
The 1947 amp probably sounds really bad when it is clipping because it will not only be clipping in the output stage but the driver will have trouble slewing as well. However, up to the point of clipping, it would sound fine.
EDIT: I did an experiment once, built an amp that had really good overload recovery and no slew problems. I listened to music which I could visibly see was clipping. It didn't sound too bad. The stuff that sounds bad is the blocking distortions, slewing distortions, IMD, etc.
The general rule of thumb in DIY is to use 5x what is needed, though because slew rate induced distortions sound really bad.
The 1947 amp probably sounds really bad when it is clipping because it will not only be clipping in the output stage but the driver will have trouble slewing as well. However, up to the point of clipping, it would sound fine.
EDIT: I did an experiment once, built an amp that had really good overload recovery and no slew problems. I listened to music which I could visibly see was clipping. It didn't sound too bad. The stuff that sounds bad is the blocking distortions, slewing distortions, IMD, etc.
Last edited:
You have corroborated what others have calculated over and over on this board and others; yet the debate continues....
2A3
One of the shocks of my life was listening to a 2A3 amplifier used by the boss of Japans FAL speaker company . The sound was not only first class it was almost home cinema in dynamics . I stopped reading spec sheets that day and trusted my ears . I am told PX25 is similar . Build it and try it . Make sure the bias is correct naturally .
One of the shocks of my life was listening to a 2A3 amplifier used by the boss of Japans FAL speaker company . The sound was not only first class it was almost home cinema in dynamics . I stopped reading spec sheets that day and trusted my ears . I am told PX25 is similar . Build it and try it . Make sure the bias is correct naturally .
Actually do need a little tweaking (input cathode resistor should be smaller or plate resistors larger to get the current to match plate curves) but basics are there.
Member
Joined 2009
Paid Member
Cold feet?
...more like anaysis-paralysis. I should be ordering some parts but I'm reluctant to order a huge pile of bits just to try out every combination.
I have found simulations invaluable when I have eningeered a SS amp design. But I haven't found them to be that useful in replacing my ears - the listening tests often throw up some surprises.
They were driving a 2A3 through a Hammond IT, and if I remember they were both running at 30mA to 35mA, the plate voltage was around 200V.
I also expected them to sound the same. The 6C45P was the cleanest. And my friend (who had asked me to do the listening tests for his amp design) chose this tube for his final design. There was clearly a difference in the sound. The 6E5P was triode wired with a 100 Ohm screen-plate resistor. It was smooooth and a bit more mellow. Perhaps the sound depends on how it is triode wired.
this begs the question - how does the Fi 2A3 sound when overloaded ?
Have not double checked values, but should be close. Rod's or Tentlabs DC 2.5V filament supply a good option.
Wait, that's a 6SL7 cathode follower driving the 2A3. Hey, that's supposed to work, even...
Of course, there are better tubes to make a cathode follower out of. If you were making a stereo amp, you could use 6SL7 as the voltage amplifier into a cathode follower made from a high-gm tube like 5687, D3a or 6688 (triode or pentode, I suppose). Or a (heaven forbid!) MOSFET emitter follower, maybe DC-coupled to the 2A3 grid.
Somehow, I thought you were talking about a high-mu, high-rp tube like 6SL7 or 12AX7 used as a common-cathode amp driving the 2A3. Now, that would be controversial.
--
It's irrelevant w/respect to driver slew rate. rp doesn't enter the picture, only stage standing current. See this if you have any questions.
"Better", to what end? Given the low load, the high mu tube works fine. At a higher load, yes there are better choices. Horses for courses. And yes, an FET would work fine (and for this amp, direct coupled, of course). Or an FET into a current source, (or whatever level of complexity you want, to drive the follower specs. to the nth degree.)
Sheldon
Last edited:
That's only part of it. The other part is that it's a bootstrapped follower, The input tube plate sees essentially a current source, since the voltage across the bottom of the two plate resistors is held constant. Net result is that full mu is realized, even with an unbypassed cathode resistor. Also, the plate resistor values can be changed without changing the load line.
Sheldon
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.