I find the way bandwidth limiting is done changes much sonically . I have Parkinsosns so tubes are my new hobby as soldering is less critical . Otherwise you would never have heard form me . I am the same about motorcycles , love them yet usually keep away from them .
In a MOSFET amp using the FET cg to limit bandwidth sounds far better than seemingly identical measuring alternatives which I wont bore you with ( I use 220R as a starting point if Lateral FET's ) . It strikes me that exactly the same happens here . Setting a grid leak critically will mater as will RC filter of the grid stopper ( do we want grid current assistance and roll off ) . My theory is manipulate what is difficult to change . Do not add unnecessary mechanisms out of laziness . More importantly , as the man says any good cat is a good cat .
In a MOSFET amp using the FET cg to limit bandwidth sounds far better than seemingly identical measuring alternatives which I wont bore you with ( I use 220R as a starting point if Lateral FET's ) . It strikes me that exactly the same happens here . Setting a grid leak critically will mater as will RC filter of the grid stopper ( do we want grid current assistance and roll off ) . My theory is manipulate what is difficult to change . Do not add unnecessary mechanisms out of laziness . More importantly , as the man says any good cat is a good cat .
Last edited:
you're right! My appologies. I was thinking bandwidth the whole time and wasn't till I sat in the car on my way home I suddenly realized it was slewrate you were talking about. Oh well, just goes to prove everything we read on the www must be taken with a grain of salt.
You can of course slew a given capacitance using a CCS just as well as if using a voltage source, (but the high impedance of the ccs will give a lower cutoff frequency).
In some cases cathode followers don't sound so good. Could this be when they are tasked to swing large voltages? When idle current is too low cathode followers can introduce slewrate distortions in one direction (when going down towards off since for large swings the cathode resistor ends up being the only 'driver') while when they swing upwards they source whatever current necessary so we end up with a distorted waveform due to slewrate limiting. In such a case the low output impedance of the CF will give a high bandwidth stage, but the low current can't slew the capacitance. This will sound much worse than a higher impedance common cathode stage where bandwidth is limited before slewrate gets to set in. 12AX7s and similars will probably act such a way.
I used a transformer volume control sent to me to play with ( sorry it was given back and I forgot to write the make down ) . It was an inexpensive one . Suddenly a passive pre amp worked . Doubtless the Sowter version would be better . That even offers 6 db gain which would be useful . I suspect it has a little bandwidth control and more current at low volumes as advantages .
Member
Joined 2009
Paid Member
You mean they are better than using capacitors.... From what I've read, capacitors add their distortion to the signal when they have significant a.c. voltage across them, whereas when they have very little a.c. across them they can be very clean. So I could imagine that where they are used to limit bandwidth they are being operated just where they do worse, because the corner frequency will occur when there is a growing amount of a.c. across the capacitor.
This thread relates to a recent project - - Tuning up and improving an old project. Years ago during a serious iron fetish phase I built a stereo PP 6A3 amp with minimal capacitor usage, as in not a single electrolytic except for the heater supply for the preamp tubes.
The signal path starts with a Tamura input tranny that does the phase splitting and drives a single 407A, which is biased by a pair of AAA batteries on the CT of the secondary. The plate load for the 407 is a CT audio choke, cap coupled to the grids of the 6A3s. So we've got one little tube driving the pair of outputs and having no trouble doing so. The frequency limitations are in the iron, this starts to roll off at 15K, but that's apparent at the grid of the 407. It does go all the way down to 20Hz, but the waveform gets a bit wobbly looking below 30. Mid-band this puts out a solid 10watts very cleanly and about 13 if we allow a little crunchiness.
Anyway, the two main points that relate are: driving the 2A3 (or 6A3) is pretty easy. It's a friendly load despite needing a lot of swing to make full power. And despite it's obvious frequency limitations, this amp sounds excellent and nobody has ever noticed the FR limitations on either the top or bottom end. I suspect the added harmonics on the low end fool the ear into thinking the bass goes lower than it does.
Anyway, a few thoughts ....
The signal path starts with a Tamura input tranny that does the phase splitting and drives a single 407A, which is biased by a pair of AAA batteries on the CT of the secondary. The plate load for the 407 is a CT audio choke, cap coupled to the grids of the 6A3s. So we've got one little tube driving the pair of outputs and having no trouble doing so. The frequency limitations are in the iron, this starts to roll off at 15K, but that's apparent at the grid of the 407. It does go all the way down to 20Hz, but the waveform gets a bit wobbly looking below 30. Mid-band this puts out a solid 10watts very cleanly and about 13 if we allow a little crunchiness.
Anyway, the two main points that relate are: driving the 2A3 (or 6A3) is pretty easy. It's a friendly load despite needing a lot of swing to make full power. And despite it's obvious frequency limitations, this amp sounds excellent and nobody has ever noticed the FR limitations on either the top or bottom end. I suspect the added harmonics on the low end fool the ear into thinking the bass goes lower than it does.
Anyway, a few thoughts ....
Member
Joined 2009
Paid Member
Do you need to limit BW? Are you experiencing slew induced distortions?
I also like the sound of interstage transformers over RC stages, and perhaps it is b/c of BW but I don't think so since BW is > 50kHz with interstage tranny.
If you must limit BW I'd rather use the internal capacitance of the tube and connecting parts and place a series resitor to give some HF roll-off, or use high impedance 'wimpy' driver. That is unless the corner frequency is so low that it becomes impractical.
Or use 'manly' high current driver (6W6 in triode?)
My first really serious own design was a SE driver into a PP phase splitting transformer which started to roll off at about 15k. It was universally liked as a very respectable performer with a sweet but detailed presentation. I decommissioned it eventually because it relied on parafeeding the IT. Since then I have used input phase splitters with a step down to extend the bandwidth - with DC coupling to the outputs. They generally perform well and extend out to about 30Khz.
What you have to be careful of with IT's of any type is if they are rolling off low then they will tend to peak out at 60khz or higher. This tends to artificially compensate for the the rolled off top end - but is very tiring in extended use. I have also found that reversing the direction of the winding can very often kill this peaking - whilst extending bottom end.
The use of Input transformers can make DC coupling a lot easier to implement with negative rails.
The only reason to be worried excessively about IT roll off is if if it is cascading from one transformer stage to the next - then you can expect very poor high frequency performance.
When you get to two stage zero global feedback designs running into fullrange speakers at 100db/w, its amazing how much any caps in the signal path tend to dominate the overall sound. I really didn't believe it until I heard it myself. To my ear an IT or DC coupling will always sound better - even when slightly rolled off.
Shoog
What you have to be careful of with IT's of any type is if they are rolling off low then they will tend to peak out at 60khz or higher. This tends to artificially compensate for the the rolled off top end - but is very tiring in extended use. I have also found that reversing the direction of the winding can very often kill this peaking - whilst extending bottom end.
The use of Input transformers can make DC coupling a lot easier to implement with negative rails.
The only reason to be worried excessively about IT roll off is if if it is cascading from one transformer stage to the next - then you can expect very poor high frequency performance.
When you get to two stage zero global feedback designs running into fullrange speakers at 100db/w, its amazing how much any caps in the signal path tend to dominate the overall sound. I really didn't believe it until I heard it myself. To my ear an IT or DC coupling will always sound better - even when slightly rolled off.
Shoog
Last edited:
I am a fan of the high mu high gm space age tubes D3A-triode, 6c45pi,we417 etc, for spud low power head-amps.
But times have changed, we now have orthos, the ones I like need at lease 2 watts. That means swinging 50+V rms across a primary. You need to swing about 9Vrms across the 38 ohm dielectric.
Recently spent a lot of time simulating the 6c45pi and the others like it driving a 6S4S/2A3 SET. According the the simulations the performance is not good even at only 1 watt. The 6N6p and many others were beating those tubes by nearly -20db 2H and 3H.
So I am a little hesitant to follow my intuition and use a 6c45pi. I can't explain why they perform so poorly in simulation driving 2A3 type tubes but notice a common trait of a subsonic oscillation on the output. Hope its just my skill level with LTSpice because the 6n6p is a bit short of gain.
Member
Joined 2009
Paid Member
That almost certainly indicates oscillation. That's what it sounds like.
You are going to damage your hearing listening that loudly
Most of my amps are being used with LCDs, and thus far nobody has complained about low volume.
Member
Joined 2009
Paid Member
where is hypothetical limit of improving tube amps parameters ?
what i try to say, is possible to build amp with modern framegrids,ten thousands of µhmos tubes, low pri OPT, or even OTL... really wideband stuff
but, why not to build rather classD with some "tube fx" instead
Have not double checked values, but should be close. Rod's or Tentlabs DC 2.5V filament supply a good option.
I'm wondering how this circuit manages to pass a 75Vpp signal to the 2A3. I mean, the CF has circa 3 volt bias... is its grid driven positive?
Member
Joined 2009
Paid Member
As the current changes through the cathode follower with the signal, it's cathode rises and falls due to the voltage drop across the resistance in the cathode circuit This allows the follower to move its cathode voltage over a wide range whilst still keeping the cathode more positive than the grid.
But the 2A3 is typically operated over a range of 45Vp to 50pp not 75Vp. You'll see that the grid-cathode voltage of the 2A3 is the difference between 120V and 160V, so only 40V and if you try to swing the grid more than 40V positive it will clip the signal due to the 6SL7 being unable to provide very much grid current to the 2A3.
But the 2A3 is typically operated over a range of 45Vp to 50pp not 75Vp. You'll see that the grid-cathode voltage of the 2A3 is the difference between 120V and 160V, so only 40V and if you try to swing the grid more than 40V positive it will clip the signal due to the 6SL7 being unable to provide very much grid current to the 2A3.
Well, it depends on the allowed distortion, see this graphic:
If my math is not too rusty an output power of 4,5W equals 6,0 Vrms into a 8 ohm resistor. The 8:2500 transformer has 1:17,67 turns ratio, so our 6,0 Vrms comes from at least 106Vrms on the primary. That equals 296,8Vpp coming from our 4,2 times amplifying triode, which would take 296,8/4,2= 70,7Vpp on its grid.
An externally hosted image should be here but it was not working when we last tested it.
If my math is not too rusty an output power of 4,5W equals 6,0 Vrms into a 8 ohm resistor. The 8:2500 transformer has 1:17,67 turns ratio, so our 6,0 Vrms comes from at least 106Vrms on the primary. That equals 296,8Vpp coming from our 4,2 times amplifying triode, which would take 296,8/4,2= 70,7Vpp on its grid.
Or ~ 35Vpk which is consistent with what Bigun noted in the previous post..
(My comment: 150Vpk on the primary of the output transformer, mu assumed 4.2 gives 35.7Vpk on the grid, with losses likely more like 40Vpk)
I have never gotten more than 3Wrms with acceptable distortion levels out of a 2A3.
(My comment: 150Vpk on the primary of the output transformer, mu assumed 4.2 gives 35.7Vpk on the grid, with losses likely more like 40Vpk)
I have never gotten more than 3Wrms with acceptable distortion levels out of a 2A3.
OK, after sorting out the misunderstanding about peak-peak and peak back to the circuit.
The CF has its cathode riding on the audio signal. That same audio signal is applied to the CF grid, but now at half the voltage. As a result the CF stage has no longer unity gain but amplifies while output impedance is still low. Is that correct?
The CF has its cathode riding on the audio signal. That same audio signal is applied to the CF grid, but now at half the voltage. As a result the CF stage has no longer unity gain but amplifies while output impedance is still low. Is that correct?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.