Guys,
The point I'm making here is that with a mosfet VAS the so called degeneration resistor is not doing any degeneration whatsoever ( if my spice is not lying )
The open loop gain does not decrease as the resistor value increases.
So I suggest that you try removing it and see what you think of the difference it makes to the sound.
If you decide to leave it there you can call it a stability impairing resistor - that's all it does !
Thanks Mike,
I must admit I have never tried this as spice suggested that the THD was lower with it in place so I just left it there. You may like to confirm this though as I can't recall exactly what my circuit looked like when I was spicing this.
I'm sure it will sound better without it, the bjt version most definitely does. Unfortunately my boards are done, it is boxed and final as far as this amplifier is concerned. I am already working on the next design.
Now, can I ask, have either you or Hugh tried no compensation and a big choke yet?
I found for my version the secret was no compensation, then make a choke as big as necessary to get suitable overshoot performance. This may be big or small depending on what you like and are willing to tolerate. With my tinny speakers I found the big choke was best, but the sound was by no means bad with no choke at all!
Cheers,
Greg.
I must admit I have never tried this as spice suggested that the THD was lower with it in place so I just left it there. You may like to confirm this though as I can't recall exactly what my circuit looked like when I was spicing this.
I'm sure it will sound better without it, the bjt version most definitely does. Unfortunately my boards are done, it is boxed and final as far as this amplifier is concerned. I am already working on the next design.
Now, can I ask, have either you or Hugh tried no compensation and a big choke yet?
I found for my version the secret was no compensation, then make a choke as big as necessary to get suitable overshoot performance. This may be big or small depending on what you like and are willing to tolerate. With my tinny speakers I found the big choke was best, but the sound was by no means bad with no choke at all!
Cheers,
Greg.
another variable . . . .
I just noticed that stability improves greatly as the drive impedance decreases so comparing what amount of compensation capacitance we are using really does not mean anything unless we control for factors like drive impedance & CLG . . .
another issue . . . .
I think the 500R in series with the miller cap sounds dreadful ! so forget that
I just noticed that stability improves greatly as the drive impedance decreases so comparing what amount of compensation capacitance we are using really does not mean anything unless we control for factors like drive impedance & CLG . . .
another issue . . . .
I think the 500R in series with the miller cap sounds dreadful ! so forget that
Last edited:
Yeah, I just don't need a gain of 21 but I will try it to see if it sounds better, it does help with stability but I can get a greater increase in stability by simply reducing the drive impedance.
I'm wondering, how you a guys are driving this amp ? passive or active pre-amp ? Potentiometer or TVC ? If it's a pot I wonder what value ?
I have now switched to a digital attenuator - when I compared the options it was clearly better - so it I could easily set my drive impedance as low as 50 or 100R - but who knows if it will it would sound better
I am now using 22pF of miller compensation, no resistor and it sounds good again.
I think phase angle at 100kz is probably more significant to the sound than frequency of 180 degrees.
I'm wondering, how you a guys are driving this amp ? passive or active pre-amp ? Potentiometer or TVC ? If it's a pot I wonder what value ?
I have now switched to a digital attenuator - when I compared the options it was clearly better - so it I could easily set my drive impedance as low as 50 or 100R - but who knows if it will it would sound better
I am now using 22pF of miller compensation, no resistor and it sounds good again.
I think phase angle at 100kz is probably more significant to the sound than frequency of 180 degrees.
I am using a computer soundcard at max volume through a 10k pot. I would like to use something more like 20k or 50k but I haven't thought that much about it. Although pulse characteristics change greatly depending on pot position... Probably should use something like the Pedja buffer.
- keantoken
- keantoken
I have posted my excel of the Cherry modified version of Thiele's output Network three or four times.
It was originally written to use either the R+C followed by the R//L
or
R//L followed by the R+C across the speaker terminals.
After using the spreadsheet that way in half a dozen amps over a decade or more, I came to realise that using the Pi version seems to work even better.
I automatically built up nFetzillas and Fetzilla with the full Pi Network without even trying the simpler versions.
I located the first R+C (10r+47nF) on the PCB, the R//L (5r0//0.7uH) off the PCB and the second R+C (5r0+68nF) across the speaker terminals.
Last edited:
Hi all,
I have been doing some serious building over the last few weeks building all sorts of versions of fetzilla. Some had a bjt vas and input, and some had long tail pair inputs. Some were all fet, some were not.
If I had to attribute the fetzilla sound to one key aspect, it would be the singleton input. I believe it is this aspect of the amp that provides the engaging sound and excellent bass response. The LTP sounds dry in comparison.
I don't believe the fets have as much to do with it as we would like to believe. Though they do add a little more enjoyment factor and warmth, a majority of this effect comes from the singleton input itself, not the device. Adding the fets is the icing on the cake, but a very comparable sound can also be had from a tweaked bjt version.
Over the coming days I will post a schematic for an inverted bjt version that I quite like the sound of. It's a little drier sounding than the fet version, but brings out more detail and better handles complex music.
If you want to make fetzilla stumble, try playing "The Battle" from the "Gladiator" soundtrack. The detail in the low end is completely lost, though it does compare admirably to most amps I have heard. I suspect mike's low gain version might do quite well here however.
I have been doing some serious building over the last few weeks building all sorts of versions of fetzilla. Some had a bjt vas and input, and some had long tail pair inputs. Some were all fet, some were not.
If I had to attribute the fetzilla sound to one key aspect, it would be the singleton input. I believe it is this aspect of the amp that provides the engaging sound and excellent bass response. The LTP sounds dry in comparison.
I don't believe the fets have as much to do with it as we would like to believe. Though they do add a little more enjoyment factor and warmth, a majority of this effect comes from the singleton input itself, not the device. Adding the fets is the icing on the cake, but a very comparable sound can also be had from a tweaked bjt version.
Over the coming days I will post a schematic for an inverted bjt version that I quite like the sound of. It's a little drier sounding than the fet version, but brings out more detail and better handles complex music.
If you want to make fetzilla stumble, try playing "The Battle" from the "Gladiator" soundtrack. The detail in the low end is completely lost, though it does compare admirably to most amps I have heard. I suspect mike's low gain version might do quite well here however.
BJTzilla
Hi All,
Here is the schematic as promised.
Starting from the front:
1) KSA992 low noise, high gain BJT input being biased by highly filtered LED voltage source.
2) Low capacitance CRT driver BJT VAS designed to be linear with high voltage/current swings (thanks Hugh).
3) Simple dual BJT VAS CCS. BD140 was chosen to keep it a bit slow and "doughy" to ensure stability without resorting to base stoppers. Current fixed at about 10-11mA. I experimented with a bootstrap here but ultimately preferred the sound of the bjt ccs.
4) Lateral fet output stage with asymmetric source resistors. This actually reduces distortion by matching the gain of the fets (see Cordell's book for more info). The difference is audible and sounds good!
How does it sound? Well, it's definitely not as engaging as a pure fet fetzilla, but it's a very honest amplifier. It is not as sterile sounding as the distortion spectrum might suggest. In fact, I question the spice profile given that it has both a single ended input and VAS. It is clean and sharp but still definitely more musical than most. Plenty of air around voices and lots of deep detailed bass. Clean treble. More musically capable than the fet version I would say.
In a pinch it can be made with BC560 and BD139 transistors and still sound quite good.
Hi All,
Here is the schematic as promised.
Starting from the front:
1) KSA992 low noise, high gain BJT input being biased by highly filtered LED voltage source.
2) Low capacitance CRT driver BJT VAS designed to be linear with high voltage/current swings (thanks Hugh).
3) Simple dual BJT VAS CCS. BD140 was chosen to keep it a bit slow and "doughy" to ensure stability without resorting to base stoppers. Current fixed at about 10-11mA. I experimented with a bootstrap here but ultimately preferred the sound of the bjt ccs.
4) Lateral fet output stage with asymmetric source resistors. This actually reduces distortion by matching the gain of the fets (see Cordell's book for more info). The difference is audible and sounds good!
How does it sound? Well, it's definitely not as engaging as a pure fet fetzilla, but it's a very honest amplifier. It is not as sterile sounding as the distortion spectrum might suggest. In fact, I question the spice profile given that it has both a single ended input and VAS. It is clean and sharp but still definitely more musical than most. Plenty of air around voices and lots of deep detailed bass. Clean treble. More musically capable than the fet version I would say.
In a pinch it can be made with BC560 and BD139 transistors and still sound quite good.
Attachments
Hi!
I think that your sim settings in LT spice are incorrect,
you should have visible peaks of the harmonics.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.