As I understand it, Pathos makes two types of output stages. For
class AB working, they use complementary MOSFETs in push-pull.
For single ended (or balanced SE) class A, they use their "INPOL"
circuit, described on their web site at www.pathosacoustics.com.
From descriptions I've read, I gather that INPOL is basically a
source follower loaded by a choke and coupled to the load by a
capacitor. The choke and cap together form an energy storage
system that supplies most of the dynamic output current, while
the source follower MOSFET controls the output voltage.
The key claim for this approach is that it allows the follower to
work under nearly constant-current conditions, so it can control
the output voltage with great accuracy.
class AB working, they use complementary MOSFETs in push-pull.
For single ended (or balanced SE) class A, they use their "INPOL"
circuit, described on their web site at www.pathosacoustics.com.
From descriptions I've read, I gather that INPOL is basically a
source follower loaded by a choke and coupled to the load by a
capacitor. The choke and cap together form an energy storage
system that supplies most of the dynamic output current, while
the source follower MOSFET controls the output voltage.
The key claim for this approach is that it allows the follower to
work under nearly constant-current conditions, so it can control
the output voltage with great accuracy.
Then perhaps I have misconstrued the claims made on their webNelson Pass said:
site. Unfortunately I can't locate a copy of their patent, from which
one might better understand and assess these claims.
In any case, the schematic posted by scottnixon above seems to
agree with the basic description given by Pathos and others.
modification....
Couldn't we choke load the 12AX7 cathode follower.... then use this to direct couple to the mosfets gates?
output operating current then dependent on the voltage drop across this choke.
Instead of a 12AX7... use a 6BX7 or something like that.
This would get rid of a coupling cap at least...
thoughts?
Couldn't we choke load the 12AX7 cathode follower.... then use this to direct couple to the mosfets gates?
output operating current then dependent on the voltage drop across this choke.
Instead of a 12AX7... use a 6BX7 or something like that.
This would get rid of a coupling cap at least...
thoughts?
I knew I wasn't making this up
Doing some more digging on this, I unearthed a review of the
Pathos Twin Towers amp by Dan Sweeny in Issue 116 (Feb '99)
of The Absolute Sound. In a sidebar on INPOL theory, he explains:
"Under no-signal conditions, the choke allows direct current to
flow unimpeded from the positive supply to ground [while] the
parallel capacitor ... provides additional energy storage while
equalizing the current draw for either half of the wave cycle, i.e.,
when the [output] MOSFET is sourcing or sinking current ...
"Impedances are set so that the total series impedance of the
capacitor plus the loudspeaker circuit is somewhat higher than
the impedance to ground; therefore current through the output
transistor never varies appreciably even though the internal
impedance of the MOSFET drops with increasing signal voltage."
In an effort to see if this could possibly be true, I quickly modeled
the INPOL output stage in Electronic Workbench. In this model at
least, the current through the MOSFET does vary by about 2:1 at
full output, just as it would with a constant current source load.
This appears to negate the claim of constant current operation.
This kind of choke loading does offer the benefit of cutting the
idle dissipation of the output stage in half when compared to a
CCS loading, as there is no need for a corresponding negative
power supply rail.
Doing some more digging on this, I unearthed a review of the
Pathos Twin Towers amp by Dan Sweeny in Issue 116 (Feb '99)
of The Absolute Sound. In a sidebar on INPOL theory, he explains:
"Under no-signal conditions, the choke allows direct current to
flow unimpeded from the positive supply to ground [while] the
parallel capacitor ... provides additional energy storage while
equalizing the current draw for either half of the wave cycle, i.e.,
when the [output] MOSFET is sourcing or sinking current ...
"Impedances are set so that the total series impedance of the
capacitor plus the loudspeaker circuit is somewhat higher than
the impedance to ground; therefore current through the output
transistor never varies appreciably even though the internal
impedance of the MOSFET drops with increasing signal voltage."
In an effort to see if this could possibly be true, I quickly modeled
the INPOL output stage in Electronic Workbench. In this model at
least, the current through the MOSFET does vary by about 2:1 at
full output, just as it would with a constant current source load.
This appears to negate the claim of constant current operation.
This kind of choke loading does offer the benefit of cutting the
idle dissipation of the output stage in half when compared to a
CCS loading, as there is no need for a corresponding negative
power supply rail.
"Impedances are set so that the total series impedance of the
capacitor plus the loudspeaker circuit is somewhat higher than
the impedance to ground; therefore current through the output
transistor never varies appreciably even though the internal
impedance of the MOSFET drops with increasing signal voltage."
That statement makes no sense to me at all, particularly
in reference to the Pathos schematic.
capacitor plus the loudspeaker circuit is somewhat higher than
the impedance to ground; therefore current through the output
transistor never varies appreciably even though the internal
impedance of the MOSFET drops with increasing signal voltage."
That statement makes no sense to me at all, particularly
in reference to the Pathos schematic.
The scheme I posted
Is the Pathos Twin Tower. Minus some values and supply / housekeeping. In another section ( I believe Tubes) I posted a scheme for a 1952 Japanese SE OTL amp, again choke loaded cap coupled follower. While it's not the best way to couple, it's certainly better than a big 'ol honkin' resistor
Is the Pathos Twin Tower. Minus some values and supply / housekeeping. In another section ( I believe Tubes) I posted a scheme for a 1952 Japanese SE OTL amp, again choke loaded cap coupled follower. While it's not the best way to couple, it's certainly better than a big 'ol honkin' resistor
choke loading
At first, thanks Scottnixon for the schematic!
Now to the technical details. If the coil was replaced with a resistance, there wouldn't be any question not for the role of the resistance (typical class A, with 25% theoretical efficiency), nor for the requirement of the big capacitor (capacitive=electrostatic charging of the "-"capacitor lead with the ac signal of the "+"lead, plus stoping of the dc to the speaker). But now that we have a coil, i think that exactly the same applies if (and only if) the frequency of the signal is such that the coil behaves as an impedance (i.e. freq.>DC). I do not see any mystery here, nor i believe that there should be any pattent granded for this coil application. (Unless i miss something very crusial). Tuned amplifiers utilise the same principle for decades.(There, the freq. is fixed and high, so the coil does not have to be big. But here, it has to be huge, in order for it's reactance to be roughly( in proportion) equal from 20 Hz to 20KHz.) I have not heard (PS*)how a coil loaded follower sounds (i do recall now that there was a class A amp coil loaded design published in Wireless World some years ago, if i will find it i will bring it to this thread) so i would like to ask the people who had such an opportunity. Does it worth the price? And one step further: Have they heard the same circuit with resistor or ccs in place of the coil? (of course the power efficiency is an other issue)
Regards
George
(PS*) Wrong! I have heard Pathos and i loved it. So, i would better write that i have not heard another coil loaded amp...
At first, thanks Scottnixon for the schematic!
Now to the technical details. If the coil was replaced with a resistance, there wouldn't be any question not for the role of the resistance (typical class A, with 25% theoretical efficiency), nor for the requirement of the big capacitor (capacitive=electrostatic charging of the "-"capacitor lead with the ac signal of the "+"lead, plus stoping of the dc to the speaker). But now that we have a coil, i think that exactly the same applies if (and only if) the frequency of the signal is such that the coil behaves as an impedance (i.e. freq.>DC). I do not see any mystery here, nor i believe that there should be any pattent granded for this coil application. (Unless i miss something very crusial). Tuned amplifiers utilise the same principle for decades.(There, the freq. is fixed and high, so the coil does not have to be big. But here, it has to be huge, in order for it's reactance to be roughly( in proportion) equal from 20 Hz to 20KHz.) I have not heard (PS*)how a coil loaded follower sounds (i do recall now that there was a class A amp coil loaded design published in Wireless World some years ago, if i will find it i will bring it to this thread) so i would like to ask the people who had such an opportunity. Does it worth the price? And one step further: Have they heard the same circuit with resistor or ccs in place of the coil? (of course the power efficiency is an other issue)
Regards
George
(PS*) Wrong! I have heard Pathos and i loved it. So, i would better write that i have not heard another coil loaded amp...
I believe the schematic as posted is fairly primitive.
There is no provision to improve the linearity of the mosfet.
Since we all agree stage current will vary widely regardless of current sources/inductors and various other trickery, then clearly the Vgs will also alter appreciably as the signal hikes up and down, and this will introduce lashes of second harmonic distortion.
If some way could be found to increase the transconductance, for example by using a triple darlington a la Nelson's own Class A circuit from the late seventies, then I can't see how this could ever be described as low distortion.
Mind you, with lashes of H2, it might sound like a tube!!
Cheers,
Hugh
There is no provision to improve the linearity of the mosfet.
Since we all agree stage current will vary widely regardless of current sources/inductors and various other trickery, then clearly the Vgs will also alter appreciably as the signal hikes up and down, and this will introduce lashes of second harmonic distortion.
If some way could be found to increase the transconductance, for example by using a triple darlington a la Nelson's own Class A circuit from the late seventies, then I can't see how this could ever be described as low distortion.
Mind you, with lashes of H2, it might sound like a tube!!
Cheers,
Hugh
Re: choke loading
could get a patent awarded based simply on the schematic and
"explanation" shown in this thread. I may have given Pathos too
much credit already, but I continue to hope that their actual
patent makes more sense than is apparent here.
believe that it is not the total change in Vgs with Id, but only the
nonlinear extent of the change, which creates distortion. Over
a 2:1 variation in Id, peak H2 would be at a small fraction of a
volt, so I would expect to see less than 1% for a 30W amplifier,
which seems acceptable, no?
Even in these jaded times, I also have trouble believing that yougpapag said:
could get a patent awarded based simply on the schematic and
"explanation" shown in this thread. I may have given Pathos too
much credit already, but I continue to hope that their actual
patent makes more sense than is apparent here.
Hugh, this may indeed be what you meant, but just to be clear: IAKSA said:
believe that it is not the total change in Vgs with Id, but only the
nonlinear extent of the change, which creates distortion. Over
a 2:1 variation in Id, peak H2 would be at a small fraction of a
volt, so I would expect to see less than 1% for a 30W amplifier,
which seems acceptable, no?
Any idea why the Schiebold patent is valid? I'd practically swear I've seen this circuit in transistor manuals from the 60's. Here's a pic from a JLH article some 25 years ago albeit with the speaker across the transistor.
From the patent: "Alternatively, a high impedance current source can be used instead of the choke" Aren't we pretty much talking Zen here?
From the patent: "Alternatively, a high impedance current source can be used instead of the choke" Aren't we pretty much talking Zen here?
Attachments
Mike,
The Schiebold circuit posted earlier differs from your examples in
having no DC blocking cap in series with the load. On account of
this, the choke in the Scheibold circuit must have low resistance
so there is minimal DC voltage across the speaker at idle. I've not
been able to access the patent, so I can't say if this difference
constitutes the circuit's sole claim to novelty.
You could put the same arrangement in the Source circuit instead
of the Drain, as is done in Circlotron's "my first ever class A amp"
(see thread by same name). This is also what Pathos does, but
they further include a DC blocking cap, albeit for other purposes
which are controversial, to say the least. I hope this helps.
The Schiebold circuit posted earlier differs from your examples in
having no DC blocking cap in series with the load. On account of
this, the choke in the Scheibold circuit must have low resistance
so there is minimal DC voltage across the speaker at idle. I've not
been able to access the patent, so I can't say if this difference
constitutes the circuit's sole claim to novelty.
You could put the same arrangement in the Source circuit instead
of the Drain, as is done in Circlotron's "my first ever class A amp"
(see thread by same name). This is also what Pathos does, but
they further include a DC blocking cap, albeit for other purposes
which are controversial, to say the least. I hope this helps.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.