Now that I have functional Aleph-X amps, the whole absolute DC offset issue has been bothering me. I have successfully adjusted the differential pair CCS so that the absolute DC offset is no more than a tenth of a volt or so once the amp has reached thermal equilibrium. However, when the amp is turned on from a cold start the absolute DC offset is of the order of a few volts, and slowly declines to lower values as the amp warms up. This places a disproportionate fraction of the amps power dissipation on the output gain transistors, and even though this occurs at a time when the heatsinks are also cold, increasing the transistors tolerance to operate at higher power levels, I would much prefer to keep the power dissipation evenly distributed between all of the output transistors. The idea of a servo to correct the DC offset of the Aleph-X has been mentioned a few times in the forum, but I haven’t seen anyone actually post an idea for one, so I came up with an idea that I’d like to toss out there for consideration.
The current method for adjusting the absolute DC offset is to adjust the source resistance of Q6, which changes the bias current for the differential pair, thus changing the voltage across R23 and R25 which sets the conductance of Q2 and Q11. I started thinking about trying to create a circuit that would perform the same bias adjustment, because it would keep the servo adjustment outside of the signal path.
My idea is based on two concepts:
1) A small level of absolute DC offset is acceptable; and
2) Opamps area available that can operate at high gain levels.
My thought is to remove D1 (the voltage reference) and R17, and replace them with the opamp based circuit I have attached, so that the Vgs of Q6 is now set by the opamp output. If the opamp is used to invert and amplify the absolute DC offset voltage by a large amount, and then the amplified signal is applied to the gate of Q6, then a loop is formed which will set the differential pair bias such that the absolute offset is maintained as a very small negative fraction of the voltage between the gate of Q6 and ground. The will not trend to a zero offset, but it will continuously adjust it to a small value.
For an example, let’s assume an Aleph-X with +/- 15V rails, a Vgs for the CCS mosfet (Q6) approx. 4V generating a current of 20mA, and a CCS source resistance (R24, R26 and VR2) of 150 ohms. Next, let’s hypothetically assume that the DC offset is currently +1V, and the opamp gain is set to 10X and is inverting. The opamp then pushes the gate of Q6 to –10V relative to ground, resulting in a very large Vgs for Q6. Q6 then responds by increasing the bias current, which causes a decrease of the DC offset. As the offset becomes more negative the opamp output voltage decreases, thus limiting the increase in the bias. Likewise, if the DC offset is –1V, then the gate of Q6 is pushed towards +10V relative to ground, decreasing the Vgs of Q6, and decreasing the bias current. For the above assumed conditions, the circuit should maintain approximately a –0.08V absolute offset.
In the attached proposed circuit, I have set the gain of the opamp to 160X (~44dB) based on the concept that for a max ouput of the opamp of +/-15V relative to ground (i.e. the voltage rails of my Aleph-X), then the max DC offset under any condition will be less than 0.1V, and the circuit should force a set absolute DC offset of approximately –50mV. I have also low pass filtered the opamp input to keeps its corrective operation real slow (i.e. down below the audible range).
Does this idea seem workable???
Cheers, Terry
The current method for adjusting the absolute DC offset is to adjust the source resistance of Q6, which changes the bias current for the differential pair, thus changing the voltage across R23 and R25 which sets the conductance of Q2 and Q11. I started thinking about trying to create a circuit that would perform the same bias adjustment, because it would keep the servo adjustment outside of the signal path.
My idea is based on two concepts:
1) A small level of absolute DC offset is acceptable; and
2) Opamps area available that can operate at high gain levels.
My thought is to remove D1 (the voltage reference) and R17, and replace them with the opamp based circuit I have attached, so that the Vgs of Q6 is now set by the opamp output. If the opamp is used to invert and amplify the absolute DC offset voltage by a large amount, and then the amplified signal is applied to the gate of Q6, then a loop is formed which will set the differential pair bias such that the absolute offset is maintained as a very small negative fraction of the voltage between the gate of Q6 and ground. The will not trend to a zero offset, but it will continuously adjust it to a small value.
For an example, let’s assume an Aleph-X with +/- 15V rails, a Vgs for the CCS mosfet (Q6) approx. 4V generating a current of 20mA, and a CCS source resistance (R24, R26 and VR2) of 150 ohms. Next, let’s hypothetically assume that the DC offset is currently +1V, and the opamp gain is set to 10X and is inverting. The opamp then pushes the gate of Q6 to –10V relative to ground, resulting in a very large Vgs for Q6. Q6 then responds by increasing the bias current, which causes a decrease of the DC offset. As the offset becomes more negative the opamp output voltage decreases, thus limiting the increase in the bias. Likewise, if the DC offset is –1V, then the gate of Q6 is pushed towards +10V relative to ground, decreasing the Vgs of Q6, and decreasing the bias current. For the above assumed conditions, the circuit should maintain approximately a –0.08V absolute offset.
In the attached proposed circuit, I have set the gain of the opamp to 160X (~44dB) based on the concept that for a max ouput of the opamp of +/-15V relative to ground (i.e. the voltage rails of my Aleph-X), then the max DC offset under any condition will be less than 0.1V, and the circuit should force a set absolute DC offset of approximately –50mV. I have also low pass filtered the opamp input to keeps its corrective operation real slow (i.e. down below the audible range).
Does this idea seem workable???
Cheers, Terry
Attachments
I mentioned a discrete servo in the very beginning of the Aleph-X thread. I don't recall that I ever posted the schematic, but someone else posted one. All it takes is a pair of transistors (I seem to recall that I used MPSA18s, but just about anything will do) arranged as a differential.
Grey
Grey
Grey,
Thanks for giving me a point in the general direction. After some further more focused searching and reading I found what you were referring to along with the long discussion about why it is better not to use a servo with the Aleph-X.
Moderators: In the interest of saving forum storage space and bandwidth, I suggest that this thread should be deleted.
Cheers, Terry
Thanks for giving me a point in the general direction. After some further more focused searching and reading I found what you were referring to along with the long discussion about why it is better not to use a servo with the Aleph-X.
Moderators: In the interest of saving forum storage space and bandwidth, I suggest that this thread should be deleted.
Cheers, Terry
preheated heatsinks
metalman
"Now that I have functional Aleph-X amps, the whole absolute DC offset issue has been bothering me. I have successfully adjusted the differential pair CCS so that the absolute DC offset is no more than a tenth of a volt or so once the amp has reached thermal equilibrium."
maybe you could preheat the heatsinks?
the add a led so the tube guys would be happy?
sorry
just a weird sense of humor
wheres the icon for falling off the chair laughing?
metalman
"Now that I have functional Aleph-X amps, the whole absolute DC offset issue has been bothering me. I have successfully adjusted the differential pair CCS so that the absolute DC offset is no more than a tenth of a volt or so once the amp has reached thermal equilibrium."
maybe you could preheat the heatsinks?
the add a led so the tube guys would be happy?
sorry
just a weird sense of humor
wheres the icon for falling off the chair laughing?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.