A pentode is a voltage controlled current element. If it is driven by a high impedance current source , a feedback resistor linking the plate to the grid transforms it into a current controlled, low impedance voltage element. The impedance of the plate becomes inverse function of the tube's transconductance. As the transconductance varies with the current , it behaves as a diode . If a 6L6 is used ,the impedance becomes a near square law function as this tube is designed to have a linearly varying trasconductance (g2=250v). At 80ma operating point the plate behaves as a resistance of 160 ohms. The tube now can deliver it's maximum power 12Watts with high damping factor and low distortion ,which is mainy of even order harmonics ,easily canceled by the driver .Overall closed loop feedback is no more necessary.
When the driver provides an ac current , the grid being high impedance , the plate counters the ac current developing an ac voltage across the feedback resistor proportional to the input current , similar to an inverting opamp. If the amp is not loaded , the grid voltage remains near constant , allowing the plate to have the same voltage excursion as the feedback resistor with low internal impedance when loaded.
The EF86 can provide about +/-1ma linear excursion operating at 3ma. Using a plate-grid feedback resistor of 330kohms, an inductor is also necessary to provide the excess current to the driver and precise bias to the power tube. This choke is simulated by a mosfet To92 transistor 2n7000 protected at startups by a zener diode.
The driver can also be made of a second mosfet 2n7000 operating at 2.5ma.
Other pentodes can also be used as EL84 with 90ohms internal resistance to give 5W with high demping factor or KT88 for 18W.
WARNING:Believe it or not the 2N7000 has a defective conception. If the transistor is heated by soldering ,even by an unpluged iron,the isolated gate brakes down by a gate-source leak.This why it should be used only on socket. I tried motorola and philips both have the same problem but the philips has much lower drain resistance that makes it unusable with this amp.
KOKORIANTZ
When the driver provides an ac current , the grid being high impedance , the plate counters the ac current developing an ac voltage across the feedback resistor proportional to the input current , similar to an inverting opamp. If the amp is not loaded , the grid voltage remains near constant , allowing the plate to have the same voltage excursion as the feedback resistor with low internal impedance when loaded.
The EF86 can provide about +/-1ma linear excursion operating at 3ma. Using a plate-grid feedback resistor of 330kohms, an inductor is also necessary to provide the excess current to the driver and precise bias to the power tube. This choke is simulated by a mosfet To92 transistor 2n7000 protected at startups by a zener diode.
The driver can also be made of a second mosfet 2n7000 operating at 2.5ma.
Other pentodes can also be used as EL84 with 90ohms internal resistance to give 5W with high demping factor or KT88 for 18W.
WARNING:Believe it or not the 2N7000 has a defective conception. If the transistor is heated by soldering ,even by an unpluged iron,the isolated gate brakes down by a gate-source leak.This why it should be used only on socket. I tried motorola and philips both have the same problem but the philips has much lower drain resistance that makes it unusable with this amp.
KOKORIANTZ
Attachments
bias issues
Thank you for being interested to this outstanding amplifier .
The grid bias voltage is provided by a zener diode refrenced to the cathode along with a source follower.It is totally independent from the plate current. It is precise and robust to have the tube used near maximum power dissipation. Cathode bias is also possible.
At startup , the 6L6 heats up faster than the driver , within three seconds it passes from zero to 95/100ma plate current, when the driver's plate current pulls it suddenly, back to zero.For the next 25seconds the amp remains dead ,waiting the gate-source capacitor to be charged by the driver to 2.4v threshold .Suddenly the amplifier starts singing reaching a bias current of 90% within a second ,it will require a few more minutes of further heating to be stabilized at cruising biased current. In total it requires 40 seconds to start up.
If the driver compensates perfectly the 6L6, the current does not pump with the program at high power.
I am using two versions of this amplifier since more than five years. One of them is used to test loudspeakers under design. It is truly a heavy duty material.
KOKORIANTZ
Thank you for being interested to this outstanding amplifier .
The grid bias voltage is provided by a zener diode refrenced to the cathode along with a source follower.It is totally independent from the plate current. It is precise and robust to have the tube used near maximum power dissipation. Cathode bias is also possible.
At startup , the 6L6 heats up faster than the driver , within three seconds it passes from zero to 95/100ma plate current, when the driver's plate current pulls it suddenly, back to zero.For the next 25seconds the amp remains dead ,waiting the gate-source capacitor to be charged by the driver to 2.4v threshold .Suddenly the amplifier starts singing reaching a bias current of 90% within a second ,it will require a few more minutes of further heating to be stabilized at cruising biased current. In total it requires 40 seconds to start up.
If the driver compensates perfectly the 6L6, the current does not pump with the program at high power.
I am using two versions of this amplifier since more than five years. One of them is used to test loudspeakers under design. It is truly a heavy duty material.
KOKORIANTZ
Yes, its Schade with a couple of twists.
The driver sees an infinite plate load resistor due to the "bootstrap" mosfet follower. The only load that the driver sees is the "Schade" feedback resistor. I have tried this and it works well with some driver pentodes, and not so well with others. I did not try the EF86 since I only have one of them.
The zener string replaces the resistor stack often seen in the "free lunch" type of DC coupled amps allowing for fixed bias rather than cathode bias often used in those designs. It also allows for independent adjustment of the current through each tube without interaction. The lunch / monkey designs require several iterations to make them stable which need to be repeated when tubes are changed.
The driver sees an infinite plate load resistor due to the "bootstrap" mosfet follower. The only load that the driver sees is the "Schade" feedback resistor. I have tried this and it works well with some driver pentodes, and not so well with others. I did not try the EF86 since I only have one of them.
The zener string replaces the resistor stack often seen in the "free lunch" type of DC coupled amps allowing for fixed bias rather than cathode bias often used in those designs. It also allows for independent adjustment of the current through each tube without interaction. The lunch / monkey designs require several iterations to make them stable which need to be repeated when tubes are changed.
- Status
- Not open for further replies.