I have some questions, hope people here will help me.
+) I want to build discrete active crossover. Can I use some kind of buffer like diamond buffer in place of the opamp, with the same performance achieved?
+) How to calculate load resistor of differential pair (no current mirror) in BJT amp?
+) I want to build discrete active crossover. Can I use some kind of buffer like diamond buffer in place of the opamp, with the same performance achieved?
+) How to calculate load resistor of differential pair (no current mirror) in BJT amp?
To the second question: It's not easy to calculate on paper and get it right without actual measurements or at least simulation to get in the ballpark, but it goes like this:
- Decide on your tail current, let's say 3mA. You want half (1.5mA) through each leg of the diff pair. Let's ignore for now the error that the VAS base current introduces.
- Your diff pair load resistor (Rload) is going to see a voltage drop equal to the Vbe of the VAS plus the drop on the VAS emitter resistor, if there is one. Let's say that you have a 22R resistor there and a VAS current of 10mA: that's 0.22V on the emitter resistor, and let's say that the VAS Vbe is 0.65V, so the total drop is 0.22+0.65=0.97V.
(Note that if you have a beta-enhanced VAS, you should also add the Vbe drop of the extra transistor).
- For the current through Rload to be 1.5mA, its value should be 0.97V/1.5mA=646.7R. Since we actually want the current through Rload to be slightly lower, i.e. 1.5mA minus the VAS base current, use the next standard value up, 680R, or use a trimmer pre-set close to that value, adjust for best balance, measure and replace with the closest standard value, or two of them in parallel if needed for accuracy.
If you know the VAS beta, you can do more accurate calculations considering its base current. For example if it's beta=100, in the example above the VAS base current will be 10mA/100=0.1mA, so the current you need through Rload is 1.5-0.1=1.4mA, R=0.97V/1.4mA=692.9R, so the closest standard value is still 680R. As you can see the error isn't huge but it could be significant if the VAS has a low beta.
Finally, the alternative is to do a rough calculation as above, use a fixed standard resistor as Rload and then adjust the tail current instead until you get balance. In the example, the actual current through Rload would be 0.97V/680R=1.43mA, so the total current through that leg will be 1.43+0.1=1.53mA, so you want the tail current to be 2x1.53=3.06mA (ok, maybe not a great example because by coincidence the standard value gets you very close to the required one already, but you get the idea).
Cheers,
Cabirio
- Decide on your tail current, let's say 3mA. You want half (1.5mA) through each leg of the diff pair. Let's ignore for now the error that the VAS base current introduces.
- Your diff pair load resistor (Rload) is going to see a voltage drop equal to the Vbe of the VAS plus the drop on the VAS emitter resistor, if there is one. Let's say that you have a 22R resistor there and a VAS current of 10mA: that's 0.22V on the emitter resistor, and let's say that the VAS Vbe is 0.65V, so the total drop is 0.22+0.65=0.97V.
(Note that if you have a beta-enhanced VAS, you should also add the Vbe drop of the extra transistor).
- For the current through Rload to be 1.5mA, its value should be 0.97V/1.5mA=646.7R. Since we actually want the current through Rload to be slightly lower, i.e. 1.5mA minus the VAS base current, use the next standard value up, 680R, or use a trimmer pre-set close to that value, adjust for best balance, measure and replace with the closest standard value, or two of them in parallel if needed for accuracy.
If you know the VAS beta, you can do more accurate calculations considering its base current. For example if it's beta=100, in the example above the VAS base current will be 10mA/100=0.1mA, so the current you need through Rload is 1.5-0.1=1.4mA, R=0.97V/1.4mA=692.9R, so the closest standard value is still 680R. As you can see the error isn't huge but it could be significant if the VAS has a low beta.
Finally, the alternative is to do a rough calculation as above, use a fixed standard resistor as Rload and then adjust the tail current instead until you get balance. In the example, the actual current through Rload would be 0.97V/680R=1.43mA, so the total current through that leg will be 1.43+0.1=1.53mA, so you want the tail current to be 2x1.53=3.06mA (ok, maybe not a great example because by coincidence the standard value gets you very close to the required one already, but you get the idea).
Cheers,
Cabirio
The B1 and DCB1 make a very simple Buffer.
Just two jFETs for each Buffer. You don't need to use sk170.
There are many high and medium gm with a suitable Idss Nchannel that fit the duty.
Read Daniel Feucht for more info on this two device Buffer.
You can use this Buffer for the input and for the output and they take up very little space.
If you are carefull with SOA you can use SOT23 devices for even smaller and cheaper Buffers.
Just two jFETs for each Buffer. You don't need to use sk170.
There are many high and medium gm with a suitable Idss Nchannel that fit the duty.
Read Daniel Feucht for more info on this two device Buffer.
You can use this Buffer for the input and for the output and they take up very little space.
If you are carefull with SOA you can use SOT23 devices for even smaller and cheaper Buffers.
A feedback style Buffer will attempt to hold down distortion until it runs out of feedback capability. Then it suddenly distorts and takes time to come out of clipping and may oscillate while in clip.
Whereas a no global feedback Buffer that is effectively a single ended amplifier moves gradually into increasing distortion and develops gross 2nd harmonic as maximum output is approached. To avoid this distortion one should limit the maximum signal to <50% of the stage buffering capability. eg. for a 2.2mVac (6Vpp) maximum signal one would use ~12 to 13Vdc as the minimum supply rail voltage to avoid getting close to the distortion generating region. The PassB1 uses 18Vdc and the DCB1 uses 20Vdc (+-12Vdc).
Whereas a no global feedback Buffer that is effectively a single ended amplifier moves gradually into increasing distortion and develops gross 2nd harmonic as maximum output is approached. To avoid this distortion one should limit the maximum signal to <50% of the stage buffering capability. eg. for a 2.2mVac (6Vpp) maximum signal one would use ~12 to 13Vdc as the minimum supply rail voltage to avoid getting close to the distortion generating region. The PassB1 uses 18Vdc and the DCB1 uses 20Vdc (+-12Vdc).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.