It is common to have a current sink (or source) biasing the long tail pair. Has anyone studied the lowest noise method in the 1ma range? I've seen two NPN's, LED biasing and NPN, and of course, a degenerated JFET. Has oneone made the alternatives and measured (or simulated) the lowest noise solution?
Thanks for comments,
Steve
Thanks for comments,
Steve
My guess is that the lowest noise would be from a bipolar transistor with a small emitter resistor, biased by the base-emitter junction of another bipolar transistor at high current.
I would also guess that the effects of common-mode current noise in a differential stage would be reduced significantly by a current mirror load.
I would also guess that the effects of common-mode current noise in a differential stage would be reduced significantly by a current mirror load.
Hi,
normally only the high gain of a BJT input LTP allows emitter de-generation.
Adding emitter de-generation adds noise.
Increasing the feedback resistor values increases the noise.
Allowing supply rail noise through to the output (poor PSSR) increases the noise.
Usually decreasing Rs decreases noise and increasing LTP tail current allows lower value of Rs to better suit low noise.
If I have got any of this wrong please correct my errors.
normally only the high gain of a BJT input LTP allows emitter de-generation.
Adding emitter de-generation adds noise.
Increasing the feedback resistor values increases the noise.
Allowing supply rail noise through to the output (poor PSSR) increases the noise.
Usually decreasing Rs decreases noise and increasing LTP tail current allows lower value of Rs to better suit low noise.
If I have got any of this wrong please correct my errors.
http://www.national.com/an/AN/AN-222.pdf is a good tutorial on bjt noise limits
for a low noise current souce/sink the "best" approach is the highest V, quietest Vsource you can afford, if just hanging a big R from that Vsource doesn't give you high enough impedance then you have to add the transistor/current sense resistor which adds a little of the transistor's Vbe voltage noise - but less with higher base voltage
higher degeneration (from higher value current sense emitter resistor) reduces transistor noise contributions to the output noise current
In a diff pair tail current source the added noise from even a noisey ccs can be small if the output is well balanced by a collector current mirror (with large enough degeneration R) or a differential second stage or VAS - since the noise current equally divides in a balanced diff pair, it differences out in the differential output
for a low noise current souce/sink the "best" approach is the highest V, quietest Vsource you can afford, if just hanging a big R from that Vsource doesn't give you high enough impedance then you have to add the transistor/current sense resistor which adds a little of the transistor's Vbe voltage noise - but less with higher base voltage
higher degeneration (from higher value current sense emitter resistor) reduces transistor noise contributions to the output noise current
In a diff pair tail current source the added noise from even a noisey ccs can be small if the output is well balanced by a collector current mirror (with large enough degeneration R) or a differential second stage or VAS - since the noise current equally divides in a balanced diff pair, it differences out in the differential output
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.