DC current(or voltage) supply has to keep constant(DC) a current (or a voltage respectively) whatever the load ie evolution(AC) of the voltage (or the current respectively). This means that even DC, the supply has to react to any load fluctuation. So, this leads to an AC parameter, which is the bandwidth of this ability to compensate for fluctuations (?). have you any experience/example of power supply specifications showing the bandwidth of the regulations?
What happens, behind this bandwidths. What consequences on the biased (sensor, amplifier ...) system?
We have to design low noise current sources and we think about adding a passive current filter (with a parallel capacitor followed by a series inductor) to limit the noise bandwidth. Apart from the regulation stability, have you any recommendation/experience in such filtering. Assuming that a series inductor reacts to any currents fluctuation, we expect "passively" increase the regulation bandwidth, even if the active regulation is at the same time inhibited by this filter.
There are some questions:
How low noise do you want to get? i.e. what current do you want to generate and what is the rms amplitude of the noise current you can tolerate? 1 part in 1000 is easy, 1 part in 1000000 is very hard and 1 in 10^9 is a real challenge. Ensure that you're not trying to exceed the thermal limits e.g. E^2 = 4 k T B R, I^2 = 4 k T B / R etc.
thanks Tony for your comment. We are trying to develop current-DACs with a typical output current of hundreds of microamps and a current noise of about (and even lower) than 1pA/√Hz. This is use for the DC biasing (of low noise front end cryogenic electronics), thus the "useful" bandwidth is below 1 Hz -> leading to a RMS noise < 1 pA. So, the ratio that you are discussing is > 10^8. However, to reach a noise bandwidth lower than 1Hz, filters must be implemented, and that's the core of my present question. Indeed, the filter allows us to suppress noise, but what happens if the DAC load is changing fast? Is the filter modify the DAC correct operation as a current source?
Concerning noise sources, shot-noise √2qI_0 is one of the fundamental noise of any circuit using a diode as a reference. Fortunately, dividers and filters allow reducing their contribution at the last stage. So, the last current mirror, a large MOS transistor, is finally the main noise contributor : a Johnson/thermal fundamental noise of the MOS channel. And even sizing (W and L) of this last MOS stage allows to reduces the noise, few pA/√Hz are anyway always present at the end. Moreover, reducing as we do the white noise (Johnson + shot-noise) bring-out the 1/f noise (flicker) at few 10 pA/√Hz at 1 Hz leading to absolute needs of a filter down to 1 Hz to reduces the noise to few pA/√Hz at these low frequences.
If your load is changing rapidly, then if your source current is truly a current with infinite effective impedance then it shouldn't matter. But to get to the current levels you want, you will have to use passive filtering with an inductor. Now your problem will be shielding the inductor from stray magnetic fields including the earth's field to ensure the current is sustained accurately to better than 1 pA. But don't worry - 1 pA is still millions of electrons per second so your goal should be very attainable!
In most cases indeed passive post filtering (either RC for low currents or LC for high ) is the best option.
The voltage DC supplies are usually equipped with large output capacitor ( they are designed to be stable with such load). Then, that capacitor defines the transient behavior. Similarly, high current DC sources might be equipped with output inductor. However inductor only makes sense for rather high current sources, otherwise series resistor is used.
Important to note that some current sources are maintaining AVERAGE current - then they can be equipped with output capacitor too - and that makes usually very stable and low noise. Examples are current sources used for various filaments (heaters), Hall sensors, high precision electromagnets.
The noise measurement of high performance DC sources is b a bit tricky. One may look into e.g. :
https://www.edn.com/the-quest-for-quiet-measuring-2nv-%E2%88%9Ahz-noise-and-120db-supply-rejection-in-linear-regulators-part-1/
- Badges
- Science topic
- Similar topics
- Control Systems Engineering
- Filtering