| Electronics > Projects, Designs, and Technical Stuff |
| whats in a keysight ultra low noise DC supply filter? |
| (1/3) > >> |
| coppercone2:
N1294A-021 http://www.farnell.com/datasheets/1936689.pdf?_ga=2.79753044.2109743412.1546101534-1593084816.1538441214 https://www.keysight.com/en/pd-2265516-pn-N1294A/b2900a-series-accessories?cc=PL&lc=eng Its meant for a 6.5 digit power supply. What do you think is in this low noise filter? |
| rx8pilot:
Some very impressive specifications! The cost likely matches the performance. Hopefully, Keysight will provide one for a Signal Path teardown/experiment presentation. Examining the design would be an education for most viewers. Typet purly on my fone. |
| David Hess:
We can tell a few things from the specifications. Noise within the power output stage is reduced by the excess gain of the error amplifier to the level of the error amplifier's input noise which is why the output noise rises in direct proportion to the range switched feedback divider. The noise density graph indicates unremarkable low frequency performance but I think this reflects designing toward the practical resolution. The external low noise module is a damped LC filter with a 50 ohm impedance which can be used in either 2 wire mode or 4 wire mode to reduce the effective output resistance. This explains why the ultra low noise external filter is limited to lower currents and much slower settling time performance; the low cutoff frequency requires higher inductance with the associated higher series resistance and lower saturation current. The not-ultra low noise filter has higher bandwidth, lower series resistance, and supports a higher current. I wonder how the current control loop is implemented. |
| EEVblog:
I love how it's got 100 nV resolution, but 10 μVrms noise ;D That's 100 times more noise than resolution. |
| David Hess:
--- Quote from: EEVblog on December 30, 2018, 04:28:26 am ---I love how it's got 100 nV resolution, but 10 μVrms noise ;D That's 100 times more noise than resolution. --- End quote --- It is not quite that bad because it is 10uVrms of noise over the entire bandwidth and this is reflected in the 5uVpp from 0.1 to 10Hz specification which would be about 1uVrms. At low frequencies, the noise is consistent with the resolution. 100nV resolution has no meaning above the measurement bandwidth which will be very low. The noise specification is not particularly noteworthy. It is low for a regulated power supply and perhaps for a source meter but attainable with a low noise operational amplifier, low noise reference, and practically any pass element with some care. |
| Navigation |
| Message Index |
| Next page |