I have two kiwis, two Rpis, and the requisite ethernet hub all up and running wsprdaemon 24/7. Once finances permit, I'd like to add a third Kiwi in a more traditional role as the standard 4 channel webSDR.
My present setup has maxed out two small Chinese 5v linear power supplies at US $70-80 each (though the constant draw is nowhere near the maximum). Adding more would require yet another one. Building a stack of little linear power supplies seems expensive and messy to me.
How do other people power their Kiwi farms?
I don't see a lot of 15+ amp constant 5v linear power options out there, and the ones I do see are expensive and would require a mess of wiring work to utilize. Am I being obtuse by using small 5v linear supplies?
Should I instead use switching supplies and try to quiet them down with ferrites? Or use step downs from 12v?
I would love to hear what other people are doing, or if you think I'm being moron (if you can say it nicely and not bruise my ego).
I found it almost impossible to eliminate lines introduced by the much more common 50-150Khz SWPSU, but this 1 Mhz unit's fundamental line at 950 Khz is almost invisible and occurs in the AM broadcast band where the AM signals bury it. The harmonics are so small that they are buried in the noise from the antenna.
This photo shows a 1 Mhz SWPSU with the in/out leads wrapped around a toroid, but in later installations that toroid has been shown to not be needed.
Much, much more important than the PSU is eliminating ground loop induced SWPSU noise, which is a very, very long topic I don't have time to discuss.
After some years experiencing a lot of different solutions I've decided to design my own board.
You can see here a picture of the board
You can supply from 10 V up to 15 V, then the TVS will clamp the extra voltage.
The main supply has been split in two ways: the main 12 V, protected against over-voltages and with a low pass filter, for the fan and a second path, devoted to the 5 V, is composed by a LM350 as first stage followed by 4 x LT3045 supporting 2 A at 5 V.
A measurement of rejection in the frequency range from 10 kHz to 20 MHz has been performed without reliable results: the crosstalk between injection of the signal and probes was higher than signal on the signal path under analysis.
Three PSU are on duty with two Kiwis and one Perseus since July. No noise source is now related to the power supply.
If you are interested in, I have some spare PCB.