I recently designed a power supply circuit for an inflight controller board. The circuit is a stand along unit that uses a LiPo battery and generates a set of DC outputs, 3.3, 3.7V, and a 5.0-7.0V output (not sure yet which).
The board uses switching power supplies. I put a 3.3V supply to drive the higher current digital units, specifically an FPGA co-processor and the RF telemetry data unit.
The 3.7V powers some LDO regulators on the digital and analog boards (these are separate) in an attempt to keep the "noise level" down on these boards.
I used a lot of capacitors, both ceramic and tantalum, and chose ones with the lowest ESR I could afford. I also ordered a bag full of ferrite beads in a surface mount package. Unfortunately, things got behind schedule, and I did not end up adding the ferrites to the circuit.. I'll have to use these next time.
A question that always comes up in these circumstances is how to keep the noise from the analog circuits. I've never found a definitive guide on this subject, but have used various types of inductor/capacitor combinations in an attempt to low-pass filter any noise. I also prefer the higher frequency switching regulators based on my perception that the higher frequency noise should be easier to low-pass filter. This might be an EMI issue however.
I would love to hear some comments on what different techniques are used, how to isolate analog/digital grounds (or if that is even an issue), etc. One final comment.. when I worked for large companies, 14 layer boards were the norm. Plenty of power planes, signal layers, etc. Now that I am doing smaller scale home/academic projects, 2 and 4 layer are far more common, so fewer it becomes more difficult to just separate everything with a plane.
Ed
The board uses switching power supplies. I put a 3.3V supply to drive the higher current digital units, specifically an FPGA co-processor and the RF telemetry data unit.
The 3.7V powers some LDO regulators on the digital and analog boards (these are separate) in an attempt to keep the "noise level" down on these boards.
I used a lot of capacitors, both ceramic and tantalum, and chose ones with the lowest ESR I could afford. I also ordered a bag full of ferrite beads in a surface mount package. Unfortunately, things got behind schedule, and I did not end up adding the ferrites to the circuit.. I'll have to use these next time.
A question that always comes up in these circumstances is how to keep the noise from the analog circuits. I've never found a definitive guide on this subject, but have used various types of inductor/capacitor combinations in an attempt to low-pass filter any noise. I also prefer the higher frequency switching regulators based on my perception that the higher frequency noise should be easier to low-pass filter. This might be an EMI issue however.
I would love to hear some comments on what different techniques are used, how to isolate analog/digital grounds (or if that is even an issue), etc. One final comment.. when I worked for large companies, 14 layer boards were the norm. Plenty of power planes, signal layers, etc. Now that I am doing smaller scale home/academic projects, 2 and 4 layer are far more common, so fewer it becomes more difficult to just separate everything with a plane.
Ed
