On 11/10/22 5:53 AM, Manfred Mornhinweg wrote:
Jim, yes, ready-made DC-DC converters are inexpensive these days, and are a good option in many applications. But for RF switching, I don't see the one you linked as very usable. 15V is definitely not enough reverse bias in any solid-state RF switch that has to handle the typical power used by hams, exccept with some types of true PIN diodes. Also the DC isolation provided by these converters is rarely required in RF switches. RF isolation is indeed required, but these ready-made converters don't provide it, having 20pF of capacitance between input and output. So the RF isolation has to be provided by means of RF chokes and/or high value resistors, in any case.
You can stack them for higher voltages. The DC isolation lets you do that.
Yeah, filtering for a specific application would be application specific.
As for RFI from voltage converters, I love to run mine at a very low frequency, such as 1kHz, taking advantage of the fact that only very low power is needed for reverse biasing RF switches. So even at that low frequency the transformer can be tiny. Where the interesting radio spectrum begins, we are already at several hundred times the switching frequency, and the harmonics up there are weak and very easily suppressed.
Yes, that works.
Elecraft went another path, and used an RF oscillator to generate the high voltage needed to bias 1N4007 and 1N5408 diodes pressed into PIN diode service. They placed the oscillator between two ham bands, locking it to a crystal to make sure it won't wander away. They also made sure that its harmonics don't fall on ham bands. A brave design, I would say! But later design changes seem to suggest that they still got RFI from these converters, maybe in the form of IMD with strong broadcast signals.
Locking the converters to an internal oscillator is a common strategy. It's used in most spacecraft for instance (it's even in JPL's "Design Principles") - I have a system that is locked to the divide by 64 from a 50 MHz reference oscillator, and we just accept that there's a noise spur every 781 kHz. Other systems with lower sample rates have locked the switching rate to the sample rate. The noise appears at DC, then.
