 | Enter Your Project for a chance to win a Spectrum Analyzer for the Most Innovative RF Project! | Project14 Home |
| Monthly Themes | ||
| Monthly Theme Poll |
Strange Bedfellows - RF and digital switching
I was wondering how the Spread Spectrum option in some Computer Motherboard BIOS actually works, at the time I couldn't find an explanation.
It sounded a lot like a FM modulator, only connected to some form of sweep or white noise generator instead of an audio input.
After much searching, I found this is exactly how it works.
This opens further questions, just how much RF technology has made it's way into computer and Switch Mode Power Supply (SMPS) design ?
Then you have things like modern digital radios, and Software Defined Radio (SDR), where technology has crossed in the other direction.
In the RF field, spread spectrum broadcasts use any of the following technologies:
Frequency-hopping spread spectrum (FHSS), direct-sequence spread spectrum (DSSS), time-hopping spread spectrum (THSS), chirp spread spectrum (CSS)
The Projects
So my projects are to design, and hopefully build, RF friendly Switchmode Power Supplies.
These could be anything from a 5W LED driver, to a 5000W offgrid solar inverter.
To the best of my knowledge none of this exists.
First we need to understand the problem why is a simple squarewave such a huge problem for radio, TV and RF in general.
If you look at a 1kHz squarewave with a Spectrum Analyzer you get this.
Every second harmonic is skipped, so 1kHz gets harmonics of 3kHz, 5kHz, 7kHz etc. This skipping is the result of the 50% duty cycle, the even (divides by 2) frequencies are cancelled out.
However, SMPS work by varing the duty cycle to get a stable output voltage, so what frequencies are cancelled alter varies.
In addition, a second main frequency is generated, as ontime is time, and frequency = 1 / time.
An FM signal looks like this on a Spectrum Analyzer, note the peak is -12dB lower.
Nothing is perfect, this will create more wideband noise, and may raise the noise floor on some RF recievers.
I happened to find this document on TI's website, the PFC chip was mentioned in one of their recent videos.
Frequency Dithering with the UCC28180 and TLV3201
https://www.ti.com/lit/an/slua704/slua704.pdf March 2014
The UCC28180 is a dedicated Power Factor Correction controller, a type of boost converter for charging a capacity with DC from AC as a continuous load.
So we have an external circuit that can vary the frequency of a PFC controller, just need to apply it to a generic PWM controller.
Simple right ?
Looking at the standard PWM controllers (including the ones I have lying around), they all use a RC oscilator to set the frequency and the UCC28180 just makes do with a single resistor.
It's not the same, so we need to find PWM controllers with the same single resistor to set frequency input.
So far I've found LM5020, LM5021, LM5025 using the search term "Single Resistor Programmable Oscillator" together with "PWM". There will be others, but not found them so far.
In detail the specs are:
LM5020 <630kHz
LM5021 <1MHz
LM5022 <2.2MHz 6-60V input, 1A mosfet gate drive
LM5025 <1MHz 13-90V input, 1A mosfet gate drive
LM5022 describes itself as having "Single Resistor Oscillator Frequency Set", *sigh* standards TI.
It's suitable for Boost Converters.
The frequency pin on the LM5022 is around 2V, so need to increase the voltage to the Frequency Dithering circuit to 4V.
May also need to alter R5, to alter the magnitude of the output. Not looked yet at changing the amount of Frequency Dithering, as 7kHz may not be enough.
Another thought, can this approach also work with a traditional RC network where a resistor charges a capacitor ?
I theory it should be possible, some experiments and component changes will be needed.
Another idea
Skip pulses, rather than varying the duty cycle.
eg 50% duty cycle, ignore every second pulse by not generating it.
Problems I see with this idea:
1. More lumpy output voltage on SMPS.
2. Requires higher frequency and voltage to make up for the reduced transfer in some situations.
3. Possibly less efficient.
4. The intermittent RF emmissions might be more problematic to filter/ignore in RF equipment.
However if the SMPS has multiple parallel converters, for increased power output, there is another option.
Disable individual sections (phases) to control the output voltage.
More complex RF friendly controllers
A Microcontroller using Fourier Transform and a table of known frequencies to avoid, to create RF safe PWM signals.
Use SDR (Software Defined Radio) methods to create a more complex frequency control.
A further idea, allow external input to one of the Microcontroller or SDR controller, from a RF device itself.
So the frequency(s) being transmitted/received can be avoided by the SMPS controller.
Top Comments
Linear Technology (Analog Devices) make Spread Spectrum Oscillators:
eg LTC6908
They have a good article on the web:
https://www.analog.com/en/technical-articles/spread-spectrum-frequency-modulation-reduces…
I didn't have time to look up all the possible chips from different companies, as briefly mentioned they sometimes use different terms.
I choose the LM5022 early in the search, as it's capable of taking…