4.4 GHz Frequency Synthesis Made Easy

How hard is it to create a synthesizer to generate frequencies between 35 MHz to 4.4 GHz? [OpenTechLab] noticed a rash of boards based on the ADF4351 that could do just that priced at under $30. He decided to get one and try it out and you can find his video results below.

At that price point, he didn’t expect much from it, but he did want to experiment with it to see if he could use it as an inexpensive piece of test gear. The video is quite comprehensive (and weighs in at nearly an hour and a half). It covers not just the device from a software and output perspective but also talks about the theory behind these devices. Β [OpenTechLab] even sniffed the USB connection to find the protocol used to talk to the device. He wasn’t overly impressed with the performance of the board but was happy enough with the results at the price and he plans to make some projects with it.

We’ve seen a lot of lower-frequency synthesizers used for everything from transmitters to antenna analyzers. While they are handy, it is important to realize that handling RF usually requires more than just wiring up a chip and connecting it to an antenna.


18 thoughts on “4.4 GHz Frequency Synthesis Made Easy

  1. I agree with the video author that it doesn’t make a lot of sense for the official eval boards to be so expensive. It does seem to be a trend that would seem to go against the logic of getting more people to evaluate and adopt the products for future designs. Also, I would comment that the better reference clock on the official eval board not only provides better frequency stability as he pointed out, but possibly more importantly is that it provides lower jitter and thus lower phase noise. Also, it is usually a good idea to start with a higher reference frequency and divide it down to get better phase noise performance when possible.

    1. Totally agree, a decent tcxo or similar wouldn’t of been too expensive but of course this goes against the cheap as possible way the Chinese do these things and extra $10-20 would have yoa board with the best of both worlds.
      Failing that for the price difference you could invest in a lab frequency reference and sort the entire lab out ;-)

    2. Oh, how I loathe overpriced dev boards. Unfortunately, the main buyer of these “authorised” dev boards are still big companies, looking to put this into their next product, so AD seizes the opportunity to make some more $$. I have seen other manufacturers do the same. Sometimes they even charge ridiculous prices for half baked dev boards and incomplete software, just because they are “authorised”.

      1. The big companies often get dev boards for free from the sales rep, if the declare the project. The supplier is interested, to get his product into a 100000k pieces or more series product.

      2. I work at ADI and have made a considerable effort to reduce the dev board prices. To us, they’re not profit makers and are actually sold close to cost.

        So why are they so expensive? It really comes down to three things
        1 – Materials. We make these dev boards to withstand comprehensive customer testing across the full temperature range of the product and often over specify the boards for quality. That means ENIG finishes, extra copper weight and high temperature substrates. This all adds cost. Likewise all the supporting components are high spec – e.g. caps are usually X7R rated across the full temp range and we use precision resistors.
        2 – Quantity. We typically only make batches of a few hundred to reduce the burden of keeping stock. Distributors usually only hold one or two boards (and they want to make a mark up when they sell them). Low quantity means higher prices
        3 – Quality. Every board is machine and hand tested to ensure it works exactly as it should. That’s perhaps 15 mins of an engineers time on each PCB to run a full set of tests.

        When you compare to some of the eBay and Aliexpress routes, there are cheaper materials, often no test and high volumes (let alone the risk of counterfeit or low quality components). Whilst these can be great for hobby use, they aren’t ideal to evaluate the performance of our components which is what our eval boards are intended for!

        1. Thankyou for the insight. Would ADI ever consider potentially lower cost dev boards for the hobbyist? I mean ADI are pretty damn good when it comes to free samples, but lower cost dev boards?

          1. I’ve been suggesting to some teams that they should try and work with companies like Adafruit and Sparkfun for that aspect as the business model they have better supports that.

            ADI is developing a considerable capability in systems engineering at the moment and as part of that, expect to see more support with components such as code examples / drivers with open source licensing.

            Most teams are happy to supply Gerbers and design files for the evaluation board PCBs – just ask on the EngineerZone forums. DirtyPCBs and OSHPark will happily run off boards to then populate with samples which may be a lot cheaper.

            The gerbers will be useful but we use Cadence Allegro for PCB design so chances are that the other files are of less interest.

    3. Manufacturer’s reference boards for microwave stuff are always super expensive, much more than the bare IC and BOM.

      Remember it’s not just something you can quickly chuck some autorouted tracks down and call it good and build it with jellybean Y5V capacitors and sell it.

      The choice of discrete components matters – e.g. bias inductors and coupling capacitors which have appropriate performance at microwave freequencies, high enough SRF and that sort of thing.

      Impedance-controlled layout at 50 ohms.

      Validating the whole reference board design to ensure that the board delivers the claimed performance – and to ensure that it’s representative of the specifications of the bare IC.

      And to do that you’re going to have to have a 6GHz spectrum analyser and VNA. Not cheap.

      Manufacturing of the reference board, in some cases, on a special substrate like Rogers, PTFE, Duroid, etc.

      This is one of those things that obviously sucks for the hobbyist, but in a commercial setting, doing RF engineering, just ordering the known $400 reference board that just works is a no-brainer compared to the labor cost of re-implementing it from scratch.

    4. There are two “strategies” for pricing the demo/eval boards.

      On the one hand, you can say: we can afford to make a loss on the demo boards as long as we sell (many) more chips.

      This is what you see with the ‘430 kits selling for $4.30 and the nucleos for about $10. But like in this case, there might also be a bunch of hobbyists wanting just one demo-board, and then never buying any (more) chips. Or companies with small series that will simply integrate the demo board.

      The other side says: We can sell the chips without a demo board, we’ve done that for decades, so people who NEED a demo board to get started should pay for their own demo board. So the demo boards are priced to pay for the development and their own (small volume) production.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s