Looking for cheap-ish generator (oscillator) module, 2.45 GHz, 10 W

[hanakp]

I'm participating in development of a portable instrument that needs to generate 2.45 GHz signal with 6 to 10 W output power. Basically, it will be a miniature microwave oven, so frequency stability, spectral purity etc. are not critical. The project leader wants me to design an oscillator based on Ampleon BLF2425M9L30 power LDMOS, but I'd rather to avoid it. I did some microstrip circuit design with Ansoft Designer (now Ansys Electronics Desktop) in the past, but it feels like over-engineering to me. Well, except when I tried searching for some off-the-shelf modules, I found nothing. Likewise, I found no published design that I could easily reuse. Do you happen to know some? The generator should preferably fit into 10x5x5 cm dimensions and cost under 500 USD. DC power input, 50 ohm RF output. It would be nice if the output RF power could be varied (our plan is to control LDMOS drain voltage), but we could just use attenuators if needed. Thanks!

[LaserSteve]

GvA-84 mmic as a driver, and a 2.45 ghz VCO module as oscillator. Dither the VCO input voltage to find cavity resonance as needed on the load..

Skyworks and Minicircuits for driver MMICs and oscillators as  needed.

AppCad to help with sizing your microstrip:

http://www.hp.woodshot.com/

Use the recommended board layout in the PA data sheet. Thus the Power FET is used as intended, as an amplifier.

AS LONG AS YOU KEEP your striplines under a quarter wavelength between the oscillator and MMIC and then MMIC to Driver you should have few problems.  Typically a 1 or 2  dB PI attenuator between the oscillator and driver amplifier would be used for preventing load pulling.

A DBM module makes for easy control.

Steve

[hanakp]

But... GvA-84+ has 20 dBm output power (0.1 W). I need up to 40 dBm.

Oh, you mean then reuse the example amplifier design for BLF2425M9L30? They do provide an example board data on their site, but they call it "test circuit" in the datasheet. Therefore, I'm not so sure that the input and output lines are matched to 50 ohm. Or at least I don't see any stepped-width microstrip sections that are normally used for matching.

https://www.ampleon.com/documents/data-sheet/BLF2425M9L30_M9LS30.pdf

[Uunoctium]

Is it an single project? Or will you produce larger quantities?
In first case i would go with ready mades from bison-electronic, PL.

https://www.ebay.de/itm/203941633749?
https://www.ebay.de/itm/203868305296?
https://www.ebay.de/itm/203949784270?
Sufficient enough for burnt fingers... 



p.S.: don't forget to implement components like these on the pink foam. Otherwise you got magic smoke
p.p.S: can imagine the real purpose

[hanakp]

It's for a specialized instrument for biomaterial preparation. Even if it actually makes it to the market, it will sell dozens of pieces per year at maximum. And right now, they need only 3 or 4 oscillators for testing purposes, that's why I'm reluctant to design it from scratch. But at the same time, they want something repeatable, with components from "respectable" manufacturers, that are likely to stay on the market for 10 years or more. That's how they came to the BLF2425M9L30, it has assured availability up to 2030. Plus, they're cheap enough so they can just stockpile them when the EOL date approaches.

Sourcing used modules from Ebay and the like is out of the question.

[Uunoctium]

Likewise, I found no published design that I could easily reuse. Do you happen to know some?

you may have a look on:
https://www.dj0abr.de/index_deutsch.html
On the left dropdown menue goto "es'hail-2"
-> 10W PA
-> 30W PA = finished layout with BLF2425M9L30

[hanakp]

Oh my, that's exactly what I need! And he recalculated it for FR4 PCB, too! Well, he doesn't specify what permittivity exactly the FR4 should have, but he offers finished PCBs for sale. He also doesn't mention exact part numbers of ceramic capacitors in the BOM (I don't think he used random general-purpose ones). I'll try to contact him about that.

Edit: hmm, can I use a hardware design released under GPL3 licence in a commercial product (even when it's a big "if" whether it will actually become a product)? How would that even work? We'd have to attach an USB Flash drive with the amplifier design files to every unit we sell?

[LaserSteve]

Generally if you contact a ham radio operator and tell him its for science you will come away with a license of sorts. Usually a good ahead email. We sell boards to have more friends on the air in a given band.  For contests at 900 and up it is not unusual to build two rigs to have a contest partner. Hes being friendly.

In the US for 2.4 Ghz we often use the " Gain is Cheap " motto.  So a GVA-84 with an 18 pf RF grade smd capacitor on input and output with short traces works just fine as a gain stage on FR4. In fact you can find 10 Ghz gear designs on FR4. It might take a few iterations but FR4  based board runs are cheap.  50 Dollars can buy you a lot of prototypes.

Many of the hobbyist board suppliers even have  characterization data for their materials on their website.  IOT applications have driven demand for that.

Your larger problem is determining if you need an isolator / circulator / matching  between the cavity and final amplifier.  1:1 VSWR is a just a dream, and the power transistor designers will anticipate having reflected power.  That said, laying out a directional coupler stripline parallel to your output trace is trivial. That and a detector diode can provide a shutdown signal.

I build gear for graduate students, and I guarantee you they will find a way to make non-standard samples all the time. One of my favorites is finding magnetic stir bars left in samples.

If your supervisor was remotely worried about board material characteristics / matching, you
would not have been told to build a self oscillating power source.

In fact he / she might be testing your design skills.

I'm not saying controlled impedance is not important, but you may find out that you are overthinking this.

Do you have access to a spec an, professional VNA,  And or dual directional coupler?

Steve

[LaserSteve]

NOTE: Prior to obtaining good gear, I used to tune my matching with a HP dual directional coupler, two diode detectors, a source , and open/short/load  for calibration at 900 and 2400.

It can be done.

Steve

[NiHaoMike]

What about use a module designed for amateur radio?

https://www.sg-lab.com/AMP2400v3/amp2400_v3.html

[hanakp]

@LaserSteve: the problems with reflected power is an excellent point, I can see it could upset the oscillator's feedback and bring the LDMOS into dangerous operating conditions. The project lead wasn't testing me, BTW - he is retirement-age guy who spent last 20 years around magnetrons and other high-power stuff. He's simply lost touch with low-power solid-state RF circuits, that's why he brought me in (even if I'm no expert myself).

@NiHaoMike: many thanks, we're going to try the bulgarian power amplifier, because it comes with case and everything. My plan is to make a small PCB with VCO, attenutator and pre-amplifier and mount it inside the PA's case. Then I will remove the input SMA and directly connect the PCBs with a short coax.

But the case is thin (19 mm) and there is not much headroom between the PA PCB and lid, so my PCB shouldn't use any tall components. I looked at metal can VCOs like ROS-2490 from Minicircuits , but all have over 4 mm. Is there any fundamental difference between these metal can and integrated VCOs like Maxim MAX2750AUA? Do they have better aging, temperature stability or something? Lower output power is no problem, since my PCB needs a pre-amplifier anyway.

[radiolistener]

you can use magnetron M-62 or M-95, it has at least 20 W power with anode current 39 mA and anode Voltage 1.35 kV.
The frequency is 2401–2499 MHz.

For 10 W output, just put anode current about 22-23 mA
For 1 W output, just put anode current about 5 mA

[hanakp]

I finally got my hands on the SG Laboratory Ltd amplifier for Es'Hail 2 satellite uplink and did some basic measurements. First I measured its frequency response with Siglent SSA 3032X generator/analyzer combo at 0 dBm input power (limit of the internal generator). I'm attaching the result below, it has 30 dB gain (1 W output power) in the target band. But it has much wider bandwidth than advertised, it employs NXP MMG30271B as pre-amplifier and Ampleon BLP9G0722-20G LDMOS as the power stage. However, I was mostly interested in how it behaves at 2.45 GHz and higher power levels. I used Agilent N9310A and R&S FSL18 for that, although I have less confidence in these measurements, because the analyzer was partially damaged a few years back (somebody put too much power to the input). At full 28 V supply voltage, the gain is:

0 dBm input => 1.3 W output, 31 dB gain
+11.1 dBm input => 10.2 W output, 29 dB gain
+13.3 dBm input => 13.5 W output, 28 dB gain
+17.0 dBm input => 20 W output, 26 dB gain

I actually missed 1 dB compression point, because I didn't expect it would lie below 10 W. That's mostly irrelevant for my application, however. I was more interested in how bad harmonics the amplifier produces at the extremes of the operational parameters - as I mentioned earlier, my plan is to control the output power via supply voltage. But my worries were unfounded, I couldn't find any 2nd or 3rd order hamonics during my experiments - the amplifier has an output microstrip filter which apparently does a great job. I'm attaching FSL18 screenshots which I took at 28 V supply/20 W power and 8.2 V/2.5 W power.

Overally, the amplifier is nicely made, the aluminium cooling base plate is 7 mm thick. It uses semiconductors from "respectable" producers, 78M05 regulator is from ONSemi and vox and PTT circuits employ Microchip MCP6001 opamps. Trimmer for vox hold time is from Bourns. It also has two identical directional couplers with detector diodes and RC filters at the output. They're denoted FWD and REF (forward and reflected?), I presume they're used to tune the amplifier during production. In the future, I may use them to detect and automatically shut down the amplifier when someone runs it without proper load. There is 7 mm headroom between components and lid, so I'll have no problem fitting my 2.45 GHz generator board inside. I plan to mount in on the lid above the vox circuits (upper left corner), to minimize the chance it would upset the PA circuits.

[hanakp]

It's been a while, but I've successfully designed and tested a small VCO PCB to drive the bulgarian power amplifier. It worked, but would be too laborious for production. So the project head contracted the bulgarian guy to produce similar VCO + PA on a single PCB for us. Which means I can share my design here.

As I previously planned, I mounted my PCB inside the amplifier case, it should be obvious from the photos. I'm also attaching Eagle 9.6.2 and Gerber files in case anyone wanted to build it. I put most design notes directly into the schematic.

I removed original input SMA connector and replaced it with U.FL inside the case. But it was a real chore to solder it in correctly, because the space was tight. If I did it again, I'd probably solder a thin coax directly to the PCB, althought I'm not sure how it would affect impedance matching. And of course I mounted the PA on a heatsink and added a fan.