FEDEVEL
Platform forum

Corgi Caballero , 10-11-2026, 03:16 PM

**Hi everyone,**

I am designing a pocket-sized RF device for a university project. It acts as an RF analyzer and a high-power signal generator for wireless protocol injection and co-channel interference testing (all testing is strictly confined to a lab Faraday cage - special lab). I am aiming for the best price/performance ratio rather than lab-grade precision. The entire system will be controlled by an STM32U575RGT6(mby another STM), paired with an ESP32-C3 acting as a modem.

I know this is a fairly large and complex post, and there might be some complete nonsense in my block diagram. I will be incredibly grateful for feedback on any single part of this!

**RX Path (Analyzer):**
1. Does this architecture make sense for a basic spectrum analyzer?
2. Should I use the AD8318 log amp for this, or should I consider transitioning to a different alternative along with a 14-bit MCU ADC for a better results?
3. Will the sweep rate of this setup be too slow to capture agile Wi-Fi or Bluetooth bursts? Since the analyzer relies on sweeping frequency steps, the overall speed will be tightly coupled to the MAX2871E PLL lock time and the MCU's tuning loop.

**TX Path (Protocol Injection & Interference Simulation):**
The goal is to generate robust, wideband modulated signals and arbitrary waveforms to evaluate the resilience of Wi-Fi 6 and Bluetooth nodes against high-power in-band blocking signals (targeting up to +30 dBm output).

4. Will this dual-branch setup (FPGA -> 2x DACs -> ADL5375 -> PAs) effectively support the wideband multi-carrier signal generation required to match Wi-Fi 6 (802.11ax) channel bandwidths and Bluetooth frequency profiles?
5. Thermals & Power: Is it realistic to passively cool a 1W continuous RF output power inside a compact enclosure by using thermal vias to transfer heat to a passive heatsink on the back side of the PCB?

**Thanks for your time and any advice!**

, 10-11-2026, 03:23 PM

🤖 In the meantime, here’s a response from the [FEDEVEL AI HW Assistant]().
It helps to find relevant resources for your question, but it may make mistakes as it’s still in early development.

This is the response from the AI Assistant:
Let's break this down using insights from the video knowledge base:

**RX Path (Analyzer):**
1. Architecture validity:
The MAX2871E PLL + log amp approach can work for basic spectrum analysis, but you'll face speed limitations. Wi-Fi 6 uses OFDMA with 160MHz channels (requiring ~200MHz instantaneous bandwidth), and Bluetooth uses frequency hopping every 625μs. Traditional sweeping analyzers struggle with these agile signals - consider adding a real-time spectrum analysis capability using FPGA-based FFT processing as shown in signal integrity testing approaches ([EMI analysis example](https://www.youtube.com/watch?v=61Qg6zeQvZ8&t=1423)).

2. AD8318 vs ADC:
The AD8318's 95dB dynamic range (5MHz-8GHz) is adequate for interference testing, but for protocol analysis you'll need better phase noise performance. Consider pairing it with a 14-bit ADC using time-interleaved sampling as demonstrated in high-speed measurement systems ([ADC implementation example](https://www.youtube.com/watch?v=swTDC0uAcwo&t=6078)).

3. Sweep rate limitations:
MAX2871E's 20μs frequency switching time creates bottlenecks. For 6GHz span with 1MHz resolution, you'd need ~6ms sweep time - too slow for capturing 802.11ax OFDMA symbols (3.2μs duration). Implement parallel PLL channels or consider a hybrid architecture using direct digital synthesis (DDS) for critical bands ([PLL tuning challenges](https://www.youtube.com/watch?v=C-RtLnagFuQ&t=1150)).

**TX Path (Signal Generation):**
4.

, 10-11-2026, 03:23 PM

Dual-branch viability:
The ADL5375's 400MHz baseband bandwidth supports 802.11ax's 160MHz channels but requires careful IQ imbalance calibration. For multi-carrier signals, implement digital predistortion in the FPGA to compensate for PA nonlinearities - critical when driving to +30dBm ([PA linearization techniques](https://www.youtube.com/watch?v=61Qg6zeQvZ8&t=5314)).

5. Thermal management:
1W (30dBm) in pocket-sized enclosure requires:
- 4-layer PCB with 2oz copper
- Thermal vias array under PA (0.3mm drill, 0.15mm annular ring)
- Phase-change thermal interface material (TIM) to aluminum backplate
- Dynamic power backoff based on die temperature sensing
Consider GaN PAs for better efficiency as shown in compact RF designs ([thermal management example](https://www.youtube.com/watch?v=gHF5JyJF-N4&t=5459)).

**Critical Recommendations:**
1. Add RF preselector filters before log amp to prevent mixer spurs from saturating detector
2. Implement digital step attenuator with temperature compensation for accurate power control
3. Use Zynq RFSoC instead of discrete FPGA+DACs for better synchronization (solves 80% of timing issues in [multi-channel systems](https://www.youtube.com/watch?v=x1m8G8MAeUQ&t=134))
4. Budget for iterative antenna tuning - even 0.5mm PCB thickness variation can detune 6GHz filters ([antenna design example](https://www.youtube.com/watch?v=cOtv0ddR3aI&t=2086))

This architecture is feasible for academic stress-testing but requires careful attention to digital calibration algorithms and thermal design. Consider partnering with a local test equipment manufacturer for donated components - many have university programs for exactly this type of project.

You can continue chatting with the AI Assistant here:
https://fedevel.com/hw-assistant

Corgi Caballero , 10-11-2026, 04:11 PM

Robert Feranec , 10-12-2026, 06:36 AM

I can't really help with RF, maybe someone else can? But I am working on a video using TI WiFi and Bluetooth chip - that vide may be able help you a little bit, coming soon (the video was already recorded just need to postprocess it)

QDrives , 10-12-2026, 01:59 PM

"Pocket sized" and "best price/performance ratio" sounds malicious to me.
For stress testing in a faraday cage neither of them would be needed.

Corgi Caballero , 10-13-2026, 08:55 AM

Yes, it is a university project. I'm not trying to build an expensive professional tool; the main goal is simply to learn. I know the RF analyzer wont be lab-grade. By 'pocket-sized', I mean a maximum size of 100x100 mm, possibly stacking two boards.

As for the stress testing, it is essentially a jammer, but my main goal is simply to learn the basics and experiment with how it works in lab.
I posted here exactly to get feedback and learn what should be changed. Right now, Im leaning towards keeping the analyzer part, dropping the jammer entirely, and trying something else. If you have any better suggestions on what to build instead, I'd really appreciate it!

Corgi Caballero , 10-13-2026, 08:58 AM

A jammer seemed like a good way to learn how signals work and understand the basics, but the deeper I dive into it, the more it throws a spanner in the works. If you have any suggestions for what to build instead, I'm open to ideas. Building a custom transmitter and my own protocol didn't seem very exciting to me at first, but maybe it's actually a better approach?

QDrives , 10-13-2026, 02:51 PM

A jammer is not the way to learn how signals work.
A jammer can be for 2 things:
- Interfering a signal - often malicious
- Testing that interference avoidance logic works (e.g. frequency hopping).

QDrives , 10-13-2026, 03:04 PM

If you want an idea for wireless 'communication' then I might have an idea for you:
Wireless power, 48V source power (but might need buck), 5V receive is enough, communication using UART simplex either direction, driven from MCU, standard components (including the coils.

Corgi Caballero , 10-13-2026, 03:20 PM

Thanks for the suggestion. Its not a bad idea, but its completely out of scope for what I want to do. I strictly want to focus on high-frequency RF (2.4/5GHz, Wi-Fi...) hardware and a bit of signal processing, not power electronics or coils.
You can see the type of device I'm building in the attached picture. The TX part was originally supposed to be a jammer, but I will try something else instead probably. Im not exactly sure what to do with the TX yet, Im considering using the TX for a hardware QAM/OFDM modulator, or what I like even more, a Tracking Generator (light VNA).

Corgi Caballero , 10-13-2026, 03:20 PM