From Testing Components to Testing Systems
News
Most wireless breakthroughs look elegant in simulation. Far fewer survive contact with hardware. In this interview, XRComm CTO and Co-founder Kodanda R. Engala explains why simulation alone can no longer validate today’s RF designs, why calibration is becoming a strategic capability rather than a maintenance task, and how the SAT 100T platform is designed to carry engineers from component characterization through system-level testing to in-field deployment — all on the same hardware.
Many wireless breakthroughs look elegant in simulation but struggle in hardware. What are the most common hardware limitations that slow down innovation in real-world radio systems?
Once simulation is done and you want to bring a design to life, the biggest hurdle today is the lack of a good prototyping and testing platform. Most hardware needs two or three revisions before it hits the field — and modern product cycles simply don’t allow that anymore. This is exactly where the XRComm SAT 100T fits in: a single platform that, with just a software change, becomes either a prototyping environment or a test platform that brings real field scenarios into the lab.
Why is simulation alone not enough for developing RF hardware platforms for non-terrestrial and terrestrial networks — especially with 5G Advanced, 6G, and the integration of NTN and TN?
Simulation lacks the real-world test signal. You can’t capture a signal in the field and then inject that same signal into your amplifier in the lab. That’s where the SAT 100T works really well: you capture the signal in the field as a test platform, bring it back, then — with the same hardware — switch to prototyping mode and transmit or receive that same signal to evaluate amplifier performance, power consumption, and all the typical hardware features.
Why is it more important than ever today to capture real data and bring it back into the simulation? What’s the value for development?
Hardware traditionally goes through two or three revisions before deployment. Today’s product cycles don’t allow that. We need the hardware to do exactly what the simulation predicts — first time. That’s why this is such an important moment for a good prototyping platform and test setup: so your design is ready in the first shot and deployable.
Most hardware needs two or three revisions before it hits the field — and modern product cycles simply don't allow that anymore.
Calibration is often treated as a maintenance task rather than a strategic capability — instruments get sent out for calibration when their tag expires, and that’s it. Why is calibration becoming central to next-generation test and measurement platforms?
The SAT 100T has three modes of calibration: factory, startup, and runtime (or ongoing). To understand why this matters, consider the two reasons calibration exists. First, you calibrate so the equipment performs within the limits set by 3GPP or FCC requirements. Second — and this is often overlooked — if you’re supposed to transmit X dBm and you transmit more, you’re wasting power. Overshooting drives up power consumption and operating expenses; undershooting compromises performance. Calibration ensures you do exactly what the scenario calls for. Among the three modes, the startup scan gives a strong baseline of performance across the whole SAT 100T.
In modern systems with massive MIMO, phased arrays, and many channels, calibration isn’t only about meeting a standard — it’s also about aligning channels to each other, in gain over frequency, time, and synchronization. How do XRComm’s three calibration methods handle that?
Good point. From 5G onwards, wherever there’s massive MIMO, phase calibration across the TX and RX chains — what’s called TX-to-TX calibration — is one of the key performance parameters. It’s essential for massive MIMO to function correctly. The same applies to NTN, where phase calibration across all channels enables proper beamforming. And for newer 6G applications like ISAC, this phase calibration directly determines the accuracy of location finding and geolocation.
Traditional instruments are calibrated at the factory; the customer turns them on, waits 30 minutes for warm-up, and then loads the calibration coefficients before measuring. With XRComm’s adaptive calibration, the startup and runtime calibration adjust before the instrument has even fully warmed up — and then continuously adapt as the temperature changes. Can you talk more about that?
What you’re describing is the runtime calibration. It uses the calibration circuitry in our hardware to compensate for temperature variations of the components in real time. As part of the runtime calibration, the hardware is characterized in the factory, and that characterization runs in conjunction with AI/ML. In the field, we can assess how the components are responding to environmental temperature, and the AI/ML algorithm — trained back at the factory — kicks in to compensate for changes in gain and phase. That model runs across all our hardware platforms.
Traditional instruments often deal with this problem using lookup tables at specific temperatures. How is XRComm’s runtime calibration different?
Traditional lookup tables are usually based on a single parameter — say, the temperature of one component or one section of the hardware. In the SAT 100T, we have multiple temperature sensors. Different channels’ performance might vary based on different sensors, and different parameters might vary on different temperatures. By bringing all of this into AI/ML, we can use 10, 20, even 30 sensors. The performance of each parameter — and each channel — can be adjusted independently, based on a complete snapshot of all those temperatures. Compared with traditional one- or two-variable compensation, this gives you a much wider dynamic range of compensation and significantly better accuracy.
Traditional T&M vendors focus on instruments; XRComm focuses on systems. What’s the difference between measuring signals and characterizing a complete radio system — especially in the context of 5G Advanced, 6G, and NTN/TN coexistence?
Traditional engineering labs work with each component individually — the FPGA module, the low-power section, the power amplifier, the transceiver. You characterize each, write down the requirements, put them together, and trust that the simulation says they’ll work. That approach works, but it requires multiple iterations. Now fast-forward to system-level integration and testing: all those components are integrated, and you can pull individual pieces in and out as devices under test. What you’re then exposing each component to is a system-level parameter — which means you can optimize each component for what the system actually needs, rather than for component-level metrics that don’t translate to system performance. That’s the advantage of system-level testing.
It seems we should think of system-level testing not only at the radio level but at the network level — including NTN satellites, user equipment, direct-to-cell devices, and customer terminals. Is there a need for testing at that level too?
ven more important. A complete NTN test cycle has the device under test, the satellite, and the channel model in between. It’s important to test this whole thing as a system — including the feeder loop back to the ground — so that link-budget analysis and resource allocation are optimized as a whole, rather than over-designing one part to compensate for another.
This also enables root cause analysis and failure-mode identification. Can you address that?
Exactly. Once you’re doing system-level analysis, it’s much easier to pinpoint the root cause of a failure or a lack of margin in a particular component. Once you identify it, you can fix the overall system performance sooner — through an iterative method if needed. That’s why root-cause and margin analysis at the complete-system level are so important: you understand each component’s performance and the corresponding margins under stress.
A satellite operator may optimize their satellite at the system level on the ground — but then it goes through launch: vibration, mechanical stress, temperature swings. Once it’s deployed, it’s accumulated a lot of bruises. How does system-level testing help adjust and correct for stresses introduced during launch?
Excellent question — let me address it from two sides. First, the SAT 100T as test equipment: once the satellite is deployed, the SAT 100T tests it in the field for the first time. You undo the Doppler, and then you can measure all the other impairments the satellite has accumulated during launch. That becomes a golden truth — a measurement of how your satellite is actually performing after the stress of launch, which may be very different from how it performed perfectly in the lab.
Second: how do you fix it? You take those same field measurements back into the lab, and now the SAT 100T becomes a prototyping platform — same hardware, different personality. You design version 2 of the satellite with extra margin for the parameters that drifted during transport — gain, phase, beam durability — and you can use telemetry to tune the satellite that’s already in orbit, bringing its parameters back in line with simulation. That’s how you close the loop: measure in the field, bring data back to the lab, build margin into the next design, and tune the existing satellite.
That seems unique. You mentioned the SAT 100T can measure true Doppler — is that something nobody else does today?
Most current equipment claims to do it — but here’s the catch. In the cellular bands (40 to 200 MHz) or the wider Ka- and Ku-bands (1 to 3 GHz), most equipment measures Doppler by taking snapshots — they don’t process the entire bandwidth simultaneously. While they’re processing one snapshot for Doppler, they miss the rest of the signal. By the time they take the next snapshot, that signal has passed. Any error during that gap is missed, and you’d assume everything was fine.
The XRComm SAT 100T is real-time: when we say we measure 1 GHz of bandwidth, we measure every sample of that 1 GHz. You can set anomaly thresholds for Doppler, sampling-frequency offset, and so on. The XRComm Doppler is continuous — every nanosecond, not every few milliseconds, not every few seconds. That’s the key differentiator.
And once you know which part of the signal is normal, you don’t have to store it. You only record the anomalies — with a small buffer before and after — so you can analyze them. We do offer full satellite-pass recording, but otherwise you’d just fill your hard drive with healthy data, which is useful to no one.
The XRComm Doppler is continuous — every nanosecond, not every few milliseconds, not every few seconds. That's the key differentiator."
What’s the advantage for developers and test engineers of measuring true Doppler over the full bandwidth, compared to traditional equipment where it’s often simulated or inferred?
Full bandwidth is critical because any change in the satellite’s parameters — typically driven by the LO — can be detected more reliably. A drift might not appear in the specific narrow samples you’d otherwise be measuring. Measuring the whole bandwidth gives you a much higher chance of root-causing the problem, which accelerates the design cycle.
I keep hearing about PIM and interference. What is passive intermodulation, why does it happen, and how does the SAT 100T help with PIM cancellation?
Passive intermodulation is the beat tone of two carrier signals. Under high-power scenarios, even passive elements like filters and connectors can generate intermodulation distortion. The intermod levels are very low, but still within the receiver’s sensitivity range. When the intermodulation frequencies fall in the RX band, the receiver sees both the desired signal and intermod products created by the transmit signal — and the desired signal is degraded.
PIM is only a problem in full-duplex scenarios — in half-duplex, the transmitter is off when you’re receiving — but it’s a serious problem for FD radios. It increases bit error rates, reduces cell size, drops calls, and creates other practical issues. The SAT 100T lets you simulate PIM and develop cancellation algorithms, by emulating the PIM through TX feedback into the RX. You can build and optimize cancellation algorithms in the lab and then validate them in the field.
Interference is adjacent to PIM, in the sense that another signal is reaching your receiver. This becomes a major issue in 6G, where the spectrum will be more crowded and ISAC is one of the headline features. In ISAC, the radio acts as both transmitter (like radar) and receiver (sensing reflections — from a drone, an aircraft, vegetation, whatever you’re trying to detect). You have to cancel the self-interference from your own transmission, otherwise you can’t tell what’s a reflection and what’s just feedback.
PIM and interference cancellation algorithms run on similar principles. Prototyping them on the SAT 100T at a system level — at universities, in industry, across different applications — gives you a strong foundation. From there you can characterize the signatures with AI/ML: this looks like a drone, that looks like an aircraft, that’s vegetation. PIM cancellation and interference rejection are major advantages for 6G ISAC.
Does PIM become more critical with massive MIMO and multiple beams?
It depends. PIM drops drastically with output power. For traditional and bigger radios at 160 or 320 watts, PIM was a major concern. In massive MIMO, each antenna runs at lower power, so PIM is less of an issue — though you still have to watch out for it. But interference becomes a much bigger problem in massive MIMO. The same receiver algorithms used for ISAC apply to interference cancellation, so the SAT 100T platform supports both.
When characterizing power amplifiers, LNAs, or RUs, what parameters are most critical for real-world deployment, and why are they often misunderstood?
If you look at the evolution of RF in 3GPP — 2G, 3G, 4G, 5G, and now 6G — you’re trying to put more data through the same channel. That means higher modulation schemes and more power to maintain reliable communication. Higher modulation — 64-QAM, 128, 256, 1024-QAM, and Wi-Fi already at 4096-QAM — demands much better sensitivity: better EVM, better ACLR. The nonlinearities in traditional RF chains, both transmit and receive, now have to be scrutinized at much tighter specifications, because these components are operating at far more demanding modulation schemes. The key parameters are EVM, ACLR (and spectral regrowth), and noise figure.
Is the SAT 100T a platform for doing that kind of characterization?
Yes. As a prototyping platform, the SAT 100T is well-suited to this because you can inject any signal you want — real-world signals, abrupt signals captured in the field — and test your components on the bench in the lab. You can directly observe how they behave on EVM, noise figure, and ACLR, and optimize for power consumption.
It seems the SAT 100T is essentially an RF generator and an RF signal analyzer combined — a powerful, integrated stimulus-response system across multiple channels.
That’s exactly right. The SAT 100T’s transmit side does what very expensive signal generators do — it can generate any waveform you need: LTE, 5G NR, SAT DVB, mixed signals, custom waveforms. On the receive side, it’s a fine signal analyzer. With 8 transmit and 8 receive channels — plus 2 wide-band receivers — it’s both a strong signal source and a strong signal analyzer, working simultaneously.
So as a designer, I can use the SAT 100T to characterize components, then to test my radio at the system level, then to test the system in the field, and finally as a prototyping platform for next-generation improvements — the same hardware, with different software personalities for each phase. That’s powerful.
Absolutely. The SAT 100T is highly flexible across test, field test, and prototyping. With the flip of a switch — really, a software download — it converts from a test platform to a prototyping platform. And within each of those modes, there’s room for a lot of custom testing: in prototyping, you can test like a radio, run PIM cancellation, test digital pre-distortion for the power amplifier, replay real-world traffic captured in the field, and optimize your PA for power consumption. With 8 transmit and 8 receive channels, the platform is built to be flexible.