FAQ
TL;DR: For 100+ MHz peak-current viewing, use a very low‑inductance 1 Ω shunt and keep leads ultra‑short—“Low inductance is key to this measurement.” [Elektroda, DAVID CUTHBERT, post #21657616]
Why it matters: This approach lets you see IC supply-current spectra without corrupting the waveform, helping you debug emissions fast; it’s for hardware/EMC engineers measuring high‑speed IC current with scopes or spectrum analyzers.
Quick Facts
- Target band discussed: 150 kHz–1 GHz for regulatory scans; measure current as voltage into a spectrum analyzer. [Elektroda, Cody Miller, post #21657614]
- Off‑the‑shelf current probes cited: CT1 1 GHz/450 mA(rms), CT2 200 MHz/2.5 A(rms), CT6 1 GHz/120 mA(rms). [Elektroda, Bob Casiano, post #21657612]
- Desired measurement bandwidth: at least 100 MHz for IC supply current. [Elektroda, Cody Miller, post #21657611]
- Low‑inductance shunt method: parallel 10 Ω parts to make ~1 Ω close to the IC; avoid probe loop inductance. [Elektroda, DAVID CUTHBERT, post #21657616]
- Readout topology: shunt → 50 Ω series → SMA → scope 50 Ω; add shunt C to set a pole with 25 Ω Thevenin. [Elektroda, DAVID CUTHBERT, post #21657618]
What’s the fastest practical way to view IC supply current without disturbing operation?
Use a low‑inductance current‑viewing resistor near the IC power pin(s). Parallel small 10 Ω chips to make ~1 Ω. Keep the interconnect to the pin extremely short to minimize inductance. Avoid clamp‑on current probes in this path, as added inductance corrupts fast edges. “Low inductance is key to this measurement.” [Elektroda, DAVID CUTHBERT, post #21657616]
Can I just use a commercial current probe for 100–1000 MHz?
You can, but inserting a probe like a CT1 in series adds too much inductance for real‑time IC supply current viewing. That extra inductance alters the waveform you want to measure. For highest fidelity, the shunt‑resistor method outperforms series probes in this use case. [Elektroda, DAVID CUTHBERT, post #21657616]
How do I connect the shunt to my oscilloscope or spectrum analyzer?
Terminate the shunt into an SMA through a 50 Ω series resistor. Drive the instrument’s 50 Ω input directly. The shunt capacitor at the SMA can create a compensating pole with the 25 Ω Thevenin source seen by the input. This stabilizes the frequency response for analysis. [Elektroda, DAVID CUTHBERT, post #21657618]
What shunt value should I start with for DRAM or other higher‑current ICs?
Use a smaller shunt to reduce intrusion. A 0.1 Ω current‑viewing resistor is a practical starting point for higher current devices like DRAM. Keep layout inductance very low, and maintain short paths to the power pins. [Elektroda, DAVID CUTHBERT, post #21657618]
How wide is a 1 Ω transmission line on FR‑4?
Approximate the 1 Ω line as about 50× the width of a 50 Ω stripline. Since a typical 50 Ω has width ~2× dielectric height, the 1 Ω line width is about 100× the height. Use a field solver for accuracy. [Elektroda, DAVID CUTHBERT, post #21657618]
High‑side current sense vs. transimpedance amplifier—which is better here?
For 100+ MHz peak‑current viewing, a high‑speed transimpedance amplifier is feasible but harder. A low‑resistance, low‑inductance current‑viewing resistor is the more straightforward, robust approach for minimal disturbance and wide bandwidth. [Elektroda, DAVID CUTHBERT, post #21657620]
What bandwidth should I aim for to debug emissions up to 1 GHz?
Users targeted at least 100 MHz to observe fast supply‑current dynamics. This helps correlate spectral peaks with IC activity before full 150 kHz–1 GHz compliance scans. Start at 100 MHz, then push higher as needed. [Elektroda, Cody Miller, post #21657611]
How do I avoid my ground plane becoming an antenna during these tests?
Contain high‑frequency loops locally. Create local power and ground nets under the IC with tight decoupling. Connect these nets to the main planes at a single point, preferably with a larger second‑tier capacitor across the join. “The only difference between a ground plane and a center‑fed patch antenna is what you’re driving it with.” [Elektroda, Olin Lathrop, post #21657615]
Could a loop antenna plus spectrum analyzer replace direct current viewing?
Yes. A small loop near the board reveals radiated magnetic fields and can localize noisy areas. It complements current viewing when you only need relative emissions and frequency fingerprints rather than exact supply current waveforms. [Elektroda, Per Zackrisson, post #21657622]
Is there an off‑the‑shelf option if I don’t want to build the fixture?
Commercial current probes exist. Examples include Tektronix CT1 (1 GHz/450 mA rms), CT2 (200 MHz/2.5 A rms), and CT6 (1 GHz/120 mA rms). Verify insertion method to avoid extra inductance in series with the IC supply during time‑domain viewing. [Elektroda, Bob Casiano, post #21657612]
How do I compensate the shunt’s inductance to keep a flat response?
The shunt’s inductance creates a zero. Add a shunt capacitor at the SMA to form a matching pole with the 25 Ω Thevenin source (from the series 50 Ω and the scope’s 50 Ω). Tune C for a flatter response across your band. [Elektroda, DAVID CUTHBERT, post #21657618]
What if I must measure on the ground pin instead of the high side?
You can place the shunt in the ground return if it does not disturb operation. Keep interconnects extremely short and maintain local decoupling so high‑frequency current loops remain confined near the IC. [Elektroda, Cody Miller, post #21657614]
Any quick build steps for a wideband current‑viewing fixture?
- Parallel small 10 Ω resistors near the IC to make ~1 Ω with minimal inductance.
- Route a short, very‑wide low‑Z line from IC power pins to the shunt.
- Feed shunt through 50 Ω to an SMA and 50 Ω instrument; add shunt C to set the pole.
[Elektroda, DAVID CUTHBERT, post #21657618]
Will measuring current frequency content guarantee I fix EMC failures?
No. Even if you capture spectra, worst‑case behavior varies with operation, lot, and temperature. Design for containment: short loops, local power/ground nets, and proper tiered decoupling to reduce emissions regardless of measured spectra. [Elektroda, Olin Lathrop, post #21657615]
Could a difference amplifier help as a front end?
A difference amplifier with sub‑unity compensation can offer high bandwidth. However, stability at gains below one is challenging. Many integrated diff‑amps are too slow for 100 MHz goals, so validate device bandwidth and stability first. [Elektroda, Bruce Carter, post #21657619]
What’s an edge case that breaks this setup?
Placing a current probe in series with the IC supply adds loop inductance and can destabilize or mis‑represent fast current transients. Keep the measurement path low‑inductance and avoid added series hardware during time‑domain viewing. [Elektroda, DAVID CUTHBERT, post #21657616]
