Bench modeSteps, parts, and safety only. Big type for a phone at the bench.

Electrical safety, with the numbers

What IEC 60601 actually limits, why a battery satisfies it and a wall supply must be designed for it, how isolation works, and what the 390 kΩ resistor is doing.

AssumesThe safety charterProject A: The ADS1299 boardSpineAnalog / mixed-signal hardwareClinical / regulatory / human factors

You are skimming: the title, the first figure, and the short version. Switch to Read in the header for the full page, or Deep to open every deep dive.

The safety charter gave you a rule: batteries only. This page gives you the numbers behind it, so that when you design something that cannot run on batteries, you know what it has to meet. The standard is , and the number at its centre is a few tens of microamps.

What current does to a body

Current through the body, not voltage, is what hurts. Around 1 mA at 60 Hz through the skin is the threshold of sensation. Around 10 to 20 mA, muscles contract and you cannot let go. Around 100 mA across the chest for a second or two, the heart fibrillates. Those are figures for current entering through dry skin at a hand; an electrode with conductive paste has a far lower contact resistance, and a wire in the heart (a pacing lead) needs only tens of microamps to fibrillate, which is why the strictest limits are for equipment that can touch the heart.

The body’s resistance from hand to hand is typically 1 to 2 kΩ. Ohm’s law says 120 volts across that is 60 to 120 mA. That is the whole danger in one line.

What the standard limits

IEC 60601-1 classifies patient-connected parts. For a part that touches the skin (type BF, “body floating”), the allowed through the patient is 100 µA in normal conditions and 500 µA with a single fault. For a part that can touch the heart (type CF), it is 10 µA normal and 50 µA fault. “Single fault” means one thing has gone wrong: a ground wire broken, an insulation barrier failed, a supply shorted.

Those numbers are why the 390 kΩ resistor exists. The bias drive output is a low-impedance amplifier that could, if the amplifier failed, sit at a supply rail. With ±9 V rails and 390 kΩ in series, the worst-case current into the body is 9 / 390,000 ≈ 23 µA. Under the fault limit even for a cardiac-class device. The AD8232 datasheet, the ADS1299 reference design, and every clinical amplifier have this resistor for this reason.

Why a battery passes and a wall supply must be designed

A battery has no connection to anything else. The only path for current from the battery through the patient is through the amplifier’s own circuits, which the series resistors limit. A single fault (say, a shorted input protection resistor) still leaves the other resistors and the battery’s low voltage; there is simply no 120 V anywhere in the system.

A wall supply has 120 V on one side of a transformer and your circuit on the other. The transformer is the barrier. Its windings have a small capacitance between them, through which a tiny 60 Hz current flows to the secondary and then through anything connected to it, including a patient. Medical-grade supplies are built with reinforced insulation and low inter-winding capacitance so that this leakage stays under 100 µA even with one fault. A phone charger is not built that way, and its leakage is often hundreds of microamps in normal operation. That is the difference between “medical” and “consumer” power supplies, and it costs about ten times as much.

Predict before you look

Your amplifier runs from batteries. You connect its output to a laptop's USB port, and the laptop is on its charger. Is the patient connection still isolated?

No. The USB cable’s ground pin ties your amplifier’s ground, and therefore its reference and bias electrodes, to the laptop’s ground, which the charger ties to the wall. Any leakage in the charger, or any fault in the laptop, now has a path through the electrodes. This is the most common way the battery rule gets broken, and it is why the plotter page says to unplug the laptop or put a USB isolator in the line.

Isolation devices

A USB isolator puts a digital isolation chip (capacitive or magnetic coupling) and an isolated DC-DC converter in the cable, so data and power cross a barrier with no conductive path. Good ones are rated for a few kilovolts. Optical links and wireless (the ESP32’s Wi-Fi or Bluetooth) are isolation by default; there is no wire to carry a fault. Isolation amplifiers do the same for analog signals. Every clinical EEG system has one of these between the patient-connected front end and the computer, and the ADS1299 board design put one in from the start.

The other half: single-fault thinking

The standard’s habit of asking “what if one thing breaks” is worth adopting for everything. What if the bias amplifier saturates? The resistor limits the current. What if an input protection resistor shorts? The other one still limits. What if the battery is reversed? Add a diode. What if the isolator fails shorted? Use one rated for the voltage with margin, and consider a second barrier. This is the reasoning that in Phase 5 formalizes into a table.

Deep dive Why frequency matters 2 min

The body is most sensitive to current at mains frequencies, 50 to 60 Hz, which is unlucky. At DC and at frequencies above a few kilohertz, the thresholds for sensation and fibrillation rise, which is why the standard’s limits are frequency-weighted and why some stimulation and impedance-measurement systems deliberately use kilohertz currents. The ADS1299’s lead-off detection injects nanoamps at 31 Hz, far below any limit, and its impedance measurement is safe for that reason.

Deep dive Defibrillator protection 2 min

Clinical amplifiers must survive 5 kV from a defibrillator applied to the patient they are connected to. That is what the large series resistors, gas discharge tubes, and clamp diodes on clinical front ends are for. Your board does not need this, but if you ever wonder why a clinical ECG front end has parts your design lacks, this is usually the answer.

Recall
What are the IEC 60601 patient leakage limits for a skin-contact (BF) device?
100 µA in normal condition, 500 µA under a single fault. For cardiac-contact (CF) devices, 10 µA and 50 µA.
Recall
What does the 390 kΩ resistor in the bias drive path guarantee, numerically?
With ±9 V rails, a worst-case fault current of about 23 µA into the body, below the single-fault limit even for cardiac-class equipment.
Recall
Why is a consumer phone charger not acceptable for a patient-connected device even though it has a transformer?
Its inter-winding capacitance and insulation are not designed to medical limits, so its leakage current can be hundreds of microamps in normal operation; medical-grade supplies use reinforced insulation and low capacitance to stay under the limits with one fault.
Explain it to your roommate

Explain what this page was about to your roommate in three sentences. No jargon they would not know.