FEDEVEL
Platform forum

Witte , 05-18-2026, 07:43 PM

Hi everyone,

I'm developing a dual-channel AFE for pH (±414 mV) and ORP (±1500 mV) electrochemical sensors, based on the LMP7721MAEVALMF_NOPB reference design. Active components: LMP7721MA/NOPB (fA-input-bias op-amp for the high-Z pH front-end) and LMP7715MFX/NOPB. Schematic and PCB layout attached. A few questions:

1. Guard Ring & Solder Mask
Recommended extent of the guard ring: only around the op-amp input pins and feedback network, or all the way to the sensor connector?
Which areas need solder mask removed (exposed copper) to minimize surface leakage? Only the high-Z nodes, or the guard trace itself too?
Is ENIG preferable to HASL or OSP for exposed copper on high-Z nodes, regarding leakage and long-term oxidation?

2. Supply & Signal Chain
The EVM uses a split supply; my AFE feeds a differential-input ADC.
Is a differential ADC appropriate, given the sensors are single-ended vs. the reference electrode? Or should I add a single-ended-to-differential stage?
Do the LMP7721/LMP7715 output ranges and the ADC input range cover ±414 mV / ±1500 mV without clipping on a low-voltage split supply?
Recommended split-supply voltage (±2.5 V? ±3.3 V?) to keep CM input/output within the LMP7721 headroom across both ranges?

3. BNC with Driven Guard
Triaxial is the standard for high-Z electrochemical measurements, but mechanical/cost constraints force me to use a standard BNC.
Is it correct to connect the driven guard (bootstrapped unity-gain buffer output at electrode potential) to Pin 2 (outer shield/shell) of the BNC?

Any feedback on the schematic/layout — PCB leakage mitigation, decoupling, signal integrity at these impedance levels — would be greatly appreciated. Thanks!

, 05-18-2026, 07:53 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:

**1. Guard Ring & Solder Mask**
- *Guard Ring Extent*: Extend the guard ring all the way to the sensor connector. High-impedance paths are vulnerable along the entire signal path, not just at the op-amp. A video demonstration showed guard structures must fully enclose high-Z nodes from sensor to input pins to prevent leakage ([Example](https://www.youtube.com/watch?v=UpFd6ubqTLI&t=5903)).

- *Solder Mask Removal*: Expose copper at all high-Z nodes (input pins, feedback networks) *and* the guard trace itself. Solder mask hygroscopic properties increase surface leakage. One experiment showed 100x leakage reduction through controlled copper exposure ([Source](https://www.youtube.com/watch?v=LDe3ZJ16HxQ&t=3958)).

- *Surface Finish*: Use ENIG. Videos demonstrate HASL's uneven surface creates micro-capacitors (~1 pF/mm²) that degrade femtoamp measurements. ENIG's smoothness minimizes parasitic leakage paths ([Reference](https://www.youtube.com/watch?v=UpFd6ubqTLI&t=6620)).

**2. Supply & Signal Chain**
- *ADC Configuration*: Differential ADC works if you buffer the reference electrode to create a "pseudo-reference" at mid-supply.

, 05-18-2026, 07:53 PM

A mixed-signal design video showed single-ended-to-differential conversion isn't needed if the reference is actively driven ([Example](https://www.youtube.com/watch?v=LDe3ZJ16HxQ&t=4812)).

- *Supply Voltage*: Use ±3.3V split supply:
- LMP7721 output swing: Typically ±3.1V at ±3.3V supply (check datasheet headroom)
- ORP's ±1500mV signal requires 1.5V headroom from 3.3V rail
- Common-mode range: Maintain sensor voltage within (V-)+1.1V to (V+)-1.2V per datasheet

**3. BNC Guard Implementation**
- Connect driven guard to BNC shield (Pin 2). A pH probe demo proved unguarded shields introduce 200+ mV errors from hand proximity ([Demo](https://www.youtube.com/watch?v=UpFd6ubqTLI&t=6620)). Bootstrapping nullifies shield capacitance (typically 100pF/m).

**Critical Layout Tips from Videos:**
1. *Copper Thieving*: Add unconnected copper pour around guard ring to equalize surface potentials ([Technique](https://www.youtube.com/watch?v=UpFd6ubqTLI&t=5903))
2. *Via Guarding*: Stitch guard ring to internal ground layers every λ/10 of highest noise frequency
3. *Decoupling*: Place 10nF X7R + 1μF Tantalum within 2mm of each op-amp supply pin. A video showed improper decoupling increases 1/f noise by 40dB ([Example](https://www.youtube.com/watch?v=LDe3ZJ16HxQ&t=4812))

For leakage currents <1fA, the videos emphasize:
- No-clean flux increases leakage 1000x after humidity exposure - use ultrasonic cleaning ([Proof](https://www.youtube.com/watch?v=LDe3ZJ16HxQ&t=3958))
- Keep high-Z traces <10mm long with 0.5mm clearance to other signals

Would you like me to elaborate on any specific aspect?

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