FAQ
TL;DR: For clean DC blocking, a 100 kΩ + 220 nF high‑pass (≈7 Hz) works, and “A simple RC filter would be a lot more reliable.” Use a DC servo only when direct coupling is mandatory. [Elektroda, Rohit Dubla, post #21668251]
Why it matters: This FAQ helps engineers pick between a coupling capacitor and a DC servo to remove offsets without killing low‑frequency signals, fast.
Quick Facts
- Case A signal: 800–1.2 kHz, ~300 mVpp with ~400 mV DC; Case B: 15–60 Hz, ~50 mVpp with ~4 V DC. [Elektroda, Walter Li, post #21668242]
- Example design: R=100 kΩ, C≈220 nF gives fc≈7 Hz; increase C up to ~1 µF to relax loading. [Elektroda, Rohit Dubla, post #21668253]
- Passive RC is simpler and more reliable; active servos add complexity and failure points. [Elektroda, Rohit Dubla, post #21668251]
- DC servo is preferred when you must keep direct coupling or parallel amplifier outputs. [Elektroda, Rohit Dubla, post #21668255]
- An AC‑coupling capacitor blocks DC; the following stage’s input impedance often serves as the ground return. [Elektroda, stephen Van Buskirk, post #21668249]
What’s the practical difference between a coupling capacitor and a DC servo?
A coupling capacitor forms a simple high‑pass filter that blocks DC with one pole. A DC servo actively senses offset and injects correction, allowing direct coupling while nulling DC. Servos handle very low frequencies without huge capacitors but add op‑amps, stability concerns, and design effort. Use capacitors when you can tolerate a defined low‑frequency cutoff; use servos when you need the passband to include near‑DC content or paralleled outputs. [Elektroda, Rohit Dubla, post #21668255]
Which method should I use for Case A (≈1 kHz, 300 mVpp, 400 mV DC)?
Use a passive RC high‑pass. Set the cutoff a decade or more below 1 kHz to minimize phase and amplitude error. The capacitor easily blocks 400 mV DC while passing 1 kHz with negligible loss. Add a buffer after the RC to prevent loading. This meets stability and simplicity goals without an active servo. [Elektroda, Rohit Dubla, post #21668251]
Which method should I use for Case B (15–60 Hz, 50 mVpp, 4 V DC)?
Design the RC cutoff below the lowest signal frequency. A target near 7 Hz preserves 15 Hz while blocking the 4 V DC. Use a buffer to maintain the intended pole and gain. If you must extend response closer to DC without large capacitors, consider a DC servo, accepting added complexity. [Elektroda, Rohit Dubla, post #21668253]
How do I pick RC values for DC blocking?
Choose fc below the lowest frequency of interest. Example: with 15 Hz content, set fc ≈7 Hz. Using R=100 kΩ gives C≈220 nF. Larger C (e.g., 270–330 nF, up to 1 µF) further reduces passband ripple but increases settling time after steps. [Elektroda, Rohit Dubla, post #21668253]
Do I always need a separate resistor to ground after the coupling capacitor?
Not always. Many inputs already provide a path to ground via their input impedance. In that case the added resistor becomes an extra load and may be unnecessary. Check the next stage’s input resistance before adding components. [Elektroda, stephen Van Buskirk, post #21668249]
Does an RC to ground remove DC? I heard it doesn’t.
A series RC to ground (Zobel/boucherot) shapes AC impedance and doesn’t pass DC through the capacitor, so it won’t remove a DC offset on the signal line itself. For DC blocking in series signal paths, use an AC‑coupling capacitor into a return impedance. [Elektroda, stephen Van Buskirk, post #21668245]
What’s a DC servo, in simple terms?
A DC servo is an active feedback loop that measures output DC offset and feeds a slow corrective signal to null it. It behaves like a very low‑frequency integrator tied into the amplifier, keeping the signal path DC‑coupled while canceling offset. Use when direct coupling is required. [Elektroda, Rohit Dubla, post #21668255]
When is a DC servo clearly better?
Use a servo when outputs of similar amplifiers are paralleled into one load or when you must keep direct coupling to pass near‑DC content without huge capacitors. In these cases, a blocking capacitor may be impractical or risky. [Elektroda, Rohit Dubla, post #21668255]
Are passive RC solutions really more reliable?
Yes. “A simple RC filter would be a lot more reliable,” because it avoids active parts that can fail or oscillate. The trade‑off is a finite low‑frequency cutoff and larger capacitors for very low‑frequency work. [Elektroda, Rohit Dubla, post #21668251]
Do I need buffers around the RC high‑pass?
Often yes. Source and load impedances shift the pole and cause amplitude/phase errors. Add a buffer after the RC to minimize loading and keep the cutoff where you designed it. A front‑end buffer may also help with high source impedance. [Elektroda, Rohit Dubla, post #21668251]
How do I design a quick passive DC blocker (3 steps)?
- Pick fc at ≤½ of your lowest signal frequency (e.g., 7 Hz for 15 Hz content).
- Choose R from available values; compute C = 1/(2πRfc).
- Place the capacitor in series with the signal and ensure the next stage provides the return impedance; add a buffer after. [Elektroda, Rohit Dubla, post #21668253]
What is an AC‑coupling capacitor?
It’s a series capacitor placed between stages to block DC while passing AC. Audio amplifiers commonly use it to remove DC bias and protect speakers. Your next stage’s input resistance provides the return path that sets the high‑pass pole. [Elektroda, stephen Van Buskirk, post #21668243]
What is a Zobel (Boucherot) network and is it for DC removal?
A Zobel is a resistor and capacitor in series to ground, used to stabilize impedance and tame high‑frequency behavior, often across speakers. It does not block DC in the signal path. Use it for AC noise shaping, not for offset removal. [Elektroda, stephen Van Buskirk, post #21668245]
Any gotchas or edge cases I should watch for?
Large RC time constants mean long startup settling; signals near the cutoff will tilt or attenuate. Active servos can misbehave if loop dynamics are wrong. Verify stability and ensure step responses meet your application’s timing. Keep component stress within ratings to maintain reliability. [Elektroda, Rohit Dubla, post #21668251]
Can I minimize attenuation at 15 Hz while still blocking DC?
Yes. Set the cutoff well below 15 Hz—about 7 Hz was suggested. With R=100 kΩ and C≈220 nF, attenuation at 15 Hz is small while DC is strongly suppressed. This balances phase shift and passband flatness for low‑frequency signals. [Elektroda, Rohit Dubla, post #21668253]
