Okay, here is the abstract and summary for the provided transcript.
Abstract:
This video documents extensive experiments aimed at developing a low-noise amplifier stage suitable for a parallel crystal oscillator configuration, intended as an alternative approach to phase noise reduction compared to previously explored cascaded filters. The core focus is the iterative design, testing, and optimization of a low-input-impedance amplifier circuit centered around a BFR91A transistor and a custom-wound transformer. Key investigations include impedance matching, selection of passive components (metal film resistors proving superior to carbon film), transistor choice and operating conditions (including the benefits of cooling), the critical impact of RF base decoupling on phase noise, and detailed transformer design using different core materials, winding techniques, and core stacking, encountering challenges with enamel insulation integrity. Noise performance is evaluated throughout using Allan deviation (TinyPFA), phase noise measurements (PN2060, Linrad), and noise figure analysis, revealing significant improvements but also highlighting potential measurement system artifacts. The video concludes with a refined amplifier design deemed ready for implementation in a three-channel parallel crystal oscillator system planned for a future video.
Detailed Amplifier Optimization for Low-Noise Parallel Crystal Oscillators
0:00:04 Cascaded Filter Recap: Previous work using two cascaded crystal filters yielded only a 3dB noise improvement (due to √2 increase in Q), deemed insufficient.
0:01:15 Parallel Oscillator Concept: References a paper showing √N noise improvement by paralleling N phase-locked oscillators (coherent signal, incoherent noise). Aims to apply this principle differently.
0:02:52 New Parallel Crystal Approach: Proposes paralleling multiple crystals, each driving its own low-input-impedance amplifier. The goal is a √3 noise reduction with three units.
0:03:13 Input Impedance Analysis: Calculates that a ~10 ohm amplifier input impedance is a reasonable target to balance close-in phase noise (dominated by loaded Q) and the far-out noise floor (dominated by insertion loss) for the 80-ohm R_s crystals.
0:06:20 Amplifier Impedance Characterization (Initial): Measures the input and output impedance of the initial amplifier design (using BF240). Input impedance is shown to be highly dependent on the output load (e.g., 10.35 ohms in with 25 ohms out).
0:11:04 Power Level Calculation: Determines the expected power levels based on the maximum crystal drive (0.1mW), calculating the power delivered to the subsequent stage. Note: An initial calculation error is corrected at 0:33:33.
0:13:47 Allan Deviation Measurements Begin: Uses TinyPFA to measure phase stability (Allan deviation). Initial results with carbon film resistors show significant noise.
0:15:57 Environmental Sensitivity: Demonstrates sensitivity to airflow (temperature changes) and minor sensitivity to supply voltage variations.
0:20:32 Transistor Comparison (BCY59): Replacing the BF240 with a BCY59 increases noise, contrary to expectations.
0:22:10 Component Upgrade (Metal Film Resistors): Replacing carbon film resistors with metal film resistors significantly reduces noise, identifying them as a major previous noise source. Key Takeaway: Resistor type is critical for low-frequency noise performance.
0:23:30 Cooling Benefit: Cooling the BF240 transistor improves its noise performance, especially at higher operating currents, suggesting thermal effects.
0:26:48 Transistor Upgrade (BFR91A): Switching to a cooled BFR91A transistor yields the best Allan deviation results so far and lower supply voltage sensitivity (0:28:00). Key Takeaway: BFR91A (cooled) appears superior for this application.
0:29:52 Transformer Redesign: Modifies the output transformer using Material 67 (lower loss than Material 61), adding taps and tuning capacitors.
0:31:15 Improved Crystal Filtering Ratio: The new transformer design improves the impedance ratio between the crystal's off-resonance and on-resonance states, suggesting better potential noise filtering.
0:34:02 Potential Measurement Artifacts: Observes periodic "wavy" structures in phase measurements (Allan deviation) using the TinyPFA, particularly with the new transformer. Subsequent tests (0:40:07, 0:41:34) suggest these might be artifacts of the measurement system (TinyPFA) rather than the amplifier itself.
0:42:37 Phase & AM Noise Measurements: Uses an interferometer for carrier suppression and measures phase noise (PN) and amplitude noise (AM) with PN2060 and Linrad.
0:45:11 AM/PM Noise Correlation: Finds AM noise is significantly higher than PN close to the carrier and highly correlated, suggesting flicker noise in the transistor modulating both gain and capacitance.
0:50:29 Bias Resistor Value Impact: Increasing the impedance of the base bias resistors increases AM noise but doesn't significantly affect PN.
0:52:48 Base Decoupling Importance (Electrolytic): Adding a large electrolytic capacitor to the base significantly reduces AM noise and improves PN. Key Takeaway: Base decoupling is crucial.
0:54:02 Base Decoupling Optimization (RF Capacitor): Replacing the electrolytic with a 0.1uF ceramic capacitor (chosen for 10MHz resonance) further improves PN and reduces AM/PM correlation. Key Takeaway: Proper RF base decoupling is critical for phase noise.
0:57:34 Noise Figure Measurements: Measures the amplifier's noise figure under different load conditions using a complex setup and corrections for the non-standard source impedance and receiver noise. Estimates NF around 3dB.
01:05:24 Performance Summary & Load Choice: Summarizes measured impedances, gain, noise figure, and expected noise performance vs. load. Decides to proceed with a design targeting a 25-ohm load impedance.
01:08:23 Transformer Winding Experiments: Explores various winding techniques for the transformer (litz wire vs. thick solid enamel wire, wire diameters). Finds thick solid wire performs better than litz wire for the secondary winding (0:11:26).
01:14:23 Transformer Core Stacking: Stacking two ferrite cores improves coupling (indicated by lower inductance when secondary is shorted).
01:16:13 Transformer Failure & Diagnosis: The optimized transformer fails (increased loss). Suspects a shorted turn caused by enamel damage from sharp ferrite edges. Key Takeaway: Careful construction is needed to prevent enamel damage on ferrite cores.
01:17:37 Transformer Rebuild & Protection: Rebuilds the transformer, taking precautions by smoothing/coating the ferrite edges (using nail polish then UV glue) to protect the wire enamel.
01:18:17 Final Amplifier Test & Potting: The rebuilt amplifier works correctly, showing improved impedance characteristics. Plans to pot the core in epoxy for stability.
01:19:04 Conclusion & Next Steps: The amplifier design is finalized. The next video will focus on building three such amplifiers and interfacing them with crystals for parallel operation.