I have owned my venerable Phase Linear D-500 amplifier since 1978. It was the last amplifier that Bob Carver designed before he left the Phase Linear Corporation to start his new business.
Over the years, I have done several maintenance items on this amplifier, mostly related to re-capping. The majority of problems were due to capacitors drying out on the board that controls the amplifier's functions. I've replaced these over the years.
Just recently, the transistor that controls the latch circuit which powers the main relay for switchings mains power to the large power transformer had developed an intermittent breakdown where the amp would occasionally turn itself on with no one pressing the front panel button. Since I didn't have an MPSA42 on hand, I replaced with a 2N3440, with just as good as new results.
Integrating this amp into my racks a number of years ago brought about a problem: hum. The amp threw a lot of magnetic energy into everything made of metal that was nearby. Especially with direct conduction from the D-500 chassis to the rails of the rack.
I decided to redo my rack configuration this month in an attempt to fix the problem by isolating low level systems from power systems. While I was at it, I did a little bit of needed maintenance on several pieces of gear.
While inside the D-500, it occurred to me what an awful job of engineering went on as far as wiring and grounding topology was concerned. These guys wired the amp in the most wasteful, noise-attracting way possible!
I found several wiring no-nos and gradually worked on fixing them. Before I get to details, I should point out that this amp always showed a bit of instability in the form of ultrasonic oscillation, when driven to clipping, since it was bought new. (I love to put all my amps into dummy loads and put a scope and signal generator on them to measure the real power output.)

Getting back to details now, I ended up cutting out about 60% of the wiring lengths where B+/- voltages were running around the chassis. They had the DC, AC and low level audio wiring bundled together. No wonder I would hear a buzz when a light dimmer was turned part way on in the house. Every wire went into one harness.
Even more unbelievable was the fact that the ground for the driver board and audio inputs was taken off the negative terminal of the left and right speaker outputs! No wonder I was measuring crosstalk garbage in the undriven channel when the driven (other) channel was driven to clipping.
I proceeded to correct these deficiencies, observing the "star topology" of grounding methods. I moved the driver board ground to the power supply ground, at the junction of the two big capacitors. I unbundled all the leads carrying AC, DC and audio and shortened the DC power leads by directly routing them to the capacitors from the output stage. I shortened the ground leads in the same manner.
The next thing I did was add a second 25 amp bridge rectifier to the power supply. I ran separate parallel wiring from that to the Low/Hi Z relay and parallel wiring to the +/- capacitor terminals.
At an earlier time, I had replaced the original electrolytic capacitors with much larger 45,000uF capacitors. Recent research made me aware that the ESR of electrolytics is not so good above 1000hz, so I added 4.4uf of poly capacitors across each power supply rail, to improve the ESR at higher frequencies.

I shortened the audio wiring from the rear apron, routing it directly to the front panel, instead of letting it run parallel to 110 vac wiring in a bundle. To cure a slight left channel hum, I ended up placing/gluing a strip of copper clad PCB on the left edge of the driver PCB, right over the 5" trace of unshielded audio input, grounding that strip to audio ground. That quieted the left channel hum.
Over the course of these modifications, wiring was shortened, leads carrying diverse signal types were segregated, and grounding was corrected so that all systems share a star topology.

The result? The crosstalk between left and right channels is gone. The amp doesn't oscillate when driven to clipping anymore. The power supply is stiffer and doesn't sag under load.
Subjective impression? The noise is gone. Music emerges from a blanket of silence. There seems to be a greater clarity and detail and the soundstage is wide and deep. Transient response is crisp. The amp sounds better today than it did when it was new.
In fact, now that I isolated the amp from the rest of the equipment by insulating it from the rack rails electrically, the pervasive hum that plagued me is gone to inaudible on my ultra-efficient speakers.
I'm sure there are other things I could further do to make subtle measureable improvements, but the changes made so far have resulted in pretty radical improvements. I'm quite surprised that Phase Linear would design such marvelous circuit topology, but completely screw up the wiring topology. But having corrected that, the true sheen of this amplifier is realized.