The master playback ran in a continuous loop. The recorded spools of tape, depending on the program length, could have anywhere from 12 -30 units recorded on it. Handling the spools took some practice, especially the cassette tape, but you eventually got the hang of it. The recorders had servo-controlled motors providing the correct amount of tension. On the take-up side there was an arm with a wheel guide which held and aligned the tape to the take-up reel. We had to check and adjust tensions frequently.
We ran test tones daily to check the levels and EQ'd all recorders and the Master playback unit once a week. We used a real-time frequency analyzer on the master while playing a white noise flat level test tape. We had a custom designed computer which check the levels and EQ on the recorders. Our specs were +- 0.5db for the recorders and +-.3db for the Master.
I was also responsible for EQ of our QA and A/B rooms which were used to test the masters against product. We used mostly McIntosh and Crown amplifiers, Nakamichi cassettes, huge JBL Studio Monitors, one in each corner of the room. This was just after the Quad phenomenon. We still had a quadrophonic version of Backman Turner Overdrive on an 8-Track tape. Instead of 8 separate tracks it switch a four-track head to two positions for Quad. Very cool at the time.
So yes the duplicated tapes sounded very close to the masters. They were compared constantly. The capstan motors were a constant 60Hz motor, so its speed was constant. If there was any variation in speed it was adjusted by adjusting the tension roller, adjusting the tension on the take-up motors, or in severe cases, different diameter capstans were used. We were very sensitive to specs.
Occasionally we "farmed out" replication to other plants, Ampex in Indianapolis was one of them. To ensure quality they sent us test tapes from their recorders which we tested once a week. They were not as good at making consistent adjustment as we were. It was common to see distortion in the 2.0 range, or +-3db out of spec in the 2K, 6K, and 10K readings of their EQs. WIth our specs being +-.5db for recorders and +-.3db for Master playback, QA and A/B rooms.
Working there was a blessing and a curse. A blessing because I was exposed to some very high quality gear. And a curse because I was exposed to some very high quality gear. I was taught what to listen for. What was good/bad. Afterwards I had a very low tolerance for Hisssss, distortion, sibilance, and scrape flutter. It made it very difficult to listen for enjoyment while hearing artifacts.