trouble making this work

[jonorthwash]

I'm having trouble getting my Tascam recorder to respond to signals as generated by this code, even when shifting the serial signal to 3.3V. I was wondering if @abbrev or others might have any insights.

I have the following configuration:

• Tascam DR-60DmkII
• KOOKYE Pro Micro ATmega32U4 microcontroller
• 4-pin 2.5mm TRRS plug, with sleeve wired to microcontroller GND (I've tried the lower ring as well) and tip wired to microcontroller TX. I've confirmed multiple times that each lines is conducting appropriately.
• I have tried also placing a level converter between the microcontroller and the plug, using the 3.3VDC from the recorder to power the low-signal side.
• Various permutations of the following code (simplified somewhat):

  #define START_MASK  0x80
  #define REPEAT_MASK 0xC0
  #define END_MASK    0x00
  
  static const uint8_t TASCAM_RECORD = 11 ;
  static const uint8_t TASCAM_STOP = 8 ;
  
  void setup() {
    Serial1.begin(9600, SERIAL_8E1);
  }
  
  void loop() {
    if (condition) {
      Serial1.write(TASCAM_RECORD | START_MASK);
      delay(100);
      Serial1.write(TASCAM_RECORD | REPEAT_MASK);
      delay(100);
      Serial1.write(TASCAM_RECORD | END_MASK);
    }
  }
  

• I know that condition is being met, because I have tested other outputs on it (powering an LED, and simply writing the data to the serial monitor instead using Serial.println; I get e.g. 136 for the start command, etc.). Furthermore, the TX LED flickers when the write commands are called.
• I've also tried to send start and end bytes with no delay between them.
• I am able to measure a 5V signal from TX when not writing, which I understand is normal for TTL.

The next thing I plan to check whether there is an appropriate TTL signal coming from the TX pin. EDIT: I've now checked, another microcontroller reports receiving the following signals on sending record: 139, 203, 11; and the following signals on sending stop: 136, 200, 8. But (obviously, perhaps) only for 5V signals, not 3.3V signals.

Beyond that, my only guesses are, in order of likelihood, (1) that I'm not doing something right, (2) that there's a hardware compatibility issue, or (3) that there's a problem with the reverse engineering of the protocol. I'm also likely going to buy an actual RC-10 to verify that the recorder actually responds to signals from it.

In the meantime, I'd be interested in hearing from someone who has experience making this work. Does anything I'm doing stand out as wrong?