TTL output teensy 3.5

[y33]

Hello,

I've setup a TTL generator on arduino uno that I'm trying to port to teensy 3.5.
The generator will need to drive two different kind of devices.
The first is an AMPI iso-flex isolation unit (SIU). From what I understand, the input to that box is directed to an 6N135 optocoupler.
The second is a prizmatix led power supply (PRIZ) inside is a 500mA buckpuck luxdrive.

when the output is connected directly to pin5, the teensy drives the SIU (which seems to accept the 3.3V logic)
However, the PRIZ is not driven by these pulses.

I therefore added a logic level shifter - Great the PRIZ now works.... but the SIU doen't work anymore!
all connections are direct

teensy 3.3V----->low level High level--->teensy Vin
teensy pin 5---->A1 B1---> BNC for TTL output
A2 B2
A3 B3
A4 B4
teensy GND---->GND GND--->teensy GND

with an oscilloscope, the output signal is 5V!

any ideas why the SIU refuses to respond when logic level shifter is used?
(on arduino, which outputs directly 5V, both the SIU and LPS respond to pin change).

thanks in advance

EDIT: the SIU in fact does reply, but its output signal is highly dampened, something like 50 times less than should be

 

[y33]

Another problem which may be related:
On arduino, I can use the following ciruit to drive the SIU:

pin.5_____220ohm___LED____GND
| |
|___ TTL OUT ___|

This does not work on teensy 3.5 (replacing 220ohm with 60ohm)

any help? or alternate solution to convert 3.3V to 5V ?

 

[PaulStoffregen]

any ideas why the SIU refuses to respond when logic level shifter is used?

I do not know anything about a "SIU", but I can tell you not all level shifter circuits are created equal.

Many people have experienced problems with the very cheap bidirectional ones using 1 mosfet and 2 resistors. Their low output is strong, but the high output is very weak. Low to high transitions are much slower than high to low, which distorts PWM and pulse-width protocols like WS2812 LED data. These mosfet-based circuits are meant for I2C and other "open collector" signals. Unfortunately, many of the sites that sell them falsely advertise them as useful for all kinds of digital signals. Really, they're terrible when you need high speeds or correct pulse widths or strong high output to drive other circuitry.

The designed using bidirectional chips are usually better, but still problematic. They also have weak high level output. They also can get confused if you have noise or interference.

When unidirectional will work, chips like 74AHCT125 or 74HCT245 are recommended. They perform very well.