4
\$\begingroup\$

I am trying to downshift a 5V TTL signal to 3V logic signal using NTB0102 level shifter. The connection is described below:

enter image description here

However, my oscilloscope measurements are distorted. Even the input signal (out of the microcontroller pin-pin connected) does not look like a true TTL 0-5V signal:

enter image description here

Placing caps on VCC does not help. Reversing the propagation way (0-3V input to 0-5V output) also produces ugly picture. Pulling UP/DOWN microcontrollre's output pin does not make a difference. Any ideas about what's going on?

The layout:

enter image description here

After someone suggested using 10x probes, I have an improvement in the output signal, but it still does not like a square wave:

enter image description here

\$\endgroup\$
  • \$\begingroup\$ Did you connect the probe grounds properly? \$\endgroup\$ – Erik Friesen May 14 '14 at 16:25
  • 4
    \$\begingroup\$ Is the uC connected to the same ground as the translator? \$\endgroup\$ – The Photon May 14 '14 at 16:33
  • \$\begingroup\$ @ErikFriesen I hope I did. They are connected at the place all grounds. \$\endgroup\$ – Nazar May 14 '14 at 16:46
  • 1
    \$\begingroup\$ @Naz . you do need both capacitors. My point was to bring the capacitors ground legs all together on NTB pin 2. Your capacitor ground lead for pin7 capacitor was going to bottom rail, and capacitor ground lead for pin3 capacitor was going to top rail. \$\endgroup\$ – Marla May 14 '14 at 19:34
  • 1
    \$\begingroup\$ This is turning into 20 questions. Stack Exchange isn't really well suited to troubleshooting like this. My advice, however, would be to take everything out of the breadboard, move up ten or fifteen rows, and build it again. Also, at 2 MHz, is the reactance of the breadboard significant enough to cause problems? I'm imagining capacitors between all the lines, and a little bit of mutual inductance to boot. \$\endgroup\$ – Jason_L_Bens May 16 '14 at 1:30
1
\$\begingroup\$

Two more guesses:

  1. The scope probe compensation is set badly. The blue trace in your bottom scope plot looks like the poster case for this.

  2. You have the "coupling" for that traces set to AC instead of DC.

\$\endgroup\$
  • \$\begingroup\$ Yes. The probe was at 1x. However, changing to 10x did not completely fix the problem. But the problem was indeed in the measurement technique. The signal was clean after buffering the NTB shifter. \$\endgroup\$ – Nazar Jun 25 '14 at 18:14
0
\$\begingroup\$

This is pure speculation, and I know nothing about this chip, but perhaps the NTB0102 output pin-5 needs to have some kind of load on it to produce a clean signal. After looking at the spec sheet for this device, the "test circuit" designates a load resistance to ground of 1 Meg ohms for measuring wave forms, and 50K ohms for some other types of measurements. Maybe the load will dissipate some of that "noise".

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.