UCC27511DBV: Buffer to drive PWM into 50 ohms

Hey Greg,

I've split this thread since we don't support that part in the Logic forum. Here's a link to the original thread for the engineers over here.

Hello Greg,

I've alerted the people in the drivers group about your question. Someone there should be able to help you assess the needs of your application and the appropriateness of a particular driver to it. Look for more contact on this soon.

Regards,
Ulrich

Hi Greg,

I am one of the applications engineers in the gate drivers group at TI. I have used the UCC27511 before, and I actually recall a conversation about using it as a drive for a digital signal in the way you are considering. At first glance, yes, this driver will work for what you need (drive 50ohm, 0-5V, at 80kHz). Another option at a similar price point is the LM5112. Do you happen to know the node capacitance that you are driving? And what rise/fall time spec do you need to meet?

Regards,

- Daniel

Thank you Daniel and Ulrich. The output is general purpose, so unfortunately I don't know exactly what the end customer will connect it to or use for cabling. I do know the load impedance will be 50 Ohms and is 0 to 5V PWM. The rise/fall time spec probably isn't critical, and of course slower is better to reduce high-frequency components. I'm going to guess about 10 to 30 ns or so would be good. Thank you for suggesting the LM5112. I looked at the specs and the ROH spec (Output resistance high) is too high. The typical value is 30 Ohms, so in the DC condition a very large portion of the output signal would be dropped across the driver. The UCC27511 has a worst case value of 11 Ohms, resulting in an output voltage of 3.9V across 50 Ohms, which is acceptable. I get nervous using parts outside their intended usage, but it looks like the UCC27511 should work.

I just read through the layout guidelines for the UCC27511 and the warnings make me a bit nervous. The guidelines are with respect to very short traces on a PCB, which is a lot different than driving a cable. But at the end of the cable is a 50-Ohm termination, whereas the gate of a MOSFET is high impedance. If the cable is 50 Ohms and is terminated properly, the signal needs to have a high DC current in order to maintain a logic high. In the case of the MOSFET, very little current flows once it reaches steady state. I've never designed anything for a 50-Ohm load resistance. If I'm not mistaken, on a PCB a 50-Ohm transmission line does not have to be terminated with a 50-Ohm load (though it can, and sometimes capacitively coupled to block the high DC current once it reaches steady state). In that case the source impedance needs to match the 50-Ohm transmission line so that when the signal reflects off of the high-impedance termination it gets absorbed at the source and doesn't reflect back down the line. The load sees the full voltage in the steady state because it isn't conducting very much current. If that description isn't correct, please let me know. If you have any advice regarding how to approach the design of driving a 50-Ohm load termination (which is a very common thing in test equipment), please let me know. I'm struggling with this a bit. It might be that my expectations are a bit out of line, and I will need to go to extraordinary measures to meet the specs I stated.

Thanks and regards,

Greg