LMH7324 Differential output

Hello Jin,

I think you are asking why some of the LMH7324 output toggles / transitions look "normal" while the one you've circled in yellow is low amplitude / looks-incomplete.

If the input signal pulse duration is too fast, I can see that the output would not have the time to complete the low-high and high-low transitions and that particular portion of the pulse waveform may appear to have a different amplitude. Do you think that this is a possibility here? The parameter that should affect this the most is probably the LMH7324 output waveform rise / fall times. Have you made sure that the measured output rise / fall times are adequate to support the pulse duration you are applying?

Regards,

Hooman

Hooman

I measured the waveform you asked and the duration time is below 600ps. It means that the measured pulse duration meets performance listed on the LMH7324 data sheet. In case of the low amplitude waveforms, the outputs Q+/Q-, are not differential. I attached the waveform file.

3465.LMH7324 rev2.doc

Thanks

Hi Jin,

I see that you've measured the input pulse duration of 600ps which meets the LMH7324's minimum pulse width of 290ps. However, in your scope photos, your output pulse 10%-90% rise times look to be around 1.5ns or so (your fall times is slightly faster!). You should get a rise time close to 140ps with small (2pF) load:

I suspect that your issue is that your output(s) are not transitioning fast enough and the low-duration-inputs are not wide enough to allow the output(s) to reach their final value.

• Are you driving an extra heavy cap load, beyond the 2pF stated? If your cap load is doubled, for example, I'd expect your rise time to double to 280ps, etc.

• Have you biased the output stage with the proper DC current (~14mA nominal)? You have not shown the connections to your output(s) so I cannot tell.

In Figure 30, if VT= VCCO - 2V (or 0.5V in your case with your VCCO= 2.5V), you'd use RT=50ohm (2 places) which sets the DC output(s) current at 14mA:

I_out_DC= (2V - (1.1V + 1.5V) / 2) / 50ohm = 14mA

You should bias your output(s) similarly. You could use any VT of 0.5V or lower, and adjust RT for 15mA current. You can increase the current (by lowering RT) within reason, for faster output rise / fall time.

• You have stated measured outputs with fast duration inputs "is not differential". I don't know what you mean by that! The output(s) you have shown are transitioning but they just don't reach the final value, and they are still differential (i.e. one increases while the other decreases, and vice versa).

So, please double check your output connections / biasing and see if you can find what is slowing the output. Disconnect the load (if any, leave the biasing resistors connected with as little lead length / trace length as possible) and see if you can get a response with your 600ps input duration. Also make sure your board layout is not adding much capacitance (e.g. ground plane spacing, keep controlled impedance traces, etc.).

Regards,

Hooman

Hello again,
I forgot to mention that the scope probe itself (even if its an active scope probe which I'm sure it is) you are using to capture the output waveform, could be adding enough cap load to slow your rise time.

Is there anyway you can "observe" the LMH7324 operation without a scope probe attached?

Regards,

Hooman