LMH7324: Minimum pulse width

Hi Chuck,

Thank you for your response.

Please find the case below, I would like to know the behavior in the Case B.

Hi Joe,

thank you for your explanation.

Please find out attached my simulation results.
Do you know why is the output signal changed?

Thanks.

wave.pdf

Hi Joe,

Thank you.

Yes, I know that it has a propagation delay of about 730ps.
However it is a different and an independent from the input toggle rate.
I guess its point is how it fast switches the input threshold voltage in the internal circuits.

This typical rate is 3.72Gbps. The interval is about 269ps.

I attached my comments on wave2.pdf file.

wave2.pdf

Hello,

I now understand your question and I will work on getting you a response by the end of business tomorrow (12/11) and by the latest (12/14).

Regards,

Joe

Hi Joe,

Thank you so much for your help.

waiting your feedback.

Regards,

Hello,

Can you please attach your simulation schematic? I would like to view the PSpice code for the LMH7324 and I want to see your circuit configuration.

Thanks,

Joe

Hi Joe,

Thank you.

please find the attachment.

comp_2.TSC

Hello,

I have attached an improved model for the LMH7322. I disabled the latch functionality of the LMH7322 so that the device is always on.

The main difference between the LMH7322 and the LMH7324 is that you set the hysteresis value of the LMH7322 with an external resistor. To maintain the internal hysteresis of the LMH7324 of approximately 20mV, I used a 5k resistor in the attached TINA schematic.

I hope this resolves the issue you were experiencing.

Regards,

Joe

LMH7322_FIX.TSC

Hi Joe,

Is there any update?

please find the attached files in the above reply.

Thanks.

Hi Joe, Chuck，

Thank you so much.

I think I missed your response.

I will confirm it.

Thanks,

Hi Chuck,

Thank you for your supports.
I checked the behavior of modified model (LMH7322_FIX).

In case of the difference voltage is 10mVpp, it is okay. However when its voltage is more than 10mVpp (like 16mVpp), its result is not correct. The hysteresis voltage is around 20.8mV (22.5mV@typ) in the LMH7324.

I tried to change the hysteresis resister to 2kohm. But the result was incorrect.

Hi Frank,

Sorry about the issues you are seeing with the models. If I am understanding correctly, you want to see this device exhibit the 22.5mV of hysteresis that the LMH7324 actually has?

If so, I made a fix to control the internal hysteresis of the part by adding this line into the macro and disabling the Rhys/Rhref control for hysteresis.

V_V16        N741641 0 22.5m

Please let me know if this helps solve your issue.

Regards,

JonnyLMH7322_VhystFixed.TSC

Hi Jonny,

In the last, please let me know the actual product behavior of LMH7322 and 7324.
In this time, you modified the spice library to meet my expectation.

We've made the evaluation boards by using the LMH7324 and LMH7322 both. However, it did not work correctly, then I tried to run these spice simulations to understand the behavior.

Thus, I'd like to know the actual design specifications. When I input the signals in the same as simulation, can I get the same results in the actual products?

Thank you.

Hi Frank,

I am glad that Jonny was able to modify the model to meet your expectations. 

From what I can tell, your evaluation boards using the LMH7324 and the LMH7322 are producing results different from what you are seeing in simulation?

Do you have scope captures of the results you are getting with your evaluation boards? I would like to see the input waveforms and the output all on the same scope capture to better understand your issue. 

Regards,

Joe

Hi Joe,

Hi Jonny,

Before I sent our evaluated waveforms, I'd like to make sure the problem point of the simulation library. When I looked at the library (macro), there is the following sentences:

EVH VH 0 VALUE = { ( V(VHYS)/2) }

I guess it does not need the division by 2 since the threshold is below:

V(INP)>{V(INN) + (V(VH))} or {V(INN) - (V(VH))}

Please inform me your comments or feedbacks.

Thank you.

Hi Frank,

EVMID VMID 0 VALUE = { ( V(VDD) + V(VSS) )/2 }   ** Midpoint value to detect when output is high or low
EVH VH 0 VALUE = { ( V(VHYS)/2) } **VH =Needs Vhys divide by two to split hysteresis to either add to the threshold or subtract voltage to the threshold
EINNNEW INNNEW 0 VALUE = { IF( ( V(OUT_OUT) < V(VMID) ),(V(INN) + (V(VH))),( V(INN) - V(VH) ) ) } ** If OUT is low, add VH to IN-, If OUT is high, subtract VH from IN-
EOUT OUT 0 VALUE = { IF( ( V(INP) > V(INNNEW) ), V(VDD), V(VSS) ) } **If IN+ is higher than IN- with hysteresis thresholds added/subtracted, output high.

For example, lets say that Vhys of the device should be 20mV. Depending on if the output is high or low, the comparator will add or subtract 10mV from the threshold - covering the hysteresis range of 20mV. Without the divide by two, the comparator would be adding or subtracting 20mV of hysteresis in each direction from the threshold, resulting in a hysteresis range of 40mV.

In regards to as if you would see the exact same results as the EVM from the model, we have not done a 1:1 comparison on that and have only spec'd the model to the datasheet. I will definitely take a closer look into this model to ensure it is up to datasheet specifications.

Regards,

Jonny

Hi Jonny,

Thank you for your feedbacks.

In case of 20mVpp hysteresis, the V(INNNEW) would be V(INN)+10mV in original.
Then the V(INP) is compared with the V(INNNEW). I guess it means the hysteresis range is 10mVpp.

When will the out be changed?
As you mentioned that it is adding VH or subtracting VH.
VH is a half of the hysteresis value in your case.

Thus, it has only 10mV to change the output state even if it sets 20mV hysteresis (vhys).

e.g.,
rhys = 5kΩ, about 20mVpp hysteresis
in+ = 100mV to -8mV
in- = -100mV to +8mV
out = high to ? ... high or low

I expect the out is not changed since the difference is 16mVpp.
But in your case, the out is changed because of divided by two.


Regards,

Hi Joe, Jonny, Chuck

Thank you for your support on these questions.
Have a great day!

Kind Regards,