LMH6703 Enable Disable Characterization

Hi Jose,

When you turn off the LMH6703, the output goes high impedance and the device supply current drops to ~200uA in 10ns. So, if you monitor the output voltage, you are a little bit at the mercy of the exact moment when disable is initiated to see the instantaneous output voltage decaying away through the available external impedance at that node. So, you may get different decay profiles for when output is near V+ compared to mid-point or near V- (relative to the edge of the disable waveform). To get a measurement as to the exact moment when the quiescent current has dropped and when the output is high impedance, you could:

a) Monitor the supply current

b) Measure the output impedance. An example would be a setup where the output is driven with a source and resistance and then the high-impedance state is easily detected. Note that it is a lot easier to detect that point in time if you have a large RF so that the loading is minimal.

Sorry about the general nature of my response, but I don't have a setup available to try anything more specific right now.

Regards,

Hooman

Hooman,

Thanks for your response.  You said that "when disable is initiated to see the instantaneous output voltage decaying away through the available external impedance at that node"  - is the external impedance the load resistance RL = 100ohms or the high impedance state left by the disabled part?

Thanks again for your response,

JD

Hooman,

Sorry for my persistance, but I am trying to set it up just like you stated and I still don't get what the datasheet is stating the part should be doing.  You state to measure current changes, but the datasheet is showing voltage changes in the 10ns range.  On page 7 of the datasheet it is showing a voltage output with a sinewave being shut off by the part. 

Is that a source on the output, or is it an input sinewave measured on the parts output? 

Thanks for your response,

JD

The datasheet plots usually are done with a high speed input signal, the scope is set to capture a single trace and triggers on the enable logic and then a single sweep is plotted.  We usually use around 10MHz for the repetition rate of the enable signal. 

Your test setup is watching the 1V output voltage decay through 100 Ohms of resistance.  There are probably even thermal effects showing up with such a slow enable signal.  The amp is heating up and cooling down during the different enable states. 

Loren,

Correct me if I am wrong, but your response is contradicting the way Hooman states TI characterizes the enable/disable time. 

When I first created this post I stated that the part was shutting off at about 140ns.  Your response is telling me that I am doing everything correct, and that the part is not responding the way it should. 

Thanks for your response,

JD

Hello Jose,

Hooman gave you an alternate way to measure the response.

Can you attch some plots of what you are measureing and a schematic of your test setup? 

Did you understand that the output goes into a high impedance state when disabled? 

What do you require from the circuit? 

Loren,

I can't send you schematics because our software only works with the layout of our PCB's. 

Yes, I understand that the output goes into a high impedance state when disabled.  I also understand that the datasheet is showing a signal shutting off at 10ns with a 100ohms at the output.

"what do you require from the circuit?"    I was hoping to get an answer as to why my opamp is not shutting off the way the datasheet is stating it should.

Thanks,

JD

In your test the amplifier has been biased up to 1V at the output.  You disable the amplifier output.  Then you wait for the voltage to fall to 0.5V. 

Can you make the same measurement, but put one marker at the instant the SD pin is asserted and then the other marker at the beginning of the voltage drop (as illustrated above)?  Once the output voltage begins to fall the amplifier is off.  The slope that happens afterward is the response of the RLC equivlaent circuit of the amp output, the PC board and your test fixture. 

Loren,

Your right it is an RLC issue, but on the out put I have eliminated all of the copper from around the op amp up to the resistor's (input and output load)  the traces are not impedance matched because I eliminated the ground plane from underneath.  The output load is about 325 mils from the part connected through a 6 mil thick trace on a 1ounce copper FR4 board.

I take the timing measurement at the 0.5v point - 50% of the signal, and the scope gives me a PDelay of about 140ns.

Should I take the measurement at 10%?

Thanks for your help,

JD

I was trying to see how National Semi characterized this part, and I found that the parts response as it is disabled is still active for a good 140ns at 50% of the signal.  The datasheet is misleading when it states that the part can be disabled at about 10ns.  If this part is driven slower than 50MHz the disable pin will not shut off the output.  I saw the same pulsed input at 50MHz and at 1Mhz.  At 50MHz, there is a 10ns Pdelay because of the 10ns pulse from the input signal going to 0v at the half way point of the period.  The 1MHz signal shows a full pulse  decaying to the end of its pulse at 1us and then a fast drive to 0v.  This is well below the 50%, but it is still active.

THE DATASHEET IS MISLEADING!!!!

jd

If you solder a 100 Ohm resistor from the amplifier output to ground how does the measurement change? 

Can you upload a schematic of your test setup?  Can you send me your scope model and setup parameters?  Are you using a high impedance setting on your oscilloscope? 

Do you have an application that requires the output to clamp to 0V immediately upon disable? 

One method to get this result would be to  use 2:1 MUX.  One Mux input can be used for your signal, and the other mux input would be set to 0V.  When the mux is switched from the input signal to the 0V input the output will be driven to 0V much faster than can be accomplished by letting a disabled output stage drift down. 

I am posting an exerpt from the datasheet to help clarify.  I did not make this graph, but normally when we measure disable time we use a 50 Ohm oscilloscope and there is a 50 Ohm termination resistor on the evaluation board.  We use the shutdown signal as the scope trigger.   The scope is set for a single sweep and we then plot the results.  Although the transmission line is doubly terminated at 50 Ohms when the amplifier is active in the disabled state the transmission line is only singly terminated (by the oscilloscope) and the load from the amplifier output is 100 Ohms. 

Loren,

I have a 100 ohm resistor on the output.

Scope is a TDS 3034B set at High impedance - not using 50ohm termination.

I don't have an application that requires the output to clamp to 0v immediately upon disable. 

I am trying to characterize the part to the spec of the datasheet.

Loren,

That is not what the datasheet is stating.  The datasheet only states the 100ohm load. 

The pulse on the output is measuring 10ns because the signal is a 50MHz sinewave. 

That is MISLEADING.

Set this part up and put a 1MHz signal in the input of the device.

THE PART WILL NOT SHUT DOWN LIKE IT IS STATED ON THE DATASHEET.

Please post a schematic of the test setup.  I really cannot help without more detail. 

For example.  Is the 100 Ohm resistor in series with the test probe or is it connected to ground?  What is the total amplifier load resistance, including the oscilloscope.  Are you still using a DC input as the input signal?