LMH6739: LMH6739 output R Value

Hi Hugo,

figure 28 has nothing to do with driving a 75R load. It only discusses how a capacitive load (CL) should be driven by the LMH6739 without causing the OPAmp to oscillate. This can only be done by inserting an "isolation" resistor, here called "Rout".

Figure 28 of datasheet must be seen together with figure 8, which shows the recommended Rout for various capacitive loads:

An example: If the capacitive load at output of  LMH6739 is 47pF, Rout=24R is recommended.

Why all this hassle with Rout?

Connecting a capacitive load directly to the output of an OPAmp can introduce a dangerous phase lag in the feedback loop, which can erode the phase margin so far that the OPAmp can become instable and oscillates. Rout "isolates" the capacitive load from the feedback loop and ensures stability.

Kai 

Hi Sir,

Our customer have output level issue,
the RGB input is 0.7V, But the output Change to 0.5V level.
so we are checking this issue new.
and Thanks for your explanation.

Hugo.

Hi Kai,

Thanks for your reply.

In LMH6739 Fig.9 indicated the Rout vs. Capacitive load, however connecting a different VGA cable which mean has different terminated and capacitive load. In our system design, we used LMH6739 as a VGA signal buffer on Projector’s VGA output port. If we designed a Rout value between 20~33Ω on LMH6739 output, will it cause the signal reflection and any quality issue on VGA output image(ex: Image phase noise) when connected with a standard (ex: 75Ω) or a longer VGA cable respectively. Please help advise how to choose a correct Rout value to meet all kinds of these scenarios?

Hi Kai,

[The cable must always be terminated with its characteristic impedance to prevent improper operation. So, a 50R cable needs 50R impedances at both ends and a 75R cable needs 75R impedances.]

--> This made me confused, since in your previous reply, we should refer to Fig. 28 to design a relevant Rout value with the different capacitive load. However, in the actual application, we would never know what kind of cable (50R/75R) and capacitive load will be connected by the end-user, right?

Kent

 Hi Kai,

Thanks for your reply.

According to your detail explanation, our application is more similar to the scenario 2(driving low input impedance).

And, here is our test condition & problem:

** Test condition**

(1). Pattern generator: Quantum 804B

(2). VGA input timing: 1920x1200-60Hz (RB)

(3). VGA input pattern: Cross test pattern which has one pixel wide white details was used to observe the VGA transient response.

(4). 1.8m VGA cable characteristic impedance: 75 ohm

(5). Test jig: R/G/B signal with 75R to GND.

** Test result**

(1). In Fig. 1, if we directly put R,load(75R) terminal resistor on the Main board side, the output amplitude of R/G/B signal measured on Projector’s VGA-out port was fine (Vout=1/2*Vin).

(2). In Fig.2, if we connected Projector’s VGA-out port to the test jig through a 1.8m VGA cable, the output amplitude of R/G/B signal measured on test jig: the amplitude of High frequency^*a was insufficient(Vout<1/2*Vin), but the amplitude of Low frequency^*a don’t have this problem.

Also, if we reduce the VGA input timing to 640x480-60Hz, the output amplitude of R/G/B signal measured on test jig: both High and Low frequency^* don’t have the amplitude problem, too.

Note:

a. Below shown the measurement waveform of the Cross test pattern, that includes two parts of frequency (High frequency: 2.5MHz; Low frequency: 1KHz).

This issue has been inquired and post in the below link: http://e2e.ti.com/support/amplifiers/high_speed_amplifiers/f/10/p/720761/2658805#2658805

We have try to follow your suggestion to remove all the ferrite bead, protection diodes and filter cap from R/G/B signal trace on the LMH6739 input and output path, but there’s no help.

Hi Kai,

Yes, we have tried to use others 1.8M VGA cable, but get the same test result.

We connected the R,G,B,H,V signal to the test jig through the VGA cable, there signal have the terminal resistor on the test jig. [R,G,B signal:75R to GND and H,V signal: 2.2K to GND].

[8/22: We have try to follow your suggestion to remove all the ferrite bead, protection diodes and filter cap from R/G/B signal trace on the LMH6739 input and output path, but there’s no help.]

--> Update: To remove the protection diodes and filter cap from LMH6739’s R/G/B output path, it can improve the amplitude of R/G/B signal measured on the test jag (raise 20~30mV), but still not good.