LMH6574: Crosstalk present only while actively switching channels?

Hi Ramkuma,

this is a video multiplexer, so the inputs and outputs should be wired for 75R signals as shown in section 7.1 of datasheet. Also, choose a gain of factor 2.

By the way, is the input of your scope set to "DC" or "AC"?

Kai

Hi Kai,

Thank you for your reply. The time durations of interest being indicative of relatively low frequencies (the horizontal segment of the ramp waveform is almost a uS) I felt the input termination of 75 ohm was not necessary. However I will try this suggestion along with a gain of 2. My scope input is set to DC.

Best - Ram

Hi Ramkumar,

Dr. Ramkumar Ramaswamy said:

I felt the input termination of 75 ohm was not necessary.

What I wanted to say with this is, that the LMH6574 might need low impedances at the inputs. This must not necessarily be 75R resistors. Eventually, it's enough to drive the inputs very low ohmically by using a very short cable.

Kai

Hello Dr. Ram,

  That is very strange issue. Rather than acting like a multiplexor, it looks like there is some sort of combination at inputs IN1 and IN3. Does this occur as well at IN2? Would you be able to share a schematic which includes connections to all the pins on the device?

Thank you,

Sima 

Hi Sima,

Thanks for your message. My test jig (on which the photos were taken) consists of my target PCB with only the 6574  soldered in. So the schematic is quite simple on the jig - I have bridged pins 1 and 2  as well as pins 4 and 5. So IN0, IN2 are grounded. There is a ground plane below the 6574. Unity gain with 2k2. !EN, SD, pin 2 and 4 are all grounded. A0 held constant, A1 gets a switching waveform which switches between IN1 and IN3. due to the tight nature of the PCB it will not be easy to try another input, without introducing jumpers etc, but I will try to get to it if required.

To answer Kai's question, the input to IN1 (ramp) is coming from an emitter follower. The input to IN3 in the original circuit where the problem was observed is via a 200 ohm resistor from a TL074 opamp; in my jig it is via a 10k pot.

Due to this issue, when there is an AC waveform at IN3, the ramp waveform moves up and down very rapidly resulting in a fuzzy output ramp This caused me to believe that it was noise and/or oscillation. However after 2 weeks of digging I now conclude it is neither!

It has the nature of crosstalk but the degree is way above what the crosstalk specs state in the datasheet.

Best - Ram

The input to IN1 (ramp) is coming from an emitter follower. The input to IN3 in the original circuit where the problem was observed is via a 200 ohm resistor from a TL074 opamp; in my jig it is via a 10k pot. I think these resistances should be low enough. Why should the gain of 2 be required? The 6574 is specced to also work with G=1.

Hi Ramkumar,

my recommendation is always to start with a circuit which is exactly what the manufacturer recommends and from this setup to make subtle changes. That's why I recommended to set a gain factor of 2 and to use 75R wiring.

What happens when you apply plain DC voltages to the inputs of LMH6574?

In such a moment when very strange things are happening, it might be wise to take a fresh chip (and a fresh board) and to start from scratch.

Kai

1. I am unable to use a gain of 2 because the input waveform will start getting clipped. I have tried with a gain of upto 1.5.

2. It was precisely because I saw this strange phenomenon happening in my original circuit that I took a different 6574 and assembled it into the second board. So the problem I am reporting has been seen in two different chips already.

3. What is even more curious is that if I turn off the channel switching and keep Pin 13 low, so as to exhibit only the ramp, then varying the DC at IN3 has no effect - appropriately enough. The signal at IN3 seems to "leak through"  only while channel switching is taking place. (Original post updated with this statement.)

I guess it looks like I just have to try another IC altogether - OPA4872 or AD8174 - to see if this is specific to the 6574.

Hi Ramkumar,

correct me, if I'm wrong, but the DC shift observed by you is very small, only 20mV. This would still be within the input offset voltage specification of +/-20mV of LMH6574.

I guess that other 500MHz multiplexing chips aren't much better. Maybe you expect too much?

Kai

Hi Kai,

Note that this is not an offset because it does not remain constant but varies with a signal. The appropriate spec to look at is not Offset but Crosstalk which in Fig 25 of the datasheet is specified as better than -90dB at 1MHz. Offset of even 1V would be easy to handle. Crosstalk of 20mV ruins the design.

Best - Ram

Hello Dr. Ram,

  Quick clarification: the problem disappears when you do not have the channel switching waveform which means you keep A0 and A1 constant? And, the problem appears only with AC signals applied to IN3 (not DC signals)?

Thank you,
Sima

1. "The problem disappears when you do not have the channel switching waveform which means you keep A0 and A1 constant?"

-->Yes?

2. And, the problem appears only with AC signals applied to IN3 (not DC signals)?

--> No. The photos in my original post show the problem occurring with DC signals applied to IN3. The problem occurs with AC signals as well. The DC was used to sharpen the observation and show it was not noise or oscillation.

Best - Ram

Hello Dr. Ram,

  Sorry for the delay in reply. It doesn't look like a crosstalk issue; it would be more subtle and not as clean as shown from image 1 to image 2 then to drastic shift in image 3 especially at the top horizontal segment of the signal. The issue might be if there are any capacitance at the selection A0/A1 that might be holding the voltage or resistance at these pins. Is that the case here? Would you be able to probe the voltages at these pins (9 and 8)? 

Thank you,

Sima 

Thank you Sima. That's an interesting comment. I do think the stray capacitance could be a culprit. I will try to probe or check the capacitance reported by ORCAD as soon as I get a chance.

Best - Ram