LMH6628QML-SP: Redundant Buffer

Hello Markus, you don't want to use diodes as the loops will open up in one polarity and peg to the rail. Just sum the two output through resistors outside the loop - their value will depend on your intended load. 

Hello Michael, thanks for your very fast reply.

The load current is in the range of 2mA and the error that is introduced by the series resistor should be less than 50mV. So the value for the series resistor should be lower than 25 ohms.

The output signal is in the range of 5V, so the resistor would limit the current into the unpowered output to 200mA. Is this suitable? Is current limitation necessary when an voltage is present at the output of an unpowered device?

Regards Markus

I would help to know your load, 

I am assuming you are trying to parallel two op amps to ensure operation. What we call the ballast resistor in a wired or configuration will limit the standing DC currents between the outputs via the output offset voltage mismatch. 

A non space qualified part had a lot of work done on this issue - the OPA837 was intended for this where a wired or mux app was supported by including a switch in the inverting input under disable control - similar not exactly the same as your app, here is a reference design talking through the issues - some of which might apply to you, 

https://www.ti.com/tool/TIDA-01565

Hello Michael,

regarding the load: It is a reference voltage for a sensor. When board 1 (with op amp 1) fails, board 2 (with op amp 2) shall provide the reference voltage. So your assumption on paralleling two op amps is correct.

Thanks for the link to OPA837. I will have a look at this and will get back with my further questions.

Regards Markus

So there is basically no DC load, but a cable and input capacitance - so you you do need to think stability, estimate your total C and set the ballast R accordingly. 

Actually, I assume that it is a DC load. The sensor reference input states an input current of typically 1.6mA (2mA,max). At the moment I do not have any information about the internal circuitry of the sensor.

I am not sure if I correctly understand the function of the ballast R. At a time only one op amp is powered, the other one is unpowered. So I assume that there will be some current that flows into the output of the unpowered op amp. There is no information on the internal circuitry of the op amp in the datasheet. Sometimes there are clamping diodes present on inputs and outputs of components. Then it is necessary to limit the current. 

Well that was unclear, I was guessing both would be on at the same time. 

If you know the current is coming out only, yes - a diode inside the loop seems possible, I worry about turn on with a non-linearly element in the forward path wether it will find an operating point. 

The OPA837 topolgy is exactly for this function (but the inactive channel would be powered, but disabled), but not SP level. 

The series output resistors will also limit the current back into the off channel to a non-zero value back into the esd diodes if the supplies are 0 on the inactive channel - the next obvious question is a reliability confirmation under those conditions - good luck with that. 

Ok Michael, this sounds that I should search for an alternative solution. I was already afraid of that.

Maybe I could add a signal switch that disconnects the path of the unpowered op amp. But I dont know how to compensate the voltage drop at the signal switch (for an on resistance of 50 ohms * 2mA = 100mV).

Do you have any idea for a possible solution?

Hello Markus,

I agree with Michael's assessment, you may wish to explore other options.  I am taking a look to see if I or my team are aware of any.

Hello again Markus,

In addition to Michael's suggestion, I consulted my team for ideas.  One option is to use an amplifier with power-down functionality: for the OPA858 this disables the amplifier and sets the output at a high impedance state.  This feature is controlled via a pin on the device.  Some amplifiers with power-down may not enter a high impedance state, so be sure to check the datasheet.

Could you elaborate on your device rating and bandwidth requirements?

-Alec 

Not sure that will work Alec if he is turning power off to the backup channel. 

Good catch Michael, I overlooked that detail when I discussed this with my team.  

Markus, 

As Michael mentioned, use of the power-down function on an op-amp like the OPA858 will not work if the whole board has power turned off.  

I do not yet have an idea for how to help compensate the anticipated voltage drop if you proceed with the option in your reply.  If you would like me to investigate more, I would ask you share any additional details about your device requirements.

In the meantime I will properly relay your situation to my team and see what all we can discuss.

Alec