non inverting LT1028 oscillates

[el00]

Hello, the circuit in figure is an amplifier with 26dB of gain in the range 10Hz to 100kHz (approx).
The opamp is LT1028, powered between -6 and +6V, and of course there is a decoupling capacitor.
R4 was not installed initially. For some reason this seems to be very unstable. Sometimes it works, while sometimes it oscillates. Other behavior is that is shows an oscillation superimposed to the signal.
I also added the compensation capacitor as per LT1028 datasheet (I tried several values), it did not solve the issue.
But if I add R4=100k, however, it becomes stable. I don't understand why.
Any clue?

 

[danadakk]

Again, R4 allows the input bias current path for OpAmp input. Post #7

Inputs require a DC path for bias current needed for input diff stage.

 

[el00]

Again, R4 allows the input bias current path for OpAmp input. Post #7

Inputs require a DC path for bias current needed for input diff stage.

why not R1? I mean, that requirement should be fulfilled by R1. I already asked this question but I don't think there was an answer, unless I missed it

 

[dick_freebird]

I have not noticed any physical construction pics / info and a wireball or solderless breadboard would potentially add many couplings that are not within the model's comprehension.

Sounds like the part is a bit "edgy" and so might want to check that.

 

[el00]

I have not noticed any physical construction pics / info and a wireball or solderless breadboard would potentially add many couplings that are not within the model's comprehension.

Sounds like the part is a bit "edgy" and so might want to check that.

it's a 6 layers pcb, with 2 GND planes, 1 plane for 6V and 1 plane for -6V.

 

[FvM]

Again, R4 allows the input bias current path for OpAmp input. Post #7

Inputs require a DC path for bias current needed for input diff stage.

As shown in post #1, bias current path is through source impedance, that's o.k. if source hasn't hidden capacitive coupling.

Circuit with input capacitor C1 but no R4 does in fact not work.

 

[el00]

As shown in post #1, bias current path is through source impedance, that's o.k. if source hasn't hidden capacitive coupling.

Circuit with input capacitor C1 but no R4 does in fact not work.

oh, I see. Then I might want to move to another schematics, i.e. inverting configuration. That's because I don't want to have large input series resistors, which would void choosing of a very low noise amp.

 

[danadakk]

why not R1? I mean, that requirement should be fulfilled by R1. I already asked this question but I don't think there was an answer, unless I missed it

The inputs go to their own transistor(s) in the diff stage, so each needs a DC path.

Look at the schematic diagram in the data sheet of the LT1028. Towards back of datasheet.


Regards, Dana.

 

[yefj]

oh, I see. Then I might want to move to another schematics, i.e. inverting configuration. That's because I don't want to have large input series resistors, which would void choosing of a very low noise amp.

Hello, Klauss said some thing very integresting, -1 means inverting configuration.
+1 stable means non inverting configuration.
what is the meaning "only on -1 which stability wise equals to +2"
What is +2 in stability manner?
Thanks.


As mentioned in the first line of the LT1028 datasheet description ... it is not stable with gain of +1.
(only on -1 which stability wise equals to +2)

 

[danadakk]

why not R1? I mean, that requirement should be fulfilled by R1. I already asked this question but I don't think there was an answer, unless I missed it

The inputs go to their own transistor(s) in the diff stage, so each needs a DC path. But qualifier,
this parts input stage looks like it has a DC return on the Inv input so unlike many OpAmps,
that do not have this, I admit not expert here (or anywhere ). But could be balancing of
current sources / sinks that causes that pin to still want additional bias current to keep
stage in active region. I would post a question at Analog Devices and ask them/describe
your issue, see what they say.

Notice the q1 transistor takes the same current as the q7 current source, and any the
Darlington sink only function when the difference between inputs > ~ 1.2-1.4V. So
that seems to indicate the NI input is floating (high Z) and no source driving pin
leaves it floating, if ac coupled, so leakage dominates due to lack of some bias path.


Look at the schematic diagram in the data sheet of the LT1028. Towards back of datasheet.


Regards, Dana.

 

[FvM]

oh, I see. Then I might want to move to another schematics, i.e. inverting configuration. That's because I don't want to have large input series resistors, which would void choosing of a very low noise amp.

Purpose of C1/R4 is still unclear. If the source acts as bias current DC path, why not directly connect it to N.I. input?