Trouble working out why some resistors are needed at opamp input

[fryingpan]

I was looking at the following preamp (which I don't intend to build, it's nothing special, just a 4-stage clean opamp preamp with tonestack and some diode clipping with no sort of pre-emphasis or filtering):



I'm stumped at R4 and R9. What are they for? They are low value and right before a FET-input stage opamp. What gives?

[Rob Strand]

Resistors like that have multiple reasons for being there.   The actual reason that motivated the designer to put them in isn't always clear.   Fending off RF is one reason.  Notice that only the OP2134 has the 1k and not the TL072.

https://sound-au.com/project88.htm

The text under figure 2
"Earlier boards included RF interference suppression by adding a small capacitor between the two inputs of U1 (the space for this cap can be seen in the PCB photo, which is of an early board).  This has now been removed, as it caused more problems than it solved - in particular, opamp oscillation with some devices.

Although shown with 10k resistors for gain, you can reduce these for lower noise if you prefer.  With the suggested OPA2134 (or NE5532) opamps, these resistors can be as low as 1k.  The opamp input resistor (R2 L+R) can also be reduced in value, but at the possible expense of lower RF interference immunity.  I wouldn't recommend less than 220 ohms. "

[garcho]

it's nothing special, just a 4-stage clean opamp preamp

Nothing special?! It's a Rod Elliott design and schematic which means it works and the schematic has no errors. That's very special, even by 2021 internet standards. Give the man his propers! 

Here is the link to the project of the OP schematic on Rod's site ESP.

[antonis]

Respective resistors are also present in Rod's earlier pre-amp designs.. 

[fryingpan]

Resistors like that have multiple reasons for being there.   The actual reason that motivated the designer to put them in isn't always clear.   Fending off RF is one reason.  Notice that only the OP2134 has the 1k and not the TL072.

https://sound-au.com/project88.htm

The text under figure 2
"Earlier boards included RF interference suppression by adding a small capacitor between the two inputs of U1 (the space for this cap can be seen in the PCB photo, which is of an early board).  This has now been removed, as it caused more problems than it solved - in particular, opamp oscillation with some devices.

Although shown with 10k resistors for gain, you can reduce these for lower noise if you prefer.  With the suggested OPA2134 (or NE5532) opamps, these resistors can be as low as 1k.  The opamp input resistor (R2 L+R) can also be reduced in value, but at the possible expense of lower RF interference immunity.  I wouldn't recommend less than 220 ohms. "

How does that work? If it works together with the input capacitance, wouldn't it be better to have lower series resistance in order to lower the lowpass frequency? In fact, I remember from my lab days that we used 10:1 probes because these allowed larger bandwidth (and had to be compensated by adjusting the probe capacitor). When we used 1:1 probes, bandwidth was much lower in fact.

[This post was accidentally edited by pressing Modify instead of Quote]

[fryingpan]

it's nothing special, just a 4-stage clean opamp preamp

Nothing special?! It's a Rod Elliott design and schematic which means it works and the schematic has no errors. That's very special, even by 2021 internet standards. Give the man his propers! 

Here is the link to the project of the OP schematic on Rod's site ESP.

I greatly admire the man, and I'm planning to build an amp with his 100W power amp; what I meant with "nothing special" was that it's a very clean pre-amp with not much in the way of character. I might get his preamp too as it seems very well designed, as a clean channel, and I'm basing my idea of a dirtier preamp on his (but I'd add some pre-emphasis and post-saturation correction, to yield better results with overdrive).

[Rob Strand]

How does that work? If it works together with the input capacitance, wouldn't it be better to have lower series resistance in order to lower the lowpass frequency? In fact, I remember from my lab days that we used 10:1 probes because these allowed larger bandwidth (and had to be compensated by adjusting the probe capacitor). When we used 1:1 probes, bandwidth was much lower in fact.

The 1k resistor forms a low-pass filter with the input capacitance of the opamp,



The C shown inside the opamp triangle is not a capacitor in the circuit but is the capacitance of JFETs (and possibly protection diodes) inside the opamp.   The input capacitance of the opamp is a few pF, say 5pF.  With 1k's the cutoff is fc = 1/(2 pi R C) = 32MHz.  That cut-off would be for an RF signal getting onto the input side of the 1k.

As far as the whole signal path goes the cut-off is somewhat lower since you have 68k and 100k from R1 and R2.  And for the volume control VR4 it could look as high as 25k to 30k.

    Respective resistors are also present in Rod's earlier pre-amp designs..  :icon_wink

Interesting he dropped the resistor on the last iteration.    Also the output circuit is slightly different.

[fryingpan]

How does that work? If it works together with the input capacitance, wouldn't it be better to have lower series resistance in order to lower the lowpass frequency? In fact, I remember from my lab days that we used 10:1 probes because these allowed larger bandwidth (and had to be compensated by adjusting the probe capacitor). When we used 1:1 probes, bandwidth was much lower in fact.

The 1k resistor forms a low-pass filter with the input capacitance of the opamp,



The C shown inside the opamp triangle is not a capacitor in the circuit but is the capacitance of JFETs (and possibly protection diodes) inside the opamp.   The input capacitance of the opamp is a few pF, say 5pF.  With 1k's the cutoff is fc = 1/(2 pi R C) = 32MHz.  That cut-off would be for an RF signal getting onto the input side of the 1k.

As far as the whole signal path goes the cut-off is somewhat lower since you have 68k and 100k from R1 and R2.  And for the volume control VR4 it could look as high as 25k to 30k.

Of course, the resistance is at the denominator, the larger it is, the smaller f_c is. Stupid brainfart (even though I clearly remember 10:1 probes having much greater bandwidth than 1:1 probes, ie. probes with no internal resistance) but that said, there is plenty of resistance before the input. What is 1k gonna change? It seems strange to me that you need a resistor close to the input, any RF is going to affect the whole circuit realistically. And what about a capacitor far from the input, maybe right after the input resistors (68k and 100k)? That would solve the issue, wouldn't it?

[antonis]

And what about a capacitor far from the input, maybe right after the input resistors (68k and 100k)? That would solve the issue, wouldn't it?

The further the cap from the input the more the space left for RF interference ..

[R.G.]

Two reasons come to mind, one opamp-related, one FET related.

It is possible that the resistor is there to limit current from transients. If there is an internal diode clamp on the inputs the inputs are protected in the voltage sense, but there are often pin-current limitations on opamps - well, on many ICs, but this doesn't get a lot of press. Maybe.

The RF theory is probably better. They could be there for the same reason that it's good practice to put a 100 to 1K resistor right at the gate of every MOSFET used as a linear amplifier. It damps the trace inductance and parasitic capacitance feedback paths that can lead to self oscillation. The input is nominally a high impedance, at least at DC and audio, but its impedance changes as frequency gets higher. The issue is especially notable in followers. High impedance followers can oscillate fairly easily.

Just a guess. Two guesses, I guess. 

[composition4]

Just a guess. Two guesses, I guess. 

Isn't that three guesses then?

[Rob Strand]

Of course, the resistance is at the denominator, the larger it is, the smaller f_c is. Stupid brainfart (even though I clearly remember 10:1 probes having much greater bandwidth than 1:1 probes, ie. probes with no internal resistance) but that said, there is plenty of resistance before the input.

The probe bandwidth thing is quite complicated.  If you think about it why should the 1:1 range, which is a wire, have any band-limiting effect at all.   The probe bandwidth is measured with a 50 ohm source.    One thing about 1:1 range is it has more capacitance, so the C is larger.   If you calculate the low-pass cut off the 1:1 will give a lower frequency but what you will find is that frequency is somewhat higher than the probe bandwidth spec.   In practice it's a different story.  The high capacitance of the 1:1 range will load down circuits.   Putting a 1:1 probe on  a 100k source resistance can drop the low-pass cut-off below 20kHz!  somewhat lower than the 60MHz probe bandwidth.

What is 1k gonna change? It seems strange to me that you need a resistor close to the input, any RF is going to affect the whole circuit realistically. And what about a capacitor far from the input, maybe right after the input resistors (68k and 100k)? That would solve the issue, wouldn't it?

By putting a resistor close to the opamp's input pin you are reducing the chances of RF getting into the input pin.   The input resistor not only filters RF it filters everything.  If high frequency signals in the circuit itself loop back onto itself it can cause high frequency oscillation.  The faster the opamps are more prone to parasitic oscillation.   If you remove the resistor at the opamp input and rely on the 68k at the audio input  you have a lot of tracks upto the opamp input pin which are essentially metal plates that can promote coupling of external signals, like RF.    Capacitors themselves are exposed metal plates.  The outer plastic coating doesn't do much to stop signals getting in; a 100nF poly cap has quite large area.

Those are some iof the reasons why it can make a difference and why some designers chose to use an input resistor.    With the exception of difficult opamps which have a tendency to oscillate, if you took those extra resistors out it probably would not make any difference.    You could only see a difference if your area had strong RF at particular frequencies.   It might even help stop you mobile phone pips from come through the audio.

You could summarize by saying the resistors are bullet proofing the circuit against adverse events at the expense of extra parts.   Like any design there's nothing forcing you to set the bar at any particular level, other than the circuit doesn't work.

Other opamps circuit will have input resistor for different reasons.   For example NE5534 have input diodes and it's a good idea limiting the current through the diodes with an input resistor.  LM324's sometimes require input resistors.

So yeah, a simple resistor but not so easy to explain why it's there.