Opamp question regarding unity gain

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi all,

On my search for why my system was so darn loud now (half volume was like full volume on the old Cambridge), I discovered I hugely underestimated the voltage that my sound card puts out. I thought it put out 0.3v max, but it actually puts out more like 1.4v max. The preamp was set to 0.3v = 1v out!

I am using P88 as a preamp: http://sound.westhost.com/project88.htm but with my own PCB's. Near the bottom of the page is a "gain table", I figured I would try the 706mV = 1v, which would involve making the first opamp have no gain (unity gain if I am correct?). The text says:

"As shown, gain is 6dB, and to reduce it to 0dB just leave out R4 in each channel.".

R4 in this case is the resistor from the feedback to ground. But this leaves the 10k resistor from the output to -ve feedback input, which didn't seem right. I thought unity gain was just connecting the output straight to the -ve feedback input?

I went ahead and removed R4 anyway, as the text says, but left the feedback resistor in. It does "work".... up until 3/4 volume, but then bursts in to oscillation until I turn it down a smidge.

Would I be correct in thinking that I should infact be removing the 10k feedback resistor too, to get a proper unity gain behaviour? I'm using OPA2134's, so it should be unity gain stable I believe....

Thanks,
 
I'd probably reduce R3 or replace with a link; certainly it's worth trying.

The oscillation is probably due to layout considerations, being 'set off' by unwanted feedback above a certain signal level. It could be worth trying adding a small cap from pin 3 (+ input) of each opamp to ground to provide a touch of RF filtering here. 2.2nF would give -3dB at 72Khz, and so have negligible effect on the audio band. Small polystyrene or polypropylene caps are recommended.
 
Ex-Moderator
Joined 2005
For your application you should directly connect the output to the inverting input (which is commonly known as a unity gain buffer). The 2134 should be stable in that configuration.

When talking about gain, dB (logarithmic scale) or V/V (voltage out/voltage in) are commonly used units. 0dB=1V/V, 6dB=2V/V, 12dB=4V/V, 18dB=8V/V, etc.
 
In the scematic you posted I they are in Non-inverting configuration so yes you would remove Both 10k resistors (R3 and R4) and change R1 and R2 to 100R...

It is what is called a Unity Gain Buffer and if you would have googled Unity Gain Buffer you would have seen that it looks pretty much as described in Figure 2, (You don"t even need R2 and R1 or even C1 but they should probably stay for RF Interferance reasons )

Cheers
 
martin clark said:
I'd probably reduce R3 or replace with a link; certainly it's worth trying.

The oscillation is probably due to layout considerations, being 'set off' by unwanted feedback above a certain signal level. It could be worth trying adding a small cap from pin 3 (+ input) of each opamp to ground to provide a touch of RF filtering here. 2.2nF would give -3dB at 72Khz, and so have negligible effect on the audio band. Small polystyrene or polypropylene caps are recommended.


BWRX said:
For your application you should directly connect the output to the inverting input (which is commonly known as a unity gain buffer). The 2134 should be stable in that configuration.

When talking about gain, dB (logarithmic scale) or V/V (voltage out/voltage in) are commonly used units. 0dB=1V/V, 6dB=2V/V, 12dB=4V/V, 18dB=8V/V, etc.

Thanks for the confirmation.

Just quickly whipped out the 10k resistor and replaced with a wire link, it's now fine. The oscillations have disappeared.


Turns out I actually don't really need any gain at all. My sound card output is actually enough to give full volume, and then some. I guess I could make the second stage unity gain too...... or try a passive preamp :rolleyes: It does seem a tad pointless having two opamps and another power supply just for controlling the volume.

More adventures!

Thanks all.
 
markiemrboo: " ... I thought unity gain was just connecting the output straight to the -ve feedback input? ..."

" ... I'd probably reduce R3 or replace with a link; certainly it's worth trying. ..." [I would have said remove R3, rather than reduce ...]

" ... For your application you should directly connect the output to the inverting input (which is commonly known as a unity gain buffer). The 2134 should be stable in that configuration. ..." [Yes!, that's the ticket = same as removal of R4 & R3]

" ... they are in Non-inverting configuration so yes you would remove Both 10k resistors (R3 and R4) and change R1 and R2 to 100R ..."

" ... google Unity Gain Buffer ..." [ :D ... might also check out some of the many op-amp cookbooks out there ...]

markiemrboo: " ... whipped out the 10k resistor and replaced with a wire link, it's now fine. The oscillations have disappeared .... Turns out I actually don't really need any gain at all. ..." [I assume you mean R3 was "whipped" out = Unity Gain = 1 to 1 gain = no increase in gain, no loss either ...]
 
markiemrboo said:
The text says:

"As shown, gain is 6dB, and to reduce it to 0dB just leave out R4 in each channel.".

R4 in this case is the resistor from the feedback to ground. But this leaves the 10k resistor from the output to -ve feedback input, which didn't seem right. I thought unity gain was just connecting the output straight to the -ve feedback input?

I went ahead and removed R4 anyway, as the text says, but left the feedback resistor in. It does "work".... up until 3/4 volume, but then bursts in to oscillation until I turn it down a smidge.

Would I be correct in thinking that I should infact be removing the 10k feedback resistor too, to get a proper unity gain behaviour? I'm using OPA2134's, so it should be unity gain stable I believe....

Thanks,

Yes... and no...

Typical op-amps operate on voltage feedback. In other words, you feedback a portion of the voltage to the input. Normally, you feedback a small portion (like 1/20th) of the output back into the opamp and the gain happens to be the reciprical of that. So, if you had 1k and 19k feedback resistors, the amount of feedback would be 1/20th of the output and the gain would be 20/1 = 20. Let's say you applied the full output (or in fractions 1/1 of the output) back to the input. Then the gain would be 1/1 = 1 = Unity gain.

So, if you had a straight wire from the output to the input, you would see that the voltage at the input is the same as the voltage at the output and this would lead to the Unity gain situation I just described. Now, opamps are designed to consume virtually no current at their inputs. So, lets say you put a 10k resistor between the output and one of the inputs. Since the input to the opamp consumes no current, you voltage drop across the resistor is V= IR = 0 * 10k = 0. In other words, it acts just like the straight wire.

There are probably other reasons why leaving the 10k resistor might be fine or might not be fine. Bottom line, you can leave it in or take it out and it should be just about the same. All things being equal, Rod says leave it in, may as well listen to him.
--
Danny
 
AX tech editor
Joined 2002
Paid Member
martin clark said:
The only issue with leaving the 10K resistor in is that you've made the -ve signal input at the opamp have a 10K input impedance (opamp draws tenths of microamps, 10K connects to nearly-zero output impedance).

That's enough to make sure that stray signals can couple-in - almost certainly the reason for the spurious oscillation the OP had.


You nailed it. Most probably even a few pF at the -V input (which is in practise always there), together with the 10k gives enough hf phase shift to cause oscillations.

Jan Didden
 
Hi,
some opamps manufacturers recommend that the output to -input resistor be left in place. Some opamps require it (see lm318) to be a minimum value.

The complete circuit is, I believe, a parallel resistor and cap of a few pF to give immunity from oscillation.

I think the purpose of the resistor is to limit the current into the -ve input in pulsed (transient) signal conditions.
 
" ... So, if you had a straight wire from the output to the input, you would see that the voltage at the input is the same as the voltage at the output and this would lead to the Unity gain situation I just described. Now, opamps are designed to consume virtually no current at their inputs. So, lets say you put a 10k resistor between the output and one of the inputs. Since the input to the opamp consumes no current, you voltage drop across the resistor is V= IR = 0 * 10k = 0. In other words, it acts just like the straight wire. ..."

Of course keeping in the back of your mind that any and all real world op-amps make some (modest) distortion contribution ... so the fewer stages of amplification, the better ... sometimes unity gain stages are counterproductive. Daisy chaining op-amps is usually a no no ... unless you want the chain to "latch and hold" or turn every signal into a square wave for some reason ( :eek: ).

" ... some opamps manufacturers recommend that the output to -input resistor be left in place. Some opamps require it (see lm318) to be a minimum value. ..."

Typically this used to be done with a simple pair of 10K resistors, acting as a voltage divider with an op-amp input connected in the middle. (LM318 = man I havn't seen reference to that for a couple of decades except as "virtual" capacitance generator or as a poor boys' voltage regulator = a very old Fairchild original design executed by Nat..)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.