I am not sure if this is the right forum for this issue but if not please redirect me.
I made a simple unity gain buffer with 5 parts for an interface between dut’s with a few hundred ohms output impedance and an e1da cosmos adc which also happens to have low impedance (350 ohms in mono mode @2.7v setting). So I wanted a buffer with high input impedance and low output impedance for the obvious reasons.
This circuit uses just an lm4562 with + input to ground by a 100k ohm resistor. Output has a 10 ohm series resistor.
There are 2 10uf caps as bypass from V+ to ground and V- to ground. It is unity gain so no resistor in the negative feedback loop.
That’s the whole thing. Using 2 9v batteries to power it all is good. However when I try powering it with a +-15v bench supply or a SilentSwitcher(+-15v) with no source or load the output and of course the - in tied to it got to -14.xxx v.
One even overheated when I accidentally left it on.
I am a 70 year old noob but If the op amp can’t make both inputs equal I think it saturates and goes to one of the rails but why is it doing this? The Lm4562 has power supply max at +-17v. It’s not rail to rail but +-14v swing. But why is it needing to do more than that? Sorry for the dumb question. Just trying to learn. Thx for any help.
I made a simple unity gain buffer with 5 parts for an interface between dut’s with a few hundred ohms output impedance and an e1da cosmos adc which also happens to have low impedance (350 ohms in mono mode @2.7v setting). So I wanted a buffer with high input impedance and low output impedance for the obvious reasons.
This circuit uses just an lm4562 with + input to ground by a 100k ohm resistor. Output has a 10 ohm series resistor.
There are 2 10uf caps as bypass from V+ to ground and V- to ground. It is unity gain so no resistor in the negative feedback loop.
That’s the whole thing. Using 2 9v batteries to power it all is good. However when I try powering it with a +-15v bench supply or a SilentSwitcher(+-15v) with no source or load the output and of course the - in tied to it got to -14.xxx v.
One even overheated when I accidentally left it on.
I am a 70 year old noob but If the op amp can’t make both inputs equal I think it saturates and goes to one of the rails but why is it doing this? The Lm4562 has power supply max at +-17v. It’s not rail to rail but +-14v swing. But why is it needing to do more than that? Sorry for the dumb question. Just trying to learn. Thx for any help.
There are a few possibilities...
1/ Could the LM4562 be a fake device? Only ever buy semiconductors from recognised outlets.
2/ The LM4562 might be oscillating. That can be because of layout problems.
3/ The LM4562 might need some RF filtering at the input.
It might help to see a picture of what you actually have
Also try a different opamp like a jelly bean 4558 just to see if the circuit then works correctly.
1/ Could the LM4562 be a fake device? Only ever buy semiconductors from recognised outlets.
2/ The LM4562 might be oscillating. That can be because of layout problems.
3/ The LM4562 might need some RF filtering at the input.
It might help to see a picture of what you actually have
Also try a different opamp like a jelly bean 4558 just to see if the circuit then works correctly.
I once had a problem with an oscillating NE5532. The IC was legit in a batch from Mouser and it was part of a proven design. It should not oscillate but it did. It was mounted in a DIL-socket, I replaced it with another NE5532 from the same batch and the oscillation was gone. Was it a bad IC or was it badly mounted in the socket? I still don't know. Lesson learned, always test your circuitry, preferably with a scope.
For a proven "cookie cutter" design I usually turn to Rod Elliots pages. Nick the pieces you need like bypassing the op-amp supplies and make sure your op-amp is unity gain stable. Rods article series Designing With Opamps got a lot of practical hints and tips.
For a proven "cookie cutter" design I usually turn to Rod Elliots pages. Nick the pieces you need like bypassing the op-amp supplies and make sure your op-amp is unity gain stable. Rods article series Designing With Opamps got a lot of practical hints and tips.
Modern (fast, high gain) opamps require high speed decoupling, the 10µF electrolytics are not able to provide this (too low a self-resonant frequency), but a 100nF ceramic capacitor from V+ to V- directly at the chip's pins will give this required decoupling. Its may be oscillating violently at HF with the higher supply voltage and saturating its output.
(A tip for detecting oscillation if you don't have a 'scope: monitor the supply current - this will jump up in value if oscillations are happening.)
The LM4562 in unity-gain follower mode is best done with 100 ohm resistor as feedback rather than a direct wire/trace, note (I forget why, Doug Self explains this somewhere in one of his books).
High gain high bandwidth devices always tend to be quite sensitive to layout, you might get some improvement from tightening up the layout (shorter leads and wires).
(A tip for detecting oscillation if you don't have a 'scope: monitor the supply current - this will jump up in value if oscillations are happening.)
The LM4562 in unity-gain follower mode is best done with 100 ohm resistor as feedback rather than a direct wire/trace, note (I forget why, Doug Self explains this somewhere in one of his books).
High gain high bandwidth devices always tend to be quite sensitive to layout, you might get some improvement from tightening up the layout (shorter leads and wires).
I believe it is currently built on one of these so layout issues are a real possibility:
D Self actually says that the resistor should be no more that 1 kohm and it was accompanied by one more resistor in series with the + input.He called it common mode distortion cancellation resistor.Nakamichi CR4 used 100k ohm resistors in its headphones amplifier on njm4556 which has much lower input bias current. NAK CR4 was the only good deck to have a good sounding headphones amp as all the others used class b amps driven by same njm4556.Apparently this technique might not be recommended with op amps that employ internal input bias current cancellation same as when using a resistor to ground from +input in the inverting circuits can completely unbalance op-amps using internal bias current cancelation techniques.I used 1 kohms resistors in unity gain followers with lm4562 in my cd players and they sounded just fine.
))))))) Best joke till 5 pm today!This is one of those rare moments where I regret suggesting that likes wouldn't be useful features on diyaudio...
@Mooly I terminated the unused side by connecting in+ to ground and connected out to in- per everything I have read.
I want to thank everyone for their comments and apologize that there were stupidly some errors on my part in how good or bad connections on the breadboard were. I started over and the circuit works correctly at all voltages. I am only losing .5db in thd+n running through the buffer compared
to direct connect to the e1da cosmos adc.
Again my apologies.
I want to thank everyone for their comments and apologize that there were stupidly some errors on my part in how good or bad connections on the breadboard were. I started over and the circuit works correctly at all voltages. I am only losing .5db in thd+n running through the buffer compared
to direct connect to the e1da cosmos adc.
Again my apologies.
Well after more testing I declared victory too soon. So it is true that with no input or output load it does not overheat at +-9v( batteries I was using to test). Also the output does not saturate to the rails. Even above 9 v with no loads it has periods of not overheating, but going up in increments on the bench supply I can usually make it overheat. Also if the supply is above 9v and it happens not to be overheating, it will when I start to run a 2v sine wave through it.
I have run it for 2 hours at 6-9v with no problem and the thd+n is almost identical to the direct connection.
It didn’t appear to be changing output voltage or dc( about .05mvdc at the output) when it started to overheat.
I will try looking at the fft on the scope tomorrow.
It must be oscillating as so many had suggested, right?
I also tried 100 then 1k ohm resistors in the feedback to no effect. Also I tried 100nf caps across ground to each of the op amp supply pins instead of the 10uf. No improvement.
What might be happening to cause oscillation above 9v supply? Literally running at 9v it’s fine for hours. When I switch to 10v it immediately starts overheating( like I can’t keep my finger on it for more than 15 seconds).
Could anything besides oscillation cause the overheating? Thx for your support!
I have run it for 2 hours at 6-9v with no problem and the thd+n is almost identical to the direct connection.
It didn’t appear to be changing output voltage or dc( about .05mvdc at the output) when it started to overheat.
I will try looking at the fft on the scope tomorrow.
It must be oscillating as so many had suggested, right?
I also tried 100 then 1k ohm resistors in the feedback to no effect. Also I tried 100nf caps across ground to each of the op amp supply pins instead of the 10uf. No improvement.
What might be happening to cause oscillation above 9v supply? Literally running at 9v it’s fine for hours. When I switch to 10v it immediately starts overheating( like I can’t keep my finger on it for more than 15 seconds).
Could anything besides oscillation cause the overheating? Thx for your support!
I usually use option C for bypassing on a PCB (source from Rod Elliotts Designing With Opamps - Part 1, Figure 2 - Bypassing The Opamp Supplies). One pair of 10uF per board and one 100nF between the supplies of each opamp (or opamp package). Make sure the 100nF is mounted as close as possible to the supply pins.
And to quote Rod:
And to quote Rod:
Note on input noise: if you need such a high input impedance (100k), a FET input opamp is preferable. With 1Pa/wHz input current noise on a 100k resistor you get an input noise voltage of 100nV/wHz, which is very high, much higher than the input voltage noise from the opamp (4.7nV/wHz). With something like OPA2134 noise should improve by 20dB.