And I probably have better to do than asking questions to "engineers" who think one has to study 5 years to replace an opamp in a 20euro second hand crossover. I'm trying to learn on the field, but I guess I will go waste my time on other forums, such is the snobbery and condescending tone on here.
Enjoy your day.
Well at least 3* people who actually know what they are talking about have told you this is not a good idea. You chose to ignore all of the advice, so you were told straight. I think that is perfectly fair and not condescending. You asked a question and got an answer. The fact that it didn't fit with your world view on op-amp rolling is your problem not theirs!
* and probably 100 years of combined experience between them!
* and probably 100 years of combined experience between them!
I don't recall this particular member to give me any advice on the matter.
Jumping in a post and making that kind of comment is condescending in my book.
Besides I never chose to ignore all of the advice as you said. And I'm fully aware one can encounter problems when "op amp rolling", but I'm also aware it is not ALWAYS rocket science and i managed to use LM4562 in the past without having to get a masters degree first (sigh). There are some practical recipes that just work.
Jumping in a post and making that kind of comment is condescending in my book.
Besides I never chose to ignore all of the advice as you said. And I'm fully aware one can encounter problems when "op amp rolling", but I'm also aware it is not ALWAYS rocket science and i managed to use LM4562 in the past without having to get a masters degree first (sigh). There are some practical recipes that just work.
Last edited:
read post #12 where he points out what he does to prevent problems in LM4562. That is a very practical recipe. I am sorry to say this, but shoving a very fast bipolar op-amp into a circuit designed for a slow FET-input op-amp without checking things carefully is a practical recipe for oscillations. Take the chip off your shoulder and listen to the people who have been there an done it. Op-amps are not universally swappable!
If the chip is hot at idle, you have a problem.
I made a thread about the 4562 oscillating before and I got some great advice that helped me gain confidence in using this chip:
Example of op amp oscillation with pics...
I haven't had a problem in using this chip since this experiment.
I made a thread about the 4562 oscillating before and I got some great advice that helped me gain confidence in using this chip:
Example of op amp oscillation with pics...
I haven't had a problem in using this chip since this experiment.
I think the thing you have to remember when you compare an LM4562 (or any of the newer high gain opamps) to the older types like the Ne5532/4, TL07X are:-
1. Gain bandwidth 10x higher and more
2. Loop gain 100x more or greater (1000x in case of TL07x)
3. Consequently loop bandwidths and noise gain, orders of magnitude greater.
4. In the case of the 4562 types, very low input bias currents of 10nA - this is almost JFET level performance
Layout and decoupling become absolutely critical. Example: a bad layout that allows you to pick up just 10 nV of noise between the feedback summing junction and the inverting input on a unity gain buffer becomes a 100mV noise problem on the output. Bring your hand close to the circuit and capacitive coupling will cause noise on the output. Keep layout very compact and feedback resistors close to the opamp inverting input - SMD resistors help in this regard. Putting your circuit in a metal box will prove shirlding from this type of capacitive coupling (think about SMPS noise from LED lamps for example).
Very high gains on modern opamps don't always translate to the load capacitance drive capabilities the older opamps have. Always isolate the load with a 47 to 100 ohm resister - and especially so if you are driving a cable or other load where you may not have full control over the load.
Decoupling - follow the guidelines closely. Make sure you do not have long traces between the decouple caps and the opamp. I use 1uF 1206 for each rail and place them along the top,edge and bottom edge of the IC package. Cross rail decouple as well - place the cap under the opamp on the copper side. You can get away with sloppy decoupling on older opamps - absolutely not on the LM4562 types.
RFI - can be a big issue - but careful layout, decent L pad and screen to chassis 1nF RF coupling cap will solve the problems. Again, modern opamps have bandwidths that run well into the MHz region with plenty of gain, so if you do not take precautions, you are going to have problems. FET input opamps do have better RFI, but few offer the distortion performance of the LM4562 types i.e. Sub 1ppm.
For general opamp rolling, I'd recommend you stick with the older lower gain types unless you can ensure some of the stuff listed above can be addressed.
1. Gain bandwidth 10x higher and more
2. Loop gain 100x more or greater (1000x in case of TL07x)
3. Consequently loop bandwidths and noise gain, orders of magnitude greater.
4. In the case of the 4562 types, very low input bias currents of 10nA - this is almost JFET level performance
Layout and decoupling become absolutely critical. Example: a bad layout that allows you to pick up just 10 nV of noise between the feedback summing junction and the inverting input on a unity gain buffer becomes a 100mV noise problem on the output. Bring your hand close to the circuit and capacitive coupling will cause noise on the output. Keep layout very compact and feedback resistors close to the opamp inverting input - SMD resistors help in this regard. Putting your circuit in a metal box will prove shirlding from this type of capacitive coupling (think about SMPS noise from LED lamps for example).
Very high gains on modern opamps don't always translate to the load capacitance drive capabilities the older opamps have. Always isolate the load with a 47 to 100 ohm resister - and especially so if you are driving a cable or other load where you may not have full control over the load.
Decoupling - follow the guidelines closely. Make sure you do not have long traces between the decouple caps and the opamp. I use 1uF 1206 for each rail and place them along the top,edge and bottom edge of the IC package. Cross rail decouple as well - place the cap under the opamp on the copper side. You can get away with sloppy decoupling on older opamps - absolutely not on the LM4562 types.
RFI - can be a big issue - but careful layout, decent L pad and screen to chassis 1nF RF coupling cap will solve the problems. Again, modern opamps have bandwidths that run well into the MHz region with plenty of gain, so if you do not take precautions, you are going to have problems. FET input opamps do have better RFI, but few offer the distortion performance of the LM4562 types i.e. Sub 1ppm.
For general opamp rolling, I'd recommend you stick with the older lower gain types unless you can ensure some of the stuff listed above can be addressed.
Last edited:
The graphic below shows the problem more clearly and why layout and the discrete components around the opamp are so critical if you want good performance with modern opamps.
For the demonstration, I used 3 opamps from the LTSpice library - LT1115 (very high OLG OPA as a proxy for the LM4562), LT1055 (sort of souped-up TLO7X type JFET OPA) and a low OLG micro power opamp - LT1112) that in OLG terms is a reasonable proxy for the NE5532/4 and a TLO7X type OPA's.
At 100 KHz, you still have 60 dB of noise gain on the LT1115, about 35 dB on the LT1055 and only about 15 dB on the LT1112.
If you capacitively couple the voltage (noise) sources to the non-inverting input with a 10pF cap - not an unreasonable value in a typical layout using leaded components - the output on the LT1112 peaks at out 0dB at 300kHz and then drops off, 14 dB at 800 kHz on the LT1057 while for the LT1115 is peaks at 2MHz and ~24dB gain
You would not notice any noise problems with the first opamp with a sloppy layout. You would have serious noise issues with the LT1115. Many circuits that have problems show lots of mains noise on the output. This is usually caused by capacitively coupled HF noise being modulated by mains EM field noise.
If you want the best out of modern OPA's, you have apply them carefully and think about the layout and the physics (L, C, R) involved.
(Ignore the diodes across the opamp inputs - they make no difference to the results)
For the demonstration, I used 3 opamps from the LTSpice library - LT1115 (very high OLG OPA as a proxy for the LM4562), LT1055 (sort of souped-up TLO7X type JFET OPA) and a low OLG micro power opamp - LT1112) that in OLG terms is a reasonable proxy for the NE5532/4 and a TLO7X type OPA's.
At 100 KHz, you still have 60 dB of noise gain on the LT1115, about 35 dB on the LT1055 and only about 15 dB on the LT1112.
If you capacitively couple the voltage (noise) sources to the non-inverting input with a 10pF cap - not an unreasonable value in a typical layout using leaded components - the output on the LT1112 peaks at out 0dB at 300kHz and then drops off, 14 dB at 800 kHz on the LT1057 while for the LT1115 is peaks at 2MHz and ~24dB gain
You would not notice any noise problems with the first opamp with a sloppy layout. You would have serious noise issues with the LT1115. Many circuits that have problems show lots of mains noise on the output. This is usually caused by capacitively coupled HF noise being modulated by mains EM field noise.
If you want the best out of modern OPA's, you have apply them carefully and think about the layout and the physics (L, C, R) involved.
(Ignore the diodes across the opamp inputs - they make no difference to the results)
Attachments
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.