Legend has it that the reason Moody Blues recordings late 1960s-early 1970s had a muffled muddy sound because of the bad slew rate of the op-amps in studio mixing consoles, and perhaps not because the producer wanted a concert hall effect. when you listen to "Nights in White Satin" you especially notice the drums lack impact. Though these recordings are well loved I wonder how everything would have sounded if they had used vacuum tube equipment...
That's a great little test rig you've set up there. You've inspired me to make my own for sorting and matching my op amps. I usually only buy from digikey and sometimes mouser, but 😳 damn, seeing the nasty slew rate on those chinesium parts really drives the point home! I have yet to be blessed with knock offs, but if I see any in my digikey stock, I'll have a few choice words!
- Opamp slew rate measurement should be done with unity gain (Av=1), and not Av=-10. - Output should be loaded with 2kOhms resistor parallel with 100pF capacitor. - Close to the DUT you should have 4.7uF capacitors between ground and both rails. - Unused opamp should be terminated. You should show the reference input signal, and not only the trace from the output.
Those are interesting points. Do they only matter (or almost only?) when trying to measure "absolute slew rate"? IOW, since the video was mostly focusing on comparing slew rates between different parts, is the relative comparison he's making still valid?
@@largepimping It does matter. If you have a much larger capacitance, for example, you end up measuring the output impedance of the opamp, not its slew rate.
So, did you have the resistor AND capacitor go to ground AND\BUT did you load-up\include a 100Ω-180Ω resistor FIRST, at the OP-AMP OUTPUT, before the 2kΩ resistor AND capacitor? AND most OP-AMP manufacturer schematics show ecommend a min. of 47uf capacitor, followed by a 0.1uf capacitor and then the IC +\- pins. Why did you choose a miniscule 4.7uf capacitor instead? Thank you for your time.
@@christophero1969 Well of course you have to have the resistor AND capacitor go to ground, otherwise they wouldn't be a LOAD and IN PARALLEL, would they? And no, you wouldn't include a 100R resistor at the opamp output, because that is ONLY necessary to STABILISE THE LOOP by improving the phase margin when driving large capacitive loads. 100pF is NOT a large capacitive load, but you might use an isolation resistor, although it is by no means necessary. If you look at the DATASHEET for a common opamp like the LM358 by TEXAS INSTRUMENTS (e.g. SNOSBT3I -JANUARY 2000-REVISED DECEMBER 2014), not one of the example circuits shown for testing parameters indicate the need for a 47μF capacitor. The layout guidelines require the supply to be bypassed by "a low ESR capacitor". I certainly wouldn't use 47μF for that because of its parasitic inductance which will become significant at higher frequencies. I think you'll find a low ESR capacitor of between 100nF and 1μF is usually used close to the supply pins as is also common practice in digital circuits. A larger electrolytic capacitor can then be present somewhere on the supply lines and its value and characteristics would then no longer be critical.
In this context, linear is irrelevant. Slew rate is the current limiting charging the compensation capacitor; constant I=C dV/dt makes a constant slope in V(t).
The TI, ST, and Diodes datasheets for the LM358 specify 0.3 Volts per microsecond, so that one with over 10V/microsecond might be a fake (a reject of some faster op-amp?). The readings for the TL072 and LM4562 were also surprising. The LM4562's datasheet specifies 20 V/microsecond typical, minimum 15--although that is at unity gain. The 5532 was designed for audio mixers and amplifiers so equivalent input noise and harmonic distortion in the audio band were the key parameters, with phase linearity and bandwidth/slew rate as important secondaries. Most likely all the recorded music and film audio you hear has been through a few dozen 5532s by the time it gets to you.The LM4562 is "a better 5532"--it "only" took 30 years to improve on the 5532! The 358 has horrendous distortion caused by a poor output stage design and it's noisy. It's never used for audio, which seems to be the common theme with the other parts. The TL072 is best described as "OK for undemanding applications". It was used for a few years in mixers before the 5532 arrived, and it's apparently still used in some low-end gear. I'm less familiar with the Japan Radio parts and the LF353. I think the latter's main claim to fame (back in the day) was that it could be made to work quite well as a comparator, back when comparator choices were rather limited. Other JFET-input op-amps like the TL082 and TL072 were no good for that--I forget why.
JFET-input op-amps generally don't make good comparators and you only have to consider what happens when the gate junction gets forward biased to see why.
You shouldn't put much store in the non-linearity of the edges that you were observing. Some of the ramps clearly resembled the charge curve for a capacitor though a resistor - and that's exactly what they were. You were observing the effect of a significant output impedance (often the current limiting circuit) charging all the stray capacitances connected to the opamp output on your prototype board. You may find that a cleaner layout with less capacitance on the output and improved supply decoupling could improve those non-linear edges noticeably.
My two current favorites are the dual LME49720 still available in DIP but I think TI is obsoleting it. Faster and quieter than the NE5532 but a little pricey too. And the OPA1611 single but can't drive as heavy a load and only in surface mount. I solder them to so-8 to Dip adapters.
Hi, what's the supply-rail? If you are swinging close to the rail, extra delays may be expected. Also the 1v 'source' wave would be handy to see simultaneously, esp. as a 'trigger' source. Furthermore - you use the term "non-linear" (literally: not a straight line, which is valid), however: a "linear" system (from signals and systems) is defined differently, whereby an overshoot or an exponential rise would be perfectly acceptable as 'linear'. A slew, on the other hand, definitely is not. The most basic 'razor' is that you should not get any frequency out that you did not put in; each frequency may individually have a different gain. An RC-lowpass is, for example, a 'linear' system.
My Texas Instruments datasheet says about the LM358: "Slew rate at unity gain - 0.3 V/μs". With the NE5532 (TI datasheet), on the other hand, the slew rate is specified as 9 V/μs. Very interesting, what is shown here.
Sometimes I have the impression that such old parts are no longer originally manufactured, but the old designations are stamped on a more modern successor with better values. Maybe the newer manufacturing processes are a bit cheaper too?
having 2 much fun! happy day! good info on amps! Texas Instruments still has the dominant amps, in my opinion! but 4 under 20 dollar builds, this is the place!! do more, do more!! lookout! that's some super comments below!! I wish ya'll could make a you tube video about all that stuff, in the comments!! OP-AMP encyclopedia!!
In my mind, slew rate tells you how fast an op amp is. Also bandwidth tells me if an op amp is faster or slower than another one. But bandwidth and slew rate can't simpley be inverses of each other... What is the difference between bandwidth and slew rate?
Bandwidth is the maximum frequency that you can use for small signals. Slew rate determines the maximum usable frequency for large output signals and will be considerably less than the bandwidth, as well as being inversely proportional to the voltage swing at the output.
thank you for all videos! and thank you for this triger function! you one from all what i'm saw before who show this!
And nice that you’re multitasking too, with your refinements to the train whistle effect just faintly audible around 4:38 and on.
Legend has it that the reason Moody Blues recordings late 1960s-early 1970s had a muffled muddy sound because of the bad slew rate of the op-amps in studio mixing consoles, and perhaps not because the producer wanted a concert hall effect. when you listen to "Nights in White Satin" you especially notice the drums lack impact. Though these recordings are well loved I wonder how everything would have sounded if they had used vacuum tube equipment...
That's a great little test rig you've set up there. You've inspired me to make my own for sorting and matching my op amps. I usually only buy from digikey and sometimes mouser, but 😳 damn, seeing the nasty slew rate on those chinesium parts really drives the point home! I have yet to be blessed with knock offs, but if I see any in my digikey stock, I'll have a few choice words!
- Opamp slew rate measurement should be done with unity gain (Av=1), and not Av=-10.
- Output should be loaded with 2kOhms resistor parallel with 100pF capacitor.
- Close to the DUT you should have 4.7uF capacitors between ground and both rails.
- Unused opamp should be terminated.
You should show the reference input signal, and not only the trace from the output.
Those are interesting points. Do they only matter (or almost only?) when trying to measure "absolute slew rate"? IOW, since the video was mostly focusing on comparing slew rates between different parts, is the relative comparison he's making still valid?
@@largepimping It does matter. If you have a much larger capacitance, for example, you end up measuring the output impedance of the opamp, not its slew rate.
So, did you have the resistor AND capacitor go to ground AND\BUT did you load-up\include a 100Ω-180Ω resistor FIRST, at the OP-AMP OUTPUT, before the 2kΩ resistor AND capacitor? AND most OP-AMP manufacturer schematics show
ecommend a min. of 47uf capacitor, followed by a 0.1uf capacitor and then the IC +\- pins. Why did you choose a miniscule 4.7uf capacitor instead? Thank you for your time.
@@christophero1969 Well of course you have to have the resistor AND capacitor go to ground, otherwise they wouldn't be a LOAD and IN PARALLEL, would they?
And no, you wouldn't include a 100R resistor at the opamp output, because that is ONLY necessary to STABILISE THE LOOP by improving the phase margin when driving large capacitive loads. 100pF is NOT a large capacitive load, but you might use an isolation resistor, although it is by no means necessary.
If you look at the DATASHEET for a common opamp like the LM358 by TEXAS INSTRUMENTS (e.g. SNOSBT3I -JANUARY 2000-REVISED DECEMBER 2014), not one of the example circuits shown for testing parameters indicate the need for a 47μF capacitor. The layout guidelines require the supply to be bypassed by "a low ESR capacitor". I certainly wouldn't use 47μF for that because of its parasitic inductance which will become significant at higher frequencies. I think you'll find a low ESR capacitor of between 100nF and 1μF is usually used close to the supply pins as is also common practice in digital circuits. A larger electrolytic capacitor can then be present somewhere on the supply lines and its value and characteristics would then no longer be critical.
In this context, linear is irrelevant. Slew rate is the current limiting charging the compensation capacitor; constant I=C dV/dt makes a constant slope in V(t).
The TI, ST, and Diodes datasheets for the LM358 specify 0.3 Volts per microsecond, so that one with over 10V/microsecond might be a fake (a reject of some faster op-amp?).
The readings for the TL072 and LM4562 were also surprising. The LM4562's datasheet specifies 20 V/microsecond typical, minimum 15--although that is at unity gain.
The 5532 was designed for audio mixers and amplifiers so equivalent input noise and harmonic distortion in the audio band were the key parameters, with phase linearity and bandwidth/slew rate as important secondaries. Most likely all the recorded music and film audio you hear has been through a few dozen 5532s by the time it gets to you.The LM4562 is "a better 5532"--it "only" took 30 years to improve on the 5532!
The 358 has horrendous distortion caused by a poor output stage design and it's noisy. It's never used for audio, which seems to be the common theme with the other parts. The TL072 is best described as "OK for undemanding applications". It was used for a few years in mixers before the 5532 arrived, and it's apparently still used in some low-end gear.
I'm less familiar with the Japan Radio parts and the LF353. I think the latter's main claim to fame (back in the day) was that it could be made to work quite well as a comparator, back when comparator choices were rather limited. Other JFET-input op-amps like the TL082 and TL072 were no good for that--I forget why.
Greg, i think you are right.
JFET-input op-amps generally don't make good comparators and you only have to consider what happens when the gate junction gets forward biased to see why.
You shouldn't put much store in the non-linearity of the edges that you were observing. Some of the ramps clearly resembled the charge curve for a capacitor though a resistor - and that's exactly what they were. You were observing the effect of a significant output impedance (often the current limiting circuit) charging all the stray capacitances connected to the opamp output on your prototype board. You may find that a cleaner layout with less capacitance on the output and improved supply decoupling could improve those non-linear edges noticeably.
My two current favorites are the dual LME49720 still available in DIP but I think TI is obsoleting it. Faster and quieter than the NE5532 but a little pricey too. And the OPA1611 single but can't drive as heavy a load and only in surface mount. I solder them to so-8 to Dip adapters.
Hi, what's the supply-rail? If you are swinging close to the rail, extra delays may be expected. Also the 1v 'source' wave would be handy to see simultaneously, esp. as a 'trigger' source.
Furthermore - you use the term "non-linear" (literally: not a straight line, which is valid), however: a "linear" system (from signals and systems) is defined differently, whereby an overshoot or an exponential rise would be perfectly acceptable as 'linear'. A slew, on the other hand, definitely is not. The most basic 'razor' is that you should not get any frequency out that you did not put in; each frequency may individually have a different gain. An RC-lowpass is, for example, a 'linear' system.
+/-15 rails
That is exorbitant faster then the manufacture said.
At least AN6551, LM833 and M5218 are still available out there.
I’m curious now about how you wired this up. Do you have a schematic of how your setting it up?
Lm358 should have a slew rate of 0.5 V/us. What is it that you have there?
My Texas Instruments datasheet says about the LM358: "Slew rate at unity gain - 0.3 V/μs". With the NE5532 (TI datasheet), on the other hand, the slew rate is specified as 9 V/μs. Very interesting, what is shown here.
Sometimes I have the impression that such old parts are no longer originally manufactured, but the old designations are stamped on a more modern successor with better values. Maybe the newer manufacturing processes are a bit cheaper too?
Where is the OP7?
having 2 much fun! happy day! good info on amps! Texas Instruments still has the dominant amps, in my opinion! but 4 under 20 dollar builds, this is the place!! do more, do more!!
lookout! that's some super comments below!! I wish ya'll could make a you tube video about all that stuff, in the comments!! OP-AMP encyclopedia!!
In my mind, slew rate tells you how fast an op amp is. Also bandwidth tells me if an op amp is faster or slower than another one. But bandwidth and slew rate can't simpley be inverses of each other... What is the difference between bandwidth and slew rate?
Bandwidth is the maximum frequency that you can use for small signals. Slew rate determines the maximum usable frequency for large output signals and will be considerably less than the bandwidth, as well as being inversely proportional to the voltage swing at the output.
Yes, determinant characteristics for applications like Analog Computers
You must have got a fake lm358😂
CA3140 and CA3240 yay..... Urm only me I guess.
Wow