That's where balanced inputs are handy: slap a reasonable resistor on each line, say 100 ohms [1] and a 1000 pf capacitor between the two lines after the resistors. The RCR makes a dandy lowpass filter with a -3 dB corner at 800 kHz and the resistor effectively isolates the driving component's output from the capacitor. The values aren't incredibly important so long as the resistors are equal to avoid messing up CMRR.
[1] for extra credit figure out the cable's characteristic impedance and match the resistors to reduce RF standing waves.
My LME49720 experiences:
EMI susceptibility: https://www.proaudiodesignforum.com/forum/php/viewtopic.php?p=11071
Start-Up behavior: https://www.proaudiodesignforum.com/forum/php/viewtopic.php?t=641
LME49720 burst noise I measured and posted at TI e2E:
I've also tried them in my ULDO-Nacho Oscillator: The NJM5532 outperforms the LME49720 when heavily loaded.
I haven't bought any in several years - perhaps TI has fixed some of the process issues WRT noise but I doubt the EMI problems have been solved.
I avoid these parts.
EMI susceptibility: https://www.proaudiodesignforum.com/forum/php/viewtopic.php?p=11071
Start-Up behavior: https://www.proaudiodesignforum.com/forum/php/viewtopic.php?t=641
LME49720 burst noise I measured and posted at TI e2E:
I've also tried them in my ULDO-Nacho Oscillator: The NJM5532 outperforms the LME49720 when heavily loaded.
I haven't bought any in several years - perhaps TI has fixed some of the process issues WRT noise but I doubt the EMI problems have been solved.
I avoid these parts.
Mark, I do agree with the statement as such, but in practise that's very hard to do.
You can't just shift the ol response to the right; there's a reason it is what it is.
What you see is that the designers attempt to increase the ol gain at the same roll off frequency so the ol gain is higher across the band.
Many naive audio guys increase the bandwidth by for instance loading a Vas stage heavily. That squashes the ol gain above say 20kHz which gives a 20kHz flat bw but all that does is throw away valuable loop gain above 20kHz.
https://audioxpress.com/article/audio-myths-why-narrow-bandwidth-may-be-better
Jan
Jan, I posted my graphs in reply to post #12, which talked about increasing bandwidth by purchasing a different opamp. For example: instead of an OPA134 (GBWP = 8 MHz), use an LME49710 (GBWP = 55 MHz).
Bandwidth is generally a small-signal spec while slew rate comes into play at large signals.
They are somewhat related, but not 1:1. You can have a low bw opamp with ample slew rate in the audio band.
If your voltage amplifying stage has enough bias current to quickly charge/discharge the comp capacitor, you get high slew rate.
But it doesn't automagically incxrease your bandwidth, that can be limited elsewhere.
So it's hard to give a one-size-fits-all answer.
Jan
They are somewhat related, but not 1:1. You can have a low bw opamp with ample slew rate in the audio band.
If your voltage amplifying stage has enough bias current to quickly charge/discharge the comp capacitor, you get high slew rate.
But it doesn't automagically incxrease your bandwidth, that can be limited elsewhere.
So it's hard to give a one-size-fits-all answer.
Jan
They often put a resistor from the Vas output node to ground and keep decreasing the resistor value until they get the desired closed loop gain or bandwidth at whatever frequency they fancy. Such a waste of fine loop gain that could be used to squash distortion below that magic frequency point.
Jan
Edit whoops I should have added not directly related but........
https://www.edn.com/rule-of-thumb-1-bandwidth-of-a-signal-from-its-rise-time/
Derived via Fourier to obtain a sine of some frequency so not exactly a rule of thumb. As it needs an infinite rise to time to test mercury wetted reed relays have been used as a source, the nearest thing around to obtain that. 10 to 90% due to damping which will round corners. I suspect that no damping is the point where the response remains flat,
Yes, rise time and bandwidth are directly related, as long as you stay in the small signal regime.
As you move towards larger signal levels, rise time (but not necessarily) can start to morph into slew rate limiting.
Jan
As you move towards larger signal levels, rise time (but not necessarily) can start to morph into slew rate limiting.
Jan
Surely rise time is slew limited by definition. rise time = step voltage / slew rate, and is dependent on signal amplitude, and is thus a non-linear phenomenon, caused principally by current-clipping in part of the amp that is capacitively loaded.
Gain bandwidth product is theoretically independent of amplitude, its part of the linear behaviour, set by the poles and zeroes in the circuit and devices. Of course the real world is messier that this and the poles and zeroes move around a bit when loading increases.
Or put simply measure slew rate with square waves, bandwidth with sine waves. Measure slew rate first to check what amplitude of sine waves you can use for gain measurements.
I’d argue rise time is the fastest slope a VFA can handle and remain within its linear operating region. Go beyond that and you have slewing distortion of some kind. If you are doing this with a significant input signal stimulus, you will see a dramatic increase in distortion if the amplifier input is not BW limited. I think Jan’s points are important here as well since rise time is almost always specified in the small signal regime (1-2 V pk-pk) while slew rate is a large signal phenomena.
A good test to do in a sim is to feed the amp input with an improbably fast rise/fall time stimulus signal and then look at how the input stage behaves - this works well for both VFA and CFA and is really useful in helping to set the input BE limiting filter.
A good test to do in a sim is to feed the amp input with an improbably fast rise/fall time stimulus signal and then look at how the input stage behaves - this works well for both VFA and CFA and is really useful in helping to set the input BE limiting filter.
Craig Sawyers, I think these are the tests.
Quite a few of the NS uber OPAs use the same topology and may have the same problems. IIRC I first encountered the start up stuff with LM4562 used by my best mate in the UK who still runs an electronic company.
"Slewing" occurs when an input transistor's current falls below 1%, or rises above 99%, of its DC value with no input signal. As James Solomon, Bob Cordell, and Douglas Self all warn: you might need to apply a surprisingly fast-rising AND large amplitude square wave to guarantee slewing. Several JFET-input amps I've designed, require more than ±3.5 volts of overderive {a/k/a more than five Vbe's !!} to guarantee slewing. Solomon (December 1974) derives the moderately elaborate EE mathematics to explain why.
"Bandwidth is generally a small-signal spec while slew rate comes into play at large signals."
"...rise time is almost always specified in the small signal regime (1-2 V pk-pk) while slew rate is a large signal phenomena."
But the fact is that MUSIC is inherently BOTH small-signal and large-signal at the same time. A 1-volt rms music piece has 10-volt peaks occurring simultaneously.
"...rise time is almost always specified in the small signal regime (1-2 V pk-pk) while slew rate is a large signal phenomena."
But the fact is that MUSIC is inherently BOTH small-signal and large-signal at the same time. A 1-volt rms music piece has 10-volt peaks occurring simultaneously.
Idon't understand your point. How can something be 1V and 10V simultanbeaously?
Jan
The distortion plot posted earlier was distortion plus noise so it would be expected to fall with increased signal output with increased drive. Maybe they give the test circuit.
The max slew rate of a sine wave is given as 2 x Π x f x V volts per second. High V higher slew rated needed. Peak usually mentioned also demonstration of the effect on a square wave that relates to the rise time bandwidth aspect as well so there are 2 effects. Does the slew rate indicate the bandwidth? Just something I'm curious about but not looked at. 😉 I sort of lost interest in amps long ago as I lacked the gear to check performance. People who had circuits published had the gear - to varying extents.
Preamp peak voltages result in rather low slew rates suggesting noise is more important In the past fairly decent designs have even been based on 741's.
The only design I have seen using a high end IC with details used a TLE2141 to directly drive an output stage. Rather large voltage swing. It makes the point that the IC supply needs decoupling rather close to it. That in turn means a very solid ground connection as well. He mentions star earthing. I suppose ideally with short tracks. This aspect has already been mentioned - decoupling. This circuit used 220uf in parallel with 330nf but the chip is intended to drive a highish capacity load. End result a pretty decent 20khz square wave into a pure 4ohm load with a peak of 20v. The only value he showed. 50w amp. +/-22v supply. Slew rate of amp >20v/usec.
The max slew rate of a sine wave is given as 2 x Π x f x V volts per second. High V higher slew rated needed. Peak usually mentioned also demonstration of the effect on a square wave that relates to the rise time bandwidth aspect as well so there are 2 effects. Does the slew rate indicate the bandwidth? Just something I'm curious about but not looked at. 😉 I sort of lost interest in amps long ago as I lacked the gear to check performance. People who had circuits published had the gear - to varying extents.
Preamp peak voltages result in rather low slew rates suggesting noise is more important In the past fairly decent designs have even been based on 741's.
The only design I have seen using a high end IC with details used a TLE2141 to directly drive an output stage. Rather large voltage swing. It makes the point that the IC supply needs decoupling rather close to it. That in turn means a very solid ground connection as well. He mentions star earthing. I suppose ideally with short tracks. This aspect has already been mentioned - decoupling. This circuit used 220uf in parallel with 330nf but the chip is intended to drive a highish capacity load. End result a pretty decent 20khz square wave into a pure 4ohm load with a peak of 20v. The only value he showed. 50w amp. +/-22v supply. Slew rate of amp >20v/usec.
"In the past fairly decent designs have even been based on 741's."
On what planet in what universe is that true?
On what planet in what universe is that true?
You can see the onset slewing as the stimulus frequency is increased. Eventually all of the tail (constant) current is flowing into the comp cap (essentially the 1% and 99% you mention) and you get the characteristic triangular output waveform. Luckily, you don't usually see this in the small signal regime.
Input BW limiting and diff-pair degeneration can be very effective in managing the onset of slewing, as well as the diff-pair tail current. However, for IC opamps, these approaches require some serious tradeoffs re noise and current consumption.