It competes well against the 741 for sure, but the phase-reversal issue and high distortion go against it. Early datasheets tout it as high bandwidth and low distortion, because this was before the really performant opamps had been designed and 3MHz was high bandwidth and 0.02% was low distortion...
The TL072 overtook the LF411 in audio gear due to the ~20dB lower distortion. I don't think it recovered from this.
These days you can get yourself for instance an OPA1652 if you want a good FET opamp that proves that old opamps are definitely not the best. It has protection against phase-reversal which is a bain in the old JFET amps. And the OPA1652 datasheet is way better than those for old devices - so many more parameters graphed and characterized.
So in what instance creates said phase reversal? It won't latch up. So what is the worst that can happen? I know I preffered it over the TL072 and 5534 in my listening tests and my oscilloscope liked it better too.
If you're happy, you're happy. Who cares what other people say? Who cares what other people measure? If you're happy, you're happy.
Sorry, no this is incorrect. The input and output impedance of the op-amps is insignificant compared to the impedance of a typical filter network, as seen at the op-amp's inputs. Typical crossover networks, using sallen-key filters and whatnot, tend to run somewhere in the 5-50K range. Bipolar inputs are much lower impedance than FET, but it's still tiny compared to the network.
But yeah, as mentioned, your biggest worry is the high GBW of modern op-amps. The board needs to be well-designed, with good power delivery, grounding, and bypass arrangement. Sometimes it helps to slow the op-amps down a little with a 50-200pf cap from output to inverting input, or to R-C limit the bandwidth to the input.
Feed it square waves, before and after and look for ringing.
But yeah, as mentioned, your biggest worry is the high GBW of modern op-amps. The board needs to be well-designed, with good power delivery, grounding, and bypass arrangement. Sometimes it helps to slow the op-amps down a little with a 50-200pf cap from output to inverting input, or to R-C limit the bandwidth to the input.
Feed it square waves, before and after and look for ringing.
I'm having trouble reconciling the post above with Douglass Self's book "The Design of Active Crossovers" and in particular the section "opamp tradeoffs in crossover design".
Here's the schematic
I changed out the first and last opamp with very small differences. However, 5532s in the last stage caused .5V momentary DC on the outputs creating a big THUMP and power down. LM4562 did almost the same thing. I put back the originals.
So far only the first and second stage opamps were changed. I can't determine a sonic difference by ear.
I changed out the first and last opamp with very small differences. However, 5532s in the last stage caused .5V momentary DC on the outputs creating a big THUMP and power down. LM4562 did almost the same thing. I put back the originals.
So far only the first and second stage opamps were changed. I can't determine a sonic difference by ear.
Your thump is probably from the fact that the 5532 and LM4562 are bipolar inputs, and have much higher input bias current, causing a DC offset. TL072 is typical 1pA, and LM4562 is 10nA, four orders of magnitude. I think you'll find a significant DC offset at your output. There doesn't appear to be any output coupling caps on this design, so your amplifier must be AC-coupled or you'll get a DC offset at the output of the amp.
Self's appraisals of opamps for crossovers runs to 10 pages or so, before even getting in to specific opamps. The gist of it is that because of different noise generating mechanisms of BJT and FET input opamps, in a circuit is optimised for one type of opamp, say low voltage noise, high current noise, the performance could be quite poor if the other type of opamp, with high voltage noise but low current noise, was substituted.
Whatever ever the approach the designer used to determine impedance levels around the opamp in each stage, it would be wise to stick to the same type of opamp so as not to inadvertently introduce a degradation in performance.
TI has an online paper discussing the tradeoffs though not specifically for crossover circuits. Trade-offs Between CMOS, JFET, and Bipolar Input Stage Technology
Whatever ever the approach the designer used to determine impedance levels around the opamp in each stage, it would be wise to stick to the same type of opamp so as not to inadvertently introduce a degradation in performance.
TI has an online paper discussing the tradeoffs though not specifically for crossover circuits. Trade-offs Between CMOS, JFET, and Bipolar Input Stage Technology
Oh, yes absolutely I agree with that. If the circuit was competently designed with a bipolar or FET op-amp in mind, then it's probably best to stick with that topology. Broadly speaking, lower impedance sources call for bipolars, and higher impedances call for FET. Crossovers usually operate in the sort-of high-impedance range.
But there is a lot of overlap, and many cases where the call between bipolar and FET is not so clear.
My point is just that the input impedances of the op-amps isn't going to cause problems, per-se, with the circuit. Noise or distortion may not be ideal, but the two kinds of op-amps will basically work fine in the circuit, with the same crossover points and such.
But there is a lot of overlap, and many cases where the call between bipolar and FET is not so clear.
My point is just that the input impedances of the op-amps isn't going to cause problems, per-se, with the circuit. Noise or distortion may not be ideal, but the two kinds of op-amps will basically work fine in the circuit, with the same crossover points and such.
Yes well tons of gear has been stuffed full of TL072, which mostly have the same ill effect, and nobody dies. If you don't clip the stage before, you don't flip your bit. If you do, as in my headphone amp for live performance, the "tik!" is very short and supersonic. It is mostly a danger to people who listen with an oscilloscope, but I'm sure it has killed a few tweeters and ruined some master wax.
Has anyone verified that?? IMHO the LF411 and TL071 are not so different, except the TL is many cents cheaper. An OoM different THD could well be that Linear had not yet got a good meter, and TI did.
The Allen & Heath Xone92 DJ mixer, expensive, and considered one of the best, is chock-full of TL072's for pretty much everything, from filters to phono preamps. I could definitely imagine scenarios where that mixer is clipping. Many DJ's have the mistaken idea that "redlining" sounds good. (Ugh, no, get a real distortion FX box plz.) A good dust click on an LP might spike high enough to get that phase reversal.
I am using bipolars and FET for where they were originally used. The ICs not specified are the 2043s. The schematic doesn't show the output board with coupling caps or the power supply, just the crossover circuit.
I'll swap back in the 5532s for IC12 and check the DC offset directly out of the opamp pins 1 and 7.
I'll swap back in the 5532s for IC12 and check the DC offset directly out of the opamp pins 1 and 7.
Agreed. I'm just trying to learn. Just in case there is something lurking in the dark. 😀
OPA1656 (plus adaptor) is the hottest newest kid on the block. Faster, quieter, cleaner....I don't think ANY opamp can touch it for overall specsmanship. Sound? That's in the ear of the listener.
At the moment a good modern opamp is any one that's actually in stock somewhere! A lot of the opamps I bought earlier in the year are out of stock in several places.
Glad I've a stock of 60 NE5532's put by, although all SMT.
Glad I've a stock of 60 NE5532's put by, although all SMT.