Gentili Membri, le discussioni sono momentaneamente chiuse per esigenze legate alla migrazione al nuovo sito. Fanno eccezione le discussioni inerenti la migrazione stessa.
High performance audio op amplifiers
  • aklaakla
    Posts: 5,286Member
    Notes on Audio Op-Amps

    After I built several simple DIY headphone amps around the OPA132/134 family of op-amps, I decided to start trying different ones to see how they would affect the sound.

    Although there are hundreds of families of op-amps on the market today, not all are suitable for DIY headphone amps. We can eliminate power op-amps, super-high-speed op-amps, high-voltage op-amps, low-frequency op-amps, and low-spec general-purpose op-amps. Audio requires a fairly high slew rate to maintain low distortion, flat frequency response from 20 Hz to 20,000 Hz, and decent output current if the chip is to drive headphones directly. A low supply voltage requirement is a plus for portable applications. We also ignore chopper-stabilized and CMOS op-amps as they're generally unsuited to audio. This eliminates the vast majority of op-amps on the market.

    This article documents my testing process, and the results for many quality op-amps, plus some jellybeans:

    jellybean adj. /jel-ee-been/ 1. Of electronic parts: cheap, generic, available from multiple manufacturers. Buy 'em by the bag and use 'em by the handful. Pejorative.

    Test Method
    The first test is a clipping test, to find the op-amp's minimum useful voltage with a given output voltage and load. Only a few datasheets give you enough information to get this number without testing for it. To do the test, I run a 1 kHz tone through the amp while it is driving a 33 Ω dummy load with the volume set to give 0.5 V output. Then I repeat the test for 2.0 V into 330 Ω. The amp is powered from a variable bench power supply, which I start off at a rather high voltage and then I lower it until I observe clipping by seeing harmonics appear on a spectrum analyzer.

    I chose these test voltages to be about 6 dB higher than "loud enough" with my benchmark headphones, the Grado SR series (33 Ω numbers) and the Sennheiser HD-580/600 (330 Ω numbers). I chose to test at these higher-than-necessary voltages to account for peaks in an otherwise quiet recording, and for the occasional "rock out" session. But, beware that voltage requirements don't always go down as headphone impedance goes down. For instance, the 120 Ω AKG K501 and the 64 Ω HD-570 require about as much voltage as the 300 Ω Sennheiser HD-580. These lower-impedance phones would have higher clipping points than the HD-580s because the lower impedance means the op-amp has to provide more current at these voltages.

    The second test is a listening test. I look for more-or-less objective things like perceived distortion level and sonic signature. I also give you my subjective value judgement on the overall experience. I try to make it clear which are the objective aspects to the sound and which are just my opinion.

    Test System
    Most frequently I use a stock Music Hall CD25 CD player and Sennheiser HD-570 headphones. I chose this player for the tests because it's the best one I have, so I'm giving the op-amp the best chance to sound good. The heapdhones, on the other hand, were selected because they're a bit on the bright side so they're quite revealing. Also, they're only 64 Ω, yet they require uncommonly high voltage. This makes them a good worst-case test for finding sonic weaknesses. When I'm looking to test detail, accuracy, and pleasantness, I switch to my Sennheiser HD-600s with the Cardas cable upgrade.

    The test amplifier is a close variant of the CMoy pocket amp on a professional PCB. The major difference between it and a stock CMoy is that I buffer the virtual ground with a BUF634, and I use larger input and rail capacitors than originally specified. (0.22 µF and 470 µF, respectively.) I use Brown Dog adapters to adapt chips that I cannot get in dual DIP-8 versions to work with the board.

    (You might recognize this amp as the forerunner of the MINT amp.)

    I use this design rather than another heapdhone amp because it is very nearly the simplest headphone amp design that could possibly work, while not making the tested chips look overly bad, or masking too many weaknesses.

    The adjustable bench supply is the B+K Precision 1710. It lets me dial in any voltage from 0 to 30V, with good accuracy. I've tested its voltage quality, and it's quieter than some linear supplies I've used.

    The dummy load is a project box with stereo pairs of 33 Ω and 330 Ω power resistors, a switch to select between them, and a headphone cable coming out.

    Translating the Listening Test Results to Other Amps
    Although I favor more complex designs in amps I listen to for enjoyment, I stand by the choice to test chips by making them drive the headphones directly. It gives the clearest possible picture of the op-amp's sound, since there are so few components in the amp to mess with the sound.

    When translating these results to the sound you'd get with the chip in a different amplifier, you will have to adjust for any components in the amp that change the sonic signature. For instance, the BUF634 output buffers used in the MINT and PIMETA headphone amplifiers add the characteristic Burr-Brown laid-back quality to the sound, so a chip that's overly aggressive with your headphones when in a CMoy amp might sound quite nice in a PIMETA. For the PPA, the transparency of the amp tends to make op-amps show everything I hear in these CMoy-based tests, plus any little faults the chip has that I can't hear in this more veiled CMoy amp. If I say a chip is a little harsh in a CMoy, it'll probably be unlistenable in a PPA.

    If you have another amp than the ones I talk about on this site, please don't email me asking how I think the chip will sound in that amplifier. If I don't talk about it here, I probably don't have the experience or interest to talk constructively about it with you. You're better off searching or posting on Headwize or Head-Fi asking for others that have the amp about their experience. Failing that, try some of the chips I review here and see how what I report coincides with what you hear. You can probably use that experience to shift my reported results to match the other chips up with what you will hear.

    Translating the Clipping Test Results to Other Amps
    These clipping numbers only apply to circuits where the op-amp is driving headphones directly. If the op-amp drives something else in your application, you'll need to do your own clipping tests to get useful numbers. For instance, in a headphone amp with an output buffer, the clipping behavior of the amp will most likely be dominated by the clipping behavior of the output stage, not the op-amp.

    Where I give power supply voltages below, I'm talking about the rail-to-rail voltage. That is, when I say "10V" I mean that V+ is 10V above V-, not that I'm using a +/-10V supply.

    Where I give signal voltages, I mean RMS values unless I specifically indicate that I'm talking about peak-to-peak (p-p) voltages.

    All op-amp prices are for single quantities in US dollars, from Digi-Key. I don't promise to keep these prices up to date, so just use them as a guide.

    Chips to be Tested
    I have several chips on-hand that I'll be testing as time permits. In their current priority order: OPA2107, OPA602, AD825, AD744, AD797, LT1028, LT1115, LT1122, AD826, NE5534, NE5532, OP275, LF356, and LM386. Chances are not good to get more chips onto this list — the list is long enough as it is! I'm still on the look-out for interesting chips, but now that I've got a wide selection of known-good chips to choose from, "interesting" is a high bar to clear.

    Burr-Brown OPA132
    Cost, single: n/a Vmin, 0.5V into 33 Ω: 5.5V
    Cost, dual: $5.40 (OPA2132PA) Vmin, 2.0V into 330 Ω: 8.3V

    This is a nice family of op-amps. The sound has the typical Burr-Brown laid-back nature. It's a bit tubby on the bottom end. This is not an exciting sounding chip, but it does tend to counteract the flaws in many low-end audio systems, especially portable ones.

    Digi-Key only carries the dual versions (2132) in DIP packages, and there are two grades, differentiated by whether there's an 'A' in the part name. I've been unable to find a case where the 2132PA performs worse for audio amplification than the 2132P. From a look at the datasheet, it looks like the advantages of the non-A grade are in DC specs, which of course aren't all that important to audio.

    Bottom Line: A very good chip to start with. Indeed, you may be so happy with it that you stop looking at other chips!

    Burr-Brown OPA134
    Cost, single: n/a Vmin, 0.5V into 33 Ω: 5.7V
    Cost, dual: $2.63 Vmin, 2.0V into 330 Ω: 8.4V

    This is the audio grade version of the OPA132 family. (That's "audio grade" in the commercial sense, not the audiophile sense. Read: "lower quality".) Digi-Key only carries the dual version (2134) in DIP packages.

    The OPA134 requires a bit more voltage than the OPA132 does. This won't matter in circuits that have plenty of voltage, but in a battery powered system a 132 can pay for itself by letting you run longer on a battery.

    I found in earlier testing that the 134 was more likely than a 132 to become unstable in marginal circuits. Sometimes raising the supply voltage was all it took to make the 134 stable, and other times only swapping in a 132 would fix the problem. If you're building your own circuit from scratch and you aren't very experienced, the extra cost of the 132 can pay for itself in a better likelihood of success.

    Bottom Line: If your circuit is solid and you have a fairly high supply voltage, the 134 is better than the 132 because it's cheaper and they sound identical to me. The 132 is better for more marginal setups.

    Burr-Brown OPA2107
    Cost, single: n/a Vmin, 0.5V into 33 Ω: 10.6V
    Cost, dual: $12.25 Vmin, 2.0V into 330 Ω: 10.6V

    Full review to come later.

    Burr-Brown OPA627
    Cost, single: $18.38 (OPA627AP) Vmin, 0.5V into 33 Ω: 6.8V
    Cost, dual: n/a Vmin, 2.0V into 330 Ω: 9.8V

    The first thing I noticed is the cleanliness of the sound. With this chip in the test amplifier, I heard known problems in a low-end portable source more clearly than with my reference for this test, the OPA134PA. The 627 also seems to do better on recordings with room ambience: it reveals details about the acoustic space that the OPA132/134 chips will hide, making them sound "flat" in comparison. But these two chips are more alike than different. Both have the characteristic laid-back and dark Burr-Brown sound, and both are very tolerant, stable chips.

    The only remaining differences are that the OPA132/134 family will work well below 9V, whereas the 627's performance falls off a cliff below the clipping points I give above. There does seem to be a bit of extra low bass impactfulness with the 627. This seems less to be "extra power" than a removal of some heavier thumpiness in the 132/134 — the 627 seems to have a truer, more refined kind of bass.

    In all my testing, I've been unable to hear a difference between the OPA627AP and the OPA627BP. The datasheet says that the differences between the grades are in the DC specs, so this is not surprising.

    Bottom Line: The sonic differences between the OPA627/637 and the OPA132/134 are of the "last 5%" variety, rather than providing a dramatically different sound. If you like the Burr-Brown sound and can stand to pay 14× as much as for an OPA2134PA, a pair of OPA627APs is a reasonable investment. I see no reason to pay extra for the B grade in an audio application.

    Other Info:

    Burr-Brown OPA637
    The OPA637 is simply the "uncompensated" version of the OPA627. This means it has a higher bandwidth, but that it won't be stable at low gain levels. The datasheet says that it is minimally stable at a gain of 5, but as with all chips, the higher the gain, the more stable it becomes. The cost and voltage performance are the same as for the 627.

    The higher bandwidth of the 637 results in a somewhat more lively sound than the 627. It still has the overall laid-back Burr-Brown characteristic, though. Given the choice between these two chips, I use the 637 when I can live within its gain requirements, but I happily fall back to the 627 otherwise.

    Other Info:

    Burr-Brown OPA227
    Cost, single: n/a Vmin, 0.5V into 33 Ω: 5.4V
    Cost, dual: $3.53 (OPA2227PA) Vmin, 2.0V into 330 Ω: 8.2V

    This family of opamps sounds very similar to the OPA132 and OPA134 families. The main difference is that the 227 isn't as tubby on the bottom end as the 132. The 227 will run to slightly lower supply voltages than the 132 family.

    Digi-Key only carries the dual versions in DIP packages. The 'A' versions are the right ones for audio, as the non-A version simply has better DC specs, which is not useful for audio.

    Bottom Line: This is a good alternative to the OPA132. It is arguably even a significant upgrade.

    Burr-Brown OPA228
    Cost, single: $2.36 (OPA228PA) Vmin, 0.5V into 33 Ω: 5.1V
    Cost, dual: n/a Vmin, 2.0V into 330 Ω: 8.0V

    As the OPA637 is to the OPA627, so the OPA228 is to the OPA227.

    Being a faster chip (33 MHz vs. 8 MHz), it wasn't stable in some of my amps, probably due to minor circuit layout problems. I got it to oscillate outright occasionally, and other times I just got a kind of "grunge" in the sound.

    When I was able to avoid instability, it sounded a bit more analytical than the OPA132. By comparison, the 132 sounded more "alive". This difference is purely subjective, so some people may prefer the difference.

    Bottom Line: I'm not happy with this chip for audio. If your audio tastes are like mine, you'll be happier with the OPA227. This chip will run to even lower voltages, though, so perhaps it's a good idea for battery-powered amps that need to drive headphones that need fairly high voltages.

    Burr-Brown OPA602
    Cost, single: $7.60 Vmin, 0.5V into 33 Ω: 10.2V
    Cost, dual: n/a Vmin, 2.0V into 330 Ω: 9.6V

    Full review to come later.

    Burr-Brown OPA604
    Cost, single: $2.36 Vmin, 0.5V into 33 Ω: 8.6V
    Cost, dual: $4.28 Vmin, 2.0V into 330 Ω: 11.4V

    The specs on this chip are very similar to that of the OPA132/134 family: some specs are a bit better, and some a bit worse. Like the OPA134, this one is specifically sold with audio in mind. It's been used in several popular bits of audio gear, especially near the low end of the audiophile range.

    For a headphone amp, the most significant spec difference is that the 604s require more voltage than the OPA134/132s to sound good. If you're going to use a battery power supply, you should use two 9V batteries or at least 8 cells for AAs or similar.

    This chip has a problem with oscillation, no doubt because it's a "fast" op-amp: 20 MHz, vs. the OPA134/132 family's 8 MHz. In one of my Hansen-type amps configured with OPA2604APs, I was able to make it oscillate by giving it a supply voltage anywhere over 18V. In other amps, I was able to give it as much as 34VDC without causing oscillation. In my main test amp, I didn't have any oscillation problems. It's hard to pin down a source of blame for this problem. The only solid lesson I have been able to draw from this so far is that this chip is simply harder to use than the OPA134/132 family.

    Sonically, the OPA604 family is almost identical to the OPA134/132. I did hear slight differences, but I was at a complete loss when trying to put these differences into words. Because of the advantages of lower supply voltage and oscillation-free operation, the OPA134/132s are still more desirable chips for headphone amps.

    Bottom Line: Use this only if you can't find an OPA134/132.

    Analog Devices AD823
    Cost, single: n/a Vmin, 0.5V into 33 Ω: 4.3V
    Cost, dual: $5.00 Vmin, 2.0V into 330 Ω: 6.1V

    Spec-wise, this chip is somewhere in between the OPA134/132 and the OPA604, with one notable exception: it will tolerate single-supply operation down to 3V. It's also a "rail-to-rail" design, which ideally means it doesn't have any headroom requirements between the supply voltage and the output voltage. In practice, output loading and other things mean a practical chip can't go quite to the rails. This chip does a lot better than any of the above chips, at any rate. This makes the AD823 ideal for battery-driven amps.

    Sonically, this chip is slightly more impactful than the OPA134/132 in the bass area, and rather more detailed. With aggressive-sounding headphones, this chip will probably sound too aggressive, unless other parts of the system compensate for it.

    The main disadvantage of this chip is that it has a much lower output current than is typical: 15mA for the AD823 vs. around 40mA for most op-amps. While few headphones actually require more than 15mA of continuous current, all headphones seem to perform better when given a "reserve" of current far above their nominal requirements. I suspect it has to do with how much you make the amplifier strain, which affects its performance. This chip will perform adequately even with low-impedance headphones, but suboptimally. Output current isn't a consideration at all in buffered headphone amps, of course.

    Bottom Line: Seriously consider this chip over the OPA132/134. Especially consider it it you aren't happy with the laid-back Burr-Brown sound and don't mind spending a bit more on the chip.

    Other Info:

    Analog Devices AD8512
    Cost, single: n/a Vmin, 0.5V into 33 Ω: 4.5V
    Cost, dual: $3.89 Vmin, 2.0V into 330 Ω: 7.4V

    This chip is very similar to the AD823. The main differences are that it has higher output current and lower supply current. The sound quality is similar, though where I'd call the AD823 aggressive, I'd call the 8512 a bit harsh. The sound isn't nasty by any means, just not as euphonic as other chips I've reviewed here.

    Bottom line: This chip is best when battery savings are the absolute most important thing and the 823's low output current is a problem. If you can tolerate higher supply current or higher voltage requirements, there are better sounding chips.

    Analog Devices AD8065
    Cost, single: $4.21 (AD8065) Vmin, 0.5V into 33 Ω: 4.0V
    Cost, dual: $5.38 (AD8066) Vmin, 2.0V into 330 Ω: 5.7V

    This chip is a good alternative to the AD823 and AD8610 in battery-powered amps. It draws more current than either of those, but it does run to lower voltages so it may last nearly as long on a set of batteries in some configurations. As you can see from the clipping numbers, the chip is running rail-to-rail in the 330 Ω test: 2Vrms is 5.656V peak-to-peak. You can't get better than that.

    This chip kind of splits the difference between the AD8610 sound and the Burr-Brown sound: not aggressive, but not laid-back, either. It's a bit veiled, which is expected given the chip's price.

    This chip is rare in that it is only rated for a 24V supply. (Absolute maximum is 26.4V.) Another oddity is that it is only available in SOIC versions, so you need to mount it on a Brown Dog adapter to use it in amps that use DIP chips.

    Bottom line: This may be the ideal chip for you if you're running a battery-powered amp and the AD8610 is too aggressive and the OPA227 too laid-back.

    Analog Devices AD8610
    Cost, single: $8.00 (AD8610) Vmin, 0.5V into 33 Ω: 5.7V
    Cost, dual: $13.33 (AD8620) Vmin, 2.0V into 330 Ω: 7.6V

    This is quite possibly the best chip for battery-powered amps, period. Its voltage tolerance is among the lowest of all the chips I mention here, it has good output current abililty, it has among the lowest supply current of any chip reviewed here, and above all it sounds good.

    What does it sound like? Well, take the AD823, and remove some of the aggressive harshness. Add a bit of detail and smoothness from the AD843. That's the 8610. It's not a smooth chip, just not harsh. It's not the most detailed chip, but not heavily veiled, either.

    This chip is rare in that it is only rated for a 26V supply. (Absolute maximum is 27.3V.) Another oddity is that it is only available in SOIC versions, so you need to mount it on a Brown Dog adapter to use it in amps that use DIP chips.

    Bottom line: This is a contender for my favorite chip of all time, especially in battery-powered amps. When paired with an aggressive or very revealing system, this chip can be unpleasant. This chip is at its best complementing a smooth, laid-back system.

    Analog Devices AD825
    Cost, single: $3.76 Vmin, 0.5V into 33 Ω: 4.7V
    Cost, dual: n/a Vmin, 2.0V into 330 Ω: 8.6V

    Full review to come later.

    Analog Devices AD843
    Cost, single: $8.56 (AD843JN) Vmin, 0.5V into 33 Ω: 8.2V
    Cost, dual: n/a Vmin, 2.0V into 330 Ω: 12.2V

    When I went into this test, I was hoping to find a chip to dethrone the OPA627 which is expensive and requires a lot of voltage. The 627 also has the characteristically mellow Burr-Brown sound, which is not always a good thing. Analog Devices chips tend to be a little snappier and more aggressive, which can help balance some systems. I wanted a chip that would fit all of these criteria while still maintaining the OPA627's incredible level of resolution and clarity. The AD843 doesn't completely fit these criteria, but it does come close.

    What this chip gets right: First, the AD843 is definitely cheaper. Like the 627, the cheaper grade is fine for audio, so a pair of 843s is about half the cost of a pair of 627s. Second, this chip does have that Analog snap and verve.

    The downside is that the AD843 requires more voltage than the OPA627. And like the 627, the sound gets very nasty very quickly when it starts clipping.

    The AD843 seems to trade smoothness for resolution relative to the OPA627. In some cases the more detailed OPA627 might be preferrable, and in others the smoother AD843 could be helpful. I'm torn on what to make of this real difference. The OPA627 isn't hyper-revealing, and the AD843 isn't over-smooth. The OPA627's detail seems genuine; it isn't grain or overemphasized real detail. I don't mind the way the AD843 ignores these details, but at the same time I don't resent the OPA627 for presenting them. The OPA627 can be accurate to a fault, if your recording has unpleasant detail in it that another op-amp would ignore or deemphasize. If you have flawed recordings, you may prefer the pleasant lie told by the AD843.

    Bottom Line: This is a serious contender with the Burr-Brown OPA627 for the title of "best op-amp I've ever heard". I view these two chips as rough equals; they're both in the same audio class, but each chip has strengths lacked by the other. Taste and system matching will be the largest factors in choosing one over the other.

    Other Info:

    Analog Devices AD845
    Cost, single: $6.40 (AD845JN) Vmin, 0.5V into 33 Ω: 8.7V
    Cost, dual: n/a Vmin, 2.0V into 330 Ω: 9.0V

    This is a very interesting chip. Sonically, you can call this a smoother, less detailed AD843. It's almost tube-like, while retaining the Analog devices snap. Compare the Burr-Brown sound which is relaxed...mellow...slower. The chip has a lot more resolution compared to lesser chips like the OPA132/134 and the AD823, but it doesn't quite match that of high-end chips like the AD843 and OPA627.

    Voltage-wise, it's a somewhat hungry chip. It's only specified to run down to 9.5V, but I was able to get it to go a bit lower. In a previous listening test, I was able to take it down to about 5V. I think what is happening here is that this chip has tolerable distortion behavior when clipping, so a bit of clipping is tolerated better than with other chips.

    The chip is also a bit of a pig when it comes to current draw. Each chip draws about 10 mA quiescent, which will of course go up in normal operation, and you need two chips for stereo. Heaven help you should use the chip in a battery-powered Hansen or CHA47 amp, where you need four chips!

    Bottom Line: If you're building a wall-powered amp and want a snappy sound and either want to save some money or smooth over some detail relative to the AD843, this is a good chip. However, I think you should make the small step up to the 843 instead. This would be okay in a high-end battery-powered amp where sound quality is more important than battery life, yet going up to the 843 is too much.

    Other Info:

    Analog Devices AD744
    Cost, single: $4.52 Vmin, 0.5V into 33 Ω: 6.8V
    Cost, dual: n/a Vmin, 2.0V into 330 Ω: 9.2V

    Full review to come later.

    Analog Devices OP275
    Cost, single: $2.13 Vmin, 0.5V into 33 Ω: 17.0V
    Cost, dual: n/a Vmin, 2.0V into 330 Ω: 10.0V

    Full review to come later.

    National Semiconductor LM6171
    Cost, single: $2.83 Vmin, 0.5V into 33 Ω: 6.1V
    Cost, dual: $3.80 Vmin, 2.0V into 330 Ω: 9.5V

    Digi-Key carries the LM6171BIN single-channel version, and the LM6172IN dual version in the DIP package.

    This is a very high-speed op-amp with bipolar input transistors. Translated, that means this op-amp is hard to use. However, it has exemplary audio performance for such a low price, so experienced builders should at least consider using it. If you decide to give it a shot, you must design the amp around the chip: you cannot just pop the chip into an existing circuit and expect it work. (For info on what's necessary to make this chip work, see the companion article, "Working with Cranky Op-Amps".)

    I made another test amplifier according to that article's principles in order to test this chip. To assure myself that the necessary design changes didn't impact the sound, I tried some of the other chips in it. This brings up an important point: the changes necessary to make chips like the LM6171 work in a CMoy amp do not prevent less picky chips from working. If you think you might want to try a chip like this someday, you might make the design changes from the start so you're free to use most any chip in the amp.

    Sonically, this chip is significantly more revealing than the OPA132/134 series, and the bass is a touch more impactful as well.

    With a high supply voltage (15V), the difference between the LM6172 and the OPA134PA is subtle, but real. At lower supply voltages, the LM6172 can greatly outshine the OPA132/134 series. The clipping numbers above don't tell the full story; when this chip clips, it does so in a very soft, rounded way, instead of the harsh clipping you get with other chips. This means you can often run the chip to lower voltages than my tests indicate by accepting a mild form of distortion.

    Bottom Line: This chip is not for tyros, but it's cheap, it performs very well at low voltages, and at higher voltages it still outshines many other chips in its price class. It requires more external components to achieve that performance, however.

    Other Info:

    TL07x (various vendors)
    Cost, single: $0.48 (TL071CP) Vmin, 0.5V into 33 Ω: 17.3V
    Cost, dual: $0.64 (TL072CP) Vmin, 2.0V into 330 Ω: 12.3V

    This is a jellybean chip that was popular for audio back in the 80's. Open up an old CD player or preamp, and there's a pretty good chance you'll find a TL072.

    The 071 is the single version, and the 072 is the dual. There is also a letter or three in there, which indicates the part's tolerance and sometimes a variation code. In increasing levels of quality, they are: C, AC, BC. (I'm ignoring the special-purpose industrial and mil-spec grades.) I can't hear the difference between the grades, so get the C grade. The price is for TI's version, since they invented the chip.

    This family of chips is not as good as the OPA132/4 family. The clipping numbers above are informative here: this is one of the few chips I've tested that required more supply voltage on the 33 Ω test than on the 330 Ω test. Bottom line, this chip does not like to drive low impedances. I've heard members of this op-amp family distort even at the chip's full operating voltage of 30V. I'm not talking about minor problems here — I'm talking about crunchy, ugly, obvious distortion. Therefore, I recommend using this chip only under duress, and then only if you're willing to give it a fairly high supply voltage.

    Bottom Line: This chip is very common and cheap, but also sonically inferior to all other chips tested so far. Use it only if you can't find anything better.

    TL08x (various vendors)
    This is a lower-spec version of the TL07x. I can't hear a difference between the two, but since the TL07x is the same price you might as well get that one. As I said above, these chips are classic audio jellybeans, mainly useful when the chip doesn't have to drive a low-impedance load.

    The only reason I've bothered to review the 082 here is that you can get them at Radio Shack. If you need an op-amp in an emergency, it's good to know that you can pay the moderately obscene sum of $1.89 each to get usable chip without having to wait for mail order.

    NE5532 (various vendors)
    Cost, single: n/a Vmin, 0.5V into 33 Ω: 5.6V
    Cost, dual: $0.80 Vmin, 2.0V into 330 Ω: 8.9V

    Another jellybean chip, but bipolar-input so it requires more care in application. Probably better than the TL072 for audio.

    Full review to come later.

    National Semiconductor LF355
    Cost, single: $0.94 Vmin, 0.5V into 33 Ω: 14.9V
    Cost, dual: n/a Vmin, 2.0V into 330 Ω: 10.3V

    Yet another jellybean chip. The full family is LF355-LF357. Specwise, these are very similar to the TL071. A related chip is the LF351, discontinued by National but picked up by makers of generic chips like ST and Fairchild.

    Notice that this chip doesn't like to drive low impedances.


    Article Revision History

    Rewrote the introductory material, describing the new test methodology.

    Re-did all the clipping tests with the new objective method.

    Added skeleton reviews of the AD744, AD825, LF355, OP275, OPA2017 and OPA602.

    Added short reviews of the AD8512 and NE5532.

    Added full reviews of the AD8610 and AD8065.

    Split all of the reviews of similar chips (132/134, 227/228, 627/637, etc.) into separate reviews, for clarity.

    Fixed all the Headwize URLs.


    Added OPA637 info to the OPA627 review.


    Added the AD843 and AD845 reviews.


    Small improvements. The big one is adding the warning about the AD823's output current issue.


    Updated the OPA627 review with A vs. B grade results.


    Added the OPA227 and OPA228 reviews.


    Added the OPA627 review.


    Finished the initial LM617x review. And, touched up the rest of the article some.


    Made two new CMoy-type amps solely for the purpose of this test and re-tested all the chips tested so far in them, then updated the article with the findings. Also, added info on test conditions, and polished the initial article text some.

    Added info on the TL071ACN, TL072ACN, and TL072BCP.

    Added info on the Burr-Brown OPA604AP and OPA2604AP.

    Added info on the Analog Devices AD823AN.


    Starting with the text from the section "A Few Words About Op-Amps" in the CMoy tutorial, created the original version of this article. The initial version covers the full OPA2132/2134 families and the TL082CP. (And of course it generalizes for these op-amps' extended families until I have an opportunity to try a broader selection of chips in each line.)


    This article is copyright © 2002-2005 by Warren Young, all rights reserved.

    Updated Sat Jun 11 2005 00:32 MDT Go back to Audiologica Go to my home page
    Nesa one the first audio ologram solution

    1 Parsec = 3.08568025 × 10(16) meters
  • Giaime
    Posts: 6,631Member
    Si Ing. Lange, si... tante belle parole.

    Poi nelle apparecchiature audio multimilionarie ancora oggi vedi i 5532 imageimageimage

    saluti termoionici image
    Giaime Ugliano
    (donazioni di apparecchi audio vecchi/inutilizzati sono benvenute)
  • Microfast
    Posts: 5,288Industry
    Si Ing. Lange, si... tante belle parole.

    Poi nelle apparecchiature audio multimilionarie ancora oggi vedi i 5532 imageimageimage

    saluti termoionici image
    Giaime Ugliano
    (donazioni di apparecchi audio vecchi/inutilizzati sono benvenute)

    E addirittura i TL072c, i JRC4558 i poco migliori JRC2068, i JRC 4580 ...............

  • Giaime
    Posts: 6,631Member
    Si beh i 5532 almeno hanno la fama di suonicchiare, quelli che hai messo tu... AAAARGH!!! Il TL072 è buonissimo nei distorsori per chitarra elettrica IMHO. Forse solo in quello image

    Sul MITICO 4558 non commento image

    Il 4580, pensa, è usato nello stadio d'uscita di un acclamatissimo DAC a basso costo di derivazione pro... image

    e... ci dimentichiamo il mitico 741? Ma lo producono ancora? Gli "operatori del settore" lo classificano con il 555 sotto la voce "generatori casuali di forme d'onda" più che "amplificatori operazionali"... dicono che sia per il rumoreimage

    Un'idea per un white noise generator fatto in casa image

    saluti termoionici image
    Giaime Ugliano
    (donazioni di apparecchi audio vecchi/inutilizzati sono benvenute)
  • Microfast
    Posts: 5,288Industry
    Si beh i 5532 almeno hanno la fama di suonicchiare, quelli che hai messo tu... AAAARGH!!! Il TL072 è buonissimo nei distorsori per chitarra elettrica IMHO. Forse solo in quello image

    Sul MITICO 4558 non commento image

    Il 4580, pensa, è usato nello stadio d'uscita di un acclamatissimo DAC a basso costo di derivazione pro... image

    e... ci dimentichiamo il mitico 741? Ma lo producono ancora? Gli "operatori del settore" lo classificano con il 555 sotto la voce "generatori casuali di forme d'onda" più che "amplificatori operazionali"... dicono che sia per il rumoreimage

    Un'idea per un white noise generator fatto in casa image

    saluti termoionici image
    Giaime Ugliano
    (donazioni di apparecchi audio vecchi/inutilizzati sono benvenute)

    Forse non ci crederai, ma il 4580 ( che comunque non e' proprio orribile ) ultimamente e' addirittura comparso nella recensione di un player Accuphase, spero si siano sbagliati, ma chissa' .........

  • Giaime
    Posts: 6,631Member
    No infatti, però se ci pensi è solo la versione low-noise del 4558, e partendo da un progetto oramai sorpassato si può fare solo un buon opamp, non un ottimo opamp.

    Ps guarda che con le attrezzature cinesi coi 4580 dentro ci fanno anche molti dischi "audiophile" image

    saluti termoionici image
    Giaime Ugliano
    (donazioni di apparecchi audio vecchi/inutilizzati sono benvenute)
  • aklaakla
    Posts: 5,286Member
    Forse non ci crederai, ma il 4580 ( che comunque non e' proprio orribile ) ultimamente e' addirittura comparso nella recensione di un player Accuphase, spero si siano sbagliati, ma chissa' .........


    Non ti sbagli è cosi'

    ©2002 Fairchild Semiconductor Corporation
    Rev. 1.0.1
    • Internal Frequency Compensation
    • Slew Rate: 8V/μs
    • Input Noise Voltage:
    • Full Power Bandwidth: 140KHz
    The NE5532 is a internally compensated dual low noise
    OP-AMP. The high small signal and power bandwidth
    provides superior performance in high quality AMP, all
    control circuits, and telephone applications.
    8nV ⁄ Hz (fo = 30Hz)
    Internal Block Diagram
    Dual Operational Amplifier
    Absolute Maximum Ratings
    Thermal Data
    Electrical Characteristics
    (VCC=15V, VEE= - 15V, TA = 25°C)
    Parameter Symbol NE5532 Unit
    Power Supply Voltage VCC ±22 V
    Differential Input Voltage V(DIFF) ±13 V
    Input Voltage VI Supply Voltage V
    Power Dissipation, TA = 25°C
    PD 1100
    Operating Temperature Range TOPR 0 ~ + 70 °C
    Parameter Symbol Value Unit
    Thermal Resistance Junction-Ambient Max.
    Rθja 110
    Parameter Symbol Conditions Min. Typ. Max. Unit
    Input Offset Voltage VIO - - 0.5 4.0 mV
    Input Offset Current IIO - - 10 150 nA
    Input Bias Current IBIAS - - 200 800 nA
    Supply Current ICC - - 6.0 16 mA
    Input Voltage Range VI(R) - ±12 ±13 - V
    Common Mode Rejection Range CMRR TA = 25 °C 70 100 - dB
    Power Supply Rejection Ratio PSRR TA = 25 °C 80 100 - dB
    Output Voltage Swing VO(P-P) RL ≥ 600Ω ±12 ±13 - V
    Input Resistance RI TA = 25 °C 30 300 - KΩ
    Short Circuit Current ISC - - 38 - mA
    Overshoot OS RL =600Ω, CL =100pF - 10 20 %
    Voltage Gain GV f = 10KHz 2 2.2 - V/mV
    Gain Bandwidth Product GBW CL =100pF, RL = 600Ω 8 10 - MHz
    Slew Rate SR RL =1K, CL =100pF, RL =600Ω 6 8.0 - V/μs
    Input Noise Voltage eN
    fO = 30Hz
    fO = 1KHz - 8.0
    5.0 - nV/ Hz
    Typical Performance Characteristics
    Figure 1. Open Loop Frequency Response Figure 2. Large Signal Frequency Response
    Figure 3. Supply Current vs Supply Voltage Figure 4. Input Bias Current vs Temperature
    Figure 5. Output Circuit Current vs Temperature Figure 6. Slew Rate
    Mechanical Dimensions
     2002 Fairchild Semiconductor Corporation
    CORPORATION. As used herein:
    1. Life support devices or systems are devices or systems
    which, (a) are intended for surgical implant into the body,
    or (b) support or sustain life, and (c) whose failure to
    perform when properly used in accordance with
    instructions for use provided in the labeling, can be
    reasonably expected to result in a significant injury of the
    2. A critical component in any component of a life support
    device or system whose failure to perform can be
    reasonably expected to cause the failure of the life support
    device or system, or to affect its safety or effectiveness.
    Ordering Information
    Product Number Package Operating Temperature
    NE5532N 8-DIP
    0 ~ +70°C
    NE5532D 8-SOP

    inoltre l' articolo dimostra come sia possibile ottimizzare parametri e
    circuitazioni e non una sola lista di prodotti
    Le caratteristiche del 5532 vanno ovviamente ottimizzate
    integrandole magari in circuitazioni Ibride
    ove si prediliga l' alto coefficente di rise time
    tenendo in debita considerazione le sovraelongazioni di risposta impulsiva
    di un serie di tone burst ad alta rapidita
    Risposta impulsiva
    Non Credi

    semplicemente prova

    Buon Lavoro

    1 Parsec = 3.08568025 × 10(16) meters



    Saluti Neutronici

    Modificato da - akla il 08/02/2006 09:26:47
  • walge
    Posts: 1,635Industry
    Consiglio la lettura di alcuni articoli apparsi anni fa su Audio Amateur (non so se sono ancora reperibili, io li ho ma non so se si possono fotocopiare senza avere grane) che riguardano gli operazionali; ovviamente sono passati molti anni ma i concetti espressi sono molto chiari e aiutano a capire come gli OP possono essere impiegati in campo audio.
    Dal mio punto di vista (volutamente limitato) considero gli AD827/847 i migliori in assoluto.


    Walter Gentilucci
  • aklaakla
    Posts: 5,286Member

    esempio tipico di un modulo ibrido direttamente ispirato alle tecnologie
    circuitali adottate negli op amps
    ovviamente si riducono le possibilita' di interazione di segnale
    inoltre e possibile concretizzare una linearita intrinseca
    agendo direttamente sulle hfe dei singoli Bjt a piacimento
    e linearizzando il Gbw in funzione delle reali esigenze in campo operativo audio
    con moderata local nfb

    1 Parsec = 3.08568025 × 10(16) meters
    There is no dark side of the moon really. matter of fact it’s all dark.


    Modificato da - akla il 09/02/2006 10:33:52
  • aire
    Posts: 1,128Member
    Bello il post sugli operazionali.
    Quando misi mano al mio paganini, che come ho sempre detto originale faceva un po' schifetto, arrivai a delle conclusioni simili, in modo del tutto empirico.

    Mi feci campionare dall'EBV (distributore Texas) tutti gli OPA possibili e immaginabili, e devo ammettere che i 627 e i 228 sono i miei preferiti.

    Non ho mai sentito gli AD8xx poiché la mia compagnia non tratta con analog, ma mi hanno sempre incuriosito.

    Non è che qualcuno vuole fare uno scambio con degli OPA627?

This discussion has been closed.
Tutte le discussioni