njm2068 4558 4580. are often used in vestax mixers and it sounded good, but i bouth a nog high fidelity phono pre-amp and still ne5534 inside and cheap caps at the input , it's a m1lps . Pretty case, heavy, but a simple design offcourse and the opamps not in a socket.
a iec input is a nice plus but why not using 6 panasonics caps and use an opa dip8 in a socket instead of the not to replace opamp... only two.... So i think i return it, since the unit weights 2 kgs and that's useless heavy.
For 6 euro would it have been nice, but diy you can make this also. The ne5532 is nice, only
there are better once . It's like companies are addicted to some opamps.
Greets richard
a iec input is a nice plus but why not using 6 panasonics caps and use an opa dip8 in a socket instead of the not to replace opamp... only two.... So i think i return it, since the unit weights 2 kgs and that's useless heavy.
For 6 euro would it have been nice, but diy you can make this also. The ne5532 is nice, only
there are better once . It's like companies are addicted to some opamps.
Greets richard
Maybe a bit small on a high-density screen, I admit. I have an old-fashioned 19" in 5:4 here.
Anyway, it's two NJM4580M-Ds in parallel with:
* the outputs combined via 680R resistors, with muting transistors on the output
* each of the 4 amplifiers having its own 10k/10k feedback network (noninverting)
* a 20k pot DC-coupled to the input, with 220k input parallel resistors
Something tells me that this one probably wasn't thought out by a seasoned audio designer... At least they put some beefy 470µ/16 caps on the rails.
This is a headphone driver circuit from a fairly recent commercial CD player.
LM324
I design a lot of audio gear for a living and I use the LM324 and LM358 a lot. For metering buffers and the like. But I would not dream for a minute using it in an audio signal path. Nor the JRC 4558. If I need a quad pack for density the LM837 has been my usual device.
For duals, we find a lot of people like the sound the OP275, MC33078 and LM833. Though some swear by the sound of the venerable NE5532.
One of our dirty secrets in the guitar pedal world is the TL072. It is still a contender in the music business in a lot of applications.
I design a lot of audio gear for a living and I use the LM324 and LM358 a lot. For metering buffers and the like. But I would not dream for a minute using it in an audio signal path. Nor the JRC 4558. If I need a quad pack for density the LM837 has been my usual device.
For duals, we find a lot of people like the sound the OP275, MC33078 and LM833. Though some swear by the sound of the venerable NE5532.
One of our dirty secrets in the guitar pedal world is the TL072. It is still a contender in the music business in a lot of applications.
A promising but decidedly cranky part that is not simply a quad version of the LM833. At 45° of phase margin, calling this one "unity gain stable" is a bit of a stretch, and linearity quickly decays at higher frequencies as the ingenious distortion cancellation circuitry in the output stage loses effectiveness.
People often seem to prefer the similar vintage MC33079 ('078 in quad) over this diva, in spite of its dislike for any kind of output loading. Unlike the output stage, the input stage does not seem to be bad at all for a cheapie, especially in terms of common-mode distortion (the Achilles heel of many a modern cheap NE5532).
If you have not seen Samuel Groner's measurements, he tested all the parts you mentioned and then some. Well worth poking your nose into for an hour or five. NJR parts are the only truly glaring omission, along with some very recent models like OPA16xx.
I have a highish power Kenwood integrated amp (ailing) that has the NEC version of the 4558 in some low level positions. I pan to oust them in favor of OPA2134 w/enhanced bypass when I get around to it. The amp probably has a dried up cap in the protection circuit, a common cause of an amp failing to start.
JRC NJM4558 is used in many hifi products. I have one in output of Denon TU660 FM tuner and this is one of the best FM tuners of all times. 4558 did not ruin the sound of it. In fact is sounds fantastic. But it is used as unity gain buffer! For low gains or buffer application it is just fine and sounds very good. In hifi products it almost always used as unity gain buffer, as some kind of impedance matching device.
Last edited:
Have an eye on surrounding circuit impedances when doing this, however. OPA213x are good parts in a number of respects, but exhibit about as much input impedance distortion as many typical FET input opamps, while I would expect a '4558 not to have more than a '5532 or possibly even less (going by Ib and noise, it runs little LTP current after all and would be expect to have very small input transistors). Also be careful when you encounter a phono stage wih an external LTP.
As stated, '4558s tend to be quite adequate for unity-gain buffers without undue output loading. Now after the volume pot, that's a different story. I'd still prefer a lowish-noise bipolar though (e.g. LM4562 family), which should given a fair bit less noise still.
Measure DC offset after startup to be sure.
4558's 8nV is not that noisy. QUAD 405 used LM301 at the input with 22k in series (opamp was used in inverting mode) and it still sounded decent. And LM301 has 25nV noise spec!
I'll have to check the service manual to make sure, but I suspect all sorts of little sins in the construction of the Kenwood amp, including electrolytic coupling caps in the low and medium level circuits. Switching to jfet input amps may allow me to use good film coupling caps that aren't the size of a coconut. Eliminating the coupling caps where possible is my preferred option. I haven't had a look at the phono stage, but that will probably get changed as well. I may even bodge in a discrete jfet phono stage.
Fixing the amp hasn't been my highest priority (far from it - I often forget I still have it), but eventually I'll do it just to get it functional and/or out of the house. It's 125W/channel, so it may get pressed into duty as a sub driver, kind of a demotion for a full-fledged integrated amp.
Fixing the amp hasn't been my highest priority (far from it - I often forget I still have it), but eventually I'll do it just to get it functional and/or out of the house. It's 125W/channel, so it may get pressed into duty as a sub driver, kind of a demotion for a full-fledged integrated amp.
I had some DD 4558s and man they were damn most unbalanced damn non linear things ever haha, just that really fake sound hard to describe, tried some NE5532 imo waay better and does come across more hi-fi clear/detailed, Need to get a hold of them Burr Brown chips see what that Hype is all about.
You are not going to change some people's mind though, many still swear by that damn 4558 haha, not going to change my mind I know it sounds ugly non-linear middy/fake sounding haha
You are not going to change some people's mind though, many still swear by that damn 4558 haha, not going to change my mind I know it sounds ugly non-linear middy/fake sounding haha
Last edited:
^ Mind showing what your test circuit looks like?
What many DIYers don't realize is that opamps don't have a "sound", per se. They'll just react to how they're being (mis)used in different ways.
For example, the trusty TL072 isn't fond of driving loads, like at all. Even at line level it may prefer >10 kOhms, with an isolation resistor as high as 470 ohms on the output in order to keep capacitive loading off. A '4558 also is on the wimpy side. I would never expect one to drive low-impedance headphones, for example. That would require a part with considerably more "oomph", like NE5532 or even NJM4556A.
If you insist on using 1970s-vintage parts, try two LF356 - while basic linearity is not quite as good as the TL072's and maximum current output still is a bit limited, the output stage runs a lot more quiescent current and hence is a lot better-behaved in terms of output loading.
Then there is the question of supply bypassing. Faster, more powerful newer parts tend to appreciate better bypassing and tighter layouts. If the whole shebang oscillates, all bets are off on how it'll sound.
If you spot some of the orange Elnas in your Kenwood, these are low-leakage types and likely to have held up well, but ordinary small ones may be begging for replacement after 30+ years, especially since Kenwood tended to be quite stingy with capacity from about 1981-ish on. A look at the service docs certainly couldn't hurt. Which model are we talking about anyway?
What many DIYers don't realize is that opamps don't have a "sound", per se. They'll just react to how they're being (mis)used in different ways.
For example, the trusty TL072 isn't fond of driving loads, like at all. Even at line level it may prefer >10 kOhms, with an isolation resistor as high as 470 ohms on the output in order to keep capacitive loading off. A '4558 also is on the wimpy side. I would never expect one to drive low-impedance headphones, for example. That would require a part with considerably more "oomph", like NE5532 or even NJM4556A.
If you insist on using 1970s-vintage parts, try two LF356 - while basic linearity is not quite as good as the TL072's and maximum current output still is a bit limited, the output stage runs a lot more quiescent current and hence is a lot better-behaved in terms of output loading.
Then there is the question of supply bypassing. Faster, more powerful newer parts tend to appreciate better bypassing and tighter layouts. If the whole shebang oscillates, all bets are off on how it'll sound.
Sounds like you're trying to fix a non-problem, tbh. There is absolutely nothing wrong with good-quality, preferably low-leakage audio electrolytics of generously oversized capacity as coupling caps. You can still get a few low-leakage series, like Elna RLB or Vishay / BCComponents 013 RLC, though ordinary Panasonic FM/FCs and similar ought to be fine, too.
If you spot some of the orange Elnas in your Kenwood, these are low-leakage types and likely to have held up well, but ordinary small ones may be begging for replacement after 30+ years, especially since Kenwood tended to be quite stingy with capacity from about 1981-ish on. A look at the service docs certainly couldn't hurt. Which model are we talking about anyway?
Around 1980 I changed out the 4558's in a mixing console, and the coupling caps.
There were ten electrolytics in each channel, and replacing the 4558's with TL072's I was able to replace the ten electrolytics with only three film caps per channel. The mic input had a 5532, that was left in place.
The difference in sound was stunning. A friend that had a $25K mixing console that came with 5532's declared he was upgrading his capacitors too, as this modified $1600 mixer now sounded better than his $25K desk.
Yes, you need to be able to evaluate the circuit for any changes that the different parts may require.
More recently I replaced a bunch of TL072's in an expensive Allen & Heath desk with OP2604's, no other changes needed, and a similar leap forward in sound quality as the 4558 to TL072 was.
I still use many old devices, it just depends on the application (and how much money you want to throw at a given device).
There were ten electrolytics in each channel, and replacing the 4558's with TL072's I was able to replace the ten electrolytics with only three film caps per channel. The mic input had a 5532, that was left in place.
The difference in sound was stunning. A friend that had a $25K mixing console that came with 5532's declared he was upgrading his capacitors too, as this modified $1600 mixer now sounded better than his $25K desk.
Yes, you need to be able to evaluate the circuit for any changes that the different parts may require.
More recently I replaced a bunch of TL072's in an expensive Allen & Heath desk with OP2604's, no other changes needed, and a similar leap forward in sound quality as the 4558 to TL072 was.
I still use many old devices, it just depends on the application (and how much money you want to throw at a given device).
Old TI TL072s yeah were pretty good before started going bad/noisey......got just some STM TL082s now and something about them really liking, not sure what the hell for. A lot more of a difference/detail than I was expecting.
So for sure there is some damn noticeable differences in opamps regardless how setup/placement/surrounding RC....but all them years/still probably to this day people still say there's no difference...I don't know what on earth you are hearing but I'm hearing something with it. Probably something beyond my knowledge/understanding.
Lets say you are forced to use something with this op amp. This very old trick can work. Some will not know it so I took the trouble to show it. By forcing the output into SE class A any crossover distortion is eliminated upto the current limit choosen. How large the resistor needs to be can be done by ear. Usually 10 K serves well. If one likes a constant current device can be used. On paper it is better, in reality time to use something better if you get that far. A 2 mA CRD might serve well, it is like a diode in looks( Semitec ). The resistor/CRD is usually to the - ve rail, when the 741 it was the better option. It seems OK to do that regardless. This is a 3 minute fix. I often solder 10 nF from V + to V- on the pins whilst about it. The better op amps become different when doing this. Sometimes better, often worse. Worse might mean not enough current or just distortion type. A nice cheap dual type is MC33078. It suits either way.
MC33078 actually is a near-ideal candidate for Class A biasing, the TI version that Samuel Groner measured at least (which may actually be a modern-day reimplementation, seeing how they don't offer a MC33079 quad, how it's more powerful than other versions and how measurements do not seem to line up with Douglas Self's who had either ST or ON Semi parts and did not even mention TI as a manufacturer). His results indicate an output stage of very modest symmetry (in fact it's a quasicomp) run at a quiescent current level best described as pitiful. Why one would design a low-noise opamp like this is anyone's guess, they tend to be used with lowish surrounding impedances after all.
I think it's worth it because the input stage actually is quite decent, with far less high-frequency common-mode distortion than found in TI's NE5532. It also slightly beats the '5532 for both voltage and current noise. (Input impedance distortion isn't that great though.) Maximum current output is quite healthy, so there's some room to spare.
Maybe that's why Beyerdynamic used an MC33078 in one of their headphone amps - the bias diodes for a discrete buffer stage need some current anyway...
I think it's worth it because the input stage actually is quite decent, with far less high-frequency common-mode distortion than found in TI's NE5532. It also slightly beats the '5532 for both voltage and current noise. (Input impedance distortion isn't that great though.) Maximum current output is quite healthy, so there's some room to spare.
Maybe that's why Beyerdynamic used an MC33078 in one of their headphone amps - the bias diodes for a discrete buffer stage need some current anyway...
Last edited:
You can do it by ear. 10 K is not bad. Nothing bad will happen if you go too far except the chiip will get a bit hot. The 2 K often stated in ideal loadings seems the limit. I find if the voltage swing is low the MC33078 is great in SE class A. It can sound better without SE class A when working hard. Also if not giving much current the chip itself might be mostly in class A due to the bias needing to be safe . I didn't say about MC 33079. I recently had some common-mode problems with an oscilator using TL074. My own fault the way I made it. Using LM324 cured the problems. Of what I had the MC33079 was the perfect answer, it was better than either.
Most op amps will say very quickly if they will work well in SE class A. If you use the worlds worse op amp the LM324 ( 358 ? ) with a 5 mA CCS and headphones it is almost OK. LM324 has survived because it works in specail ways which suit difficult engineering. It's designer I read was very dissapointed with it. However it did do it's party tricks very well. The choice was made to sell it cheaply and get the money back that way. A very good choice. It scores over LM339 in that it can both source and sink current and can often work as well as a LM339. It looks almost like a LM339 which makes me wonder if that was it's starting point? Ironically LM339 is a SE class A device , it is a comparator. It has no compenstaion so will be a difficult device and about 40 nV noise. MC33078 about 4.5 nV ( per root Hz ).
The resistor is dependant on the voltage rail. My 10 K looks to a typical 12 V. Try either -ve or + ve rail ( not both ). Often the -ve sounds more open. Even when NPN/PNP it can favour the NPN. When all NPN the theory was the emitter output favoured. That's debatable as the feedback pair can have lower distortion. Each use will have it's sweet spot.
You can try this trick with power amps. Alas the power needed makes it impractical. A TDA2040 and class A trick ( 100 R ? ) might make a headphone amp. Use the capacitor coupled version to protect the headphones I would say. Capacitors are much better than people think, even awful ones. The 4558 is worse than any is my guess. Panasonic FC series are cheap and very low distortion.
Most op amps will say very quickly if they will work well in SE class A. If you use the worlds worse op amp the LM324 ( 358 ? ) with a 5 mA CCS and headphones it is almost OK. LM324 has survived because it works in specail ways which suit difficult engineering. It's designer I read was very dissapointed with it. However it did do it's party tricks very well. The choice was made to sell it cheaply and get the money back that way. A very good choice. It scores over LM339 in that it can both source and sink current and can often work as well as a LM339. It looks almost like a LM339 which makes me wonder if that was it's starting point? Ironically LM339 is a SE class A device , it is a comparator. It has no compenstaion so will be a difficult device and about 40 nV noise. MC33078 about 4.5 nV ( per root Hz ).
The resistor is dependant on the voltage rail. My 10 K looks to a typical 12 V. Try either -ve or + ve rail ( not both ). Often the -ve sounds more open. Even when NPN/PNP it can favour the NPN. When all NPN the theory was the emitter output favoured. That's debatable as the feedback pair can have lower distortion. Each use will have it's sweet spot.
You can try this trick with power amps. Alas the power needed makes it impractical. A TDA2040 and class A trick ( 100 R ? ) might make a headphone amp. Use the capacitor coupled version to protect the headphones I would say. Capacitors are much better than people think, even awful ones. The 4558 is worse than any is my guess. Panasonic FC series are cheap and very low distortion.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.