Hi Helma
Yes the MJM4558 is the Japan version of the RC4558 a revamped upgrade of MC1458 . The noise in the audio region is 10nV/Hz^1/2 due to PNP input transistors (no super beta) nice improvement over the 741/1458 but poor for today standard . The second improvement is the slew rate 1v/µS , nothing to "wow" but twice of the 741/1458 and enough to recover the full stroke in audio bandwidth . In direct MM phono one stage this chip is not ridiculous and can be a kind of economical all runner , he beat all FET chip TLO's and so on (for phono MC stage) . But it is a 2Komhs chip classical 2 stage design with 110dB gain a little short to drive a low impedance RIAA network . The 4559 is the uncompensated version same gain but 2v/µS . The 4556 had the same gain and noise with 3v/µS and the ability to drive low loads (350 Ohms). NJM is essentially a manufacturer that repack classical commercial chip they have the same part number that the original and often the availability in the specific Japanese single in line pack . They don't make bad chip , they pack chip from other manufacturer for the japan factory market . The NE5534 is an NJM5534 , the 5532 an NJM5532 . Where also interesting the NJM 4580 or 4560 .
Here the link to the manufacturer OP amp low noise page .
Operational Amplifiers (Op Amps) | Parametric Search | Semiconducter | New Japan Radio(New JRC)Operational Amplifiers (Op Amps) | Parametric Search | Semiconducter | New Japan Radio(New JRC)
Test the SIP8 package
Your choice is wise with the 2041 with better noise and slew characteristic . He is available and not far from ideal characteristic .
Simplex
Yes the MJM4558 is the Japan version of the RC4558 a revamped upgrade of MC1458 . The noise in the audio region is 10nV/Hz^1/2 due to PNP input transistors (no super beta) nice improvement over the 741/1458 but poor for today standard . The second improvement is the slew rate 1v/µS , nothing to "wow" but twice of the 741/1458 and enough to recover the full stroke in audio bandwidth . In direct MM phono one stage this chip is not ridiculous and can be a kind of economical all runner , he beat all FET chip TLO's and so on (for phono MC stage) . But it is a 2Komhs chip classical 2 stage design with 110dB gain a little short to drive a low impedance RIAA network . The 4559 is the uncompensated version same gain but 2v/µS . The 4556 had the same gain and noise with 3v/µS and the ability to drive low loads (350 Ohms). NJM is essentially a manufacturer that repack classical commercial chip they have the same part number that the original and often the availability in the specific Japanese single in line pack . They don't make bad chip , they pack chip from other manufacturer for the japan factory market . The NE5534 is an NJM5534 , the 5532 an NJM5532 . Where also interesting the NJM 4580 or 4560 .
Here the link to the manufacturer OP amp low noise page .
Operational Amplifiers (Op Amps) | Parametric Search | Semiconducter | New Japan Radio(New JRC)Operational Amplifiers (Op Amps) | Parametric Search | Semiconducter | New Japan Radio(New JRC)
Test the SIP8 package
Your choice is wise with the 2041 with better noise and slew characteristic . He is available and not far from ideal characteristic .
Simplex
Phew, working on that A-700 phono section was tight without disassembling the whole amp into pieces, but I managed to get the ICs, 1000uF caps and 47uF bipolars changed. Also changed the 1000uF, 330uF & 680uF electrolytics on the main board. I have a full set of electrolytics for the amp, but I think I'll leave the small ones for later.
NJM2068s for the phono stage like the A-520 got. Spinning first disc now and it sounds good, but it's not a record I've listened to regularly enough to say much about the sound
I left the servo as it is, just replaced the opamp and gave the caps a refresh.
NJM2068s for the phono stage like the A-520 got. Spinning first disc now and it sounds good, but it's not a record I've listened to regularly enough to say much about the sound
I left the servo as it is, just replaced the opamp and gave the caps a refresh.
Yeah, actually I did order the necessary components for the switch, but decided to have faith in the Yamaha engineers and left it stock except changing the op-amp. That IC was one of the few components that was actually easy to get to. The other stuff, I would've had to unsolder something like 30 cable connectors to get another board out of the way, unless there's an easier way to do it that didn't occur to me.
I got wondering if the servo circuit would have any effect on noise performance? Could rising the resistor value from 6k8 to 300k+ introduce more resistor noise? It's just I wondered why they chose the components they did and if it was an engineering decision or cost based. I can imagine 6k8 could be a compromise between going low and still cutting the driver stage some slack.
There was an M5128 (or something) op-amp in the servo circuit and I noticed the same op amp was used in the A-520 phono stage but as the main IC if I recall correctly. So it could be they had mountains of those chips and used them in their cheaper gear and relegated them to servo duty in better models.
The 6k8's on each input would have been chosen as low value to keep the IC's own offset errors in check.
It works like this. A bjt (transistor input) opamp like this has a small but not negligible bias current flowing into or out of each input pin. In other words the base bias current for the input stage.
Although the current is very small, it is still enough to develop a DC voltage across the 6k8's. That current causes the DC conditions to be non ideal i.e. for a servo it would shift the operating point away from zero.
The bias current is constant and so if we make the resistor larger, then the voltage developed across the resistor becomes larger (ohms law).
eg.
100 nano amps bias current flowing through 6k8 develops just 0.00068 volts.
Change the 6k8 to a 1meg and we get .1 volts error. Massive difference.
A fet opamp has (for all practical purposes) bias currents in the pico amp or lower region and so that allows us to use very high value resistors.
Noise... there is no penalty because the opamp is configured as an integrator which effectively kills all the AC component. Remember you are feeding signal into the integrator and so the noise of the 1meg is still magnitudes below the audio anyway.
Random numbers to illustrate.
Its like adding 0.00000001 of noise for a 6k8 to say 7 of signal versus 0.00001 + 7 for a 1meg. The opamp removes the total anyway.
The big advantages. We can use a much higher quality cap for the servo which filters the signal and this is much better at higher frequencies than a larger and lossy electrolytic. This is one application where it really does make a difference.
If you use a FET opamp then the other 6k8 (to the +input) which was needed to equalise the DC bias errors must be shorted out. Leaving it in place simply creates a node that could theoretically pick up stray interference as it is at 6k8 impedance. Make it zero and we kill any possibility of that.
The FET opamp is win win win all round.
It works like this. A bjt (transistor input) opamp like this has a small but not negligible bias current flowing into or out of each input pin. In other words the base bias current for the input stage.
Although the current is very small, it is still enough to develop a DC voltage across the 6k8's. That current causes the DC conditions to be non ideal i.e. for a servo it would shift the operating point away from zero.
The bias current is constant and so if we make the resistor larger, then the voltage developed across the resistor becomes larger (ohms law).
eg.
100 nano amps bias current flowing through 6k8 develops just 0.00068 volts.
Change the 6k8 to a 1meg and we get .1 volts error. Massive difference.
A fet opamp has (for all practical purposes) bias currents in the pico amp or lower region and so that allows us to use very high value resistors.
Noise... there is no penalty because the opamp is configured as an integrator which effectively kills all the AC component. Remember you are feeding signal into the integrator and so the noise of the 1meg is still magnitudes below the audio anyway.
Random numbers to illustrate.
Its like adding 0.00000001 of noise for a 6k8 to say 7 of signal versus 0.00001 + 7 for a 1meg. The opamp removes the total anyway.
The big advantages. We can use a much higher quality cap for the servo which filters the signal and this is much better at higher frequencies than a larger and lossy electrolytic. This is one application where it really does make a difference.
If you use a FET opamp then the other 6k8 (to the +input) which was needed to equalise the DC bias errors must be shorted out. Leaving it in place simply creates a node that could theoretically pick up stray interference as it is at 6k8 impedance. Make it zero and we kill any possibility of that.
The FET opamp is win win win all round.
Last edited:
Mooly thanks for the informative reply. If the board were easier to get to, I'd definitely experiment with that, but it's so tight in there it kills all the fun. I was thinking of trying it out on the A-520, but it uses one dual op-amp per channel, where the other half does servo duty. Might try it out at some point anyway, but it would be impossible to tell what effect adjusting the servo circuit was having if any, and what is just changing the op-amp.
I've got some more hours on the amp now and the phono stage is definitely better. It used to be using class A mode was pointless when using phono as a source, because the phono pre was the bottleneck and all the extra heat really made no perceivable difference but that's not the case any more. During winter months I can certainly use the extra warmth for the room as well, so it's a win-win.
I've got some more hours on the amp now and the phono stage is definitely better. It used to be using class A mode was pointless when using phono as a source, because the phono pre was the bottleneck and all the extra heat really made no perceivable difference but that's not the case any more. During winter months I can certainly use the extra warmth for the room as well, so it's a win-win.
Probably. I love this quote I lifted from this forum elsewhere, though I don't recall where:
"when I see a 4558 or one of its variants in a piece of gear, my thoughts immediately turn to ripping that thing out and replacing it with something (just about anything) else."
Hello
The JRC4558 is not "the worst" it is only outdated for demanding application . If you use the commanded voltage , no high current output not to much gain he is OK . It is a better µA741 . Today you will find certainly a better OP Amp but not with certainly better result . It depend of the circuit design .
Your reference use low voltage but this chip is for a classical +/- 15V and can't drive low impedance .
In this design I am sure that a NE5532 equivalent like JRC4580 would be nice . He had the noise , the output ..
The JRC4558 is not "the worst" it is only outdated for demanding application . If you use the commanded voltage , no high current output not to much gain he is OK . It is a better µA741 . Today you will find certainly a better OP Amp but not with certainly better result . It depend of the circuit design .
Your reference use low voltage but this chip is for a classical +/- 15V and can't drive low impedance .
In this design I am sure that a NE5532 equivalent like JRC4580 would be nice . He had the noise , the output ..
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.