Equivalent input noise currents:
LT1028, LT1128, LT1115: 3.4 pA/√Hz at 1 kHz when measured under realistic conditions
LM4562: 1.6 pA/√Hz typical at 1 kHz, but some are known to have popcorn noise, those are probably far worse
NE5532: 0.6 pA/√Hz at 1 kHz, but looking at the internal circuit and the bias current spec, it should drop below 0.3 pA/√Hz at high audio frequencies
OPA2134: negligible
Values above 0.81 pA/√Hz exceed the thermal noise of the 100 kohm potmeter when it is set halfway. A much lower potmeter resistance would change the picture completely.
LT1028, LT1128, LT1115: 3.4 pA/√Hz at 1 kHz when measured under realistic conditions
LM4562: 1.6 pA/√Hz typical at 1 kHz, but some are known to have popcorn noise, those are probably far worse
NE5532: 0.6 pA/√Hz at 1 kHz, but looking at the internal circuit and the bias current spec, it should drop below 0.3 pA/√Hz at high audio frequencies
OPA2134: negligible
Values above 0.81 pA/√Hz exceed the thermal noise of the 100 kohm potmeter when it is set halfway. A much lower potmeter resistance would change the picture completely.
Good that you do not have an oscillation problem. Just keep that point in mind for future work 😊
Re the noise and the pot position. Noise will always be highest when the pot is set to mid position since this is when the source impedance for the following stage is highest. Although the LM4562 has low noise voltage, it’s noise current is high, and that is why the NE5532 is quieter with a 100k pot - it’s noise current is only about 0.4pA/rt Hz.
A 100k pot for level or tone control is a very high value and you will run into noise problems. I usually run all my level pots at 10k and on some occasions 20k. If you use a good opamp like an AD797, the source impedance of a 10k pot at electrical mid position (the noisiest point) is 2.5k and the result is you still get a very low noise preamp because the opamp noise current flowing through this low source impedance does not add much additional noise to the opamp noise voltage. However, if you replaced the 10k pot with 100k and compared the AD797 with an OPA1641 JFET input opamp, the 1641 would be much quieter despite being more noisy with 10k pot.
To summarise:-
1. Try to use the lowest value pot you can. 10k is a good compromise and allows the use of bipolar or JFET input opamps with similar good noise performance results
2. If using high value pots of >20k, use them with JFET input opamps. JFET opamp input noise currents are negligible so they will perform better
3. If you are using a bipolar input opamp like a LM4562 or OPA1612, keep the pot values to below 20k, 10k preferable. The NE5532/4 is s bit different because of its low noise current and will therefore tolerate 20k pots and still give good results
4. In general, good bipolar opamps have lower noise voltage than JFET types. But good JFET opamps haven negligible noise current.
Re the noise and the pot position. Noise will always be highest when the pot is set to mid position since this is when the source impedance for the following stage is highest. Although the LM4562 has low noise voltage, it’s noise current is high, and that is why the NE5532 is quieter with a 100k pot - it’s noise current is only about 0.4pA/rt Hz.
A 100k pot for level or tone control is a very high value and you will run into noise problems. I usually run all my level pots at 10k and on some occasions 20k. If you use a good opamp like an AD797, the source impedance of a 10k pot at electrical mid position (the noisiest point) is 2.5k and the result is you still get a very low noise preamp because the opamp noise current flowing through this low source impedance does not add much additional noise to the opamp noise voltage. However, if you replaced the 10k pot with 100k and compared the AD797 with an OPA1641 JFET input opamp, the 1641 would be much quieter despite being more noisy with 10k pot.
To summarise:-
1. Try to use the lowest value pot you can. 10k is a good compromise and allows the use of bipolar or JFET input opamps with similar good noise performance results
2. If using high value pots of >20k, use them with JFET input opamps. JFET opamp input noise currents are negligible so they will perform better
3. If you are using a bipolar input opamp like a LM4562 or OPA1612, keep the pot values to below 20k, 10k preferable. The NE5532/4 is s bit different because of its low noise current and will therefore tolerate 20k pots and still give good results
4. In general, good bipolar opamps have lower noise voltage than JFET types. But good JFET opamps haven negligible noise current.
Last edited:
awesome so i am on the right track
-will change pots to 10k for sure.
other than the NE5532, is there any other "on the shelf" opamps i shoud try ? i am concerned too about the remaining noise of U1 and U2 too since they seems to be a bad choice for this application
-the OPA2134 seems n/a at digikey for now.
LME4970 seems to have high input bias current too
any other suggestions is more than welcome
-i am also looking at advice for the NIKKO preamp with its weird j-fet/opamp output stage.
thanks !!!
-will change pots to 10k for sure.
other than the NE5532, is there any other "on the shelf" opamps i shoud try ? i am concerned too about the remaining noise of U1 and U2 too since they seems to be a bad choice for this application
-the OPA2134 seems n/a at digikey for now.
LME4970 seems to have high input bias current too
any other suggestions is more than welcome
-i am also looking at advice for the NIKKO preamp with its weird j-fet/opamp output stage.
thanks !!!
I would go over to OPA1642 JEFT input opamp and drop all resistor values by 10x, increase coupling and frequency shaping caps 10x and you will have a quiet system.
https://www.ti.com/product/OPA1642?qgpn=opa1642
https://www.ti.com/product/OPA1642?qgpn=opa1642
intersting but they are only available in SMD....my actual PCB is all thru hole....might re-do it then....smd to dip adapter are not cheap either, good quality ones are 5$ and up....
Unfortunately DIP is on its way out - all the new high performance opamps are only available in SMD. SMD is easy to solder - you should learn it 🙂
With 10 kohm potmeters, both NE5532 and LM4562 will produce less noise than the potmeter itself does when it is set halfway.
Another issue may be anodization of the wipers. If you use carbon potmeters, a bias current drawn out of the wiper will slowly cause a bad contact between the resistive element and wiper. That's no issue with JFET op-amps as those have very small input bias currents.
Another issue may be anodization of the wipers. If you use carbon potmeters, a bias current drawn out of the wiper will slowly cause a bad contact between the resistive element and wiper. That's no issue with JFET op-amps as those have very small input bias currents.
i solder on daily basis smd, 0402 components for RF applications and industrial applications.
Marcel, the pots are brand new. sure enough they are what digikey had, Bourns carbon linerar. 10k are on its way.
Does conductive plastic (instead of carbon) are less noisey ?
Does conductive plastic (instead of carbon) are less noisey ?
The thermal noise is exactly the same, 1/f noise might be different, but should be far from dominant anyway. My comment about carbon potmeters was related to long-term reliability, not to the hiss issue you have now.
ok so other than price, nothing seems to dictate a preference, its not something that will be played it all day long. altough my searches on this forum showed few comments about the c-plastic ones beign less noisey...
I've order the adapters directly from TI, $10.00 for a panel of adapters.
https://www.ti.com/tool/DIP-ADAPTER-EVM?keyMatch=&tisearch=search-everything&usecase=partmatches
TL07x and TL08x have more voltage noise but still widely used for low current noise and cheap.
TL072 18 nV/√Hz (typ) at f = 1 kHz 0.01 pA/√Hz
Other easy "sanity checks": install shorts across U2A pins 6 & 7 and pin 5 to ground to get Vn of U1B; install short across U1B pins 1 & 2 to get Vn of (U2A + R12).
TL072 18 nV/√Hz (typ) at f = 1 kHz 0.01 pA/√Hz
Other easy "sanity checks": install shorts across U2A pins 6 & 7 and pin 5 to ground to get Vn of U1B; install short across U1B pins 1 & 2 to get Vn of (U2A + R12).
some update
i ended up with better results
reduced gain resistor and all level potentiomenter /10
i played with opamps and ended up using LM4562 for U1 and U3 and NE5532 for U2, this gave me the lowest noise possible
didnt changed the 100k dual pot for frequency and decade capacitor. for now it is acceptable altough there is still some hiss but much lower.
The crossover is the lowest item on the picture.
The Alpha 220 powers up JBL2123h and BeymaCP21f crossed at 5k
The BGW powers up 4x Goldwood GW15PC4, crossed at 200-250hz
Thanks for all the helps !!!
i ended up with better results
reduced gain resistor and all level potentiomenter /10
i played with opamps and ended up using LM4562 for U1 and U3 and NE5532 for U2, this gave me the lowest noise possible
didnt changed the 100k dual pot for frequency and decade capacitor. for now it is acceptable altough there is still some hiss but much lower.
The crossover is the lowest item on the picture.
The Alpha 220 powers up JBL2123h and BeymaCP21f crossed at 5k
The BGW powers up 4x Goldwood GW15PC4, crossed at 200-250hz
Thanks for all the helps !!!