TL072 or LF353 in a GK 250 ML amp ? Preamp in noisy ...

[toby888]

The schematic I have seen calls for LF353 . The amp in question is loaded with TL072's ( various types ) . The preamp has a bit more white
noise than I would like . Tried a couple of new 72's - does not seem much better . Would going to an LF353 or something else ( OPA2134 / OP275 )
help much to kill some of the white noise ? The amp is not bad with the volume turned down to zero ( you can hear a bit of noise ) , but CH A is fairly noisy when turned up to 5 .

[darkfenriz]

OPA2134 is an upgrade option so may be worth trying.

On the other hand is there anything that makes you think the op-amp performance that is the bottleneck (and not, say, layout, PSU, instabilities...) ?

[toby888]

Thank you for the advice ! I am pretty sure the circuit is stock and fine ( the noise is not really random , just straight adhead white noise ) . This guy posted a video
where he compared noise level of different opamps in his ADA preamp :
https://www.youtube.com/watch?v=6xRAAWZeC9Q&t=212s

[darkfenriz]

If the excess noise is on the clean channel just try swapping U1 to OPA2134 and see. Moving on R5 through R9 arrangement looks suspicious and might be considered to have resistors upgraded, but start with U1.

[Thoriated Tungsten]

The amp in question seems to be poorly designed from a noise point of view. A (very) casual noise analysis suggests that even if you had a mathematically perfect opamp (zero noise) plugged in, the input stage alone would limit the (signal+noise)/noise ratio to around 60 dB, which is not terribly good. Then comes the rest of the circuit and adds more noise on top of this.

An OPA2134 would be less noisy compared to a LF353, but I wouldn't exactly expect miracles. The opamp isn't the sole problem here.

Of course I could be completely wrong, as this conclusion is based on just a quick 'back of the envelope' analysis using a pocket calculator.

[toby888]

Thanks ! GK just said that the LF353 would be noisy compared to the 72 or 34 . I will try the 34 first

[toby888]

I did not have an OPA2134 at the shop . Put in an OPA2604 for now . It did help a little but I think you (TTung) are correct about the design . Someone had also adjusted CH A gain + compressor settings . It works better now . The electrolytics
are from 84 but they measure fine . Will try the OPA2134 later once I get one ( American ?? ) .

[darkfenriz]

The amp in question seems to be poorly designed from a noise point of view. A (very) casual noise analysis suggests that even if you had a mathematically perfect opamp (zero noise) plugged in, the input stage alone would limit the (signal+noise)/noise ratio to around 60 dB, which is not terribly good. Then comes the rest of the circuit and adds more noise on top of this.

An OPA2134 would be less noisy compared to a LF353, but I wouldn't exactly expect miracles. The opamp isn't the sole problem here.

Of course I could be completely wrong, as this conclusion is based on just a quick 'back of the envelope' analysis using a pocket calculator.

I have no idea how you might have calculated this, but I hope you didn't just take the input 1Meg resistor as a thermal noise source, which it isn't, practically speaking.
WHat concerns me is relatively low gain of the input stage and attenuation before the signal reaches the second stage...

[J M Fahey]

Up to a point, "itīs in the nature of the beast" ... itīs an ambitious but *old* design, and it shows its age.

Improving Op Amps will help, no doubt, but some hiss will still remain.

That said, used live itīs perfectly adequate, specially if you plug it into *real* speakers, and then, stand back.

Was very popular here in Argentina ...in the 80īs of course, and I sold tons of stereo 2 x 12" cabinets for them.

[Thoriated Tungsten]

I have no idea how you might have calculated this, but I hope you didn't just take the input 1Meg resistor as a thermal noise source, which it isn't, practically speaking.
WHat concerns me is relatively low gain of the input stage and attenuation before the signal reaches the second stage...

I ignored the 1Meg resistor for my calculations, seeing that it is in parallel with the much lower source resistance in the guitar pickups, which I also ignored for my initial calculation (which I probably shouldn't have). Instead I simply calculated the total RMS resistor noise on the input side by root squaring the noise voltages of the parallel resistance of the feedback network of the input stage, 8.3K, and the 12k RFI suppressor resistor. Then multiplied the noise voltage by the stage voltage gain. The new voltage was then compared to the supplied output RMS voltage given in the service manual of 2.4mV, though that does seem kind of low. But the initial value I got was about 60dB (S+N)/N while completely ignoring the opamp current and voltage noise densities. Think I guesstimated the effective bandwidth to be 5kHz.

Today it ought to be possible to redesign the first two input stages to a point, where the total noise of the preamp would be completely dominated by the guitar pickups, though possibly not without also designing a new PCB layout

[darkfenriz]

I ignored the 1Meg resistor for my calculations, seeing that it is in parallel with the much lower source resistance in the guitar pickups, which I also ignored for my initial calculation (which I probably shouldn't have). Instead I simply calculated the total RMS resistor noise on the input side by root squaring the noise voltages of the parallel resistance of the feedback network of the input stage, 8.3K, and the 12k RFI suppressor resistor. Then multiplied the noise voltage by the stage voltage gain. The new voltage was then compared to the supplied output RMS voltage given in the service manual of 2.4mV, though that does seem kind of low. But the initial value I got was about 60dB (S+N)/N while completely ignoring the opamp current and voltage noise densities. Think I guesstimated the effective bandwidth to be 5kHz.

Sounds about right for the input stage, good methodology. Unfortunately the second stage (the other part of U1) works with higher resistances (R5-R9) and the signal as low as 0.19mV (sic!) if service manual is to be believed.
Luckily I never had such problems with my playing technique and intonation which is probably at <10dB SNR - maybe if I practice more...

[Thoriated Tungsten]

Sounds about right for the input stage, good methodology. Unfortunately the second stage (the other part of U1) works with higher resistances (R5-R9) and the signal as low as 0.19mV (sic!) if service manual is to be believed.

Yup, which is why I stated in my first post that I only considered the first stage, seeing that things would only go downhill from there...

Luckily I never had such problems with my playing technique and intonation which is probably at <10dB SNR - maybe if I practice more...

Heh. I hear ya'...

[doctor]

The OPA1642/OPA1652/OPA2140 are better, but the OPA2827 is the best.