Still rooms for improvements request for comments (RFC)s
Fourier at point Y
Low hanging fruit, alternative with fewer components and competitive performance
https://www.pcbway.com/project/shareproject/Phono_preamplifier_LT1115_add89c2f.html
Fourier at point Y
Low hanging fruit, alternative with fewer components and competitive performance
https://www.pcbway.com/project/shareproject/Phono_preamplifier_LT1115_add89c2f.html
Last edited:
LT1115 has an equivalent input noise current that is rather high for moving magnet (3.4 pA/√Hz).
The schematic in post #1 has no connection between the signal source and the amplifier and the inductance for moving magnet is about 300 times as high as what I'm used to. The moving coil inductance also seems rather high. That's assuming L1 and L2 are not coupled; I don't see anything indicating that they are.
Decoupling the base of Q2 would short the noise of R9 and R10, but that is probably far from dominant anyway if the JFETs are in saturation.
If you haven't already done so, you could reduce the distortion of the JFETs at the expense of a bit more noise by reducing the base voltage of Q2 just enough to get the JFETs into the triode region. That would make the circuit rather sensitive to JFET component spread, though.
The schematic in post #1 has no connection between the signal source and the amplifier and the inductance for moving magnet is about 300 times as high as what I'm used to. The moving coil inductance also seems rather high. That's assuming L1 and L2 are not coupled; I don't see anything indicating that they are.
Decoupling the base of Q2 would short the noise of R9 and R10, but that is probably far from dominant anyway if the JFETs are in saturation.
If you haven't already done so, you could reduce the distortion of the JFETs at the expense of a bit more noise by reducing the base voltage of Q2 just enough to get the JFETs into the triode region. That would make the circuit rather sensitive to JFET component spread, though.
Shockingly high for MM input in fact, giving the equivalent of >100nV/√Hz at 10kHz for a 500mH cartridge, making a complete mockery of its 0.9nV/√Hz voltage noise spec. LT1115 is for MC only, not MM!!
Note the current noise for the LT1115 is quoted lower than 3.4pA, but that's because the datasheet value assumes symmetrical input impedances and the chip has bias-current cancellation circuitry, often the bain of the low noise designer. If you ever seen bias current for a bipolar-input opamp quoted as +/- some value, this means cancellation circuitry and the current noise will be much worse than advertized in practice in a single-ended preamp.
In the same situation NE5534A would give equiv of 12.5nV/√Hz at 10kHz, 18dB less hiss... It has a true 0.4pA/√Hz current noise spec.
Basically for MM use either NE5534A (or other genuinely low current noise bipolar opamp) or a FET input opamp.
Note the current noise for the LT1115 is quoted lower than 3.4pA, but that's because the datasheet value assumes symmetrical input impedances and the chip has bias-current cancellation circuitry, often the bain of the low noise designer. If you ever seen bias current for a bipolar-input opamp quoted as +/- some value, this means cancellation circuitry and the current noise will be much worse than advertized in practice in a single-ended preamp.
In the same situation NE5534A would give equiv of 12.5nV/√Hz at 10kHz, 18dB less hiss... It has a true 0.4pA/√Hz current noise spec.
Basically for MM use either NE5534A (or other genuinely low current noise bipolar opamp) or a FET input opamp.
It depends a lot on how the bias cancellation is implemented and how honest the manufacturer is.
By the way, if you are interested in the weighted integral noise, you can calculate with the cartridge impedance at about 3852 Hz for RIAA- and A-weighting and at 5179 Hz for RIAA- and ITU-R 468 weighting. The conclusion remains the same: 3.4 pA/√Hz is quite high.
By the way, if you are interested in the weighted integral noise, you can calculate with the cartridge impedance at about 3852 Hz for RIAA- and A-weighting and at 5179 Hz for RIAA- and ITU-R 468 weighting. The conclusion remains the same: 3.4 pA/√Hz is quite high.
Thank you all for the input. @MarcelvdG, in the first schematic that's a step up transformer.
In addressing that noise performance alternatives are needed. Chip still looks good for MC. In the first schematic we could replace the costlier ($130+) transformer with the LT1115. Once our level is in the 5mV range we have more amplification solutions
In addressing that noise performance alternatives are needed. Chip still looks good for MC. In the first schematic we could replace the costlier ($130+) transformer with the LT1115. Once our level is in the 5mV range we have more amplification solutions
This is the more realistic graph, as in reality you never have the exact same impedance driving the positive and negative inputs.
Regarding the "RS noise only" line, keep in mind that for a given impedance, a largely inductive source that only generates thermal noise would produce less noise than a resistor. This applies to a cartridge not playing any record. Then again, it would be higher for a cartridge playing a dusty record.
Anyway, the conclusion remains that it is pretty good for MC, but not for MM.
Last edited:
@MarcelvdG,
We can use the cheaper LM4562
LM4562 2Euro6
LT1115 8Euro4
The LT1115/LM4562 solution does work, not forgetting the SNR for record pressing is even lower
We can use the cheaper LM4562
LM4562 2Euro6
LT1115 8Euro4
The LT1115/LM4562 solution does work, not forgetting the SNR for record pressing is even lower
People can try different chips on the socketed PCBs. There's also OPA1656. One could also stick a JFE150 or LSK170 infront of most opamps like the first schematic
Don't quite see the similarities with the Pearl designs by Wayne or why use the Pearl-name in the first place.
As far as I know, all Pearl designs are two stage, yours is three. The Pearl-3 run U1A at high gain and has feedback to R4-R1 junction.
Your design with two cascaded linear amp (JFET without NFB and the first op-amp) would have a rather high gain. Have you verified that you actually have the required headroom for MM signals?
What are your design improvements over the Pearl 1, 2 and/or 3 designs in the first place and why do you have R4 there if you have no feedback to this junction?
Also your choice of input JFETs would have a rather high input capacitance for a MM stage (since the input stage is running without feedback). I would look towards the LSK489 in this case.
@MarcelvdG: The input stage is is a cascoded JFET without feedback, the op-amp is the second linear gain stage. Why would opamp-input current noise and cartridge impedance matter much towards noise contribution in this case? would that not be defined by the JFET input stage? Not that I know how to calculate voltage or current noise for this stage in the first case....
The second gain stage is the inverting op-amp, It would be feed by the cascode output independence amp series resistor of 499 ohm (vs Wayne's design where this is actually a 22K ohm resistor in the middle of the feedback loop)
As far as I know, all Pearl designs are two stage, yours is three. The Pearl-3 run U1A at high gain and has feedback to R4-R1 junction.
Your design with two cascaded linear amp (JFET without NFB and the first op-amp) would have a rather high gain. Have you verified that you actually have the required headroom for MM signals?
What are your design improvements over the Pearl 1, 2 and/or 3 designs in the first place and why do you have R4 there if you have no feedback to this junction?
Also your choice of input JFETs would have a rather high input capacitance for a MM stage (since the input stage is running without feedback). I would look towards the LSK489 in this case.
@MarcelvdG: The input stage is is a cascoded JFET without feedback, the op-amp is the second linear gain stage. Why would opamp-input current noise and cartridge impedance matter much towards noise contribution in this case? would that not be defined by the JFET input stage? Not that I know how to calculate voltage or current noise for this stage in the first case....
The second gain stage is the inverting op-amp, It would be feed by the cascode output independence amp series resistor of 499 ohm (vs Wayne's design where this is actually a 22K ohm resistor in the middle of the feedback loop)
I see,
As you point out Marcel, that circuit from the Linear Tech appnote has high current noise. In my opinion it should best be used with MC or possibly HOMC pickups.
Although no mention is made of pickup impedance / inductance in the appnote, there is a section called "application information" that explains current vs voltage noise rather well. By the way, this section has been updated in the LT1028 datasheet, including a section on paralleling of amps to reduce voltage noise.
For a typical MM pickup like Ortofon 2M series the appnote recommendation would be LT1037, although I'm still not convicted for RIAA duties due to the input current cancelation network used.
For universal MM my choice of LT family op-amps would be the LT1113 that sports JFET input for negligible current noise and voltage noise only 1dB above the now obsolete NR5534.
Table from LT1115 datasheet:
As you point out Marcel, that circuit from the Linear Tech appnote has high current noise. In my opinion it should best be used with MC or possibly HOMC pickups.
Although no mention is made of pickup impedance / inductance in the appnote, there is a section called "application information" that explains current vs voltage noise rather well. By the way, this section has been updated in the LT1028 datasheet, including a section on paralleling of amps to reduce voltage noise.
For a typical MM pickup like Ortofon 2M series the appnote recommendation would be LT1037, although I'm still not convicted for RIAA duties due to the input current cancelation network used.
For universal MM my choice of LT family op-amps would be the LT1113 that sports JFET input for negligible current noise and voltage noise only 1dB above the now obsolete NR5534.
Table from LT1115 datasheet:
Main voltage amplification phono stage noise (simulated)
130nV/√Hz-155nV/√Hz - pearl X phono jfet no feedback input
360nV/√Hz - pearl 3 phono
10nV/√Hz(MM)-200nV/√Hz(MC) - LT1115 phono
In comparison the Morpheus III power amplifier in solid state does 185nV/√Hz
Pearl3 Test Rig
130nV/√Hz-155nV/√Hz - pearl X phono jfet no feedback input
360nV/√Hz - pearl 3 phono
10nV/√Hz(MM)-200nV/√Hz(MC) - LT1115 phono
In comparison the Morpheus III power amplifier in solid state does 185nV/√Hz
Pearl3 Test Rig
It's quite noisy for moving magnet when you simulate or measure with a realistic source impedance.
See "Noise and moving-magnet cartridges", Electronics World October 2003, pages 38...43, https://worldradiohistory.com/UK/Wireless-World/00s/Electronics-World-2003-10-S-OCR.pdf Mind you, Electronics World drew one of the sections of the gain switch in the wrong state in figure 5 and I mixed up the terms spectral density and power spectral density.
Last edited: