My Commercial pre-amp has a RIAA based on 2-stage amplification using 5532 op amps from TI.
I am happy with the RIAA as it is, comparing favorably to other amps and external RIAA that I've tesed. However, beeing a DIY'er I have difficulties leaving things alone and I suspect that there are improvements to be had with amps of better PSRR and comparable or better noise characteristics.
On closer inspection it seems that no EQ is applied at the first stage, except maybe a 50 kHz roll-off. First stage has amplification switchable between 3x and 27x depending on MM/MC setting. The second stage is an "all in one" type, ref attached diagram.
Now I wonder if replacing the 5532 with something like LME49720 / LME49723 or OA2104 would bring about a real improvement to the circuit. Power supply decoupling seems to be good as far as passive RC filtering of the supply rails goes (no active regulators close to the RIAA stage).
Since the current design is for a dual-surface mount op-amp, I am also considering the use of a Brown-Dog (or similar) adaptor allowing use of two single amps instead instead of one double. This will allow different single op-amps for first and second stage. For example LME49710 on the input and AD797 for the equalizer.
Is this a worth-while project or am I better of making an external RIAA from scratch?bIf going forward with this - What amp would you propose, and are there any precautions necessary (Stability, noise, stray Capacitance etc.) if I decide to use the Brown-Dog adaptors? My main source is a MM pickup (Rega Super Exact), but I do not want the amp to be any less suited to MC pickups after the mod.
I am happy with the RIAA as it is, comparing favorably to other amps and external RIAA that I've tesed. However, beeing a DIY'er I have difficulties leaving things alone and I suspect that there are improvements to be had with amps of better PSRR and comparable or better noise characteristics.
On closer inspection it seems that no EQ is applied at the first stage, except maybe a 50 kHz roll-off. First stage has amplification switchable between 3x and 27x depending on MM/MC setting. The second stage is an "all in one" type, ref attached diagram.
Now I wonder if replacing the 5532 with something like LME49720 / LME49723 or OA2104 would bring about a real improvement to the circuit. Power supply decoupling seems to be good as far as passive RC filtering of the supply rails goes (no active regulators close to the RIAA stage).
Since the current design is for a dual-surface mount op-amp, I am also considering the use of a Brown-Dog (or similar) adaptor allowing use of two single amps instead instead of one double. This will allow different single op-amps for first and second stage. For example LME49710 on the input and AD797 for the equalizer.
Is this a worth-while project or am I better of making an external RIAA from scratch?bIf going forward with this - What amp would you propose, and are there any precautions necessary (Stability, noise, stray Capacitance etc.) if I decide to use the Brown-Dog adaptors? My main source is a MM pickup (Rega Super Exact), but I do not want the amp to be any less suited to MC pickups after the mod.
Attachments
LME49720 and AD797 both have an equivalent input noise current that is a bit high for moving magnet. Using one of those in the first stage would reduce the noise floor for moving coil, but increase it for moving magnet. So all in all, you have to decide whether you want to optimize the amplifier for moving magnet, moving coil, or make a compromise that works reasonably well with both.
To find a good compromise input op-amp for a combined moving magnet/moving coil amplifier, look up the 1 kHz noise density values and pick the type for which the largest of the voltage noise and the current noise multiplied by 1.2 kohm is minimal. That is, each pA/sqrt(Hz) counts as 1.2 nV/sqrt(Hz). The idea behind that is explained in this thread: https://www.diyaudio.com/forums/analogue-source/338620-opamp-phono-noise-4.html#post5811627
Examples:
NE5534A: 3.5 nV/sqrt(Hz) and 0.4 pA/sqrt(Hz) -> 3.5 nV/sqrt(Hz) and 0.48 nV/sqrt(Hz)
AD797: 0.9 nV/sqrt(Hz) and 2 pA/sqrt(Hz) -> 0.9 nV/sqrt(Hz) and 2.4 nV/sqrt(Hz)
LME49720: 2.7 nV/sqrt(Hz) and 1.6 pA/sqrt(Hz) -> 2.7 nV/sqrt(Hz) and 1.92 nV/sqrt(Hz)
LT1028: 0.9 nV/sqrt(Hz) and 3.4 pA/sqrt(Hz) -> 0.9 nV/sqrt(Hz) and 4.08 nV/sqrt(Hz)
(The current noise value for the LT1028 is the value for unequal source impedances measured by Hans Polak, not the useless value for equal impedances specified by Linear Technology.)
So from these four, the AD797 is the best compromise op-amp. For an amplifier optimized for moving magnet, the current noise would count ten times as hard and the NE5534A would win. For moving coil, the current noise would do almost nothing and the AD797 and LT1028 would win.
To find a good compromise input op-amp for a combined moving magnet/moving coil amplifier, look up the 1 kHz noise density values and pick the type for which the largest of the voltage noise and the current noise multiplied by 1.2 kohm is minimal. That is, each pA/sqrt(Hz) counts as 1.2 nV/sqrt(Hz). The idea behind that is explained in this thread: https://www.diyaudio.com/forums/analogue-source/338620-opamp-phono-noise-4.html#post5811627
Examples:
NE5534A: 3.5 nV/sqrt(Hz) and 0.4 pA/sqrt(Hz) -> 3.5 nV/sqrt(Hz) and 0.48 nV/sqrt(Hz)
AD797: 0.9 nV/sqrt(Hz) and 2 pA/sqrt(Hz) -> 0.9 nV/sqrt(Hz) and 2.4 nV/sqrt(Hz)
LME49720: 2.7 nV/sqrt(Hz) and 1.6 pA/sqrt(Hz) -> 2.7 nV/sqrt(Hz) and 1.92 nV/sqrt(Hz)
LT1028: 0.9 nV/sqrt(Hz) and 3.4 pA/sqrt(Hz) -> 0.9 nV/sqrt(Hz) and 4.08 nV/sqrt(Hz)
(The current noise value for the LT1028 is the value for unequal source impedances measured by Hans Polak, not the useless value for equal impedances specified by Linear Technology.)
So from these four, the AD797 is the best compromise op-amp. For an amplifier optimized for moving magnet, the current noise would count ten times as hard and the NE5534A would win. For moving coil, the current noise would do almost nothing and the AD797 and LT1028 would win.
Well the switchable feedback network is a bit of a compromise anyway, better to have a separate MC amp as a head-end that can be jumpered in, since the noise issues are very different.
The NE5534A will serve well as a first stage in an MM amp due to the low current noise, as will JFET and CMOS opamps.
A generic MC amp wants very low voltage noise, which typically means an opamp with poor current noise specs (ie one opamp cannot be ideal for both MC and MM). Also discrete front ends can be used for even lower voltage noise (such as the ZTX851 et al).
If the MC cartridge isn't super-low impedance, ultra-low voltage noise becomes less important. The range in impedance is large for MC so its hard to be optimal for all.
A passive transformer stage for MC is another possibility.
The NE5534A will serve well as a first stage in an MM amp due to the low current noise, as will JFET and CMOS opamps.
A generic MC amp wants very low voltage noise, which typically means an opamp with poor current noise specs (ie one opamp cannot be ideal for both MC and MM). Also discrete front ends can be used for even lower voltage noise (such as the ZTX851 et al).
If the MC cartridge isn't super-low impedance, ultra-low voltage noise becomes less important. The range in impedance is large for MC so its hard to be optimal for all.
A passive transformer stage for MC is another possibility.
Yes it can be improved, though the values for the MC setting aren't awful, perhaps reduce all resistors by a factor of 4 or so.
390 ohm means about 2.6 nV/sqrt(Hz), which to my opinion is pretty awful for an MC stage. At least it is much greater than the thermal noise of the cartridge and greater than the voltage noise of two of the four exampe op-amps.It will in most cases still be somewhat smaller than the record surface noise, though.
When you reduce all resistances by a factor of ten and increase the capacitance by a factor of ten, at least you will get standard values for everything.
When you reduce all resistances by a factor of ten and increase the capacitance by a factor of ten, at least you will get standard values for everything.
I would replace the 2nd op amp as it performs RIAA eq. the first one is flat amplification for low noise.
The result may not be audible, 5532 is a good op amp. I opted not to replace the HA1457 in mine as it's similar to 5532 performance wise. I used a low noise transistor for input gain rather then an op amp.
if you reduce the FB resistors of the first stage 1/10 the cap has to be 10x bigger.
LJT, How does your turntable sound next to a CD recording of the same track?
The result may not be audible, 5532 is a good op amp. I opted not to replace the HA1457 in mine as it's similar to 5532 performance wise. I used a low noise transistor for input gain rather then an op amp.
if you reduce the FB resistors of the first stage 1/10 the cap has to be 10x bigger.
LJT, How does your turntable sound next to a CD recording of the same track?
Last edited:
The design uses a dual Op-Amp per channel, and I am likely to stay with my current MM pickup for another few years.
I can only guess that the high impedances used on the Inverting_In is a compromise between dual MM/MC use and to keep switch noise and potential for warranty issues to a minimum.
The resistor of 4K7 is permanently in the circuit and when switched to MC mode it is shunted by 390R.
The amp used by the OEM is a surface mount 5532.
I do not fully grasp the calculations as shown by MarcelvdG above, but it seems I should assume about 12K source impedance for the MM calculations?
From the postings above it is my understanding that the 5532 may be the best Dual Op-amp for MM operation unless I install a Brown Dog adapter and use different devices for first and second stage, in which case a 5534 will give slightly improved performance. Improvements that may only be realized if I also make changes to the feedback network.
Looking at the stage 2 Equalizer. This is now being feed from an op-amp instead of a pickup. What are the criteria when looking for an alternative op-amp here?
I can only guess that the high impedances used on the Inverting_In is a compromise between dual MM/MC use and to keep switch noise and potential for warranty issues to a minimum.
The resistor of 4K7 is permanently in the circuit and when switched to MC mode it is shunted by 390R.
The amp used by the OEM is a surface mount 5532.
I do not fully grasp the calculations as shown by MarcelvdG above, but it seems I should assume about 12K source impedance for the MM calculations?
From the postings above it is my understanding that the 5532 may be the best Dual Op-amp for MM operation unless I install a Brown Dog adapter and use different devices for first and second stage, in which case a 5534 will give slightly improved performance. Improvements that may only be realized if I also make changes to the feedback network.
Looking at the stage 2 Equalizer. This is now being feed from an op-amp instead of a pickup. What are the criteria when looking for an alternative op-amp here?
Cross-posted with stocktrader-200. His post address some of my questions.
As for comparison with digital source: Vinyl is giving a much more lifelike presentation in terms of "space". This compares not only to "CD" and "Master" quality sources from Tidal but also to CD from my Raysonic CD128.
The pre-amps DAC is better than that of my Bluesound Node-2 streamer, but still not as good as Vinyl.
The Phone input has an anomaly in that it gives a muted "thump" about 2 seconds after needle has been lifted of the record.
Other than this I have no grevings with the sound quality of the Phono Input as it is, other than that I Would like to make it better if I can.
I have compared by using an old 80's Denon pre as RIAA only (Connected tape out of the Denon to line in on my pre). That is an map using 8 2SK389 in an LTP on the input. The Denon RIAA gave a much better reproduction, but unfortunately it did not have the same resistance to scratches, making blemishes that went un-noticed on other amps clearly audio-able and anoying.
Adjusting the resistive network around the input stage may be a good place to start.
As for comparison with digital source: Vinyl is giving a much more lifelike presentation in terms of "space". This compares not only to "CD" and "Master" quality sources from Tidal but also to CD from my Raysonic CD128.
The pre-amps DAC is better than that of my Bluesound Node-2 streamer, but still not as good as Vinyl.
The Phone input has an anomaly in that it gives a muted "thump" about 2 seconds after needle has been lifted of the record.
Other than this I have no grevings with the sound quality of the Phono Input as it is, other than that I Would like to make it better if I can.
I have compared by using an old 80's Denon pre as RIAA only (Connected tape out of the Denon to line in on my pre). That is an map using 8 2SK389 in an LTP on the input. The Denon RIAA gave a much better reproduction, but unfortunately it did not have the same resistance to scratches, making blemishes that went un-noticed on other amps clearly audio-able and anoying.
Adjusting the resistive network around the input stage may be a good place to start.
I have only reverse engineered the circuit layout, not taken out and measured the SMD capasitors. However, resistor values in the feedback circuit are the same as in the LME49710 datasheet (attached here). inverting input to ground resistor is 1200 instead of 390, so slightly higher gain even with the pre-amplification of stage-1. I assume this is to level mach the phono input with the internal DAC of my unit.
Attachments
Last edited:
LJT, it sounds like your current preamp is performing very well seeing as you prefer the sound to your Tidal and CD sources played from quality equipment. If you want to lower the noise a bit you can replace the 1st op amp feedback network parts. I would use 2.2k for 10k, 1k for 4.7k , 68 for 390 and Cfb 5x larger. The input resistor 47k and 100p capacitor remain unchanged. Metal film 1% resistors should be used.
This does change the differential loading of the op amp which could be noticeable.
This does change the differential loading of the op amp which could be noticeable.
Last edited:
Yes.
In 2003, I wrote an Electronics World article that shows that minimizing the RIAA- and A-weighted noise integrated over the band is roughly equivalent to minimizing the spot noise at 3852 Hz. A typical moving-magnet cartridge with an inductance of about 0.5 H has an impedance of about 12 kohm at 3852 Hz.
I later extended it with ITU-R 468 weighting, see Linear Audio volume 8. The magic frequency then becomes 5179 Hz, if I remember well, so the impedance gets even higher. Then again, many cartridges are a bit lower in inductance than 0.5 H and noise optima are rather broad anyway, so I always kept 12 kohm as a rule of thumb for moving magnet cartridge impedance.
Not necessarily, by neglecting the current noise term altogether, the people at Elektor have managed to make a moving-magnet amplifier that is so far from the noise optimum that its noise is of the same order as record surface noise, so you lose several dB of signal-to-noise ratio. That is, they have used lots of expensive ultra low noise op-amps to make the highest noise moving-magnet amplifier in history:
Supra 2.0 - High End Preamp for Record Player [150616-I] - Elektor LABS | Elektor Magazine
Regarding THD and IMD, you really have to do your best to make those worse than the distortion of a record, which to the best of my knowledge is usually of the order of a percent. Maybe that will be Elektor's next project, to use ultra low distortion op-amps and still somehow end up with a higher distortion than a record or a cartridge.
Supra 2.0 - High End Preamp for Record Player [150616-I] - Elektor LABS | Elektor Magazine
Regarding THD and IMD, you really have to do your best to make those worse than the distortion of a record, which to the best of my knowledge is usually of the order of a percent. Maybe that will be Elektor's next project, to use ultra low distortion op-amps and still somehow end up with a higher distortion than a record or a cartridge.
Last edited:
Methinks that all this points to the fact that we STILL have not achieved measurements that clearly reflect the true SOUND of audio gear. Why does a Fender tube guitar amplifier sound so GOOD, with all its high THD? Why do some amps sound so different when they are both equally rated at much less than 0.1% THD(the supposed threshold of audibility)? Why do vinyl records and magnetic tape sound better than CDs?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.