I think that's a good choice. I wasn't really interested in the OPA1642 that I first got because the common mode distortion wasn't much improved compared to its predecessor, the OPA2134. But it wasn't until a year and a half ago that I noticed that subsequent changes had made a huge improvement.
Yes, I know the details well because I have seen them before. I also built a phono amplifier with a similar configuration (J-FET + OPAMP (LME49860)). configuration.
Bonsai use LSK389 in parallel with an emphasis on low noise, but when I install the old Shure V-15 Type 3/Type 4 (1.38kΩ 500mH) that I regularly use, there is not a big difference, so I use 2SK2145 as a differential pair. (In the attached simulation circuit, it is 2SK117, but it is actually SMD dual type 2SK2145.)
Bonsai may be familiar with this since you lived in Japan, in my country we can buy it two for 100 JPY at Akizuki Denshi.
Also, despite having very low distortion, the LM4562 and its variants are known to be very sensitive to EMI. However, this configuration (FET+LM4562) makes it much more tolerant of EMI. This is theoretically clear. First, front end is a FET that is more tolerant to EMI, and even if RF propagates through the input stage and demodulation occurs in the LM4562, disturbances generated in the middle stage of the feedback amplifier are compressed by the gain of the previous stage.
I used it for DC servo. Originally, the LME49860 is not suitable for DC servos, so the output coupling cap cannot be omitted. It would have been nice if dual op-amps were available, like an LME49860 on one side and a precision J-FET (like an OPA140) on the other.
Mark's circuit is very similar to mine, but I have a few questions.
1. Why did you choose a single power supply instead of dual supplies? Is there any reason? For example, you wanted to use wall wart…
2. J-FET Input stage load resistance is 120Ω and the degeneration resistance is 470Ω, the gain will only be 1/4 (-12dB) no matter how large the gm device is used. It seems that using only OPAMP has a larger overall loop gain and is advantageous in terms of distortion etc.
I think the reason for doing this is to avoid oscillation.
(The AD797 is unstable at unity gain depending on the conditions, and the LM7171 is recommended to have a noise gain >2.) Have you considered applying the following phase compensation? My phono circuit is a little complicated because it uses 2-pole compensation, but I think that even with 1-pole compensation like the one shown in the left figure, the distortion will be sufficiently low because the phase compensation itself becomes the feedback of the minor loop.
Or prioritize circuit simplification?
Or, if there is a description of the concept of that pre-amp circuit published somewhere, could you please provide the link?
Attachments
Good points mason_f8.
The thing about low noise MM phono amps is there is very quickly a point reached where lower amplifier noise adds no benefit. I used a 110 Ohm source load resistor in my circuit vs 10 Ohms in the original LP 797 Collins design. His input referred spot noise with the input shorted was about 4x lower. However, connect a cartridge (1350 ohms plus 500 mH) and the noise difference between the two amplifiers was just 0.2 dB. So, you really always have to consider the source noise contribution.
The thing about low noise MM phono amps is there is very quickly a point reached where lower amplifier noise adds no benefit. I used a 110 Ohm source load resistor in my circuit vs 10 Ohms in the original LP 797 Collins design. His input referred spot noise with the input shorted was about 4x lower. However, connect a cartridge (1350 ohms plus 500 mH) and the noise difference between the two amplifiers was just 0.2 dB. So, you really always have to consider the source noise contribution.
Weird. Mouser Germany list 130k of BC850C and 28k of BC860C, both from NXP and they are orderable in any quantity from singles to 10k reels.
THAT300 series, again readily available at Mouser, even in some DIL14 packages ;-)
It's perhaps too easy to ask, "please tell me fact X so I don't have to exert any effort myself". But, in my opinion, that is not an excellent way to build permanent, long term knowledge and deep expertise. So as a rule I often do not tell people fact X.
Instead, I prefer to say that some of the answers to some of the questions posed in post #101 above, can be found by searching for the PCB named "Bon Homme Richard" which is the name of the circuit displayed in post #96 of this thread.
You will discover that Bon Homme Richard is one of twelve compatible and interchangeable Front End PCBs, which all work with the Lottery VFET amplifiers (tranche P and also tranche N) and also work with the Ship Of Theseus amplifier. These twelve compatible and interchangeable Front End PCBs are called:
Instead, I prefer to say that some of the answers to some of the questions posed in post #101 above, can be found by searching for the PCB named "Bon Homme Richard" which is the name of the circuit displayed in post #96 of this thread.
You will discover that Bon Homme Richard is one of twelve compatible and interchangeable Front End PCBs, which all work with the Lottery VFET amplifiers (tranche P and also tranche N) and also work with the Ship Of Theseus amplifier. These twelve compatible and interchangeable Front End PCBs are called:
- Scourge (sold in the diyAudio Store HERE)
- Bulwark (sold in the diyAudio Store)
- Marauder (sold in the diyAudio Store)
- Dreadnought (sold in the diyAudio Store)
- Lottery Front End NP
- Nimitz (discussed in Ship Of Theseus post #1 found HERE)
- Missouri
- Lexington
- Bon Homme Richard
- Kitty Hawk
- Hornet
- Pequod
See, I don’t do that. When asked a question, I try to provide as much relevant information as possible commensurate with their question. In doing so, I hope to prevent the audience from making assumptions and mistakes.
Mason_f8 is not asking grey questions, his query has definite answers.
The LME49720 is the same IC as the 4562 - and you can easily get them in SOIC format: https://www.ti.com/product/LME49720/part-details/LME49720MAX/NOPB
The 1656 may work as well, if not better - but make sure to take ESD precautions with the CMOS inputs... (don't trust the datasheet - CMOS is CMOS...)
The 1656 may work as well, if not better - but make sure to take ESD precautions with the CMOS inputs... (don't trust the datasheet - CMOS is CMOS...)
Sure but this is not available in my neighborhood, only DIP version. Ok so I'll try with OPA1656.
You'd think that, but once the stylus is out of the groove and silence reigns again electronic noise can be noticed even if much quieter than surface noise. Its also nice not to have a big jump in noise when switching input selection between AUX and PHONO for instance, it's a luxury true, but one that can be appreciated!
On the X-Altra MC/MM phono preamp, I used 2 JFETs in parallel with active RIAA EQ around the whole gain stage (JFETs + opamps), coming back to the sources of the paralleled input JFETs. Because the EQ components would be unwieldy if I dropped the source resistance to 10 Ohms, which is what Denis Collin did on his passive/active LP797, I set the source resistance to 110 Ohms and was able to use readily available tight tolerance passives. This meant the X-Altra input-referred noise on MM with the input shorted was 1.6nV/rt Hz - about double (6 dB) worse than Collins preamp with his preamp input shorted. However, when you connect the MM cartridge (500mH + 1350 Ohms), the difference in noise is only a 0.2 dB advantage to Collin's design. (As a benefit for the 0.2 dB noise penalty, the X-Altra offers >30 dB overload margin from <20 Hz all the way through to 50 kHz).
I don't have much movement in here. It's for drop on replacement for e-mu 404 USB sound card. To improve thd for measurements.
Generally lme49720/4562a re advised as cost effective improvement.
Last edited:
AD826, which is a more " direct" replacement for LM4562 (same DIL8 package as well...). Very nice. It used to be expensive... ($10). It will surprise you how good this OPAmp is.
AD8066 (my favourite), but Vcc/Vee is max +/-12V and it needs an adapter. Still, worth the effort of placing them on adapters, and reducing the power supply rail voltages from the usual +/-15, to +/-12V, just to accommodate that OPAmp. The price is okay... at around $8.
Of course, make sure there are no oscillations when doing OPAmp rolling.
AD8066 (my favourite), but Vcc/Vee is max +/-12V and it needs an adapter. Still, worth the effort of placing them on adapters, and reducing the power supply rail voltages from the usual +/-15, to +/-12V, just to accommodate that OPAmp. The price is okay... at around $8.
Of course, make sure there are no oscillations when doing OPAmp rolling.
But is it really faster than LME49720? Either I'm looking on wrong specs or they seem to be similar.
The AD826 (~$11) is designed to be a VIDEO amplifier, with its large bandwidth and extremely fast slew rate. But for audio, it's a very POOR choice, with its too-high THD and noise.
The AD8066 (~$8) is better, but it's still a very high bandwidth, very fast amplifier that has too much THD and noise. Neither of these are particularly suited for AUDIO, and are way too expensive. Neither will trump an LM4562 for audio circuit applications, that's for sure. Not even close.
The AD8066 (~$8) is better, but it's still a very high bandwidth, very fast amplifier that has too much THD and noise. Neither of these are particularly suited for AUDIO, and are way too expensive. Neither will trump an LM4562 for audio circuit applications, that's for sure. Not even close.
Would you say the same goes for the LM6171/2, which is also a video amp. is it a POOR choice? It's one of my favourite audio op amps, and has been for over 15 years - or back whenever everyone was wetting themselves over the OPA2604 (one of the dullest op amps of all time, IMVHO). God is it boring!
In fact, one could almost restate my OP question as "Anyone know an op amp like the LM6172, but with exemplary distortion figures?"
In fact, one could almost restate my OP question as "Anyone know an op amp like the LM6172, but with exemplary distortion figures?"
Have you seen the 1/f noise on this series, including the 1656? The "knee" is above 1kHz and only intersects the LHS of the graph at around 35nV/ because that's 10Hz rather than 1Hz. It's pretty hard to make flicker noise significant but this will do it. And it rules the chip out for active crossover use too. Get down to a 2kHz bandwidth - which covers most modern 2-ways - and you've got enough noise to have defeated the purpose of using a FET amp. (I should mention that I haven't done the maths for the whole of this assertion.. I did integrate the 1/f noise on some of the OPA165X series some months back and was pretty horrified, but it didn't actually reach significance with anything like a normal bandwidth. I'm kinda assuming that you're going to be between two stools with a 2k or 3k low pass, and the joy of not worrying about large resistors in your filter circuits won't be there.) It'll be worse still for something below 250Hz.
OPA1656 24v/uSec slew rate; LME49720 (same as LM4562) 20v/uSec slew rate: 1656 is a LITTLE bit faster. But the only REAL advantage of an FET 1656 over the bipolar 4562 is much less CURRENT noise---so any application with a source impedance of ≥3KΩ will have lower noise with the 1656.
But what is STILL surprising to me is that an LM4562 is still WAY cheaper than am LME49720 (about HALF the $$$) even though they are reported to be the same exact chip!
But what is STILL surprising to me is that an LM4562 is still WAY cheaper than am LME49720 (about HALF the $$$) even though they are reported to be the same exact chip!