Also consider the fact that the number of masks used to fabricate a BF862 is MUCH less than the number required to make a typical op amp. So the cost per area of an op amp is likely much higher.
Thx johnc124, great info.
customers seem to buy (still in production) opa627/637 and Muses01/02 at their high prices. Lots of price room for maneuvering if you consider them.
As with many offerings, it comes down to marketing rather than technical feasibility. I am certain TI and other semis weigh all these points into the product requirements.
We do. To a nauseating amount sometimes, well before UV touches photoresist. I now understand why the team developing the Mustang at Ford met in secret off-site 😉
OPA627/637 still sell at a high price because they're a pretty unique combination of DC precision and fantastic AC performance. With the exception of chopper amplifiers, DC precision will always increase the cost of a part. The DiFet process that the OPA627 is also pretty old, which increases the cost per device. But yes, people do still buy them.
I haven't played with the Muses01/02 yet to know if their performance justifies the price. Muses01 looks awfully similar to OPA2604.
OPA627/637 still sell at a high price because they're a pretty unique combination of DC precision and fantastic AC performance. With the exception of chopper amplifiers, DC precision will always increase the cost of a part. The DiFet process that the OPA627 is also pretty old, which increases the cost per device. But yes, people do still buy them.
I haven't played with the Muses01/02 yet to know if their performance justifies the price. Muses01 looks awfully similar to OPA2604.
What about adding the cascoded jFET into pins 1 & 8 of ne5534 and disabling the BJT input LTP?
Would that give us a near optimum MM stage?
Would that give us a near optimum MM stage?
I think it would be an good solution.
Care to give it a try AndrewT?
Since 2sk146 is obsolete, or very expensive and hard to get from Interfet. When i showed this ckt to the group, Scott Wurcer said that using two BF862 in parallel would be equivalent to a 2sk146. I should have asked at the time as I would like to know how he came to that conclusion?
There is also the example that Bob Cordell has used in the lsk489 app note.
Lots of options if anyone cares to try them out.
Last edited:
The IFN146 has a NF of 1 dB for a 100 ohm source impedance, which is a noise spectral density of 0.653 nV/rtHz, at 1kHz, which is pretty darn impressive! The BF862 is spec'd at 0.8nV/rtHz at 100kHz, so we can expect it to be slightly worse at 1kHz. 2 BF862s in parallel would be 0.8 * 0.707 = 0.567 nV/rtHz, at 100kHz. So they may very well be equivalent at 1kHz.
Mouser has the IFN146 for (only) $16.76, but if we buy a thousand it drops to $10.14! I think you could parallel a lot of BF862s and still come out ahead 😉
Mouser has the IFN146 for (only) $16.76, but if we buy a thousand it drops to $10.14! I think you could parallel a lot of BF862s and still come out ahead 😉
INF146
the to-71 package must be a good part of the cost, we should ask Interfet if they can put it in a sot package for us. might be cheaper to ask nxp to put adjacent dies in a sot package for us.
If you are asking $1K+ for phono preamp, like a Bews, it might be justified, to use a inf146. great for advertising, who can say they are using a Interfet part in their front end. The jfet specialists. TI won't afford to put such a good jfet in their devices 🙂
now why did Bews not used a jfet in there design? or did they? cost? performance?
Most of the high end japanese MM phono-amp designs I have looked at all used jfets.
Has Doug Self ever used a jfet in the front end for MM phono-amp?
Fun stuff, talking about obsolete media = LP's
the to-71 package must be a good part of the cost, we should ask Interfet if they can put it in a sot package for us. might be cheaper to ask nxp to put adjacent dies in a sot package for us.
If you are asking $1K+ for phono preamp, like a Bews, it might be justified, to use a inf146. great for advertising, who can say they are using a Interfet part in their front end. The jfet specialists. TI won't afford to put such a good jfet in their devices 🙂
now why did Bews not used a jfet in there design? or did they? cost? performance?
Most of the high end japanese MM phono-amp designs I have looked at all used jfets.
Has Doug Self ever used a jfet in the front end for MM phono-amp?
Fun stuff, talking about obsolete media = LP's
Using a jfet as the first stage of the NE5534 is old technology. See the link page 28.
http://www-f9.ijs.si/~margan/Audio/Ironing_RIAA.pdf
http://www-f9.ijs.si/~margan/Audio/Ironing_RIAA.pdf
From
MUSES Op Amps are assembled with Advanced Symmetry Die-Bonding (ASD) technology, which achieves higher separation between the right and left channels resulting in well-balanced symmetry.
Muses are duals and are expensive because of special technology that provides good separation. Considering that only the best and the most expensive turntable pick-ups have 30dB separation between channels, I doubt that such level of separation in opamp is necessary. Even industrial quality dual opams have separation over 80dBs. For RIAA circuit such separation is simply more than adequate. Opamp is not limiting factor for separation in RIAA circuit.
Hmm, yet Muses8920 seems to have almost identical channel separation, at 1/4th the price? As well as having lower noise? I'm not discounting the possibility that they use an expensive manufacturing technique, what I still fail to see is how the performance achieved justifies the price of $46/ea on Mouser? 🙄
Most modern op amps achieve this level of channel separation through proper IC layout engineering. For example OPA1688 is a monolithic CMOS amplifier with 140dB of channel separation at 10kHz (4dB better than a Muses01) and is around $2/ea.
But in reality, as you say, these numbers are useless. In a real system the channel separation is totally swamped by the effects of op amp PSRR (wish I could find a Muses01 PSRR curve) and the impedance of the ground plane. And furthermore, all that channel separation is ruined once sound comes out of my stereo speakers. Sound from the right speaker mixes with the left and vice versa in total chaos! 😉
In this case we did it to minimize thermal distortion effects due to the high output current capability of the OPA1622. If the output stage of one amplifier produced a thermal gradient across the input stage of the other amplifier, you could have cross-channel thermal distortion effects. The OPA1622 die is laid out along a single thermal line of symmetry to prevent this.
Okay, thanks great paper. but the ckt is recommend in this paper.
If you read p 28-29 and look at fig 22 the author proposes the same ckt to bypass the ne5534 input and use lsk389 instead.
Noise spec is important in RIAA circuit. Most MM cartridges have impedance around 1000 Ohms. That's exactly source impedance that provides the lowest noise for NE5532. NE5532 is still one of the best choices for RIAA circuit, plus it is readily available at low price and in DIY friendly package. For DIY project look no further.
For sure. I have a nice 12AU7+12AX7 stage between my phono pre and power amp. If I can get the typical 40ish dB, a single stage phono should work fine.
gbp? Sorry, new to this. OK, so it seems fine as Mac had it, which makes sense as they are typically cautious about their designs... often to a fault. Thank you kindly for looking at that and verifying! Also thanks for the very nice link... it looks quite useful. This maybe overly specific, but do you happen to have a link or info on setting an offset null?
Looks like a gem! Thank you for sharing. If you were using an SUT between the RIAA stage and a low-output MC, would the OPA1642 still be your choice for a single op-amp phono pre? I imagine that a typical SUT has between 50 and a couple hundred mH inductance, albeit that might be less critical when loading an SUT vs. a cartridge.
- Status
- Not open for further replies.