I am using the ADA4898 as a balanced MC input stage and it works well.
As Dual Fet i recommend the OPA1642. As far as i can tell it solves the problem the OPA134 has with high frequency distortion into loads lower the 2kOhm. Some of the bipolar transistors in there are silizium - germanium. I compared the sound in one of my phono stages with the LME49720. That is of cause a bipolar but both ICs work in my particular application. The OPA1642 has a somewhat bigger soundstage that is more populated and finer structured. In the bass the LME has a slight edge. It sounds more tight and dynamic with a better grip. The OPA is a bit looser and warmer down there.
As Dual Fet i recommend the OPA1642. As far as i can tell it solves the problem the OPA134 has with high frequency distortion into loads lower the 2kOhm. Some of the bipolar transistors in there are silizium - germanium. I compared the sound in one of my phono stages with the LME49720. That is of cause a bipolar but both ICs work in my particular application. The OPA1642 has a somewhat bigger soundstage that is more populated and finer structured. In the bass the LME has a slight edge. It sounds more tight and dynamic with a better grip. The OPA is a bit looser and warmer down there.
FWIW, I've used the AD797 in a number of loop filters in PLL's for microwave synthesizers. They do very well, as you might expect.
Because of the cost advantage of the ADA4898, we tried it in the same circuit. Even after making sure everything was stable and all, the AD797 performed better in providing low frequency phase noise performance.
This implies that the low frequency noise performance of the AD797 is better, at least in that circuit. The '4898 was hardly bad, but just not quite as good.
Because of the cost advantage of the ADA4898, we tried it in the same circuit. Even after making sure everything was stable and all, the AD797 performed better in providing low frequency phase noise performance.
This implies that the low frequency noise performance of the AD797 is better, at least in that circuit. The '4898 was hardly bad, but just not quite as good.
Nathan's patent has been discussed before and even simulated by Edmond in june 2009.
Multi tanh is a very interesting subject, but you will need a PTAT circuit to get the advantages out of it.
BTW: I have designed new (very linear) bjt ips that looks promising, (without the need for a PTAT circuit)
Cheers
Stein
Multi tanh is a very interesting subject, but you will need a PTAT circuit to get the advantages out of it.
BTW: I have designed new (very linear) bjt ips that looks promising, (without the need for a PTAT circuit)
Cheers
Stein
That was just an aside, I don't particularly pay much attention to the PIM stuff.
It should help with PIM, I agree. And that Stanford professor who vetted Ron Quan's paper came to his defense from people on your side. Did you know that?
What defense what sides? Did he measure anything made after 1969 yet. I have to repeat, why on earth would anyone present data solely on 30-40yr old circuits unless there was an agenda afoot. I saw Tom last year and there was not an opportunity to discuss this, I will see him in February and I will challenge him on this.
Last edited:
Nathan's patent is not, if speaking in exact terms, multitanh principle. Linearization is achieved by tail current modulation of NPN and PNP differential pair. The effect is similar as for multitanh cells, but circuit realization is different, IMO.
PMA, I would be interested it how it compares to other input stages, especially self biased comp. jfets, especially in distortion order.
some ADA4898 numbers (from the datasheet) with a little calc
if my calculation from the patent's equation was correct then the gm of the input stage is 1 ppm linear for input differential V of up to +/-4 mV
I also estimate that the output could slew ~2V/us with |4 mV| input difference (the datasheet psrr curves give a good estimate of the audio frequency open loop gain if you know Sakinger's “A General Relationship Between Amplifier Parameters, And Its Application to PSRR Improvement”)
note that this isn't the slew rate limit, or onset of gross distortion - this is for the input stage gm variation contributing less than -120 dB nonlinearity
2 V/us seems like it would cover a lot of consumer line level audio - with essentially no "phase modulation" from input gm nonlinearity
the output stage doesn't look current limited for 600 Ohm load - but does have rather high dropout from the supply V such that 10 Vrms isn't practical
that's 6 Vrms into 150 Ohms
if my calculation from the patent's equation was correct then the gm of the input stage is 1 ppm linear for input differential V of up to +/-4 mV
I also estimate that the output could slew ~2V/us with |4 mV| input difference (the datasheet psrr curves give a good estimate of the audio frequency open loop gain if you know Sakinger's “A General Relationship Between Amplifier Parameters, And Its Application to PSRR Improvement”)
note that this isn't the slew rate limit, or onset of gross distortion - this is for the input stage gm variation contributing less than -120 dB nonlinearity
2 V/us seems like it would cover a lot of consumer line level audio - with essentially no "phase modulation" from input gm nonlinearity
the output stage doesn't look current limited for 600 Ohm load - but does have rather high dropout from the supply V such that 10 Vrms isn't practical
that's 6 Vrms into 150 Ohms
Pavel,
What is there to explain? The input stage is followed by gain stages, so if these stages are not slew rate limited at all the slew rate of the IC would be 2V/us * Gain. In reality the other gain stages will actually limit the slew rate...
So the OP's point was that the input stage does not introduce any slewrate limit.
Ciao T
What is there to explain? The input stage is followed by gain stages, so if these stages are not slew rate limited at all the slew rate of the IC would be 2V/us * Gain. In reality the other gain stages will actually limit the slew rate...
So the OP's point was that the input stage does not introduce any slewrate limit.
Ciao T
I repeat again another point, low open-loop BW is meaningless. You can use positive/negative feedback to the null pins on many op-amps to make the Aol and 1/gmC break point almost anything you want (even make Aol switch sign!). I have posted pictures here on doing this. Try it PMA, you can make the 797 have 10^5 gain rather than 10^7 and lower the OLB 2 orders of magnitude. Unfortunately null pins are deprecated these days or you could take the AD825 and tweek it up to 10^7 Aol and see for your self it makes no difference.
Last edited:
- Status
- Not open for further replies.