I designed discrete op-amp and shared to local audio community. Someone made the implementation and measure it. THD is not very low around 0.0x and slew rate about 12V/uS. The output is class A single-ended. Many of them like the sound.
I think, if we cannot achieve very low distortion, then use class A because it is small signal. It is hard to make unity gain stable with high slew rate.
If you want to low noise input, use jFET low noise in LTP. For BJT to use as MM pre-amp, you can use low RBB transistor and bias the collector of LTP around 100uA each.
I think, if we cannot achieve very low distortion, then use class A because it is small signal. It is hard to make unity gain stable with high slew rate.
If you want to low noise input, use jFET low noise in LTP. For BJT to use as MM pre-amp, you can use low RBB transistor and bias the collector of LTP around 100uA each.
@hbtaudio - I can see that people are not happy without a schematic. I do not intend to publish my designs, and so I will stop mentioning them.
I will share a simulation result which anyone can replicate. This is the distortion of a differential pair running at 1.5mA total (i.e. not a phono pre-amp) with emitter degeneration, versus the peak input voltage. The distortion comes from the exponential relationship between current and voltage (Ebers-Moll).
60dB of open-loop gain means that 10mV input will produce 10V output. The distortion is only 100PPM with 75 ohms, or 50PPM with 100 ohms. Now close the loop by dividing the THD by the loop gain. The closed-loop distortion can easily be made single-digits.
Maybe some day I will show that the rest of the op-amp contributes less distortion than the differential pair. I believe this is the case since the degeneration elsewhere is greater, and everything runs in class A.
Ed
I will share a simulation result which anyone can replicate. This is the distortion of a differential pair running at 1.5mA total (i.e. not a phono pre-amp) with emitter degeneration, versus the peak input voltage. The distortion comes from the exponential relationship between current and voltage (Ebers-Moll).
60dB of open-loop gain means that 10mV input will produce 10V output. The distortion is only 100PPM with 75 ohms, or 50PPM with 100 ohms. Now close the loop by dividing the THD by the loop gain. The closed-loop distortion can easily be made single-digits.
Maybe some day I will show that the rest of the op-amp contributes less distortion than the differential pair. I believe this is the case since the degeneration elsewhere is greater, and everything runs in class A.
Ed
Not really. It is evident that low bandwidth amps are more sensitive to these radio interferences.
A slow device has no problem to demodulate the 100%AM- bursts in the GHz region sent by cell -phones .
So I did some experimental investigations concerning sensity against cell phones.
As a result I re-routed my pcbs for max emc immunity.
Additionnally you may find specific em-immune op-amps at TI, for instance.
I am shure they outperform any discrete op-amp in this aspect.
I can only agree with you completely. I can also understand and accept that you don't want to publish a schematic - no question.
Basically, this thread can end here. Because we are no longer breaking new ground. Newcomers still have to find their way, it can be rocky and arduous - but pre-chewed food is not for the hard-boiled.
"Grüße",
HBt.
Basically, this thread can end here. Because we are no longer breaking new ground. Newcomers still have to find their way, it can be rocky and arduous - but pre-chewed food is not for the hard-boiled.
"Grüße",
HBt.
The demodulation of an amplitude-modulated RF carrier could become a permanent topic of its own in the radio amateur forum.
The previous measurement was with a sine sweep, so the resolution, especially at HF, was not the best one. Below the same setup, gain 30dB amplifier, and AD829 vs. LT1028. I have many similar measurements and also noise measurements, so maybe I could open a new thread.
Below 100Hz - signs of LF capacitor distortion of the input coupling electrolytic capacitor in the ADC. This is verified to be the case.
BTW, below is the opamp that gave so bad reputation to opamps in audio (a jewel in my stock) and was a target of taunt of opamps from some designers (JC). Yes it is bad, but there was no reason to make it a representative piece of opamps used for audio.
If we're going to stay with RIAA designs, then a recent Pass related one (Wayne Colburn)- the Pearl 3- uses high open loop gain opamps in a couple of critical locations.
The 1kHz distortion is about -103dBc at 6dBv output c. 69dB gain, mostly 2nd, and -94dBc 10kHz also at 6dBv output, using OPA1656s- as measured on my test bench. Riaa compliance +/- 0.2dB, simulated as nominally about +/- 0.1dB.
It seems to be well received. No sign of any pathologies that I can discern.
Bear in mind that the OPA1656 has an open loop DC gain of nominally 154dB.
The version that I have described previously, using similar, but not identical approaches, is c. -126dBc at 1kHz under near identical circumstances, still mostly 2nd, and -109dBc at 10kHz, also mostly 2nd. It also uses OPA1656s. RIAA compliance nominally +/- 6mdB, measured and simulated limits +/- 50mdB over half a dozen or so builds.
It too seems to be well received, but by many fewer people.
No sign of any pathologies, even with a wifi-6E router a few feet away.
Again, it appears that these non-linearities that you attribute to the use of high open loop gain opamps can be sufficiently mitigated by careful design of the surrounding circuit environment.
As far as people liking harmonic distortion, particularly even, and low order. Yes, that is a thing. Not my thing, but clearly something that many do enjoy.
Yeah. There is no EMI sensitivity consequence of high OLG gain. That is a myth. There is a consequence of proper PCB design, system shielding and input stage design/technology.
There's an AD797 input stage card for M2x called "Cedarburg" -- the name was suggested by Scott Wurcer of Analog Devices, who was the circuit designer of the AD797. (link)
Many (but of course not all) M2x builders absolutely love the sound they get from Cedarburg.
Many (but of course not all) M2x builders absolutely love the sound they get from Cedarburg.
Quite decent step/HF spectrum behaviour of AD797 in the linear 30dB gain amplifier:
What is the difference between a FET and a CMOS opamp?
Is there anything of that makes anything to consider when chosing an opamp?
Is there anything of that makes anything to consider when chosing an opamp?
No idea if this is the kind of answer you are looking for, but... MOSFETs have far more 1/f noise than JFETs at a given area. It can still be made arbitrarily small by using a very big area, but that also has its disadvantages, for example that the op-amp gets more prone to instability due to resonance peaks in the source impedance.