Wow,
and until a minute ago I thought you were the biggest troublemaker in the class.
"As almost to be expected, outcome from both was identical." [Hans P.]
You astonish me. Where does this 180 degree turnaround suddenly come from?
greetings,
HBt.
If the system responses are identical, then both topologies are to be regarded as equivalent in their transmission behavior.
thx Hans P.
👍
thx Hans P.
👍
Now we can finally get to the bottom of perceived perception (auditory) systematically, i.e. methodically.
I have a hunch,
a pretty good hunch at that!
But I don't want to reveal the solution just like that.
HBt.audio
Well, after I said that I will turn my volume down, somebody has to repeat it and ask me to shut up...... Don't get it, but I will leave it here and forget.
Nevertheless wanted to conclude on my gain vs noise statement. I'm mechanical, not electrical engineer, so many things to catch up. This one I must clarify.
I myself made confusion without saying that I only used INA amps as front end, not opamps.
INA has different specification and calculation; except input related noise, also output related voltage noise is introduced (and much larger in amplitude.
Formula is :
Total INAmp Noise, referred to input =
Usual input noise (as with opamp) is amplified with gain, but output refereed noise (eno) is not.
Here is screen shot from INA217 datasheet as example
ein = 1.3, eno =90nVsqrt/Hz
When it is calculated, this is input noise vs gain vs frequency graph:
Now I can shut up I guess 😎
Thanks hbt,
I talked with Calvin quite a bit in my thread https://www.diyaudio.com/community/threads/fully-balanced-mc-phono-preamplifier-thoughts.414816/
He is indeed into them, I appreciate. I'm also in INAs, but in my thread discussions I realized that designs of Hans , Wayne Kirkwood and what Hierfi is cooking at the moment, are very advanced over normal INA application. I'm contemplating of using THAT1580 (very special INA) in transimpedance configuration, but time still did not allow to built prototype. However , these are all LOMC preamps with fully balanced configuration allowing to get rid also of externally introduced noises (IMHO worst than usual amp noise) and well under nV noise density, so not really a subject here.
Only point from that discussion that could contribute here is using balanced amp that apart of offering great CMR also doubles available Vp-p capability for clipping concerns. Just I never tough of applying feedback EQ and have no idea at moment how to do it in balanced setup. Not that I need it neither.
I talked with Calvin quite a bit in my thread https://www.diyaudio.com/community/threads/fully-balanced-mc-phono-preamplifier-thoughts.414816/
He is indeed into them, I appreciate. I'm also in INAs, but in my thread discussions I realized that designs of Hans , Wayne Kirkwood and what Hierfi is cooking at the moment, are very advanced over normal INA application. I'm contemplating of using THAT1580 (very special INA) in transimpedance configuration, but time still did not allow to built prototype. However , these are all LOMC preamps with fully balanced configuration allowing to get rid also of externally introduced noises (IMHO worst than usual amp noise) and well under nV noise density, so not really a subject here.
Only point from that discussion that could contribute here is using balanced amp that apart of offering great CMR also doubles available Vp-p capability for clipping concerns. Just I never tough of applying feedback EQ and have no idea at moment how to do it in balanced setup. Not that I need it neither.
Surely the unity gain at high frequencies of the non-inverting active RIAA topology has something to do with this?
Most still don't seem to follow the RIAA stage with a low pass filter to correct that.
Of course the active inverting, and passive, RIAA circuits have no such problem.
Most still don't seem to follow the RIAA stage with a low pass filter to correct that.
Of course the active inverting, and passive, RIAA circuits have no such problem.
All-active RIAA reduces gain wrt frequency to track the RIAA curve, so it is not the same as the INA spec you show. What is important is to put as much gain as possible up front with a good low noise amplifier stage. The reason for this is you do not want to be amplifying the thermal noise of a suboptimal front-end stage. RIAA EQ confers about a 4 dB improvement in SNR over the audio band compared to a flat amplifier with the same gain and using the same amplifying device eg an opamp, and with the midband gain set to the same magnitude and using a notional cart source of 500mH + 1350 ohms dc resistance. If you throw the cart out and use a low source Z, the improvement is about 4.7 dB in SNR
Lipshitz covers this in his paper. If the designer fails to put the secondary post LPF filter after the main RIAA EQ stage, 'they just ain't designing it right!'.
Yes, surely lacking the integration at high frequencies is measurable by all, and is audible on impulses,
compared to the proper response. Most phono preamp reviews seem to truncate the response curve at 20kHz,
which will only show the beginning of the response error. Combine that with a strong ultrasonic tweeter resonance,
now present in many high end speakers, and the result is not pretty. After all, linear distortion is still distortion.
compared to the proper response. Most phono preamp reviews seem to truncate the response curve at 20kHz,
which will only show the beginning of the response error. Combine that with a strong ultrasonic tweeter resonance,
now present in many high end speakers, and the result is not pretty. After all, linear distortion is still distortion.
Hi, I agree with all you said.
But I think its in different direction, all I wanted to confirm is that unlike in opamps, there are amps like INA where output noise is not equal input noise x gain, as so many are repeating.... Nothing else.
This is known and there are ways how to do it properly
However, it has little effect to "click response" with the cartridge. Frankly, I do not understand this continuing debate. All of this can be both precisely simulated and also measured. Tales do not help.
That's correct, however keep in mind that a simulation uses ideal components transforming the transfer function in this case into equal mathematical equations for both topologies.
That's why I didn't expect anything when comparing both topologies in a simulation composed of flawless components.
Current loops could have impact on the sound perception, loops that are IMO harder to control with a single amp solution as with a hybrid solution using passive filters.
When using a differential topology, with passive filtering in between, there is not even a gnd reference during signal transfer from input to output, thereby completely eliminating intermodulation between signal and ground currents.
But this effect will also be impossible to catch in a simulation but is clearly visible in the extremely good CMRR in posting #111.
The question is always in what way this will result in an improved sound perception.
Simulations are perfect for producing a frequency response and also for noise calculation, but faithful dynamic behavior will often be hard to simulate.
Hans
That's because it is a non-issue rather than a problem when the gain at 1 kHz is 40 dB or more. You then get an extra zero with a corner frequency above 200 kHz, while the record cutting machine and the cartridge roll off well below that.
Regarding click reproduction and overload, a passive post filter won't prevent the amplifier from clipping, as it is placed after the amplifier. It only puts an extra load on the output, which may reduce the clipping voltage a bit.
As long as nothing clips, you can get an impression of the effect of the + 1 term on the shape of the amplified click by adding both traces shown in post #2, https://www.diyaudio.com/community/...clicks-and-pops-unencoded.424462/post-7938189 Don't forget to scale down the cartridge signal by a factor of 20 because of the different vertical scales.
High frequency limiting is often better applied early on in a network, thereupon relaxing the requirements of subsequent networks to deal with such frequencies of higher amplitude.
It isn't necessarily good practice under all circumstances to include a feedback capacitor from the output of an op-amp to the inverting input terminal to create that pole, even if the device is stated unity gain stable. Unlike a highly stable NE5532 for example, many op-amps can be considered marginally stable at unity gain, often requiring careful adherence to power supply decoupling requirements. This adherence being commonly described as using "good engineering practices" to support such unity gain connections. This isn't to suggest high gain can't be equally problematic.
Nevertheless, it is sometimes instructive to connect devices without local power supply decoupling to establish the nature and extent that particular devices behave, and to investigate alternative mechanisms to support stability. For devices such as the LT1028/LT1115 for example these devices can behave better with minimum gains of perhaps 2, whereupon the addition of "local" power supply decoupling can support a more trustworthy stabilized response. The implications of such gain shelving requires investigation as well.
Sonic artifacts to clicks and pops can be impacted by poor stability margins, created by less than perfect network realities from simulated responses. As stabilities are related to gain, these can be impacted by cross-over distortions as well. In other words there is a host of potential issues that can impact sonics beyond the clarity that simulated responses suggest.
It isn't necessarily good practice under all circumstances to include a feedback capacitor from the output of an op-amp to the inverting input terminal to create that pole, even if the device is stated unity gain stable. Unlike a highly stable NE5532 for example, many op-amps can be considered marginally stable at unity gain, often requiring careful adherence to power supply decoupling requirements. This adherence being commonly described as using "good engineering practices" to support such unity gain connections. This isn't to suggest high gain can't be equally problematic.
Nevertheless, it is sometimes instructive to connect devices without local power supply decoupling to establish the nature and extent that particular devices behave, and to investigate alternative mechanisms to support stability. For devices such as the LT1028/LT1115 for example these devices can behave better with minimum gains of perhaps 2, whereupon the addition of "local" power supply decoupling can support a more trustworthy stabilized response. The implications of such gain shelving requires investigation as well.
Sonic artifacts to clicks and pops can be impacted by poor stability margins, created by less than perfect network realities from simulated responses. As stabilities are related to gain, these can be impacted by cross-over distortions as well. In other words there is a host of potential issues that can impact sonics beyond the clarity that simulated responses suggest.
Have you a couple of RIAA circuits; one which clearly audibly emphasizes CoP, and one which seems less disturbed? We can LTspice them and see if their responses to a transient are different.
I have such a case but there were loadsa other factors which might be responsible for the difference ... and it wasn't a DBLT so I'm wary of mentioning it.
Thanks for your *.ASC file Bonsai
Many phono EQ amps clip high frequencies at 3v p-p a 2 stage design can do 40v p-p with an op amp at 20khz. Pops and scratches don't affect my preamp.
I think this is an apples-oranges comparison. The MM output is not equalized, the preamp's output is. Relative to the input signal, the preamp smashes the high frequencies -- but that's OK because the transducer has a rising response to velocity. The RIAA curve is an approximation used to fix that.
There is no way that the waveforms could ever resemble one-another. If your preamp does manage to do that, most listeners wouldn't like it, unless they like no bass and screaming highs.