tanh
Indeed, most easily generated by an AD590 or AD592 (1 uA/K).
Cheers,
Edmond.
Indeed, most easily generated by an AD590 or AD592 (1 uA/K).
Cheers,
Edmond.
Ada4898
The ADA4898 has a far too low OL BW, at least for you.
For the rest of us it seems to be a beautiful op-amp.
The ADA4898 has a far too low OL BW, at least for you.
For the rest of us it seems to be a beautiful op-amp.
Unfortunately no info about the ADA4898. Anyway, thanks for the hint, the IC looks very good. Careful PCB layout and decoupling would be needed, IMO, and always a small resistor directly connected to the output pin.
No info yet, it's from the west coast anyway. You might find the output biased a little light for your taste.
I will send Stinius enough to get him started on his project, I hope you give the multi-tanh stuff a fair chance.
Scott, thanks for the input, so far. I am not against multi-tanh in principle, but from what I saw, the inherent production of higher order of higher order harmonic and IM components is out of this world, at least from looking at the transfer function that Barrie Gilbert showed. If this is universally true of multi-tanh, then the problem fixes the lower orders at the expense of higher orders. Is this the case?
The RMAF posts have been given a proper thread here: http://www.diyaudio.com/forums/showthread.php?t=152784
/Hugo
/Hugo
The key question, John, IMO, is where you tolerate higher order harmonics. -160dB? -140dB? As you know, well accepted tube amps have much higher level of high order harmonics, and also the hated 7th. I am not sure if harmonic distortion, I mean high order of some -140dB is of an importance. I do not think so, and I suggest to try to find new measurement methods.
Open loop PMA? I think not. However, I understand where you are coming from. With feedback, virtually all IC's are good enough for audio, because I bet that you can't hear .02% 3rd in any case. Of course, I am looking toward better measurements, dare I say FM? However, it would be necessary to measure any AM distortion at between .1V at 1V, where the action is, not at nearly full output. The problem could be a real compromise of distortion at standard levels, such as .1V, which translates to 94SPL in my system, 104 SPL with a K Horn, etc.
At least standard input stages are very linear over the first 10% of their range. Not so, with Tanh, from what Barrie showed in his example.
At least standard input stages are very linear over the first 10% of their range. Not so, with Tanh, from what Barrie showed in his example.
Last edited:
Hi PMA,
I think THD-20 is useful if it is done with spectral analysis, but too often the analyzer bandwidth is not enough to properly represent the high-order harmonics.
Given the availability of good spectrum analyzer tools these days, my default high frequency distortion test preference is 19+20 kHz CCIF with full spectral analysis (CCIF just looking at 1 kHz is almost useless). In this case, thirds will appear at 18 kHz, 5th will appear at 17 kHz, seventh will appear at 16 kHz, etc.
The CCIF test is numerically a bit less sensitive than THD, so the numbers we are looking for will tend to be a bit lower, and should not necessarily be compared to THD-20 numbers.
In this regard, I think it is not unreasonable to strive for 7th order CCIF to be below -120 dB.
Cheers,
Bob
John there are several ways to utilize this. I think Stinius wants to try the basic application where you create a small perfectly linear region for MC etc. Barrie extends the concept for very large input ranges with a minimum ripple transfer function. Please bear with us there is is a large potential for distraction from the basic priciples of what is going on.
John
Could you be a bit more specific?
What would you like to know about the multi-tanh?
The whole idea is that it makes a very linear region, (AKA low distortion)
Based on simulation:
It’s not a distortion spreader.
The high orders are low, also the 7th (as PMA said in the -140dB to -160dB range)
Cheers
stinius
- Status
- Not open for further replies.