Acceptance in a vacuum Esp ...
One like OPA's the other not, how are we to know who to know which is best. i can hear JC interpretation anytime i wish, it's available , how can we hear ESP's OPA setup , where are these great OPA pre-amps ..?
I'm not against , I would love to hear ......
Find anything (they are soooo many) with a slew-rate >=300 V/µs witch fill your other needs (noise, PSU and current)
By example, on My Behringer DCX2498, i was obliged to use voltage feedback (cause differential input signals and i wanted to took the output from the out of the second stage)
I was very limited because +-15V, 8SOIC, and i don't wanted to modify too much the cms parts. I found this one:
THS4012 (310V/µs, Out:110mA, vn:7.5nV/rtHz)
But, when it is possible, i prefer to chose current feedback's ones (i have a stock of nice old OPA260, + some in my Rotel preamp).
During my researches on Google, in order to find a part, i was looking at people's reports about "the sound" of various OPA they have tested: When people compare parts, they are *always* rated in the order of their slewrate. N°1 is always the highest.
Doing like-me, compare with a strait wire ? If no difference nothing can be better.
Last edited:
Meet THS4631:
- JFET input, +/15V
- 325MHz unity gain freq.
- 80dB open loop gain, -3db @ 30KHz (happens to meet JCs criteria of "open loop bandwidth")
- 1000V/uS slew rate
- 7.5nV/rtHz noise
- 0.01% distortion @G=5 into 100ohm at 5MHz and 2Vpp (much less distortions at audio frequencies).
Nice to met-you.
(it was a single, while i wanted a couple
The best voltage feedback of the market, yes.
And, current-feedback, one of the best killers: LME49713
Supply Min (Min) (Volt) 10
Supply Max (Max) (Volt) 36
Special Features Current Feedback Audio Amp, AvCl>2
Vmax (V) 36
THD + N @ 1 kHz (%) 0.0004
Vmin (V) 10
Number of Channels 1
Slew Rate (Typ) (V/us) 1900
Iq per Channel (Max) (mA) 18
BW @ Acl (MHz) 132
BW @ G=+2 (MHz) 107
GBW (Typ) (MHz) 190
Vio (Max) (mV) 2
Offset Voltage Drift (+/-) (uV/Degrees Celsius) 0.29
Vn at Flatband (Typ) (nV/rtHz) 1.9
Noninverting Input Current Noise (pA/rtHz) 16
Iib (Max) (nA) 5000
Io (Typ) (mA) 140
Pin/Package 8TO-99, 8SOIC
Approx. Price (US$) 1.45 | 1ku
Last edited:
Nice part. But needs attention when doing layout, easy to oscillate if to do it wrong.
yes it is. But --- you get low thd+n that way also. -RNMarsh
Last edited:
That looks like a nice part.
From the datasheet (see http://www.ti.com/lit/ds/symlink/ths4631.pdf ; too large to post directly here) it looks like 0.01% is a little optimistic. But the change in 2nd and 3rd distortions with lighter loading is very encouraging, particularly the 3rd behavior. Note that the figures 7 and 8 are for noninverting gain of 2, which means that common-mode distortion is not much of an issue. It would be interesting to see the curves for inverting mode and light loading. And any buffer in the feedback loop is going to have to be a very fast one.
The voltage range is refreshingly large for this fast a part as well.
Slew rates are reasonably equal, and 0.01% settling time is specified, so we know it is pretty well-behaved in that respect, which can also be partially inferred from the phase response.
Since the open-loop gain is only about 80dB we would not expect the frequency-dependent nonlinearities to be much further reduced with global feedback below the -3dB frequency, and if we want much gain we will have to live with more distortion. But still in all, a nice part. And the voltage noise corner is not too high.
The predominant 2nd suggests that if part-to-part matching is good, arrangements with multiple amps can be envisioned to mostly-cancel 2nd and leave almost-only 3rd.
No, better results come from massive feedback architectures. There is nothing technical to support these low GNFB circuits are better.
Yes, why pray tell? Why not the most parts? Or maybe this is not a parameter that matters.
Wayne, which of JC's illustrious designs do you have?
You need zillion V/us to do well in the Quan PIM test. One day, JC will conduct Quan on Blowtorch and we'll see how a truly 'hand carved from solid Unobtainium by virgins' product performs.
If we're REALLY lucky, we might be able to see the results of a Blind Listening Test of this against an evil 4558 device too.
In the meantime, JC has some Hirata test results on Blowtorch.
Assuming the 1K load , H2 amplitude curve keep flattening below 1Mhz
to the point that i would be surprised if it managed to reach the -90dB
mark at 10Khz , not counting that it s with a paltry 2V PP output level.
All in all , and despite an impressive slew rate , it will be undoubtly
outperformed at audio frequencies even by an old school 5534 linearity wise.
With high slew rates, it is my opinion too.
With transparent devices, better to use specialized ones, one with low noise, gain and little out current, followed by one with little gain and big current for output drive.
Last edited:
Why do not listen to -it, as i have done ?
5534 is a bad sounding OPA, on my taste.
Last edited:
what context?
Yes, I keep forgetting that this isnt really a Do-It-Yourself forum/site to build things. What constructor would care that it takes 40 transistors and more associated parts to get two channels of sound working. My bad. -RNMarsh
Yes, I keep forgetting that this isnt really a Do-It-Yourself forum/site to build things. What constructor would care that it takes 40 transistors and more associated parts to get two channels of sound working. My bad. -RNMarsh
Last edited:
- Status
- Not open for further replies.