NE5534 and FETs
The vexing thing for IC designers is the significant area required for JFETs, to achieve low voltage noise. Scott mentioned that the 2SK170 die are about the same size as some ADI video amps! And I noticed that Linear Technology shows recommended circuits with one of their low-voltage-noise op amps prefaced by either common-drain or common-source BF862s. That must smart a little 🙂
I'm bringing up a breadboard of a phono pre suitable for MM and MC and using four paralleled SK170 equivalents, and a low-impedance series feedback network and servos (in essence, current feedback without the complementarity of the typical bipolar realizations). Cascoding and the bootstrapping effect of the series feedback makes the otherwise-high input capacitance tractable. This is an example of something that would be virtually impossible to integrate. For an SMD version I would probably migrate to BF862s, which are quite remarkably good, and likely to be around for a while owing to their use in Chinese AM radios, iirc.
There was a pretty nice ADI JFET amp that is not recommended for new designs, which one supposes is because they really don't want to make more of them. It shows up on a chart. Nothing is said to speak of iirc about the audio performance. The nearest substitute has higher e sub n and lower maximum rail voltages.
It is amazing how good a part it is, considering its age. Self remains fond of it, although his use to realize a cooled termination shown in Small Signal Audio Design is imo suboptimal (for one thing, as van de Gevel has mentioned, you can use part of the same gain chain and dispense with the extra op amp, and get certain noise advantages).
The vexing thing for IC designers is the significant area required for JFETs, to achieve low voltage noise. Scott mentioned that the 2SK170 die are about the same size as some ADI video amps! And I noticed that Linear Technology shows recommended circuits with one of their low-voltage-noise op amps prefaced by either common-drain or common-source BF862s. That must smart a little 🙂
I'm bringing up a breadboard of a phono pre suitable for MM and MC and using four paralleled SK170 equivalents, and a low-impedance series feedback network and servos (in essence, current feedback without the complementarity of the typical bipolar realizations). Cascoding and the bootstrapping effect of the series feedback makes the otherwise-high input capacitance tractable. This is an example of something that would be virtually impossible to integrate. For an SMD version I would probably migrate to BF862s, which are quite remarkably good, and likely to be around for a while owing to their use in Chinese AM radios, iirc.
There was a pretty nice ADI JFET amp that is not recommended for new designs, which one supposes is because they really don't want to make more of them. It shows up on a chart. Nothing is said to speak of iirc about the audio performance. The nearest substitute has higher e sub n and lower maximum rail voltages.
The audio stuff was fun, ATE was the intended market (where measurable results matter).
Some good stuff. I believe Hofer and Cabot left Tektronix after that, as the company didn't see the audio test market as a good fit. Thus was born Audio Precision.
I could use a 2465. And it would be fun to have the microchannel-plate-intensified 2467, which John Addis has some choice words about in connection with his appeal to "slay the time-to-market God", in his contribution to one of the Williams compendia.
Thanks Scott. I am glad my recall was not doing violence to the history.
Having started with a more lavish mostly-discrete design, I toned things down to use 797s and some other ADI parts in the Harman switcher/gain control for the speaker mover. The plug was pulled, alas, in a budget crunch, and I don't know what they wound up using to change preset gains to manage equal SPLs in the computer-controlled DBTs. I'm almost afraid to ask.
I'ld like to add to the tv visual analogy. If some one gave you a tv with a picture of a Monet you were familiar with and the tv had its colour and brightness set to zero and you were asked to adjust the tv ( not just the bright and colour) so it looked as close to the original as possible, without looking at the original, how close do you think most will get. And how many different settings. What you'll get is what people remember and what they like, not what is accurate. The same goes for audio tweaks done purely by listening. I don't care who you are or how long you've been doing this, it's human nature. The smart ones ( and most of the pros) know this. And since there's many who prefer big boosts in low end or some even order distortion,it's hard to accept someone's claim of improvements by random mods or other claims that don't have any physical proof. I'm not saying we know everything in this field, I'm saying there is a proper scientific method for investigating these claims and without it, we'll I have moon rocks for sale that will turn your stereo into a 10 piece jazz band. You might want to move it into a bigger room.
Last edited:
Many years ago, when laserdiscs were the best thing going for video, Joe Kane decried the absence of standardization and produced a test disc sold by Reference Recordings iirc. I'm not sure how many took advantage of it.
In the audio arena Toole has described the Circle of Confusion. It's still with us, and as well the tendency to regard recordings as being reference material in listening tests---when many are anything but.
for MM OPA1644 quad could be paralleled, DiFET input gives negligible Cin modulation
http://www.ti.com/lit/an/slyt595/slyt595.pdf
http://www.ti.com/lit/an/slyt595/slyt595.pdf
Of course you did. I have routinely encountered the like, loading the VAS
resistively to ground, even with single-ended circuits, because it can
occasionally trim the Vgs/Vbe/Vgc characteristic against the non-linear
Drain/Emitter/Plate impedance and cancel the second harmonic (If that's
what you want).
😎
Bruce Hofer and the CTO of Stanford Research regularly send their regards, just say'in.
Except in my experience the second harmonic suck out region can be VERY narrow.
For someone who wanted a stepup for MC to go with his tube phono stage, I did a design with three SK170s and a folded Boxall cascode, and tuned the pullup resistor to null second. There was a little R at the JFET sources to help normalize gain. For once two channels got finished and I listen through the prototype now. The downside of the simple design is the need for very quiet power supplies.
The region of effective nulling was not too bad, and the pullup value determined from simulation was pretty close to optimal. The second channel was very close to the first.
The guy also is a GFB hater, so I avoided that. But now he is trying to do the stepup himself with tubes, despite my warning that the low-frequency noise is large and somewhat unpredictable. He's been looking at someone's similar stage with two 6922s in parallel, hence four triodes, run at a very low plate voltage.
I suggested that maybe about 100 triodes in parallel might get into JFET territory, but so far that didn't faze him. Can you say stepup transformers?
The region of effective nulling was not too bad, and the pullup value determined from simulation was pretty close to optimal. The second channel was very close to the first.
The guy also is a GFB hater, so I avoided that. But now he is trying to do the stepup himself with tubes, despite my warning that the low-frequency noise is large and somewhat unpredictable. He's been looking at someone's similar stage with two 6922s in parallel, hence four triodes, run at a very low plate voltage.
I suggested that maybe about 100 triodes in parallel might get into JFET territory, but so far that didn't faze him. Can you say stepup transformers?
The circuit Scott posted with the six different kind of actives was interesting and we should turn it into a sport. Points can be had for the inclusion of any further active device that has not yet been used before.
My turn: To put the present schematic within an op amp feedback loop. This will decrease distortion, improve PSSR and reduce Rout. The first 3 x 10 = 39 points are mine. Who's next?
(Failed to retrieve Scott's post with the schematic, but this sport is anyhow best played blind)
My turn: To put the present schematic within an op amp feedback loop. This will decrease distortion, improve PSSR and reduce Rout. The first 3 x 10 = 39 points are mine. Who's next?
(Failed to retrieve Scott's post with the schematic, but this sport is anyhow best played blind)
Given that most TVs have a 'showroom ' mode to look as bright and saturated as possible to shout 'look at me' I cannot see that analogy working at all. I doubt anyone at those bigncheap stores knows how to calibrate the picture.
some of the 4K test material you see in the shops is really quite painful to look at.
Bonsai what soundcard do you use and to what extent is the measurement noise floor influenced by the bus noise inside the PC ?
I am sure there must be external usb or firewire converters that have a better noise floor ??
I am sure there must be external usb or firewire converters that have a better noise floor ??
I am using a Focusrite 'Scarlett Solo'
The noise floor with exponential averaging and Kaiser 7 FFT is about -136 dBV.
The THD resolution is limited by the - I think - D-A to 14 ppm which shows harmonics after about 800 or 900 mV. I will need to do some more work on attenuator box to resolve it but it will have to wait a few months.
at the levels you are measuring the "cause" is not going to be obvious, simple
when I was speaking of VAS loading principles I expect the illustrative measures to be % not ppm, the differences to sometimes be orders of magnitude when unloaded you can have 40 dB more gain at 200 Hz reducing IMD difference products than if you had flat loop gain to 20 kHz with VAS shunt R loading
when I was speaking of VAS loading principles I expect the illustrative measures to be % not ppm, the differences to sometimes be orders of magnitude when unloaded you can have 40 dB more gain at 200 Hz reducing IMD difference products than if you had flat loop gain to 20 kHz with VAS shunt R loading