I bought a stepped attenuator (actually, two of them) from ebay seller hanshare-electronics. As always with eBay, I was a little worried about what would show up.

I'm happy to report that this is a nice job. The included resistors are clearly labelled, and the documentation, while spartan, is sufficient.

The only negative is the rotary switch is a bit "clicky". Better quality units I've seen are more damped.

On the positive side, though, the resistor values are very elegantly selected, to give a smooth response of about 4 dB per step from -60 to -40 dB, and 2 dB per step from -40 to -3 dB. Having been-there, done-that, I can say that I'd happily pay the asking price to get those resistors sourced and sorted for me. Next time, I'd also pay to get it assembled. Soldering was more painful than I imagined it would be, and the pre-assembled units much neater-looking than what mine ended up as.

B-board vs. 0247.

Comparison of the noise baselines, measured at the circuit output using a NI USB-6215 DAQ. Unloaded for the preamps, and with a 6 ohm load for the 0347 amplifier.

• B-board: -139 [300Hz-100kHz] 0 dB gain (-139 - 0 = -139 dB, 110 nV sqrtHz input referred). The actual B-board output noise is below this measurement threshold.
• 0247: -124 dB [300Hz-100kHz] 14 dB gain (-124 - 14 = -139 dB, 110 nV sqrtHz input referred).
• 0347: -109 dB [300Hz-100kHz] 31 dB gain (-109 - 31 = -140 dB, 100 nV sqrtHz input referred).

See the attached plot for the FFT data. Note the peak at 28 Hz is an artifact of the measurement apparatus.

By way of comparison, a typical audio opamp has an input referred voltage noise figure of 3-8 nV sqrtHz (-170 ~ -160 dB) and can be expected to return this datasheet specification in most well-designed circuits. In other words the output noise is going to be about -160 dB + the circuit gain.

The 0247...

I was fortunate to stumble across this place on a trip to Yokohama earlier this week.

https://g.co/maps/nzebr

I was walking around the area near Sakuragicho Stn. after dinner, and, through the windows I could make out a back wall filled with LPs, two turntables built into the bar countertop, and a pair of giant multi-way speakers. The sign out front said "Modern Jazz and Coffee". So, naturally, I went in.

First, the history of the place as it was explained to me:

First opened in 1933. Destroyed by American bombs during the war, 6000 SP records gone, re-opened, and was in business until the owners death in 1994, then taken over and run by the owners younger sister until 2007, when, at the age of 77 (!) she gave it up. Volunteers somehow managed to preserve the records and furnishings, including the sound system, and, some 5 years later, decided to relocate and reopen at a nearby location just, if I've got this right, a...

A real, honest-to-goodness voltage regulator has three parts: a fixed voltage reference, an error amplifier, and a pass element.

Most people only put eyeballs on the final, all-wrapped-up-in-a-tidy-IC-package version, typified by the LM7812, or with a couple of extra gain-set resistors, the LM317. These chips have a working bandwidth about about 2 kHz, as they are designed to 1) reduce 120 Hz ripple and 2) be rock stable no matter what abuse they are subjected to. As a result at audio frequencies and above they are pretty much noise generators...

Knowing this, many people have set out to build better regulators for audio work.

The most obvious route is to build a high performance LM7812 from discrete components. (Most excellent review here.) AD797 for the error amplifier, high stability, low noise voltage reference, etc. The trick though, is stability. The LM7812 is low bandwidth not because it's too cheap to manage anything better, but because...

I have been doing a web-dredge, searching for simple, single supply, transistor buffer/gain stages, preferably FET-based.

This one seems interesting.

The designer, Henry Nurdin, has kindly posted a full schematic.

Ignore the fact for a moment that it's designed as a tiny battery-operated module to retrofit into electric guitars, the basic circuit block with 5-6 dB non-inverting gain could be used as a front end for a headphone buffer stage like, I'm thinking especially, the Szekeres MOSFET buffer.

That's if the bandwidth is sufficient for high-end audio (should be!?) and, slightly more worrisome, acceptable gain matching between channels can be achieved without resorting to trim pots.

This circuit reminds me of something. ... Sziklai pair? Time to do some more dredging...

Ah, bingo!

Update:

And with some jiggery-pockery this is taking shape as a (yet another) headphone amplifier....

File this under "things-I-should-have-learnt-to-do-many-years-ago-but-was-too-lazy-to-bother".

Many pcb fab outfits that do business with hobbyists and DIYers choose to accept Eagle .brd files, which means they do the conversion to Gerber output so you don't have to. I've relied on that for far too long, but when an error showed up in one the .pdf proofs on the last batch of boards I sent out for fabrication, they asked me to send the Gerber files instead. So I bit the bullet and after a couple of false starts managed to give them what they wanted. Looking back at it, it was easy and something I should have learnt, as I wrote up above, years ago, but, for posterity, here's how it's done:

Eagle 6 on Windows

1. Make a working folder for the Gerber files.
2. Copy the Eagle board (.brd) file to this directory.
3. Open Eagle (version 6.1)
4. Select menu item "file/open/CAM job"
5. The open file dialog appears, defaulting to the Eagle CAM

...

I was asked to suggest a voltage regulator for the First Watt (Pass DIY) B1 buffer. One thing led to another and the next thing I'd sketched up a circuit board for the buffer as well as the regulator.

It's like a B-board, but with the JFET buffer instead of the diamond buffer, and with a single supply and, hence, the coupling caps front and back. Since it's using JFETs for the buffer I used a JFET for the pass device in the regulator, too.

Full credit to Nelson Pass for his design.

Eagle files do not show 2SK170 because the package is not in Eagle. All transistors 2SK170 or equivalent. Zener is 18-22V DO35 or DO41 i.e. Vishay BZX85. V++ is 5-15 V above whatever you select the Zener reference to be.

Denon DL-A100

Transparent DL-103 for Denon's 100th anniversary.

So now we have the model 0247 DAC/preamp/headphone amplifier and model 0347 power amplifier in the main system. That's the old Denon DP-2000 TT, Phonoclone 3, and Onkyo D-605SR speakers.

The gain of the 0347 is 30 dB, so with the 0247 doing duty as a preamp the overall gain is a tad high, but it's perfectly useable. I'll look at the passive preamp options (including the model 0447) at a later date.

Anyhow, the big news is it completely destroys the non-inverting LM3875 Gainclone I have been using for the last decade, so badly, in fact, I'm starting to wonder if my Gainclone was busted. In comparison with the 0347 it sounds like a distorted, uncontrolled mess.

Since my speakers are really very modest in quality (though at 82dB and 6 ohms, a bit of a rough load) I wasn't expecting much difference one way the other, so this comes as a severe shock.

Unfortunately I'm not within reach of a good reference system to give the 0347 a real evaluation....

And not even sure what a tribulation is, but I'm sure I've had at least one building this amp.

Like the 0247 it's not so much difficult as it is painstaking hard work. The output transistors have to be mounted "just so" so the boards screw to the case properly, the wiring of the thermal protection transistors and gluing them to the output devices is tricky and easy to mess up.

I triple-checked and was rewarded by, it seems, error free boards. I messed up and wired the mute switch incorrectly though (the photos show it wrong, in case you come across this blog post looking for help) and I think the switch has a fault somehow as L and R channels appear to me mixed together. Taming the offsets gave me considerable grief but I think its all OK now.

Tomorrow I'll hook it up to the speakers and see what happens.