Pete Millett NuHybrid Headphone Amp

I have not tried it with a Noise Nuke yet.

I tend to run things like this off my lab-supplies if I think it’ll help (or at least to see if it does). Though it sounds just fine via the specified SMPS so far. I haven’t gotten around to doing any proper measurements yet, nor any audition-type listening, so hard to say if there are less audible/lower-level issues there.

I like the tone and the character of the little thing enough that I am planning an “all-out” build of it. Not because I think that will be in anyway cost-effective (it won’t), or that it will necessarily noticeably improve the performance, but purely because I think it’ll be fun to do.

But I am planning for a dedicated, internal, low-noise LPS with proper filtering, capacity and regulation, digitally-controlled 128-step analog volume control, front-panel adjustable digitally-controlled bias offset (with defeatable LED metering).

Ideally I’d like to use the digital bias control as a per-channel offset to a “master” bias setting, so you can dial in each channel to account for tolerance variations, and then use a single control to ramp the final bias voltage up or down, while listening, for both channels at once.

And then when I have all of that done, which will be using standoffs/additional boards connecting to the current stock PCB via wire, if the results are fun enough I’ll draw up a new PCB to accommodate the appropriate extra elements (e.g. the digital/analog attenuators etc.) and shock/vibration isolate the NuTube, likely on a separate, shielded, carrier.

Displays/meters and controls would be on another board, isolated from the main one.

Then a shielded metal enclosure, with appropriate cut-outs, and a view-port on top on which I plan to put a nice prism, so you can see the NuTube while keeping it safely shielded on all sides except the top.

Absolutely massive overkill.

And more about fun/variety/experimentation than necessarily improving the sound.

Chassis-excepted and LPS-excepted, it’s probably another $100 to $150 in parts. And between parts selection, circuit design, and actually putting it together, testing, and tweaking probably 100-150 hours or so of work give or take the time to layout a new PCB and panels.


That is really cool! I look forward to your build of this…I’m going to put together the baseline (with suggested upgrades) as soon as it comes in the mail this week (snow might cause a delay)

I’m excited to see what you come up with! and if it is possible to recreate it in a separate build, to my OG build.


It does already have a 2H choke on the PCB… PSU input has an LC filter.

Used to the iron chunks used with regular tubes, looks like a liliput one :smile:


Just found this those thinking on “displaying” the NuTube in a project (window on the box,. etc):

Standalone “carrier PCB” is $6.


I’ve chosen my digitally-controlled analog attenuators and trimmers.

I’ve gotten the meters designed for measuring and displaying the absolute bias values for each channel on a 3-digit, 0.1v precise, 7-segment LED display (one per channel).

Now working on tying those in to a single control to alter the relative bias so both channels can be adjusted from a single control (once balanced individually).

Sadly some of this is going to wind up being SMT, as not everything I want to use is available as through-hole.



I’m “salivating” :smile: will you create a new PCB?

I love bias indicators, though I was always worried of injecting unwanted bias while measuring it.

Is it going to be single ended?

One of the things I love about the Loxjie P20 is the volume control / input selector, an universal board based around NJW1195A would be killer for the DIY scene, IMHO much superior to pots and possibly cheaper.


My first pass will be using the existing amp in its current form, so it’ll remain single-ended.

The bias indicators, and the meters that drive them, will be defeatable in two steps. The first step will disconnect the meter connections, and disable the adjustment input, but leave the displays showing the current value (in volts). The second step will turn off the displays as well.

As I was thinking about that more last night, I thought it’d be cool to actually use vacuum fluorescent displays for the meters, so they’d match (roughly) the color and appearance of the glow from the NuTube (and because they have some nice old-school charm). Unfortunately the available displays, without salvaging used parts, are either a) about $100 a shot and/or b) require a microprocessor to drive. So that will have to wait for a subsequent build.

As for a new PCB, originally I was thinking that if the end result was good, and there was enough interest, I’d lay out a whole new board. However, now I’m thinking I can do all of this on a smaller 2nd PCB that’ll just sit on headers above the existing board. That can bring the necessary connection points up off the main board, including the tube (which can then be shock-mounted). Which would make it a potential retro-fit for existing builds (though you might need a hot-air/reflow station to safely pull the tube or pot off).

Doing an additional PCB (rather than just relaying out and expanding the current one) would also help if I do want to go in a direction that needs a microprocessor to run it, as it’ll make it easier to shield and isolate, power, and allow for more flexibility in parts placement.

Right now the meters and bias adjust design I am looking at do not need complex control logic. In fact, since each attenuator is a discrete piece (one chip per channel), they simply require power, and then to close the contact on either their individual up, down and mute pins, to drive the internal “wiper”.

And they’re a simple, comparatively large-pitch, SMD with just 8 pins each. 2 for power. 3 for the wiper (pot). And 3 for the control pins.

The prototype/test-bed I’ll likely just run wires to for the extraneous/broken-out connections.


@Torq - This, or the BH Crack? Which do you like better?


As long as you’re using higher impedance headphones (150 ohms or greater), then the Bottlehead Crack would be my choice - from a “how it sounds” perspective at least.

This (the “NuHybrid”) is obviously smaller, easier to build, cheaper and doesn’t shift in tone/control with lower impedance cans. It’s signature is also more immediately and obviously euphonic, at least with bias around 11v.

I like both of them, for different reasons.


While I was away, some new parts came in for my bias controls (digitally controlled analog attenuators) and bias meters (7-segment, 3-digit LED).

I had also decided to add proper, analog, front-lit, VU meters (more vintage/submarine-hardware in style) and those have arrived too. They’ll need buffers to avoid contaminating the audio signal and a way to calibrate them, but that’s easy enough.

I was going to do push-button up/down control on the bias control, but now I’m thinking about doing a “Steampunk” style build, and using plumbing hardware in an oversize case, with taps on rotary encoders for the bias settings … which will be a major departure from the simple, black, low-profile casework I had planned before.

Either way, I’ll post the state of the electronics build once it is working before I get to casing etc.


I still need to put mine together…and the Speedball… I am a failure of time management… I blame my lil dictator! And my ADHD (hopefully that is getting helped soon lol).

1 Like