Shipping will be unavailable from Dec 21 to Jan 3. Orders placed during that time will be shipped the week of Jan 3.

Choosing an nw2s::o2 Configuration

Yes, we currently offer FIVE different versions of the nw2s::o2. Hopefully, this will offer some insight into the differences in them and why you would want to pick one over another. Of course, if you have any questions along the way, feel free to ask us directly

If you are looking for more information about why you would want a balanced. output over an unbalanced output, you may want to read some of our FAQs.

The different configurations vary in a number of very important parameters (to modular folk, anyway):

  • current consumption
  • module depth
  • electronic or transformer balanced
  • AC or DC coupling
  • discrete or IC gain stage
  • color/distortion
  • line or microphone level output

Note that all comparisons are based on unity gain unless otherwise stated. If you'd like some information about gain staging, we have some information on that as well further along. 

What's important to remember as you read this is that the option A is not "better" than option B or "worse" than option C regardless of each items cost. The pricing is based solely on parts and labor. It's all about feature set and your own requirements. The goal of this overly long comparison is to help you match what you need to what we offer.

Model Overview

Before we get too deep into the details, we should briefly cover the topology of the nw2s::o2 configurations. Broadly speaking, all of the modules contain an input attenuator (the potentiometer), a fixed gain circuit, and a balancing interface. The balancing interface can either be a transformer or an electronic device. Here are the equivalent block diagrams. 

nw2s::o2-84, nw2s::o2-990, nw2s::o2-purple, nw2s::o2-di




Electronically balanced using THAT 1606 integrated circuit

Reliable, robust, DC-coupled output stage that is fully floating similar to a transformer. This compact configuration will reproduce your audio with precision and is extremely power efficient.

The THAT1606 is used in many modern pro-audio applications. It is capable of driving long loads with minimal loss and will prevent any ground differential between your modular and studio system from impacting the audio quality.

This configuration has the shallowest of the profiles and is best suited to compact systems and some skiffs with smaller power supplies

This module runs on standard +/-12V rails with plenty of headroom for typical modular signals. 

Note that the +6dB gain is built into the 1606 IC, so there is not any additional gain circuitry inside the module.

Buy Now


Transformer balanced with a single Analog Devices ADA4084 op amp gain stage

This configuration is the ideal balance of efficient power, floating balanced outputs, and transformer color. In fact, most of what people consider the clean side of "tube" sound comes from the transformers rather than from the tubes.

It is AC-coupled using an oversized capacitor which ensures a solid response far below audible frequencies. The op amp is a recent, fast, high-voltage design with low-noise and a rail-to-rail output. The all nickel core transformer is the ideal size to allow you to operate in its clean range or gently push it into its lower regions of saturation. 

It is a fairly deep module, so. it will require a deep case. It is definitely not skiff friendly, but it is reasonably efficient, requiring only 30mA from both positive and negative supplies.

This module runs on standard +/-12V rails with plenty of headroom for typical modular signals. You can upgrade to a nw2s::18 power supply if you wish to run them at +/-18V rails for additional clean headroom.

Buy Now


All-discrete Jensen 990 op amp based signal path, high-nickel transformer balanced output

You can read about this op amp for days. From Deane Jensen's original design documents and John Hardy's updates to countless articles and posts discussing what makes this amplifier so special.

The Jensen design is a complex device. It contains not only discrete transistors, but also inductors and capacitors. This makes the 990 differ from even a good audio IC op amp in two ways.  First, since it is a discrete, transistor-based op amp, there will be slight non-linearities prior to full clipping. That plus some phase distortion due to the complex reactance will lend tracks recorded through this configuration with a bit more weight and depth. 

The 990 was originally designed to run on extremely high power rails. At +/-24V rails, it had twice the overhead of eurorack, more than API, and even Neve. Modifications over the years allow it to run as low as +/-12V, but at 12V, the dynamic range is severely compromised. This configuration should only be run with the external +/-18V supply, so an in-line supply is included which will generate a dedicated high voltage rail from your standard rack power. 

Buy Now


All-discrete Purple op amp with high-nickel transformer balanced output

There is something to be said for character. While the Purple op amp may not have the pedigree of the 990s, it is a modern design meant for microphone preamps, inductor-based EQ, optical compressors, and even in the summing circuit of large American consoles. It's awfully pretty with it's purple anodized heat sinks as well. 

When run on +/-18V rails, this configuration will provide plenty of headroom and enough character to give your tracks just a little added dynamism. While it's hard to choose a favorite among children, I may have to send a nod in this direction.

If you are not looking to run at a higher 18V power rail, and would like to be able to hit the input stage a little harder for some extra distortion from time to time, this is the ideal configuration. 

The +/-12V configuration may be ideal for lead tracks. or those needing to have a little added distortion to stand out more, but may be too much of a good thing for tracking and overdubbing an entire piece. This is available by special request only as they are not normally stocked and you will need to be able do discern the differences and be able to swap back and forth which requires just a bit of soldering.

Buy Now


Transformer-based DI with a single op amp gain stage for mic-level interfaces

Sometimes you want to run your modular through a boutique mic pre or perhaps have a direct interface that is more friendly in live situations when the only option for a balanced input is through the console's mic preamps. In those cases, the line-level outputs of the other modules will be far too hot. You will want a transformer based DI that will lower the signal and be able to drive the low impedance of a microphone input. 

Unlike the line level interfaces, however, the DI interface transformer is more robust and is not made to easily be driven into saturation. It has low phase shift and low distortion characteristics. Its job is to provide extremely high common-mode rejection ratios into low impedance preamps. 

The DI is a bit different from the other configurations, not only in that it requires a microphone preamp, but in that you have a bit more room to play with the gain staging. If you take into account each stage of gain adjustment, you will have complete flexibility in which gain stages will drive which devices for more or less.

API 512c block diagram from the API 512c manual used completely without permission

Taking for example, an API 512c mic pre block diagram paired with an nw2s::o2-di compared with simply an API 512 using the line input directly. When you route directly into an API 512c (or most any transformer coupled mic pre) you are bypassing perhaps the most colorful part of the circuit - the input transformer.

Using an nw2s::o2-di, you will get the o2 output transformer routed into the input transformer of the API. The various gain staging options will allow you to customize the signal to your liking using a combination of o2 gain, mic input pad, and preamp gain.

Available Soon!

Gain Staging

The nw2s::io is a fixed gain device. I won't go into all the reasons the 'io performs so well as a fixed gain interface, but one oversimplified reason is due to the fact that a typical VCO will output in the neighborhood of +/-5V. If you run that through a VCA or a VCF with some drive capability, then you might be able to get to +/-7V or +/-10V. With a professional-level audio interface from Avid, Apogee, UA, or MOTU, that will typically result in about -18dBFS to -9dBFS signals depending on how the interface is configured. 

-18dBFS is considered "optimal" recording level and typically maps to about +4dBu. So basically, a typical VCO routed through the -6dB gain of an nw2s::io will end up around +4dBu, give or take. 

The nw2s::o2 is a variable gain device. In order to map the "ideal" gain staging of the 'io, you can set the gain to the 12:00 position. Setting the input attenuation knob straight up (12:00 position) results in about 6dB of gain reduction. 

At the 3:00 position, the 'o2 will operate at about unity gain, and at full clockwise rotation, the gain will be +6dB. 

This means that the gain is optimized at the 12:00 position.

Above the 12:00 position, and you will begin to operate outside the parameters of various components of the system including the op amp, transformer, and your audio interface. If you are running on a typical 12V rail system, then the most likely device to give in will be the op amp and it will begin to clip. The transformer will always be in some saturation factor just because it's a transformer and that's what they do. If you are operating on an 18V rail, then the audio interface will likely clip before the op amps. 

Clip LED

I should take the time now to talk a little bit about the clip LED. The clip LED will illuminate when the output signal of the op amp reaches 2V below the power rail. For a 12V rail on the 'o2-84, this means that it glows about 1.5dB below clipping. Consider that fair warning. For the 'o2-e, it is closer to the actual clip point. Your audio interface will likely clip at or just above where the clip indicator shows on the o2. 

When running an 18V rail (with the discrete op amps), the clip LED will illuminate when the output signal reaches 32V peak to peak! Yes, that's a lot of signal. In fact, I can't imagine how you would be able to reach that level of audio from a eurorack even at +6dB gain. Your audio interface will clip well before you clip the audio path on an o2 which has an 18V power rail. Enjoy and be gentle to that poor new Apollo interface. 

Remember, you can ALWAYS make up gain in your DAW. Even with the gain set to -6dB (12:00 position), you are getting plenty of bits in your signal. Remember that green is good. You do not need to max out your signal going into your DAW. So when determining where you set the gain knob between -6dB and +6dB, don't look at the meters. USE YOUR EARS!

How do they sound?

While there's nothing like being able to get your hands on an interface to see how it responds and what it sounds like as you would use it, I've started just a simple set of comparisons. I may add more as I have time, so if there's something you're looking for, let me know.

The samples are a couple of modulated oscillators from an E370 into a pair of SEM20 VSF filters whose frequencies were also modulated. Those were then mixed to a single channel with an L-1 VCA mixer and multed using an nw2s::m OPA2134 IC-based mult. Everything was recorded into Pro Tools through an Avid HD IO with no further processing other than gain matching.

The transformer-based configurations were tracked twice at different gain levels so that you can compare the slight peak compression provided by the transformer at the highest levels. One was tracked at -6dB (a normal tracking level) and a second pass was made with the gain turned up as high as possible without the HD IO inputs clipping.

We also added an nw2s::io for comparison. This can be considered thee control as it's the most transparent of the interfaces in the mix.

You can download the original samples here. They are 48kHz, 32 bit float. Load them into your DAW and compare as you see fit. They are normalized to -15dB LUFS. 

First, here's the 'io module recorded at -6dB gain so you can hear a clean, high-fidelity representation of the sweeps. 

18V Rail Discrete Amps

This is the Purple op amp at -6dB (my personal favorite)

And this is the Purple op amp at max gain prior to clipping.

This is the 990 at -6dB

This is the 990 at max gain prior to clipping

12V Rail IC Amps

This is the ADA4084 -6dB. Note that this is an IC op amp into the same transformer as the discrete op amps, but running at 12V rails.

And this is the '84 running at max gain prior to clipping.

This is the THAT1606 electronically balanced module and is closest to the 'io and 'o16 with just an added gain adjustment. It's character will not change up to the point of clipping, so we have just the one recording at -6dB.

Comparison Table




nw2s::io 8 in, 8 out -6dB DC +/-18V included +/-12V 125mA 48mm
nw2s::o16 16 out -6dB DC
+/-12V - +/-18V
125mA 25mm
nw2s::o2-e 2 out -inf to  +6dB DC +/-12V - +/-18V
10mA 35mm
nw2s::o2-84 2 out
-inf to  +6dB
+/-12V - +/-18V
30mA 65mm
nw2s::o2-purple 2 out
-inf to  +6dB


120mA 74mm
nw2s::o2-990 2 out
-inf to  +6dB




nw2s::o2-di 2 out -inf to -16dB AC +/-12V - +/-18V +/-12V 30mA 58mm


* Note that for most modules to run at +/-12V or +/-18V versus the one they were built to support requires changing two SMT resistors. It's a simple process, but best done by those with experience or you'll ruin a perfectly good transformer PCB. Contact for more details!

Discrete Op Amp Support

Here's a conversation I am imagining happening at this moment... perhaps this though is going through your head...

Did you say "discrete op amp"? Can I swap out with other op amps that I have sitting around?
Do you know if the ones I have will work?
Absolutely Not!


Seriously though, there are many op amps floating around out there. There are a number of DIY options available as well as Forsell, API, Seventh Circle, and so forth. In general, they seem to be grouped into 990-type (Jensen) and 2520-type (API). The Purple is close to the 2520 family (although some call it Neve-ish which seeems weird to me) and the 990 is obviously the official descendent of the Jensen op amp. I love the Purple and I love the 990 for different reasons and I really feel like since they are available and represent the best of those families that there's little more to look for. However, I know that given a chance to tweak anything, folk are gonna do it. So what do you need to consider when looking for other discretes?

Different op amps have different input impedance and different levels of stability at different gain levels. The o2 circuit is a very basic circuit with a variable input attenuator and a fixed gain of 2. The compensation filter around the op amp is around a single order filter 3dB down at 33kHz - far enough out to not to impact the audible frequencies. Whether or not that's far enough out to prevent any particular op amp from oscillating can only be tested. It's a pretty safe circuit, so we will leave testing as an exercise for the listener. Let us know!