[Review] Sparkos Labs - Discrete Op amps SS3601

Sparkos Labs - SS3602 Dual Op Amp

Small but beautifully formed!

Product name: SS3602 Dual Op Amp
Manufacturer: Sparkos Labs
Cost: $79.80. (Currency conversion)
(YMMV)
Evaluation platform - Maverick D2 DAC: Maverick Audio
Cost: $250.

Reviewer: Mike Cox - TNT UK
Reviewed: May, 2014

Introduction

Many audio equipment manufacturers these days use operational amplifiers. Operational amplifiers are a simple solution for obtaining gain and low distortion. These devices cost pennies, though there are some designed for audio use costing much more. Many audiophiles reject equipment using integrated circuit operational amplifiers (op amps) preferring discrete circuits. For example, my La Voce DAC is designed without op amps. The benefit of using a discrete in my experience, is a more open and detailed sound but this solution requires careful design and is more expensive to manufacture. Sparko's Labs have designed a "Discrete Op Amp" that is a plug in replacement for most of the IC op amps aimed at the audio market, and betters the standard designs with higher gain, lower distortion and improved stability.

The design - Sparkos Labs Explanation

The differences between my devices and op amps such as OPA2604 are as follows - 1 - Open loop gain. Loop gain reduces distortion and essentially makes the output more precisely mirror the input. The amount of error between the input and output (distortion) is directly related to loop gain. My devices have 40dB more open loop gain than the OPAs (140dB vs 100dB) Keep in mind, this is not a 40% increase in gain due to the logarithmic nature of dB's.

For example 100 dB corresponds into a gain of 100 thousand for the OPA, and 140 dB corresponds to a gain of 10 million for the SS3602.

So my devices have 100X the loop gain of the OPA's .

As a side note, I know that some do not like high gain devices, however, I contend the reason for this is because their device is likely oscillating in the MHz region. In single pole compensated monolithic IC op amps, high gain always comes with high bandwidth (~50MHz or so). High bandwidth devices are prone to oscillate and become unstable when dropped into random circuits with DIP sockets and such. So its not really the high gain that is to blame, it is rather the high bandwidth and associated stability issues. In single pole compensated IC op amps, high gain and high bandwidth are essentially inseparable.

My devices use two pole compensation which is impossible to do in monolithic op amps since it takes 2 capacitors. Capacitors are very hard to make inside of ICs.

Two pole compensation is how I achieve very high loop gain and still have a reasonable bandwidth of 5MHz. The reasonable bandwidth allows my devices to easily drop into circuits without the stability and oscillation issues. Another advantage of 2 pole compensation is the input stage of the device does not have to drive the compensation network - which makes the input stage far more linear. In single pole compensated op amps, the input stage must drive the entirety of the current demanded by the compensation network. It amounts to more loading on the input stage which degrades sonics. (I talk about this compensation stuff more in my datasheet and further explain single vs 2 pole compensation)

To summarize, the Loop Gain and compensation scheme are a major difference between my devices and the OPAs.

2- OPAs use JFET inputs, my devices use BJTs. have less noise (the OPAs have 3X the broadband RMS noise than my devices 1.5uV vs ~500nV) BJTs are far more linear and have a higher transconductance than JFETS. BJTs also have better matching between devices than JFETs do which is important in differential input stages.

3 - Every stage in my device operates in class A mode with fairly high bias currents. The OPAs only draw 5mA total per amplifier. This is spread across their input stage, gain stage, and output stage. It leaves their devices somewhat current starved, and their output stage is undoubtedly class B or at best it's biased with only a couple mA.

My device's output stage alone is biased at 7.5mA. My input stage and gain stages are biased at 3mA each. In total, my devices run at ~14mA / amplifier while the OPAs run 5ma. High current biasing, especially in the output stage always sound better and make for more linear amplification.

Keep in mind, that 14mA biasing current isn't possible in monolithic op amps either due to their power dissipation limitations. This limits the amount of current that IC op amps can use for biasing.

[Maverick D2 DAC]

Evaluation Platform

As soon as I knew the Sparkos Labs discrete op amps were on the way I had to come up with an evaluation platform. I needed a device known to be good, using op amps in sockets and easily obtainable. A email to the team at Maverick Audio and the D2 DAC was dispatched.

The Maverick D2 uses the Analogue Devices AD1955 DAC with Burr Brown OPA2604 op amps, fortunately fitted in sockets. The supporting components have all been chosen for their sound quality with the USB interface delivered via the Tenor 7022L chip, a pro audio device capable of handling up to 24bit 96kHz data.

I supplied the Maverick with data from my MAC Book PRo with a Chord Silver Plus USB cable, AQVox USB power supply and finally an iFI iPurifier. I wanted to ensure the quality of the data was as good as possible. The Maverick D2 offers excellent value for money at $250 US especially when you consider it also has a valve output stage based on the 5670 double triode.

I setup a playlist with a selection of tracks, some at 24bit 96kHz, for evaluation and started by listening to the basic Maverick D2 via the op amp output stage. I kept with the op amp output for the duration of this review but for anyone interested I have added a section at the end on the valve output.

For the asking price, the Maverick D2 is outstanding value for money, the build quality is excellent and the service from the team at Maverick first rate. The sound quality is very good indeed when fed a good quality stream of data. When you consider the difference in price between the Maverick and the La Voce, my regular DAC at 6 times the price, the Maverick is amazing. Where the Maverick loses out, is in stage depth and smoothness; it has a digital edge or grain. The Maverick uses the cheap and functional 78XX and 79XX voltage regulators to supply the op amps. There are numerous options for better quality regulators so I asked Andrew from Sparkos if there was any benefit to upgrading the regulators. The response was interesting, as the Sparko's op amps are operating in class A, their demand on the supply is much more constant. With the regular op amps operating in class A/B mode, as the output voltage of the op amp increases so does the current demand and the supply must follow this. This means the regulators must have the ability to hold the voltage constant whilst at the same time deliver varying current at audio frequencies. Most regulators were not designed to meet these varying demands. I may explore voltage regulator upgrades for the Maverick in a future article.

[Inside Maverick D2 with OPA2604 op amps]

Fitting the SS3602

With the OPA2604 op amps in sockets, removing them was easy. Fitting the SS3602 is a little more difficult as the pins on the Sparkos devices are considerably fatter and really do not slide into the sockets. I expect that some sockets will be a better fit, but those in the Maverick mean the SS3602 are only partially located in the socket and are easily knocked out.

The recommendation from Andrew at Sparkos Labs is to find a better socket that will take the fat legs of the SS3602 and use that as an adaptor. I expect the turned pin variety would work well, I just don't have any to try at the moment.

How do the op amps compare?

Switching between the IC and discrete devices is quick, but my ability to memorise sound quality is shorter! Even with my poor audio memory the change in sound quality was obvious. The Maverick D2 seemed to gain weight in the lower frequencies and the detail improved considerably. On the Eagles' "Hotel California" track from the "Hell Freezes Over" album the drum on the intro was delivered with more weight and the decay much easier to hear.

Turning to the Tallis Scholars and the Allegri Miserere, I heard the best performance of this piece in my room using the Magnaplanar SMGa speakers and Emotiva XPR-2 amplifier. As with the Eagles the discrete op amps seemed to add more weight to the lower registers, more basso profundo and less castrati! It would have been interesting to partner the Maverick plus the discrete op amps with the Divine Acoustics Proxima speakers and the Encore 7 Eggshell Prestige 15WS amplifier I reviewed recently. The ability to resolve details of the setup might well have been very interesting. The Sparko's devices, as well as adding weight, picked out more detail in the recording that I had missed previously.

[Maverick D2 plus discrete op amps]

On to something with a bit more beat, Santana and the album "Milagro". The added weight and richness in the harmonics was obvious. The rhythm also seems an improvement, the album "My Way" or "A Mi Manera" from the Gypsy Kings comes across as if the players have stepped up a gear and are playing with more energy and enthusiasm. Playing the Paul Desmond/Gerry Mulligan album, specifically the track "Body and Soul" the saxophones have tremendous texture with the double bass and drums providing the perfect support. It feels as though you could reach out and touch the musicians.

Conclusions

Fundamentally, the Sparkos Labs discrete operational amplifiers are still a high open loop gain amplifier that when feedback is applied delivers low distortion. The difference is in the execution, the Sparkos devices operate in Class A mode and have much higher gain than all the regular devices that I know of. The impact of replacing a regular device with those from Sparkos Labs is significant, with added weight, tonal richness, resolution and improved timing.

The impact on the Maverick D2 DAC is to lift it from a very good $250 device to something equivalent to a $1500 DAC. In my system the Maverick D2 and Sparko's SS3602 combination costing approximately $500 is as good as, but different to, my La Voce when fed data via the USB input. The differences are difficult to describe and I expect are down to the DAC chips, the La Voce using the TDA1541A and the Maverick the AD1955 a much more modern design.

Valve Output

The Maverick D2 has a valve output stage as well as that from the op amps. I am not sure if the op amps are in the circuit between the valve and the DAC chip. Usually some form of buffer or I/V stage is required, between the DAC chip and output stage. The difference in the sound quality between the the two outputs is remarkably small. In my setup and room I find it hard to choose which I prefer. The additional output lifts the Maverick D2 even further up the DAC ladder. The listener can connect both outputs to their pre-amp inputs if available. It is now a simple matter of switching between between the two outputs as the music and mood requires.