Spring 2010

Modifying A Marantz CD-80 CD Player
A blow by blow account of taking a venerable CD player
and giving it a new lease of life.
Article By Thorsten Loesch

Difficulty Level

  Modifying and upgrading CD players has a long history. It started almost at the beginning of the CD era and some of today's "Digital Giants" (like Wadia and Meridian) started out modifying commercially made players for better sound. But surely, with the CD having long reached majority and heading for midlife crisis age, CD players too have come of age? Surely today's 35 UK Pound Wharfedale or Alba DVD player from the local supermarket should be able to play any CD as well as anything out there? Sadly no, cost constraints mean that very few CD and DVD Players actually can fulfill the undeniable potential offered by the best CD recordings available.

Yet where is this performance lost you ask? Simply said, where it costs money to do the job well. The CD and DVD transports of most players, be they high end or ultra budget are sourced from the same two or three manufacturers with the same fundamental technology. The Chipsets controlling these transports too come from very few sources. So cost for such drives and chipsets is very low. But no circuit is better than its power supply. A CD player relies on precise timing (free from jitter) to deliver the best quality, thus, it cannot be any better than the clock that beats at its heart. And while today's budget DAC chips can often deliver a surprisingly large part of the sound quality hiding on the best CD's, it is still very easy to cripple them with a poorly designed analogue stage supplied by a most basic power supply.

A great example of such skin flinting is of course found in budget DVD Players, however their casework is usually way to cramped to allow modifications that take some space. Another famous example of budget (CD) Players that were hampered by the issues discussed above are the Philips CD-71x, CD-72x, CD-75x series and their Marantz equivalents (in this case means merely dressed in Marantz cosmetics but with unchanged internals) CD-36, CD-46, CD-4000 and CD-5000.

Alas, this article will not focus on modifying such potential as these Philips/Marantz CD players nor will it cover such potential modern budget marvels as the Oppo Digital BD-83SE, even though they could serve as an excellent case study just how much performance lies untapped.

Instead we will modify an unashamed "high-end" player, the venerable Marantz CD-80. Well, at least in its own time it was considered high-end. By today's standards some of the digital engineering is pedestrian as is much of the analogue circuitry and jitter is high as well. Yet listening to a sample that is good nick or has been well restored and carefully upgraded (as was the case in the unit used for this case study) the results can be surprising. Cue up a CD and you are greeted by a warm and powerful sound with more than a few shades of classic analogue (LP) sound, soundscaping is pleasingly open (if not up to the best) and generally listening to this essentially 20 years old player is a joy.

Sure, there are many better CD players out there, but most of them are not enough better to justify the very large expenditure to acquire one, if you happen to own one of these Marantz Battleship Players. There were several similar designs in the late 80's and early 90's Marantz line-up, with exact equivalents also coming up as Philips products. This CD-80 had all the 20 years old electrolytic capacitors replaced, something that is absolutely essential to get the original performance these units where capable of, with modern equivalents (Elna Silmic and Cerefine) or with improved capacitors available since (Sanyo Os-Con).

Generic Clock

Generic Analog Stage

The NE5534 Op-Amp's in the analogue stage had been upgraded to the now discontinued Burr Brown OPA-627, something that can be omitted but does offer a pleasing improvement in smooth delicacy and detail and opens the soundscape notably over the NE5534. However all of the fundamental design and circuit was left original. It was used simply because it was at hand and because the owner was keen to see if the unit could be taken to the next level applying wholesale replacements to two key sections of the player, namely the clock and the analog stage.

Surprisingly, the power supply for the TDA1541 DAC features discrete low noise regulator circuits that leave little to criticize, so no upgrade for these was considered, however in many players this power supply especially would be another good target to be blitzed. So it was decided that this player would be fitted with an "Ultimate EZ Clock" System and a Universal Tube Stage, both available from DIY HiFi Supply and in part designed by yours truly.

See the sidebars "To clock or not to clock, that is the question here" and "The sweetest sins are analogue, even in digital" for some of the reasoning and technical background behind the decisions to replace these sections.

The Ultimate EZ Clock system comes in two circuit boards; one contains the clock, clock buffer and output circuitry and two low noise super regulators preceded by an LC filter. The other contains a 12V regulated power supply with its own small transformer with electrostatic screen, Schottky rectifier diodes etc. to make as close to a "no compromise" solution as possible.

The Universal Tube Stage is a hybrid SRPP tube amplifier (the upper current source part of the SRPP is realized with solid-state devices). The same compact circuit board also contains the high voltage supply realized with a CRC filtered power supply combined with a zero feedback regulator and the regulated heater supply. Only the oversized torroidal mains transformer is not on board, to keep the size compact and to allow for easy installation.

Installation and user manuals for both items can be downloaded at the www.diyhifisupply.com, so I'll be keeping the descriptions short here.

Installing Mod-Modules
Before I proceed to the actual article, allow a little rant on the subject of the installation of modifications by both DIY enthusiasts and by supposedly professional modifiers. Over time I have seen a shocking number of such "installations" that gave me the heebie-jeebies' badly. I mean it is one thing if you blow up your own prized high-end player modifying it and have to seek professional assistance to get it fixed again and to have the actual modification installed properly, it is, in my view an entirely different kettle of fish, if the modification is done incompetently by people that charge 50 or 100 pound an hour to do so.

I will refrain from being specific as to who did what and how badly, instead I will just give a small of "Do's" and "Don'ts".

Don't:

Use sticky backed plastic standoffs to try to hold any circuit board that has any substantial weight (best not use them at all).

Use the wiring that connects the circuit board to try to hold it in place.

Use hot-melt glue to glue a circuit board to a metal chassis, over time things will short circuit, even if they do not do so immediately.

Use double sided sticky tape to fix a circuit board (it will come loose).

Use silicone glue and paper as isolation to glue in a circuit board.

Use any glue or sticky stuff at all to attempt to hold things in place mechanically (what do you think they invented screws for?).

Use tie wrap or zip-lock ties to hold things in place.

Try to fit any circuit boards carrying mains voltage with minimal clearance into tight spaces.

Use improvised ways to hook up cables, especially ones carrying mains voltage.

Do:

Think about layout, wire routing and mechanical arrangements many times more than you think it needed.

Use machine screws and correct plastic or metal standoffs to mount circuit boards.

Use a decent electrical hand drill and sharp drill bits suitable for the metal chassis to make the necessary holes to use these standoffs.

Mount any circuit board so as if you expected to have to use FedEx to ship the unit after assembling (in my dayjob we substitute one round of insured FedEx shipping for doing drop tests on new packaging – that way we do not pay for the often inevitable damage).

Leave enough clearance to any mains voltage bearing circuit board or exposed metal piece (5mm MINIMUM to the nearest piece of metal carrying mains voltage).

Learn to solder if any connections need to be soldered, bad solder joints cause unreliable connections.

Use solid screw terminals or good quality solder connections for any connection made.

Make sure to isolate any connection carrying mains voltage appropriately.

Important!
Failure to avoid the Don'ts and do the Do's may in the best case lead to an unreliable piece of equipment which will soon fail in unpredictable and often hard/not at all/very expensive to fix ways which is bad enough. However, at the worst sloppy work by you or indeed those you pay may kill you or loved ones when the case is suddenly "live".

This is no idle chitter chatter nor just a standard disclaimer.

I have seen things that friends dragged in that where supposedly professionally installed that made my hair stand on edge and made me give way to long strings of really bad profanity. Some of the DIY installs were even worse. And as anyone who knows me personally can attest to, I am not given to gratuitous use of bad language and if they have seen me in tight situations they can attest I do not scare easily, if at all.

Please consider the methods shown in this article for fitting modifications as MANDATORY and MINIMUM STANDARDS and not as optional. Be solid, be safe, be reliable.

In addition, if at all possible try to get a service manual for the piece of equipment you are modifying, there are now many outlets on the World Wide Web selling them. If you cannot find the exact one, try one for a mechanically and electronically fundamentally similar unit. Even if your electronic knowledge is not great, the service manual usually shows how to get inside the piece of equipment without scratching it and how to put things back together afterwards.

Further, as you go along with disassembly and disconnecting wires and all that you may be best off taking as many digital photos as possible, so you can go back and see how things came apart, which screw fitted where and what cable was plugged into which socket. In addition it makes it easy for you to afterwards write an article like this one and blog about how great a modification guru you are and impress all your audio friends.

Now that I have this of my chest, bring on the CD-80!

Inside The CD-80
Okay, we removed the side cheeks and popped the top cover and removed the bottom cover. We are greeted by an impressive sight. The CD-80 has the justly legendary Philips metal chassis swing-arm mechanism (not visible in the picture), a creation that appeals to the mechanical engineer in me, a combination of precision and solidity rarely repeated in CD-Drives since. There is much copper plating in evidence around the chassis and the actual chassis is solid cast metal, an impressive sight if one considers that most ultra-high end gear still uses chassis made from bend and bolted together sheet metal.

Marantz where certainly trying to deliver serious value back then, a state changed rather to worse in the time gone by since then. The DAC is a TDA1541(A), a DAC that has over time developed a major cult following. I may not be entirely innocent of contributing to the hype surrounding this long discontinued chip, however to my ears at least all the hype is completely justified. It is used among other products in the highly acclaimed $43,000 (USD) Zanden CD-Transport/DAC combo and the 10,000 US Dollar AMR CD-77 CD player (for who's design I share some blame).

Okay, so an excellent basis for an upgrade, or I at least would say so. After much looking around to see how and where to best place the different circuit boards (keeping in mind the Do's and Don'ts from above – heal thyself physician) I decided to mount the clock circuit board and the clock power supply together with the mains transformer for the universal tube stage to the central brace running from the front to the back of the player. This places the clock circuit board close to the location of the original clock crystal.

Mod The Cat
I removed this brace and actually used a drill press to drill all the required holes after measuring/test-fitting several times. However anyone handy with a hand-drill could have made these holes equally well. Just use a sharp HSS steel drill-bit of the right size. To fix the universal tube stage mains transformer I used the original bolt and other mounting hardware, for the circuit boards I purchased hexagonal brass standoffs and M3 machine screws. The standoffs are fitted to the actual clock module.

This was the pretty easy part of fitting the parts. For the universal tube stage I decided to mount it piggyback above the solid state analogue stage. In order to make things more interesting I decided to leave the original solid state analogue stage in place and operational.

This called for a lot more effort. For the universal tube stage I was able to use one of the original fixing locations to locate one of the brass standoffs in the far back corner of the original circuit board of the CD-80. Two more standoffs I was able to fit to the after carefully drilling suitable holes in the CD-80 circuit board, making sure I did not disrupt any important circuit traces.

However, I was unable to find any way to fit the fourth and final standoff. Of course, I could have simply left it three, probably solid enough a fitting, but if you know FedEx as I do...

Well, rules exist to be broken. Remember how under the Don'ts I say "no sticky pads, no glue whatsoever!"? Well, in this case I resorted to a plastic pad and I used some two component epoxy to hold it in place. Not very elegant and looks messy (no, I'm not showing a photo of that), however, now I am certain this player will survive shipping via fed-ex, if the transport locking screws are correctly applied. The results of all this looks deceptively easy, yet this part really took quite a bit of time.

By comparison drilling the rear panel with a hand drill to accept another pair of RCA sockets (for the universal tube stage output) and a small toggle switch to switch between the original analogue stage and universal tube stage was rather easy.

Now was finally the time to break out the soldering iron and to start do the actual electrical work. First order of business was to find an easy and safe place to steal the mains power to supply both the clock and the universal tube stage. This may seem an unimportant detail, but it is paramount to making sure the results are safe to use.

Kindly Marantz provided some nice solder points directly at the mains transformer, making this job very easy, especially as there was nice plastic clip on cover to make the results safe from touching. Being obliged by so much kindness, I felt compelled to use this very easy way.

With many more modern players it will likely be necessary to solder wires directly to a circuit board, below the board, to the connections of the on-board mains transformer or other convenient points. If doing so, please, please take good care to solder reliably, cleanly and retain any clearances (distance between the chassis or other parts of the player to anything that carries main voltage without insulation – 5mm is the absolute minimum).

Still, if an easy solution presents itself, take it, just be safe.

So, I wired the two power supplies mains connections according to the schemes given in the respective manuals on the other side of these two wires, done. As the clock supply has sturdy screw terminals for the mains connections I used this to connect both the mains wires from the universal tube stage transformer and the main power connections, this makes things very neat looking.

Especially with any wires carrying high voltages and currents neatness is a bonus; please take my word for it. Make sure no loose bits of wire are hanging around and may short to anything they should not.

A simple pair of wires connected the clock supply module to the actual clock circuit board, real easy, no soldering, while this only carries 12V DC again, being neat does no harm. This part is done.

Now on to connecting the clock, instead of the crystal. Again, this has been made really, really easy. With all other clocks I know you actually need to find the exact pin of the chip you are connecting to and connect some coax cable, remove various components and sometimes add some extra ones. This does not faze me much, but for many average DIY enthusiasts all this is a bit much.

The "Ultimate EZ Clock" aims to make the process of connecting up the clock as easy as possible. It comes with several adapters that simply solder in place instead of the existing clock module or crystal. No parts to subtract, no parts to add. Plug in the supplied cable from the clock and you are done. Below a few small pictures illustrating the process step by step, those few pictures hopefully tell more than thousand words.

Well, I did not have the 150MHz HP digital storage oscilloscope at hand when I modified the CD-80, however on a different occasion I was able to compare exactly what is the difference between the original simple crystal circuit in place in 99% of all CD players and the Ultimate EZ Clock. Looking at the waveforms, i can see that rise and fall times are asymmetric, the rise time is slower than the fall time. This affects how precise other parts of the circuit are triggered by the clock. Equally, the duty cycle differs from 50%, this can affect the way certain DAC's perform. Note, this is the clock directly after the clock generator, it can only get worse downstream.

So, how does the UEZClock shape up instead?

First, the waveform is much more like a square wave, still not perfectly square, yet much more so than the original. Some of this "un-squareness" is actually the load of the ‘scope probe causing trouble. Without the ‘scope it is much more square, but we cannot see the results. The rise and fall times are absolutely symmetric and much shorter (by a factor three to four) than before. The duty cycle is also much closer to the magic 50%. I would call that a nice result. Of course, these traces say little about what the difference sounds like. To try this I actually fitted the original crystal removed from the player to a plug that could fit the new socket. So now I could compare (with a listening panel) between the player basically original and then with new clock by simply unplugging one and plugging in the other.

Surprisingly, with the player stopped it was even possible to switch between clocks without turning the player off, making very rapid A/B comparisons possible. Well, more on the results later, for now the picture of the lone socketed crystal…

Connecting up the universal tube stage was a little more involved, mainly because I elected to solder all connections and because of the switching between the two analog stages. In short, the switching involved using epoxy resin to glue (I am breaking my own do's and don'ts again) to glue a signal relay below the circuit board and switching the DAC output pins to either universal tube stage or the original analog stage.

A straightforward replacement would have been much easier, especially had I used solderless WBT RCA sockets. The universal tube stage features screw terminals that remove the need to solder to this circuit board completely. On the CD-80 circuit board I would have likely just removed the Op-Amp's thus eliminating the original analogue stage and would have used the holes vacated by the Op-Amp's to solder in my (four) wires from the universal tube stage cleanly.

Hence I have not documented the wiring up of the universal tube stage step by step, instead I'll show the hardwired approach I took. Wire is a mixture of thin solid core silver plated copper in Kynar wire wrap wire for the signal and stranded silver plated copper in Teflon for the grounds. All the jumpers of the universal tube stage where soldered across below the circuit board. The performance gains of doing this over the original jumpers and screw terminals are not huge, but as they say "Every little helps.".

All in all the whole job took maybe four hours to do, without doing so much "documentary/scientific investigation" stuff I could have easily finished in two hours. I suspect a reasonably experienced DIY enthusiast who has build a kit or two prior to attempting this could easily do it in an afternoon or two.

What remains is to show the plan view and some close-ups of the modified player and of course to wax lyrically about the improvements in sound all these modifications produced.

What Is All This Modcateering About anyway?
What Does It Sound Like?
Okay, I have bored you for over 3500 words (not counting the sidebars) with boring technical stuff, advise on drilling holes and not getting electrocuted. Yet I hear you ask – You said naught ‘bout sound. Is it worth it? Spending many 100's of Dollars to buy modules, using a rainy Sunday afternoon to fit them, come on, tell us how good the results are!

Okay, first, we started with a really good player (after the rejuvenation). In fact, we started with a player I'd personally prefer over probably around 75% or more of all CD players and related miscellanea. I personally would say that the post modification results lift the player to a new level of performance, while preserving the good sides of the original player or even increasing them. However, as I had a hand in designing the clock and tube stage this may be "proud papa" speaking.

So, over to the listening panel. The tests were done blind and level matched, however, they were simple preference tests with the panel members allowed to submit added notes. So they did not follow the ABX protocol, instead A and B where clearly identified, but the identity of A and B was kept "secret" (single blind test).

The original, if not stock CD-80 clearly showed its TDA1541 heritage. It had musicality and a warm, big analogue like sound. When it came to detail and soundscaping it was doing okay, but better had been heard on other occasions. Overall the sound of the original CD-80 was criticized as somewhat mechanical and fatiguing. The panel, having compared many units in the same system, nevertheless suggested the original CD-80 was a very accomplished player and probably one of the better to be auditioned it was given on average 70 points out of 100, where 100 was simply the best CD player/DAC heard by the panel.

Switching to the tube output produced more detail, more space and a much more realistic (one member called it "more humane") sound. One may dissemble lengthily about the benefit of only tubes after the DAC (I did so in another article), but short and sweet, it was "more of anything".

The next stop was the clock comparison. Surprisingly, the CD-80 in original trim did show not that great a difference between the stock crystal and the rather expensive new clock system, sure, there were some subtle improvements in dynamics and a little improved soundscape, but nothing earth shaking. The trend was reversed when the CD-80 was run with the tube stage instead of the (arguably close to "as good as it gets that way") OPA-627 analogue stage. Now the UEZClock significantly increased the perceived dynamics, the tactility and realism of the music. No negatives where noted.

Compared to some extreme and very costly high-end players the listening panel was familiar with, the modified CD-80 fell still a little short. In full trim but with little run in time a score of around 85 out of 100 was recorded, with some more burn in I suspect it would have exceeded the 90's mark. So no giant killing David then, but a player that can hold it's own against many other comers and shows its limitations only against both extremely expensive and refined opposition.

In some ways I would love to announce "spending a few hundred bucks on your CD player makes it knock all other players into a crocked hat" but that is not the case. Yet, given many more conventional high end CD players fail to even match the original CD-80, at least for listening pleasure and analogue sound, I would say the less than $500 (USD) plus an afternoon spend on the upgrades give a greater result than spending five to ten as much on a new player from the mainstream companies.

My own money? Not having at current an ultra high-end setup at home, though I have a very serious setup in an acoustically treated listening room at work, I have a DAC using the TDA1541 and a universal tube stage and the source has a UEZClock fitted. The results are very much like the modified CD-80, possibly a little better due to some extreme modifications not applied to the CD-80. The results provide much pleasure from music for a hard day's night and please my girlfriend and myself well enough not to want more.

Till the next piece, have fun and Enjoy The Music.

Sidebar 1
To clock or not to clock, that is the question here.
The subject of replacement clocks in CD players has a long history, reaching back into the early 1990's. It is a hotly debated topic, but has a sound founding in good old-fashioned electronics. Any DAC or CD player requires a clock to operate the digital circuits. This "clock" is simply a high frequency signal with (at least in theory) high precision of frequency and high stability of this frequency. As with all things – ideal clocks do not exist and the real clocks are subject to variations in clock frequency due to a wide range of reasons.

Such variation of the clock frequency can mean that a sample which was correctly read from the CD is reproduced at the slightly wrong time. This phenomena is called jitter and it is audibility is highly disputed. More reading on the subject of jitter can be found on the world wide web, so we shall refrain from discussing its sonic impact etc. here. Here we want to discuss the actual reasons why clocks in CD players show significant amounts of jitter and how this can be avoided.

The circuit to the left is the kind of clock generator most often found in all sorts of equipment and while others exist it is even often found in many a "superclock". In principle there is not a lot wrong with this circuit. It is cheap, effective and simple. To a large degree the quality of the quartz crystal used will determine the clock jitter, if (and that is a BIG if) the power supply to this circuit is free from noise.

An American patent I recently noticed uses a change of the power supply voltage to this kind of oscillator to construct a very low cost and rather effective VCXO (vcxo = voltage controlled crystal oscillator). This showed a pull range of around 400ppm (ppm = parts per million) for a change of the power supply from around 3V to 8V.

In most CD players the power supplies in the digital section invariably also supply this clock circuit. These supplies are usually subject to several 100 mV of noise. Measuring 500mV peak-to peak is entirely common. This means that the power supply noise will modulate the clock frequency by +/-20ppm. Of course, ppm do not directly translate into jitter, and the clock circuit itself is not the only circuit affected by noisy power supplies.

However, it is clear that having more noise on the supply for this generator translates into more jitter. One number that I remember came up in a JAES paper was 10pS Jitter per 5mV supply noise, in the case above this would mean 1,000pS jitter, much more than can be allowed without degrading CD replay performance. Another issue is that this oscillator does not like to drive long traces on the PCB or multiple IC inputs. So some form of buffering is needed. Yet buffering can again introduce additional jitter. Hence a "simple clock" is all but simple.

Nowadays a wide range of actual "clock" integrated circuits are available, with jitter performance and precision merely a question of specification to the supplier and actually paying the required fee for the minimum order (usually a few 100 pcs). This makes it quite easy for a manufacturer with fairly large volume to get the right combination of cost and specification, for individual DIY'ers this route is not open.

We need to buy what is offered to us in small (single) quantities. While most of these little metal can oscillators look the same, from the cheapest to the most expensive one, performance varies widely as does the internal technology. At work I have taken more than a few samples apart to see what we are getting for our money, as well as measuring the results. The results where quite illuminating. However, no matter how good the clock, it needs a quiet power supply and a low jitter buffer.

Making a really quiet power supply is quite non-trivial. Normal 3 pin regulators are noisy, so discrete circuits or so-called super regulators are needed. Yet these need a reference voltage to operate and many voltage references have much, much more noise than the op-amp in the super regulator. So this noise needs filtering, however to filter significant any noise at 10Hz – 20Hz means we need a lowpass with a small fraction of a single Hz corner frequency to be effective. Such a filtered reference may take many seconds to get to the minimum voltage required by the clock to work and the resulting delay in the clock coming on line can cause interesting results with some CD players. Hence when designing an "add-in" clock one has to take a rather different approach than one can take if one designs a CD player ground up.

So if the clock in your CDP runs of the basic supply with the rest of the digital circuitry and has sub-ideal buffering for the clock (I leave it to your best guess if this is the case) fixing these issues by adding a suitable "Superclock" where you can be assured the issues have been taken care of can bring major improvements. Note however that just adding a clock is no guarantee for improved sonics.

Sidebar 2
The sweetest sins are analogue, even in digital.
The analogue stages for DAC chips and CD players have traditionally been one of the weakest parts of the CD-Replay chain. From day one Philips and Sony, the main innovators behind CD used what was considered at the time (late 1970's) state of the art integrated circuits, called operational amplifiers. The "generic" circuit of the early CD players looked something like this:

This circuit is still standard for most current output DACs, as many modern DAC' have the first Op-Amp integrated on board while others use different circuitry, which nevertheless operates in a similar manner.

The problem was that most of these circuits are limited by something called slew rate, meaning the output can change its voltage only at certain maximum speed, the so-called slew rate. While the input signal demands the output not to change any faster than its slew rate, all is well and the operational amplifier does its job very well, very linear and with few artifacts.

In a DAC for CD-Replay especially the first operational amplifier is severely challenged. Many early CD players used DACs like the PCM56 from Burr Brown. This DAC can change it's output from 0 to 1mA in 0.35 micro seconds (and it can attain a more than 0.7 mA step in a much shorter time). This would be equivalent to stepping the output of the first operational amplifier from 0V to 2.83V in 0.35 micro seconds. An NE5534 in the configuration normally used for I/V converters has a slew rate of 6V per microsecond. Yet the output would be required to change at a rate of 8V per microsecond to follow the input signal.

Should you ever have a CD where the music contains a signal where (due to mastering problems, clipping etc) you get a single step from 0V to maximum signal or worse, from maximum negative signal to maximum positive signal the first Op-Amp would be badly overloaded by having it's slew rate exceeded.

Some may argue that capacitor in the feedback loop will cause a low pass effect (it does) which removes the need for the operational amplifier to react as fast as described. This is partially true, only very partially, as this filter can only operate with an operational amplifier that is not slew rate limited. Trying to low-pass filter any signal near the slew rate limit of the operational amplifier causes severe distortions. Of course, when measuring the design using 1kHz sine-waves (as is still common practice) non of this will show up in the measurement.

These distortions tend to happen only at certain times, not constantly. So highly dynamic music with large amounts of high frequency content is more prone to cause these problems to become audible due to being extremely heavily compressed music. Early CD players were routinely criticized (including by myself) for making all cymbals, especially crash cymbals, sound like air escaping a bicycle tire. I suspect that at least part of this was due to the operational amplifiers being operated near their slewing limits. Using a very linear operational amplifier with a relatively high slew rate will help to avoid this kind of distortion. One of my favorite operational amplifiers for current to voltage conversion is the now discontinued Burr Brown OPA627, yet others that are very good exist as well.

Modern DACs no longer use multi-bit architectures and instead switch the output in a very rapid fashion between the two extremes of voltage available. Usually these kinds of DAC chips have on board filters to avoid exposing external analogue stages to switching between 0V and 5V around 11 Million or more times a second, which can cause problems for very fast video operational amplifiers, never mind the pedestrian NE5534.

Even after low-pass filtering on chip and off chip much very high frequency noise remains. If the operational amplifier following this kind of DAC has linearity problems at high frequencies (and they all have) we have severe distortion for these (unwanted) high frequency signals. In most operational amplifiers distortion rises at a rate of 18dB per octave, that means every time the frequency is doubles, distortion rises nearly tenfold. Now a good quality "audio" operational amplifier may have a distortion level of 0.0003% at 1kHz. Using our estimation from above, we find that the around 11MHz many modern DACs use as switching frequency are 13 octaves above 1kHz. So our distortion would increase ten-fold thirteen times! At 11MHz our Op-Amp likely has many times more distortion products than actual signal!

A British "flat earth" HiFi magazine once had a front page with a leader of "CD – 300% Distortion". While this referred to low level signals, the wideband RF noise put out by modern DACs also causes more distortion than signal with a common "audio" operational amplifier. And while you cannot hear the harmonics of a distorted 11MHz signal, what you can and do hear is the inter modulation effect from this noise in the audio range.

In measured terms this kind of distortion shows up as so-called noise floor modulation, that is the background of the DAC/CD player is modulated by the signal, little signal = low noise floor, large signals, large noise. Unlike with the older multi-bit DACs, this RF noise inter modulation is present all the time regardless of signal. It tends to cause a certain haziness and artificial smoothness in the more benign cases and sounds outright nasty and gritty in the worse.

In the case of one fairly famous 1990's British CD player that used the rather ancient TL071 operational amplifier with a pure "bitstream" DAC we measured as much as -40dB (that is 1% of signal) background noise using a maximum signal multi-tone test! And the noise was not even "white noise", which already sounds bright and forward, but might have been called "blue noise" (not sure there is such), as it actually showed a lot more high frequency content than low frequency one! No wonder all reviewers of the time slated it as sounding bad compared to vinyl. By comparison vinyl has a steady state around -60 to -70dB noisefloor with a slope that approaches the "warm" sounding so called "pink noise", so in audibility of noise terms it was at least 40dB better than this CD player!

Little wonder then that "op-amp swapping" in the analog stages of CD players has been practiced since the 1980's and continues to be a hot topic. Of course different op-amp's will react differently to RF noise and signals near or above their slew rate limit, of course, the audible consequences will vary with music played. So the opinions are many and often conflicting.

About The Author
Thorsten Loesch has been the Director of Technology for Abbingdon Music Research (AMR) since 2003 and design consultant for DIY Hi-Fi Supply since 2001. Mr. Loesch has for a number of years, written extensively on audio equipment designs and modifications and is also an experienced audio equipment reviewer. Before joining AMR, Mr. Loesch was a financial systems manager for the UK public health service. Prior to this, he was a Tonmeister (Sound Engineer) in Germany and gained extensive experience in professional audio design, setup and operations. He holds degrees in both Electronic Engineering and in Business Information Systems.