Marantz CD67 clock mod.. my experiences

I tweaked my cd67 heavily about a year and a half ago replacing opamps with LM6172, doing a lot of decoupling, replacing capacitors, shielding digital ic-s etc. but at that time I didn't touch the clock.

Now last week I reread the -> the clock hack
again and got interested. I also found that I had missed a detail in datasheet that the two inverters responsible for clock oscillator function inside the SM5872BS have sepparate pins for ground and supply voltage...

So yesterday I took out the pcb and looked at it a bit. Then decided to go for the clock mod, but with a bit more radical approach. The pcb layout for the clock is indeed poor. The crystal and two caps that are in parallel with it were about 10mm from the ic. The supply line for the oscillator was decoupled by 0.047u cap and an electrolytic. The first cap is about 4mm from ic but the electrolytic is about 40mm... Also the grounding scheme was poor as already mentioned on above link and the 1M ohm resistor which Marantz had put in there was pointless as there already is one inside the IC.

I changed the layout of parts quite a bit. There was a jumper between inverter input and crystal that could be removed and was a very good place to use for crystal. As I removed the 1M ohm resistor (RD02) I used its place for the new series resistor (layout allowed it). The two parallel caps (CD02, CD03) I installed under the pcb and grounded them to one point so that the caps had smallest possible lead lenghts. The ic pin + crystal + caps + resistor layout now had about 10mm total trace lenght!!! I also placed the decoupling electrolytic (CD04) to a place of smaller decoupling cap (CD05) near the IC pin and the smaller cap was installed underside the board on electrolytics pins. Now I also cut the "bad" ground connection of oscillator ground and rerouted it straight to decoupling caps ground point. All other tracks that were too long or had no point anymore were also cut...

Finally I cut the supply line right after the decoupling caps and used a battery to supply the oscillator. For now its the standard Alkaline battery with 330 ohm in series to provide the supply voltage. I don't know how long it will last in there as for now I let it be connected all the time...


Now, I was sceptical about all the jitter and clock improvement until now, but the difference is there and it's quite a surprise how much so. For start when I started to listen I tough that it can't be, that the mind is playing tricks on me as the difference was that big. I listened for many hours and to the same tracks I had played a night before. There were many improvements. The depth of the soundstage improved a lot and the "air factor" around instruments got a lot better, but thats not all. With well recorded classical pieces I could clearly point even the lowest drums. This is something that I had never had experienced before. The Dave Matthews live concert was also great as the nuances and detail of quitars was on another level compared to earlier. The only record that I listened where I felt the differences were not so positive was 'the corrs - forgiven not forgotten' this became kind of cold and overly analytical.....

Anyway. I will continue to listen to the player and see if I feel the same way after a week or a month... I suppose the psychology plays it's role, but as the mod cost me about 3$ (the price of the battery) and I was sceptical about it anyway, there should be some truth to it......


2001-06-12 10:15 am


Jitter is the most fundamental problem with digital replay / recording.

Even expensive players have poor clocks, and the musical improvements from reducing jitter are massive. Changes of just a picosecond are audible.

Try a better clock with even lower jitter (my favourite is LCLock XO2, from and you'll get even greater improvements than you already have.

Costs a bit more than your mod though - well done.

I have tought about the LClock, but as the cost is much higher I decided to try a more affordable approach to see if I hear any improvement. It was more than I could ever expect even with this mod.

I'm going to mail LCaudio later too as I think that when upgrading Marantz CD63, CD67 and their modifications with LClock there would be even more potential if the inverter stages are fed with sepparate power supply (may be even LClocks own supply if it is 5V), instead of the general digital supply inside the player and the grounding of the inverter stage is done right. The inverters will be active even with LClock and the potential problem related to that otherwise stays.....

PS. At the moment I'm listening to an amp based on LCaudio 'The End Millennium' kit and like it a lot. So I know the company and have a lot of faith in them.



2001-06-12 10:15 am
LC Audio etc.


Thanks for that.

I've thought about the effect of the inverter still present and concluded that the problem of it adding jitter is related to the decision level at which it changes states.

The faster the risetime of the clock signal going into the inverter the less the effect of a change in decision level will be. Ideally one wants a clock risetime faster than the lowest jitter level one wishes to acheive. I'm sure changing the supply arrangements will be beneficial, since it could reduce cycle-cycle variations, but with a good fast clock edge I don't believe it's the biggest problem. It's cheap to solve though so worthwhile - I may look at this in my brother's CD63.

The owner of the site you obtained the mod. from would, I know, love to have details of what you've done. I've had lots of email exchanges with him, and he's a serious tweaker! Let him know of your additions, or point him here.

I modded a CD63 KI Sig that my brother still uses and it's a fab player. I reckon a better analogue supply (external 12V supply replacing the 7812 and 7912 reg's on the PCB) would be very worthwhile, it works a treat with other players I've used.

The AD825 op-amps are good as well, LC Audio seem to have a knack at producing reasonably priced high quality designs. The LCLock has risen in price a lot from the original one I bought, but it's still cheaper than most of the competition at a similar performance level. The End millenium looks very interesting - how does it sound and what have you compared it with?

I've recommended LCAudio for a long time, I'm glad to se others have had good experiences also.

Well the battery idea wasn't as good as expected. The battery ran out in less than 24h. So it is either that I will build a charger and use a chargable battery or advance the internal power supply.

For now it's back to internal power....

What comes to changes in the sound, I had too little time to listen yesterday so I will not get into it as I'm not sure...

Cheap Oscillator Upgrade

Using a gate other than the one in the DAC chip could be something to try. It would lower the noise in both the DAC and oscillator. You could build an oscillator on some blank PCB (it shouldn't be too difficult as it wouldn't be very complex and the PCB could provide a ground plane) with a surface mount high speed logic chip, maybe a 74AC04, one gate would be the oscillator another a buffer, the inputs to the others would have to be grounded. Use surface mount capacitors soldered between the ground plane and the chip's pins (you would have to bend the pins up). Unused inputs and the ground pin would be soldered directly to the bare copper. You could also include a voltage regulator on the PCB.
I also mailed the author of the original article, Martin Clark and this was the feedback I received...


Hi Ergo,

I haven't been to the forum for a while so perhaps you can report on my
behalf ;-)
I did almost exactly what you have done, building up the crystal and two
caps directly off the groundplane, and yes I think it is an improvement
of surprising magnitude.

Two things you might like to try, which I use - and seem to work:
1. A separate regulator for the logic gate, using a 7805 fed via a 2.2mH
choke from the raw supply to the main 5V regulator on the board, with
lots of distributed capacitance and- this is important - a 200ohm
resistor on the output side to load the regulator down and get it working
properly; the logic gate only draws 3mA, which is not enough to get a
3pin regulator working properly. The chip cap across the logic gate stays.

2. replace the ceramic 10pF capswith silver-mica types. I think it's

By the way, adding an aftermarket clock - LCAudio or similar - will still
have to drive this logic gate (X1 input) so these mods are worthwhile
_even if_ you use such a new clock source. I've already heard from
another that the difference between the tweak I described and adding a LC
clock is small, although the LC clock does have the edge.

Thanks very much for the feedback - and happy tweaking!

Martin Clark

jittery world

hi folks!

all of these designs spoken here are excellent!
but but but!!!

if you want to believe that we are living in a jittery
world you should try this!

the 2 most delicate parts in the cd for audio is the clock
oscillator for dac & the bit clock coming from the dsp or

if you want more difference!!!
make a clock reshape witha n 74ac00 between the dsp & dac.
in my case i only reshaped the bit clock & the difference was huge!!!!!!!!!!!!!!!!
now i am trying other designs to reshape the dasta/ l/r clock too!

if your cd player have power trnsistors or driver ic,
for pickup fokus,trak,spindle, try to regulate these
voltages.sound improvment you will hear!

if you can, in my design i have 3 power transformers for the
1 transformer feeds the cpu & regulated power servo.
2 nd transformer feeds analog section.
3rd transformer feeds 1 winding to the dac & the other
winding the digital dsp.
everything is dual regulated (dsp/dac/servo)

questions feel free to ask or suggest!!
thank you for reading this!



2001-09-05 1:54 pm
I just added the LC clock ox2 to my CD67 and tested with CD63KI (w/o mod). Overall sound .. the CD63KI sound better !
The CD67+LC clock is good ,not much improve in airly but u get good bass (a bit improve), clear instrument (overclear), depth sound stage !The only thing is lack of vocal .... backward and not as smooth as CD63KI ! If vocal is not important for u than add the LC clock is your first choice ,if not than NOT WORHT it to mod it !
Hi All,
Cmos inverter based clocks produce a lot of jitter. This was documented on the Analog Devices website. They had a picture of a Cmos oscillator with a large cross on it Don't do this! (unfortunately AD has reorganised its website and I could not find the picture anymore).
After much experimenting I designed the KWAK-CLOCK. It is a Collpits oscillator of a FET (J309) and the crystal followed by a very fast comparator AD8561. The Fet oscillator has very low jitter or phase noise. The comparator provides conversion to a square wave and buffering. All critical caps are silvermica.
I found this circuit to perform better than the LCAudio Clock or Guido Tent's can type oscillator. Many thanks to OLA on the AudioAsylum for bringing the use of the FET to my attention.
The KWAK-CLOCK will work with any crystal from 10 to 24 MHz.
<B>The schematic for the KWAK-CLOCK is available on request by sending me a email.</B>
Usually you can use the crystal from your CDP if it has two leads (most players except many Sony models) I use the +5V supply of the CDP while including a ferrite bead in the powersupply line to the clock. I also experimented with different separate supply's for the clock even NiCd batteries but it did not make any difference to the sound. The supply is best taken off the +5V regulator as is the ground.
If your need a higher frequency for example 33 MHz for DVD you will need a overtone circuit, that I also can provide.
My clock can be build for a fraction of the cost of the LCAudio-, Audiocom- or Trichord-clock.
[BTW LCAudio uses a BFR92A, NPN 5GHz transistor and a NC7SZU04 Ultrahighspeed unbuffered Cmos inverter]:)


2001-06-12 10:15 am
Hi Elso

Thanks for the input, Martin Clark, the author of the acoustica site has built one of your clocks and is very impressed indeed.

He's also built a dedicated discrete low noise regulator to deliver the few mA it needs and reports another improvement similar in magnitude to changing the clock. PSU noise will indeed increase phase noise in your design as with any other. The 5V supply in many CD players is extremely noisy, modulated by the relatively high currents and wide bandwidths of the servo amps and servos.

The CD63 BTW has seperate supply inputs on the DAC for the clock gate and various analogue / digital stuff. Fitting your clock, giving it a v.quiet supply and giving the clock gate a clean supply really transforms the CD63 into something special. The analogue supplies are worth seperating too, just a single supply is all that is required since the current drain is only a milliamp or so.

Thanks for sharing your design with us, I look forward to trying it myself soon.



2001-09-30 5:18 pm
KWAK clock

I built the clock module based on Elso's design.
I used a 8-pin socket for the comparator and bought 2 pcs. of LT1016. Strangely the ICs seem selective as for IC1, I need to connect it to PIN1 of the original XO location and for IC2, I need to connect to PIN2 for them to work (read the disc). How so?

Anyway, alot of thanks to Elso for this successful mod!
ergo said:
Well the battery idea wasn't as good as expected. The battery ran out in less than 24h. So it is either that I will build a charger and use a chargable battery or advance the internal power supply.

I don't know much about this tweaking stuff, but I would think a small switch or relay (for automated switching :cool: ) that connects between the battery and the crystal would help conserve it.

24 hrs of listening should take you more than a couple days (I really hope :eek: )

Good luck in your tweek jobs!