Shigaclone modifications - compiled guide
In this guide, I will
attempt to provide a complete and balanced overview of the known
modifications for the popular
"Shigaclone" CD player. The guide is still in
development - wear a hard-hat.
Read this first
For every advice I tried to provide the original source (a link to the relevant post in the Shigaclone thread), if I could still find it. Where I use resources previously published on diyaudio I try to credit the original poster, but I do not always remember who was the author - please help me improve this if you can. Much in this guide is based on my personal experiments, most of which I published on diyaudio.
Names is [square
brackets] are nicknames of the diyaudio forum members. Some of them,
I will refer to with their actual names for convenience:
As this guide is early in its development, and my spare time is limited, please let me know what kind of info you would like to see here first, what to expand etc. If you want to submit some info, I will greatly appreciate that - you will of course get the due credit for your contribution.
JVC RC-EZ31 (pdf, 2.90MB). RC-EZ51 is identical. CD servo schematic is on page 22.
LA9242M (pdf, 153KB)
LC78601E (pdf, 147KB)
CSC1469XH (pdf, 341KB - kindly submitted by Tibi - thanks!)
"Old" Shigaclone pictures
When planning my next modification, I often found it helpful to look at the actual PCB, so here goes:
New Shigaclone resources
All of the datasheets, drawings etc. for the new Shigaclone are available from Tibi's website: http://www.vicol-audio.ro/shiga.php
The only thing about the new Shigaclone that seems to be missing from Tibi's website is the connector layout picture:
There seems to be general agreement that split-bobbin or R-core type transformers are among the best to use, although there were at least two votes in favour of C-cores as well.
Generally, we want a transformer that will reject high-frequency mains-borne noise well. This means it should have as little as possible capacitive coupling between primary and secondary windings, and ideally, it should also have a relatively narrow bandwidth.
For toroidal transformers, inter-winding shield is strongly recommended.
US users are advised to go for a Hammond 166 series, which are inexpensive and sound well. European users can go for the Hammond 266 series (which allows 230V operation), or try to find a similar transformer on eBay or elsewhere - just look for those with split bobbins, not wound one over another. The only European sources of Hammond trafos I found, are these two Italian shops: (link #1, link#2), and this, also Italian, eBay seller. (Please let me know if you know of any other EU sources).
Alternatively, look for an R-core transformer - they are strictly superior to toroidal types, although I am unsure how they stack against the 166/266 Hammonds. The most popular EU source for these transformers seems to be selectronic.fr (in French - if you struggle, open it through Google translate).
For the best results, use as big transformer, as you can reasonably afford (as much as 120VA has been suggested, post #185).
The original Shigaclone recipe by Peter Daniel uses MSR860 diodes. My own experiments with other various (fairly random) diodes failed to find a better diode, although one came pretty close (MBR16100G) - a touch more dynamic, but ultimately less natural than MSR860.
Tibi suggests that the new silicon-carbide diodes from CREE should be better. The CSD-01060A type was recommended, to be exact. There have been no direct comparison reports, so I am including both diodes in the guide for now.
The proven "classic" shigaclone PSU by Peter Daniel uses one 2200 or 1000uF capacitor before the regulator, and one 1000uF cap after. Peter's original choices were Black Gate FK or F before the reg, and Black Gate STD after. My experiments confirm that this is the best combination.
With Black Gates becoming increasingly hard to get, I have done many hours of test with various other capacitors, in hope of finding a good alternative to BGs. They were described here. 2200uF seems to be the best value to use before the regulator, although 1000uF works just as well most of the time. It is not recommended to go much below 1000uF with either cap though as bass may diminish as a result.
I did more tests
since I published the results linked above, but there were no major
changes at the top of the list.
Remember that nearly any electrolytic capacitor can be greatly improved by wood-modding. Please note however, that wood-modding a cheap cap is NOT an alternative to buying a better capacitor - wood-mod improves the timbre and presence, but dynamics and detail level remain largely the same as in the original cap.
The "classic" Shiga PSU uses only one 8V regulator, LM7808 in this simple layout:
I tested a bunch of different versions of this reg and while there are some differences (even among the same brand), they are relatively very minor and not worth wasting the time.
There is an option to use an additional 5V regulator to power the electronics, with the 8V reg left in place to feed the motors. It is a significant upgrade, but there seems to be consensus that in order to take full advantage of it, the 5V regulator should be powered from a separate transformer. Otherwise the improvement may be small, or the mod may even worsen the sound or cause problems with the ground. Here is how you can connect the 5V on the old Shiga board:
This picture shows two alternative connection points for the 5V regulator output (use either 5A or 5B). The trace near R964 has to be cut in order to complete the modification.
Which 5V regulator to use?
Best results are obtained using a "super regulator" of some sort (Belleson, ALSWR, Salas, Teddy, Jung, Burson etc.). Simple monolithic regulators are unlikely to bring big improvements to the sound.
Some people used ALSWR reg and liked the result (for example [BMW850] in post#4210), but Peter Daniel tried ALSWR and Teddy regulators, and was not impressed by them (post#4211, post#4580), which implies his Bobken regulator might sound even better. Salas regulators are inconvenient because they require heat-sinks, but have been reported to sound very good.
The most recommended regulator seems to be the one developed by [Bobken] - the so called "Bobken regulator". Here is the schematic, including the PSU:
It is fairly simple
to build, but sourcing all the recommended parts may be difficult
and somewhat costly.
Please note, because AD811 is operating outside of its specification in this regulator, it is advisable to use the design exactly as posted here, or else numerous problems may quickly arise.
Unnecessary parts (old Shigaclone only, new Shiga owners read below)
In the classic Shigaclone, the following parts are recommended for removal:
Following my experiments, I also recommend removing the following bypasses:
Here is a picture showing the exact cap locations (click for the high-resolution version):
As a result of this mod, all but the last traces of the "digital harshness" disappear, the sound becomes much more pleasant and natural. Treble in particular opens up spectacularly and becomes positively sweet. I consider this a "must have" mod - especially that it's essentially free. My only word of caution is that, in an already treble-heavy system, this mod may tip the balance a bit too far into treble. Still, even if that happens, chances are you will decide to keep the mod and fight the treble issue elsewhere (as I did), because the improvement far outweighs the cost.
New Shigaclone owners: it is very likely that the counterparts of the abovementioned components can be safely removed from the new Shiga as well (they are only bypasses after all, and are not strictly necessary). I have not tested this myself though, so I cannot guarantee the effect. I will post a list of components to remove once I have at least one report of this mod working properly in a new Shiga.
C916 / C43
This is the power supply input capacitor. In both the old and the new Shiga, it is among the most influential parts in the entire transport, although the degree of the influence depends on your power supply arrangements (separate regulator lines will take some of its duties and sonic importance).
The following comments are mainly based on my own, extensive listening sessions, done on a JVC Shiga with a single 8V PSU.
Firstly, the optimal value for this capacitor appears to be 47uF. Lower values generally result in diminished bass response, although in other respects as low as 10uF works well. One can also omit this capacitor entirely - this has the peculiar effect of perfectly natural tonality, but with very high loss of detail. Values higher than 68uF appear to cause bass sluggishness problems, although I have not tested very many such capacitors.
There is only 5V across this capacitor, so voltage rating as low as 6.3V is sufficient.
Regarding capacitor type - after conducting very many comparisons, I could assign all capacitors into these four, roughly defined, groups:
So that's three groups. The fourth group is "Black Gate". I know this is going to be controversial among some DIY-ers, but the Black Gates really do come into their own here. The worst Black Gate I tested (33uF/16V STD) was better than the best non-BlackGate capacitor - albeit marginally.
Group 1. consists mostly of audio-grade, fairly expensive capacitors - you do get what you pay for, to an extent. The caps I list here are the ones I thought had good performance overall. If they had flaws, they were either minimal, or not annoying in character - the sort I could live with every day. So here is the list:
Group 2. The caps that were not all that great, but acceptable.
Group 4. Black Gate.
I tested a large number of Black Gates, so here they are in order of relative performance:
In order to extract even more performance from your caps, I recommend applying a wood-mod.
The digital output level produced by LC78901 is around 2V. Most DACs work fine with 2V, but some have trouble locking onto the signal as the S/PDIF standard specifies 0.5V signal level. This can be remedied using a simple voltage divider (also called L-pad) at the output:
Another issue to consider is LC78601's rated current output, which calls for a minimum of 390R impedance at the output. This is not strictly necessary as LC78901 works fine with no output resistors, although the sound quality improves with the resistors in place.
Taking the above into account, the optimal resistor values are:
R1 (series): 300 R
If you are on a tight budget, Dale is a reasonable choice for both positions.
For the best
possible sound, the 300R position should use a "nude" Vishay (a.k.a.
Z201 Z-foil, a.k.a. Texas Components TX2575), Shinkoh coming a far
second. Nude Vishay restricts dynamics somewhat (makes the sound
less exhilarating, some "soul" is lost), but the increase in
detail is so great that I consider it a fair trade-off overall. If
you hate restricting dynamics (and do not
mind sacrificing detail), consider Riken or Caddock here.
resistors seem equally suited for both positions:
standard resistors have more
or less equivalent detail to Vishays, but superior dynamics;
the Invisistor is simply better
than Vishay, in every aspect.
The above is the optimal approach if you like to do things "by the book". If you don't mind loading the LC78601 a little beyond its specification however, an even better solutions exists:
75-ohm Pi-pad (now obsolete, see below for the best solution)
In order to achieve
perfect 75Ohm input and output impedance, and 12dB attenuation (which is
required to knock down the signal from 2V to the textbook 0.5V), the
resistor values are:
This approach virtually eliminates signal reflections in the attenuator, and (unsurprisingly) yields a significant improvement in the sound; clarity in particular takes a leap. Even made with inferior resistors, the Pi-pad will generally outperform the L-pad (one notable exception being when the L-pad is made with my Invisistors, in which case it actually beats Pi-pad - see post#5888 by [woodturner-fran]).
There is one slight
issue with the Pi-pad, of course: 125.3R and 139.8R are not very
widely available resistor values. In fact, unless you make your own
resistors (or order them hand-made from me), your only option is to
buy a ready-made 75Ohm attenuator for cable/satellite TV, such as
A Pi-pad made with my Standard resistors sounds more natural than the BNC attenuator, although it does cost more. The "ultimate" solution, and what I am using, is a Pi-pad made with my flagship resistors. They are admittedly very costly, but - as Fran pointed out - they are still good value compared to cables...
50-ohm input, 75-ohm output Pi-pad
This has been suggested by marosik in audiostereo.pl forums. The idea is that the wave output impedance of the chip is actually 50 ohm, not 75 ohm, so by using a Pi-pad that has 50-ohm input impedance and 75-ohm impedance, we can beautifully impedance-match the entire transmission line. The resulting sound is deeper and more dynamic than using a 75-to-75ohm Pi-pad.
For a 50-to-75Ohm Pi-pad with 12dB attenuation, the resistor values are:
shunt): 72R (71.749R if you want to be really precise)
The attenuator layout is the same as in the Pi-pad above.
Recommended resistors are the same as with the previous attenuator types - Invisistors are best, Standard resistors second, third come Caddocks for shunt and nude Vishay for series (but keep in mind the dynamic penalty of the Vishays).
C906 / C8
This is probably the most puzzling of all Shigaclone capacitors. The stock JVC player uses a cheap-looking 0.1uF electrolytic cap here, and yet somehow it sounds better than most "audio grade" capacitors - both electrolytic and film. It is thought to be one of the key reasons behind the early Shigaclone "magic".
In my own experiments, I tested a bunch of capacitors, some in the
original value (Black Gate Nx, Multicap RTX), some in the Okapi
recommended 10nF (Multicap RTX, Vishay MKP1837, Duelund Alexander).
The 10nF value sounded consistently better to me.
My recommendation: 10nF (0.01uF)
Clock / crystal
I hesitate giving any explicit advice on this subject, as I have not done a lot of experimenting myself here, and accounts from others vary wildly. For example Trichord 4 clock: one person praises it for bringing detail and swing, but elsewhere two other users comment how they disliked its artificial and sterile sound. So until I conduct my own exhaustive experiments my only advice is to carefully audition your clocks before committing to one or another.
Mounting the laser assembly / mechanic
For a simplest
approach, you can mount your mechanism any way you like, or just
leave it lying on the floor, and it will work fine. But mounting has
a strong, and not yet fully understood influence on the sound.
Speaking of which, mass-loading. I cannot overstate how important this is. If in doubt, always go for more mass. If you can't find a thick enough slab of copper/etc., buy multiples of whatever thickness you can get and stack it (remember to attach in such a manner that it does not ring).
Mounting with two bolts has been found to sound better than with three or four, for some inexplicable reason. The original Shigaraki had its mechanism mounted with two bolts too. Soft mounting (for example with rubber gaskets) results in muddy sound / loss of detail. [Tony X] suggests that the mechanism is very sensitive to bolt alignment, and proposes using carefully turned spacers to achieve perfectly level alignment: post#6543 - which I think is a very good idea.
Ok, so you have a mechanism mounted to a tank armor-plate already, what else can be done? Mounting spikes to your "mechanism slab", and placing it on another, preferably even more massive object is a sure-fire improvement. Again the supporting slab should preferably be metal, but things like marble or hardwood work well enough here. If you are not going for another metal slab, I suggest adding at least metal "pads" for contact points where spikes meet the wood/marble/etc. of the support slab. This will aid in vibration transfer.
I did an experiment where I first placed my mechanism slab (with three spikes) on an oak board, and listened to the sound. I then placed a large 3mm thick copper plate on top of the oak board, so that the mechanism slab was now resting on copper, not wood. The sound was noticeably better with copper. I then took out the large copper plate, and instead placed three smaller copper plates (also 3mm thick), one under each spike. The sound was noticeably more precise when each spike had its own small plate for support - the difference was not massive, but it was certainly there, and was noticeable.
Puck is the round
element that holds your CD in place while it is playing. The boombox
puck is a hat-shaped piece of plastic, with a ring magnet attached
to it. The standard mechanism has a steel CD rest platform
("turntable"), so the magnet clamps to it. It has been found that
shape, mass as well as the material from which a puck is made, all
influence the sound. The current consensus appears to be that:
For those who do not have the boombox puck (new Shiga):
- The easiest option
is probably getting an entire CD laser and motors assembly with the laptop-type clamp, which
requires no puck:
- An alternative is to search for an old/broken JVC or Sanyo boombox - there is a small chance it uses the same CD mechanism, and has a compatible puck (it's free so why not try).
For those who look to improve the sound or visuals of their Shiga:
- [BMW850] got
himself a custom-made aluminium puck:
Puck-less (a.k.a. spider-mod): the idea being that with some non-slip coating on the CD
platform, there will be enough friction to allow spinning of the disc
even with no puck at all. I came up with a spider silk coating,
which I found (much to my surprise) substantially improved the
sound. Read the full story here:
Update: I have been using the spider-coating for a few months now, and it is still working well. The only disadvantage of this solution is that some care is required when lifting the CD off the platform - but spider silk is actually extremely strong, and that is a minor inconvenience anyway compared to the gains in the sound. Considering that this modification is easily reversible and essentially free, I heartily recommend trying it. Yes, it's wacky, it will attract incredulous looks from people, and it's probably best never to mention it to your wife - but the sound is so organic that you will never look back once you try. So read up the link above and go get some spider webs!
Update 2: Nearly a year after I applied, the web is starting to give up. It still sticks sufficiently for the player to work, but is now dry enough that the CD will occasionally break traction and slide on its platform.
The old Shigaclone had the remote from the boombox, but the new one does not come with any. Tibi suggests getting a universal remote, which should work fine with the new Shiga (most remotes will require set-up though).
Another option is getting a JVC RM-SRCBX330 remote control (run an eBay search), which is pretty much exactly what JVC boomboxes use. It should work with the new Shiga as well.
All of the information contained in this website is provided with the best of intentions. It is however provided "as is", and may contain inaccuracies, mistakes, outdated information, unfounded claims and personal opinions - use your own best judgement, and above all -use common sense. No claims will be accepted for any damage or loss, direct or implied, that arose due to the use of advice provided herein.
Copyright © 2009-13 Jaroslaw Biniek.