Quality CD-Mechanisms are long gone - let us build one ourselves!

Hi Mark,
Forget about the eye-pattern.
Ahhh, no. The eye pattern quality is an early indicator of problems. This is where errors are avoided, or embraced. Have you ever looked at eye patterns and correlated those with your magical machine's results? If you haven't done this, you really ought to. I have been able to correlate eye pattern quality with lower error flags. I really wish you wouldn't comment like that without solid experience. That experience would be comparing the quality of the eye pattern with the numbers coming off your test equipment. Note I have not once suggested that your instruments are wrong or give improper results. What I have said was that eye pattern quality will affect the error rate. The eye pattern not only gives clues as to what might be wrong with a CD player, in the communications industry today, the quality of the eye pattern is a key test for the health of the system. It is included with the constellation patterns and graphs of BER rates. The quality of an eye pattern is a basic indicator used from CDs and DVDs, right on up to TDM, Optical and satellite systems.

Please, give other tests their due. The eye pattern is a basic indicator of data quality before it is digitized into the data stream you know and love after the DSP runs over it.

-Chris
 
If a better eye-pattern results in lower error rates then I can only conclude that the Plextor drives must have better eye-patterns than the Philips.

I have tested many discs and you can not see the errors by looking at the eye-pattern. For me it shows pit geometry.

I have asked before. What is a good eye-pattern? What does an eye-pattern look like with C2 errors?
 
Last edited:
Mark,
And I can't tell from the data what the problems might be.

If you were to read that report (good going! I guess I'll buy TDK from now on), they did specifically test the eye patterns - didn't they?

An eye pattern with high C2 errors would most probably be showing any number of defects. Poor symmetry, low level, excessive noise, closed eye or excessive jitter can all cause unrecoverable read errors. If you look at the envelope of the eye pattern, you can see if it is modulated by a servo or disc defect.

If you would take the time to learn, you would immediately see how well looking at the eye pattern can show you problems. The most highly trained technical people know this and use eye patterns to great effect. Keysight and other test equipment companies spend a fortune making certain that eye patterns are displayed as accurately as possible. The eye pattern is in the default suite of screens a signal analyzer will show you.

It's not rocket science, bad eye pattern = digital errors. The errors vary as the eye pattern quality. Eye pattern degrades, C1 and C2 errors begin to show up. Take the time and educate yourself. You will end up with another tool to assist you in your job. I have installed scopes in CD mastering suites to check on the quality of the CDs they burn on a one-off basis. It is a valuable QC step that I would think you would use. It can give you the answer as to why your errors increased. Wouldn't you want those answers quickly? Before a bunch of crappy lots left the factory that is ...

-Chris
 
My guess is that they did not look at the eye-patterns. They measured the HF parameters. Symmetry, I3R, I11R and jitter, all telling you more than just looking at an eye-pattern alone.

Do you agree that the Plextor DVD drive must have the best eye-pattern? It does have the lowest error rates. The other option is that the Plextor eye-pattern is not so good but still contains less errors?


Your advise is becoming condescending. I would not tell you to measure error rates so that you could better repair players and that the players would actually play at low error rates before returning them to your costumers. I have probably looked at more eye-patterns than you have. I asked again in the hope of still learning more. You are not able to give me any specific information that can be used in analysing an eye-pattern.
 
My guess is that they did not look at the eye-patterns.

Of course they don't. Eye pattern requires MANUAL interpretation. It shows a lot of things, but require knowledgeable interpreter. Interpretation of correlation between "figures" and perceived sound is almost a mystery in audio field.

Symmetry, I3R, I11R and jitter, all telling you more than just looking at an eye-pattern alone.

No. See above reason. Besides, they represents the disc quality, not the machines.

The Clover has the ability to measure the HF parameters as above, but they used the result from the CD drive (not from the Clover). This, and their eagerness to show good result, tells me that the HF parameters measured by Clover was worse. If this is true, the eye pattern of the Clover could be relatively worse. But I believe that both would show equally perfect eye pattern, because before conducting the test the machines must have been calibrated or adjusted (e.g. the laser intensity).

Do you agree that the Plextor DVD drive must have the best eye-pattern? It does have the lowest error rates. The other option is that the Plextor eye-pattern is not so good but still contains less errors?

Problem is, I think what is measured here is only a very small portion of what a transport do to music (unless you send the EFM to the external DSP/DAC).

Eye pattern (before DSP) is only the early determinant of healthy signal. If it is bad, most probably the end result is bad too (GIGO principle). But I cannot see why it cannot be made perfect or calibrated in both CD driver and DVDROM driver, especially BEFORE conducting important test like this.

Eye pattern after the DSP (the squarewaves) is another measure which is more likely the bottleneck in audio processing. In DVDROM, I don't know if they do interpolation technique when there is un-recovered errors, because this seems logical only to music signal. The chips found in DVDROM are very different than the ones in CD players.
 
Hi Jay, see you live in Jakarta. I lived there for a year and a half some twenty years ago. Worked for a CD Plant in Tangerang. The people are great, love the food, still miss driving in Jakarta, so much fun.


It would have been great if they had published the results of all 50 discs or had tested other brands as well. On the other hand I would not expect big variations in the results when testing a spindle of recordable discs.
 
Hi Mark,
My guess is that they did not look at the eye-patterns.
Why would you say that? It seems to me that if you are measuring the eye pattern, a look at it would provide a sanity check.
Your advise is becoming condescending.
Sure, maybe. You continue to dismiss out of hand a test that is a communications industry standard. To listen to you, I have a problem believing that you have ever actually looked at an eye pattern to analyze anything. I also have some doubts of how many eye patterns you have looked at. In the service world, I had a very good reputation for repairing CD players and they represented a large part of my work load. You would be better served by not making so many assumptions. Certainly, you have disagreed with what is standard industry practices for service and quality control. That's your business, but passing yourself off as an expert in this field while being at odds with normal practices - proved to be the best practices in the field, it becomes a little hard to take.
It would have been great if they had published the results of all 50 discs or had tested other brands as well.
I agree, except that there is little point in providing the similar data for all 50 discs, as you have said yourself. But the results for the other brands might well have an effect on sales figures for the bottom group. Clearly those groups would rather kill publication of their standings.

Anyway, lighten up on knocking the tests the rest of the world does. I already know from first hand experience through direct observation of eye pattern quality and the C1, C2 error flags (which the test measured, the others are derived from that data) that low error rates do not occur when you have a poor eye pattern. As it turns out, I measured the same things that these tests were at the 1X rate. While looking at error flags is not the norm for the service industry, I am positive that many others have and found the same things to be true. Written service procedures reference the quality of eye patterns as the goal in servicing. You're alone signing against the choir, and it seems you like it that way.

-Chris
 
We are not discussing servicing a player, but a new design. I have asked what makes a good eye-pattern and you are unable to answer. I am starting to believe that you cant explain what you are looking at.

I posted the link to the test as it clearly demonstrates that a Plextor drive outperforms the Philips as there were some questions posted about the quality of CD/DVD drives. The analysers used in the tests are standard for disc replicators and they do not show the eye-pattern. The company I worked for build and sold an analyser that did have a scoop. Here and at CD-plants I tested thousands of discs and trained operators on how to use this analyser.
 
Hi Jay, see you live in Jakarta. I lived there for a year and a half some twenty years ago. Worked for a CD Plant in Tangerang. The people are great, love the food, still miss driving in Jakarta, so much fun.

Hi Mark, nice to know that. Tangerang was an urban city 20 years ago 🙂

We are not discussing servicing a player, but a new design

Chris is aware, that discussing the mechanism and servo is only the preliminary step of discussing a good transport. I'm agree with you that at this step, the Plextor DVDROM could be the better mechanism than the Philips. But as a whole, including reliability and DSP, I only rely on the fact that DVDROM is not intended for "high-end".

I have tried finding the best DVDROM. My main concern is that I always found switching power supply to be fatiguing (and my result was with switching supply only)
 
Hi Mark,
I am fully aware we are talking about a new design. One aspect that was introduced was that the transport should create the best eye pattern reasonably possible - you disagreed. This is how we reached this point in the discussion.

I answered your question with regard to what type of eye pattern would be observed that would create high error rates. I answered you fully and correctly given the information you were giving me. The plain fact is, many problems or defects can contribute to high error rates. In fact I was being factual when answering you and you did get the proper answer.

I am more than tired of the dance you're doing. Computer CDROM and DVD drives are, in general , not suitable for a high end application. You disagree, but that's okay since you really don't need to spend time here. Just go buy the drive of your choice, plug it into the common interface (I assume SATA) and be done with it. I don't want to go that route and neither do most the people spending time here. So if that's your goal, to push a commodity computer drive, I'm not only not interested, but I also know from first hand knowledge that it is the wrong route.

-Chris
 
Some random thoughts about CDROM drives:

- The eye pattern does not have to be perfect. It is ok if it allows reliable bit detection. As so often the rule of "diminishing returns" apply here too.

- Many modern communication system do not only look at one "eye". They calculate the correlation over sequences of bits. This allows them to even detect bits burried completlely in noise.

- It is clear that a CD-ROM drive working at 40x speed does not have a good looking eye pattern, but this is ok as long as bit detection is reliable.

- The CDROM standard builds on the CD-DA standard. A CDROM player therefore contains the complete firmware code for a CD-DA.

- The CD-DA (CD-DigitalAudio) is organized in sectors.
Each sector has 98 frames.
Each frame has 6 samples of 16 bit stereo samples = 24 bytes
and additionaly 8 byte of Reed-Solomon error code and 1 byte subcode = 33 bytes.
33 bytes * 98 subframes = 3234 bytes of raw sector size.
27 % of the raw data is used for error correction and 73% is used for
audio samples.

- The CDROM standard interpretates the 2352 audio bytes as data consisting of:
16 bytes header
2048 bytes digital data
288 bytes additional Reed-Solomon error correction
CDROM standard uses 37% of the raw data for errror correction and
would actually be a better standard for audio too...

- Have you looked at a modern CDROM drive? To me it looks like a
masterpiece in precision engineering.
Granted, is is flimsy and mostly plastic, but it reads absolutely reliable Gigabytes of data even after 5 years in my laptop.

- No consumer product manufacturer would survive if his drive would not
be reliable at least for 4-5 years given the high volume and the small profit margins.

- Vibrations seems no big deal - looking at my laptop CDROM drive. The
control loop for the servo is fast enough to deal with this.

- Building a CD Drive is a major taks. It needs knowledge from different fields:
1. Mechanical design
2. Optical design
3. Analog electronics (low level, high speed)
4. Digital design (FPGA, DSP)
5. Error correction code understanding

- No way that an individual without at least 3 millions of $ could do this.

- But if you nevertheless try to build one:
+ A lot of decisions need to be made now to cut the complexity down and focus on the system
+ A mecahnical copy of one of the first primitive philips CDM1 drives may be doable if you use an available optical pickup and available CD decoder chips.... But which one is available in 2020?

- A more realistic goal would be to build a good and useable audio CD player on top of a computer CDROM drive. Use the SATA interface to be compatible. If the drive fails or turns out to be unreliable.... change it for 30 bucks.

- A better path may be to rip the CDROMs to a harddrive and build a good USB DAC. But beware: Harddrives are not completely error free too.

Regards,
Udo
 
Some random thoughts about CDROM drives:

- The CDROM standard builds on the CD-DA standard. A CDROM player therefore contains the complete firmware code for a CD-DA.

Standard and complete, yes. But how about the quality (chip specification)?

I opened up a Samsung DVD Writer. The mechanism is good. But the chipset is so cheap. One motor driver, one servo/signal processor from Mediatek, MT1868B. I couldn't find the pdf on the net but I know it is a cheap one from Mediatek.

Compare to a cheap and old typical VCD player that uses these chipset:

Mechanism: Samsung SOH-AAN
Motor driver: Samsung KA9258D
Servo signal processor: Samsung S1L9223 ("advanced" error correction, interpolator)
Digital Signal Processor: Samsung S5L9284 (With Digital output)

Video Processor: ESS ES3210
Video Driver: ESS ES3207 (I2S, 16-bit delta-sigma, VOut DAC)
External RAM

I like the springy mechanism and how it can read well even on a non-flat surface.
Oh, and old VCDs use real EI transformer, not switching power supply with its fatiguing sound (but many people cannot hear it)
 
Last edited:
The quality is good in the sense that the bits stored on a reasonable clean CD come out on the SATA interface without errors.
This can be checked by programs like cdparanoia or Nero.
The handling is not so good as the mechanic is cheap.

So you need a microcontroller with SATA interface and a few MByte of RAM.
The microcontroller reads out the digital data from the CD and writes the samples into the RAM.

The audio part needs a state of the art digital-to-analog converter.
The jitter problem is completely solved by reading out the digital samples
with a jitter free crystal clock from the RAM.

Add a good user interface with nice buttons and a readable VFD display.

Put this in a good looking housing and you have a premium CD player.

This would still be a lot of work to do! If you do not want to go the SATA
way you can use the TEAC transport mentioned in a post here (they cost about 75 €).
They have the advantage that "PRO AUDIO" or similar is printed on the plastic and that the interface is not SATA but SPI or I2S.
The microcontroller interface is therefore much simpler - you can even use a 8 bit one 🙂

Transformers have the disadvantage that they have more idle current than the EU standby directive allows.
They need a lot of filtering to suppress the 100 Hz Ripple too and have
a strong magnetic field (especially the EI).

Switch mode power supply have near zero idle current, a wide range input (90-240V) , and you can not hear the 100 - 200 kHz Ripple.
The last point is not unimportant for audio.

If done right, a switchmode supply can have some advantages over an EI transformer.

Standard and complete, yes. But how about the quality (chip specification)?

I opened up a Samsung DVD Writer. The mechanism is good. But the chipset is so cheap. One motor driver, one servo/signal processor from Mediatek, MT1868B. I couldn't find the pdf on the net but I know it is a cheap one from Mediatek.

Compare to a cheap and old typical VCD player that uses these chipset:

Mechanism: Samsung SOH-AAN
Motor driver: Samsung KA9258D
Servo signal processor: Samsung S1L9223 ("advanced" error correction, interpolator)
Digital Signal Processor: Samsung S5L9284 (With Digital output)

Video Processor: ESS ES3210
Video Driver: ESS ES3207 (I2S, 16-bit delta-sigma, VOut DAC)
External RAM

I like the springy mechanism and how it can read well even on a non-flat surface.
Oh, and old VCDs use real EI transformer, not switching power supply with its fatiguing sound (but many people cannot hear it)
 
Last edited:
Transformers have the disadvantage that they have more idle current than the EU standby directive allows.
They need a lot of filtering to suppress the 100 Hz Ripple too and have
a strong magnetic field (especially the EI).

What is the symptoms, or the perceived sound character of this strong magnetic field which is inaudible in switching power supply?
 
Hi udok,
The eye pattern does not have to be perfect. It is ok if it allows reliable bit detection. As so often the rule of "diminishing returns" apply here too.
If you are "almost there" for eye pattern quality, it is not a case of diminishing returns to make it perfect. What we are talking about is mechanical alignment and there is really no excuse to get this wrong. This one factor really affects everything that follows. It makes sense to expend a little labour to get it all the way there.
Many modern communication system do not only look at one "eye". They calculate the correlation over sequences of bits. This allows them to even detect bits burried completlely in noise.
But my point is, they do examine the eye pattern. They do this for each channel, and also examine the constellation too. The comment about noise is a critical one. The S / N ratio is something carefully maximized. It is the difference between the desired signal and system noise that is very important. Noise will directly result in decoding errors.
It is clear that a CD-ROM drive working at 40x speed does not have a good looking eye pattern, but this is ok as long as bit detection is reliable.
But this isn't true because ...
27 % of the raw data is used for error correction and 73% is used for
audio samples.
Instead of ...
CDROM standard uses 37% of the raw data for errror correction and
would actually be a better standard for audio too...
Yes, but we are using the first method, and we are stuck with it for the short term at the very least. So better standard or not, we can't use it.
Have you looked at a modern CDROM drive? To me it looks like a
masterpiece in precision engineering.
Granted, is is flimsy and mostly plastic, but it reads absolutely reliable Gigabytes of data even after 5 years in my laptop.
Yes, as a matter of fact I have. Some consumer machines (Creek) use a computer type transport. They are a masterpiece of engineering. They have been carefully designed to be as inexpensive as possible to manufacture while still working. I also know some people who use their CDROM drives for music. They go through them quickly compared to most consumer CD players.
- Building a CD Drive is a major taks. It needs knowledge from different fields:
1. Mechanical design
2. Optical design
3. Analog electronics (low level, high speed)
4. Digital design (FPGA, DSP)
5. Error correction code understanding

- No way that an individual without at least 3 millions of $ could do this.
Okay, you've opted out as well. Just buy a controller on Ebay and a case and you are all set. No need for you to drag everyone else down that miserable road too.

If there are some folks out there who would prefer to go the CDROM or DVD route, please just go do that. Discussions along this line aren't helpful or applicable to this thread. Read the posts from the beginning. The first post defines the intention and direction desired. I would suggest that you begin your own thread with those members who feel as you do. You can leave us alone and still have your say.

-Chris
 
Hi Chris,

I have nothing against a perfect eye pattern. It is important. And i think that each "cheap" CDROM drive gets it right. Otherwise they could not put out data at 40x speed without serious errors.
But at that speed you can not expect a perfect picture on your Osci, and you need careful measurement to not disturb the picture (active probes).

And the additional 10% of error correction in CDROM over CD-DA will not safe your day if you do not get it right at the lower layers.

I don't want to buy anything at Ebay and i have not opted out. I want to discuss technical details about CD players. I am sure that you can bring something more constructive too into this discussion.

I have pointed out 5 areas where a lot of knowledge is necessary and that these areas could be possible workpackages.
Up to now i have not read a single post how to precisely solve one of these parts besides using a CDROM drive.
As in any technical system there are compromises taken into account.
I think that using a tested CDROM drive in conjunction with a quality DAC and a proven classic user interface is a good compromize, especially as audio quality will not be impaired.

If Salas still want to build a CD player he should present a concept how to proceed...

Regards,
Udo
 
Hi udo,
Fair enough. I should ask though, have you read the early posts? We did go over a fair amount of detail and there were some preferred mechanism types that had been lightly settled on. I do think that Salar does have a workable plan in mind as well.
- Building a CD Drive is a major taks. It needs knowledge from different fields:
1. Mechanical design
2. Optical design
3. Analog electronics (low level, high speed)
4. Digital design (FPGA, DSP)
5. Error correction code understanding
1. This is Salar's strong point
2. Buying some kind of head is the most easy way to deal with this.
3. This is relatively straight forward. There are enough examples of this out in the wild that it shouldn't be a problem to come up with exceptional performance.
4. There are folks here that can design something using say, the PIC micro-controllers. Control and features can be designed on this web site.
5. Not required. You use a chip set and the manufacturer has app notes to describe how the chips are connected. The error correction takes place in that chip set.

The members here are an amazing bunch of people. The SymAsym was an amplifier designed in an open thread and was very successful. Other designs were offshoots of this good base design. A very nice guy and competent designer gave us the DiyAudio amplifier, the Honey Badger.

I don't think I would be far off by suggesting that different people would gladly step up and help us through any tough spots. That and the performance will be well above average market models. Keep in mind that the goal was a very high quality CD player built to last a long, long time. This is an achievable goal. What we really need to do is to pull in one direction as a group. I can only imagine how disheartening these past few pages may have been to Salar.

Best, Chris