DIYHiFi.org

Hello all,
It’s no secret, we’ve been bragging about this for quite some time. With some effort, inspiration, more cash, your ideas (and luckily a few of our own) the two of us, my friend ozone_stink and me, we’ll be the happy owners of some nice homebrewed CD-Players. After a good start our project felt between the cracks for a while, now it looks alive again ….kinda.

We started this thread in Crafting because our questions will range from chassis design and knitting toroids to analog and digital topics. You are most welcome to post your contributions/experiences about this subject. In return we intend to document our progress to the best of our abilities so it can serve as reference to anyone…well, courageous enough to try it. (our friends called it otherwise )

We started to buy the main components, here’s a list that will also offer you an idea on the project:
1. CD-Pro2 kit from Dantimax – both tested and working fine. A few caveats concerning the control software, more details later in the thread.
2. Display : we preferred a large VFD model with blue filter over the classic LCD or PLED type. It is a 2x16 character of 8mm height: CU16029ECPB-W1J from Noritake
3. PMD100 HDCD digital filter: unfortunately discontinued, together with the manufacturer if we may say so. You may remember this thread: PMD100 for sale on eBay. To make full use of it we will use it in software mode with gain set in the analog domain.
4. As DAC we will use the AD1862JN (nothing fancy here just one per channel) which is also a discontinued part. Tried, tested and true - our thanks to Jocko.
5. TentClock at 33.8688MHz and some shunt regulators also from Tentlabs
6. Discrete IV using matched pairs from AD: MAT02, MAT03. Expensive stuff, but as far as we can tell the results look very promising. On the prototype the DC stability is excellent; I don’t think we need a servo for this stage. On the contrary we need to include in our design some sort of filter + output stage with switchable 0dB to +6dB gain. (or +1dB to +7dB, it depends) The analog outputs will be single ended.
7. There will be a S/PDIF output, most likely we will use Jocko’s board, to have at least something professionally designed.
8. Power supplies: we start from scratch with 4 primaries that we bought from Toroid. It is the first one in the list, the ‘small’ 60W model, which will allow us to define whatever secondary windings we may need in order to keep the power dissipation low. These may end up in separate enclosures; including the rectifiers and filters…we’ll see.
9. Chassis: our little nightmare…when it comes to mechanics I’m two left handed (you don’t want to see that drawing in my avatar at full scale). Our hope is to find another one like in the attached photo (from a kit sold a couple of years ago) Or to end up with at least one of these when the kits will become available.

OK then, let's get started with what we did so far.

Btw my ‘partner in crime’ is on duty call in Geneva. (or eating swiss chocolate on his employer’s money ) I’m sure if he’s seeing this he will login to say hello.

Here's the uC board which will control the digital filter in software mode: pcb is double sided and kitchen made (it shows); has a single button used to cycle between the dither modes including turning the dither completely off. Also the same button is used to put the uC to 'sleep' and to shut-off the display. The little blue digit is only 7mm high. It works very nice as you'll see later and it helped us to re-discover an unknown (obviously to us) feature of dither.

It wouldn't be complete without some convenient 'audiophile' features: the display is not multiplexed, it memorizes the last setting even after cycling the power off-on, when pressing the button for 3 sec. it shuts off the PIC (including the XO) and the display. It indicates HDCD presence with the decimal point.

We only use dither modes 1 to 6, the ones which are suitable for multi-bit DACs. We skipped modes 0 and 7 but a '0' is displayed when dither is turned off.

The schematic is fairly simple

For testing the dither modes I had to adapt an older DAC. Now it contains: YM3623B, PMD100, 2x PCM1702K, discrete I/V for the left channel (upper trace) and the original OPA627B I/V on the right channel (lower trace).

For all tests I used Track 4 from CBS Records test CD (all zeros). First screenshot was posted recently in another thread. No dither and the osc. bandwidth is not limited (100MHz).

Dither mode 1 is barely visible so we go directly to mode 2. Please note that both vertical and horizontal axis have changed and the osc. bandwith is limited to 20MHz.

Mode 3, said to give the best results (depends on actual implementation). Some pattern is starting to emerge, patience until the next post.

And mode 6, having the highest amplitude. Here is the bummer: dither is some kind of random noise but it is in antiphase between the 2 channels.

Anyone who knows the explanation wants to shed some light? I can only think of mitigating the influence on the power supply if it is common for both channels.

Thanks!

Errrr, two different output stages for left and right. One inverts and one does not ???

As for your dither measurements. Been there

I think the 16f84 must be an audio grade component, as everybody seems to use it for their designs Why no RC5 to control dither?

<edit: removed nonsense, really time for bed>

Can't remember if i did measure both channels. But if i did, i don't remember this inversion. And i'm too lazy now to get my player and open it again. Time for bed.

Hello Guido,
Yes the 16F84 is just about everywhere; but even so we were running out of pins. If you look closely at the schematic we did a neat trick: since we only need to display numbers from 0 to 6 we hook together segments ‘a’ and ‘d’ (they differ only when displaying 7 ).

Even though different, both I/V stages are in phase, the impulse response is attached. Btw good idea - I'll check the 'invert' setting on my oscilloscope, altough with both input signals being in phase on the test CD....

About implementing the RC5 protocol, I don't think it is practical since you only need to change dither mode once in a while, like when you need to reassess the setting after major changes in your system. Also we did not intend to use the other functions available in PMD100 like volume control...etc. (need bigger uC, display, extra buttons = )

It just hit me: they simply inverted the other channel because it's cheaper (they don't use extra silicon to implement two pseudorandom sequence generators) Still I think it is an elegant way to do it - in case the power supply theory holds.

So it is inverted, no idea's at this hour as to why..

RC5 is a good way to minimise buttons. There are plenty on a remote.
As for running out of pins: use i2c with some cheap expanders and you have plenty of pins/eeprom/etc.

I'm using a 16f84 to control the pmd100 by rc5, with a vfd to display the setting.

And to minimise more parts: use BCKI of the pmd as osc1 Xtal & stuff can go.

Time for zzzzz

Hi, If that is the impulse response of the PMD100 I will stay far away from it.

As I can buy a used Philips CD930/940 for just 50 Euros I feel no need to build a player.
I guess you have much spare time or do it "for fun".

Elso, as a soothing remedy I will post (haven't run out of screenshots yet ) a 20kHz full scale sinus; less likely to occur in real life so I see it as a worse case. Not bad, with a little filtering we get back the original.

How I see it, the impulse response is a single sample that may correspond to any frequency above the Nyquist frequency so I'd expect it to look bad, as the filter is not designed to 'reconstruct' those freq. Intuitively it is similar (to a certain degree) to the case when we apply say a sharp pulse to a 20kHz low pass analog filter and look at the residual.



We have little time, but YES!

Not true.
When using Bessel or passive low-pass filtering Rs and Cs no ringing occurs.
The ringing is as bad as a higher order Butterworth in ur case PLUS pre-ringing.

Indeed it's true when freq. of the input signal is lower than Fc.

OTOH, I checked the set-up last night and everything looked OK, at that time... Unless something went terribly wrong the dither thing is in oposite phase between left and right channels.