Yeah, the braid only removes the solder to about halfway, it sucks
The outputs of the DAC are symmetric: there's a positive and a negative output that work together. If one of them is not connected, you only get 'half' the signal and the output of that channel will be lower. This would explain the balance problem. You can check if both outputs are connected by measuring the voltage at the inputs of the DOS (LO/RO and LON/RON): all should measure about 2.5V DC.
Ray
Thanks Ray, the dos board is out of the player, I have to find some time to put it back in and do some more testing.
One question, is the 2.5v measurement in play mode or at anytime?
Pete
Hi Ray, thank you so much. That makes a lot of sense to me and I've got a feeling that your insight is absolutely bang on with this as always. I really appreciate you providing an emergency support hotline for me in the middle of the night*
I will find the trimmer and give it a tweak.
*(It's the age old story - woke up due to nightmare about purple octopus, got up to get glass of water, stayed up to obsess about CD player).
That's the advantage of being at the other side of the ocean, it was broad daylight over here
If that octopus is bothering you already for ages, maybe you should see someone?
Ray
Hi Pete,
I just stopped in to the thread for a rare visit and have read about your troubles. Are you using my board, or an older board? I tried to make my holes plenty big in order to accomodate any desoldering issues, if they happen.
The voltage out of the dac chips will be a constant +2.5v on all the outs, as long as the player is powered. It doesn't have to be in play mode.
Hi Steve, I'm using the older board from Ray, the one from you was a bit too large to fit in the player easily. In fact the holes on Ray's board are a bit bigger than those on yours. The blame lies with my desoldering technique though unfortunately
Regards
Pete
I should visit more often! Too many other things going on I guess.
For sure, the CFP is not a tweak, but a basic design change that is an immense improvement. My current sink is fairly primitive, so your new board and redesigned current sink should add even more.
The CD clock's oscillator is a single-ended design which is more sensitive to supply-noise. The circuit already has L-C decoupling of the supply lines. Separating the supply for the oscillator and the comparator will bring some gain. Improving the PSRR further basically means choosing an other circuit. The supply itself is not the lowest-noise, so that could be improved as well. But then it will stop being a simple circuit
Thermal noise can be reduced by submerging the clock in liquid nitrogen
Good mains cables can make a world of difference. I did an A-B test once with the standard IEC cord and some decent shielded mains cords from Supra and some other brand. The difference was clearly audible (in favour of the shielded ones). Same goes for the wiring in the wall, but that's more difficult to replace. Unfortunately the pipe in my wall doesn't allow a shielded 1.5mm2 cable. Maybe i'll drill a hole through the floor some day...
Ray
Thanks Ray,
So using split +/0v/- supply rails with a discrete voltage regulator ,of the type you have on your own website to replace the 12v regulators in the digital supplies of the marantz 63 etc, I have built these and they are good, so thanks for that too !, to improve PSRR (perhaps a 78/79Lxx ?? maybe as it's almost complete ?) which I could build or will build as it would be better with it's own dedicated supply...agreed?
I like the sound of liquid nitrogen, but I think I have just run out...damn
I have some (I think) 1% Welwyn resistors and maybe even some SMD types hanging around so that's a possibility too eh ? I was wondering what type of cap's would be best too. I am an electrician by trade so I am learning electronics as I go along with the help of yourselves. I was thinking of WIMA MKS or MKS P types, what about tantalum or are they intended for a different purpose ? Any ideas/ thoughts would be most appreciared here Ray !!!!!!!! As I have said I am learning electronics as I go, i.e. which types of cap and where to use them etc.I have two other clock designs at home for the CD63 which I hope to post on here for your perusal if you don't mind. The first utilises the same series 16.344MHz crystal feeding in series the input of a CMOS device (4007 I believe so there may be newer alternatives). It runs from a single ended 12v supply. Between the crystal and ground is a 10 to 100pF polystyrene cap and a 20 to 200 varicap between the output and ground. Above on the 12V rail is a 1Mohm resistor (all this is off the top of my head and may not be explained very well which is why I am going to post the schematic here for you !!!). The actual output is then taken from above the varicap at the CMOS output. Apparently using these two capacitors and the resistor guarantee a 0.1% or better output, not using any of the cap's or the resistor puts it up into the 1% category !! Sounds good and it is from a renowned British designer, I will post as soon as I can. I have not seen such a simple design and if the spec's are anything to go by then it sounds great no ? The other design is one of my own which is a bit more complex so I would be gratefull if you could have a look at that and give me your opinion and criticisms. These would be much appreciated.
In respect of the mains cable, then being a fully qualified sparky, I have utilised the fourth core of some spare SY cable I had knocking around and at the CD player end I connected the shield to this fourth core so the screen is in parallel with the earth and connected to earth at the mains plug end. I found this to be the best I have tried so far and now all the mains cables I use to fire up my kit is like this. Sometimes I have crimped the SY shield to the fourth core as the soldering can get a bit messy and the crimping helps stop the copper becoming so fragile too. 1.5mm2 cable can curry about 10 amps and it's voltage drop is in the lowish mV/A/m range. Off the top of my head I can't remember but I have it in the Reg's at home.
Thanks again Ray
Gareth
Yes, any clock will perform better with it's own dedicated supply. The regulator should have the lowest noise possible. A discrete design is mostly better, as the standard 78xx or LM317 regulators are quite noisy for this task. I have good experience with The Flea, the Tentlabs shunts and the Super Teddy Regulator. All of those can be used to power a clock. Please do post the schematics
As for the caps, I wouldn't use tantalum, maybe just for decoupling purposes in digital circuits. Wima MKS are just MKT a.k.a. polyester caps, but they are better than an electrolytic for decoupling. Not recommended for carrying audio-signal. For signal applications you should use polystyrene (PS) or MKP caps.
Regards,
Ray
As for the caps, I wouldn't use tantalum, maybe just for decoupling purposes in digital circuits. Wima MKS are just MKT a.k.a. polyester caps, but they are better than an electrolytic for decoupling. Not recommended for carrying audio-signal. For signal applications you should use polystyrene (PS) or MKP caps.
Regards,
Ray
Yes, any clock will perform better with it's own dedicated supply. The regulator should have the lowest noise possible. A discrete design is mostly better, as the standard 78xx or LM317 regulators are quite noisy for this task. I have good experience with The Flea, the Tentlabs shunts and the Super Teddy Regulator. All of those can be used to power a clock. Please do post the schematics
As for the caps, I wouldn't use tantalum, maybe just for decoupling purposes in digital circuits. Wima MKS are just MKT a.k.a. polyester caps, but they are better than an electrolytic for decoupling. Not recommended for carrying audio-signal. For signal applications you should use polystyrene (PS) or MKP caps.
Regards,
Ray
Yes, I have been loking at the power supplies you have mentioned and I fancy going down this road instead of just tapping off of some internal supply lines - even though, to a degree, that's what will happen but only to feed my own dedicated supply. I have been looking at various types of discrete regulated circuit designs that have low noise characteristics. There are many ways but what do I choose ?
Out of curiosity, what current rating (milliamps I guess) needs to be supplied to one of these clocks ? How musch does a series-mode crystal need ? It may seem dumb but I am not sure.
Thanks for the tips about capacitor choice too, this is an area where, apart from power supply electrolytics I was very unsure as to what the best choice is especially as you mentioned tantalum as I have not seen these specced in many designs. Thanks.
As for the clock designs, I have them here on paper and I hope to get them posted. The design using the 4007 CMOS sounds great, it's acccuracy and it's simplicity. I was wondering if it would be possible, as it is an enhancement mode n and p channel IC - paralleled, could the cd 63's (and other machines) CD Sled servo and microprocessor have their clocks fed from this device after being divided by two ?
Gareth
Most clocks will consume a modest amount of current, I think somewhere between 10 and 50mA max. will be a good guess. Clocks with CMOS and FET oscillator will need even less.
It's not needed to clock the microcontroller, because it only provides control and display functions. It is a noisy device, so I would not hook it up to the audio clock. But you can clock the servo of the CD63 with it. Using a divide-by-two is an option, but given that this clock has a low parts count, it's easy to use a second one for the servo.
Ray
It's not needed to clock the microcontroller, because it only provides control and display functions. It is a noisy device, so I would not hook it up to the audio clock. But you can clock the servo of the CD63 with it. Using a divide-by-two is an option, but given that this clock has a low parts count, it's easy to use a second one for the servo.
Ray
Yeah, I guess your right. Just being a tight wad because some crazy illegal driver has wiped out my car and getting to work from where I live at the moment is hard work.
I thought that it would have been a neat idea as most of the 4007 would be redundant and some of it could be re-utilised with a divide by two. O.K. the microprocessor needs to be left 'as is' and so I shall concentrate on the main clock and servo. So, does it follow that I could use a different type of n/p channel mosfet solely ? A discrete device as opposed to the 4007 ? as it will be mainly redundant ?
Gareth
Hi Gareth,
You need a flip-flop function for a divide-by-two. I doubt that it can be realized with just two gates of the 4007. It's not a good idea to use one of the other gates as a second oscillator for the servo clock. They share the same supply-pins, so it's very easy for noise to travel from one clock to the other. It better to use two separate IC's.
Ray
You need a flip-flop function for a divide-by-two. I doubt that it can be realized with just two gates of the 4007. It's not a good idea to use one of the other gates as a second oscillator for the servo clock. They share the same supply-pins, so it's very easy for noise to travel from one clock to the other. It better to use two separate IC's.
Ray
