CD6000OSE - tube output stage - help appreciated :-)

[ilimzn]

IMHO you do not need a separate PSU for the DAC digital section. The 'analog' section has it's own internal connection to the clock circuits and it also re-synches everything it gets from the digital section so that jitter at this point equals the clock jitter (assuming good XVdd/XVss and AVdd/AVss quality). Also, there is no need to use a separate PSU for the output stage and the analog DAC section if the analog section has it's own CCS+shunt - the DC current is so small the extra load does not make any difference, and any AC current loops are kept local by the shunt + CCS. The shunts for the analog output may be a problem because you need quite a bit of current there, so you are introducing heat. The simple solution using a FET + LEDs or TL431 will not work here. For starters, I would suggest leaving the original series regulation for the output stage, but think about upgrading the caps on the power lines. IIRC each HDAM has a small resistor in series with the output from the regulators, and a cap to ground. This is sort of a 'poor man's shunt' where it's not paramount to keep proper DC voltage, but to keep the AC current loop local. Increasing the caps and using good quality (...bypassed with small foil caps...) makes a lot of difference.
What you do need as you surmised, is a separate power supply for the clock. This alowes you to completely separate the clock PSU ground and the player PSU ground(s) and connect those in a single pint, where the clock comes to the main board. As for the clock circuit in the DAC chips, this also needs a clean supply as eny clock given to the DAC passes through that part of the circuit. It is made up of clock buffers, so a dirty supply will inject the 'dirt' into the edges, which is in other words, jitter.
This portion of the circuit consumes only a fraction of a mA. It is important that it is low noise. This is where I would use a LED shunt. The CCS for this can be fed from the analog power, just like the DAC analog power. A LED shunt has a little advantage here because i't extremely low-noise, while the current consumption is small and conctant, so it's higher internal impedance is of no great consequence (all current variations occur at clock speed here so a good cap in parallel with the shunt takes care of that).

IMPORTANT!
The SM5872 datasheet states that all power supplies should be derived from one common power supply. What it is trying to tell you, is that you have to be careful with power supply sequencing, to prevent latch-up and incorrect chip operation or even destruction. In a scenario with multiple power supplies, like you are proposing, you need to take care of that by connecting two diodes in anti-parallel between each separate power supply. In your case, that would be DVdd, XVdd and AVdd.
I've done some research on this in the CD63 - XVdd is not that crytical. If it is not there, the chip fails to get a clock signal so it just does not work until it's there. This is because the clock circuit feeds all other parts of the ship. but no other parts of the chip feed back any signal to the clock. So, if there is no XVdd, the clock remains at zero. However, since you are inputting a clock signal which has it's own power supply, which might become live before the XVdd does, put a diode from the XIN line to XVdd (1N4148). This will clamp the clock input so that it cannot become more than the XVdd line, generally such a situation, where an input is driven with voltages higher than the power supply, is not what chips like. Not at all...
What remains is DVdd and AVdd. Put an anti-parallel diode combo (1N4007 or something like it) between these two. This takes care of various caps charging unevenly alowing a situation where the internals of the chip could start driving the analog section, while it still has no power supply - generally a no-no. With the diodes, the maximum difference between DVdd and AVdd is around the safe value of 0.5V. Take care to match the voltage on AVdd and DVdd well, so that under normal circumstances the diodes work with as close to 0V across them, which will make any current that leaks through negligible. Please be sure to incorporate this safety feature as otherwise you might find your SM5872s die a quick death

[ernesternest]

.....sat a long time over original layout, thinking about which parts I have to remove, which parts I can use (done ), tested appropriate LEDs for shunts (have meanwhile 2 that show 4.94V and two with 5.01V), trying to take all the advices into account.....being a bit shitty before removing all the resistors and filmcaps ...... but I'll go on......

more to come


Ernst

[ernesternest]

well, have desoldered all the unneeded parts and soldered the new ones that found a place on the pcb. Just have to add the new ones without a chance to add them to the original pcb.
During a break I was looking over the circuit and layout to figure out how to move on after having finished and tested outputstage. Next step will be replacing psu for DAC clock support (XVDD) and analog support (AVDD) by LED shunt and I wonder if I can keep the electrolytics close to the DACs when doing this. There are 1000µF Panasonic FM in each case. I don't know if these little LED shunts can drive these or if I should change to smaller?

Thanks Ernst

[ilimzn]

Originally Posted by ernesternest

well, have desoldered all the unneeded parts and soldered the new ones that found a place on the pcb. Just have to add the new ones without a chance to add them to the original pcb.
During a break I was looking over the circuit and layout to figure out how to move on after having finished and tested outputstage. Next step will be replacing psu for DAC clock support (XVDD) and analog support (AVDD) by LED shunt and I wonder if I can keep the electrolytics close to the DACs when doing this. There are 1000µF Panasonic FM in each case. I don't know if these little LED shunts can drive these or if I should change to smaller?

Thanks Ernst

Electrolytics can be left in place - in fact, MUST be left in place, in the sense that the capacitors should be as close as possible to the DAC pins. Even larger can be installed. The only difference is how long it takes for them to initially charge through the shunt regulator CCS.
The LEDs are not 'driving' them, just limiting voltage - hence shunting all further charge current once the voltage reaches the proper value. The current in turn is limited by the CCS, and should be at least 10x higher than the current used by the DAC, but also, within operational limits for the LEDs and transformer/pre-regulator, and CCS regarding heat generation. The capacitors themselves offer the closest and lowest impedance for the AC current component of the DAC/clock. If more than one is used in parallel, at least the smallest one should be as close as possible to the DAC chip pins, to keep the AC current loop as small as possible - see the CD63 mods thread, AVdd/AVss have been shunted by SMD chip caps directly on the DAC AVdd/AVcc chip pins to accomplish this, the SMD chip caps being the smallest of 2 or 3 caps that are all in parralel across AVdd/AVss. This keeps the highest frequency components inside the smallest possible loop where it cannot interfere with other things around it.

If you have done the analog output mod, you might want to try it out with the power supply as is, to check how it sounds, as a first step - then tweak that before proceeding (for instance change the 33-39k resistor thet fine-trims the final roll-off in the frequency characteristic of the filter at the highest point of the audio band).

[ernesternest]

Originally Posted by ilimzn

Electrolytics can be left in place - in fact, MUST be left in place, in the sense that the capacitors should be as close as possible to the DAC pins. Even larger can be installed. The only difference is how long it takes for them to initially charge through the shunt regulator CCS.
The LEDs are not 'driving' them, just limiting voltage - hence shunting all further charge current once the voltage reaches the proper value. The current in turn is limited by the CCS, and should be at least 10x higher than the current used by the DAC, but also, within operational limits for the LEDs and transformer/pre-regulator, and CCS regarding heat generation. The capacitors themselves offer the closest and lowest impedance for the AC current component of the DAC/clock. If more than one is used in parallel, at least the smallest one should be as close as possible to the DAC chip pins, to keep the AC current loop as small as possible - see the CD63 mods thread, AVdd/AVss have been shunted by SMD chip caps directly on the DAC AVdd/AVcc chip pins to accomplish this, the SMD chip caps being the smallest of 2 or 3 caps that are all in parralel across AVdd/AVss. This keeps the highest frequency components inside the smallest possible loop where it cannot interfere with other things around it.

If you have done the analog output mod, you might want to try it out with the power supply as is, to check how it sounds, as a first step - then tweak that before proceeding (for instance change the 33-39k resistor thet fine-trims the final roll-off in the frequency characteristic of the filter at the highest point of the audio band).

Thanks again for advices!! Yes, meanwhile output mod has been finished and - the player is still - playing
Since there were a lot of parts to change and I wanted to use as many traces and places of the old parts of the pcb it was a real peace of work. I had to use some viewing aid due to SMD.... Getting old ....
I have been listening to it now for a couple of hours. Result? The sound changed, but it is hard to describe. I have just this CD6000 and can't compare to an older one. I can just compare to different CDPs; currently I have running a slightly modified (foils for DA, Oscons for DA and Opamps, Foils for coupling) Grundig 8400MKII, a TDA1541 player. Tonally they are pretty the same, before the mod (output and clock) CD6000 was brighter. Naja, it still is but just in the highest tones. So I will play as suggested with the 33-39k restistor to get heights a little bit further down. I like the sound of the 1541, even the soundstage of this particular player is slightly blurry. This is better now with the Marantz Deeper soundstage and better separation. However stunning related sound (A-B comparism by using 2 of the same burned CD-R's hearing the same musik swiching between players by amp remote control). Have to break in a bit more. Set an opamp AD712 in place just to give it a try. Will probably come back to HDAM.

Opamp draws 5mA, HDAM 15mA. Would it make sense to add the 6mA FET as shown in your circuit?

I'm really curious about the sound after having put in the LED shunts for XVSS and AVSS plus foil caps and SMD cap as suggested. Will do this next week.

Cheers Ernst