difficulty with 1/2 clock to TDA 1541

[tubee]

If you have success with the /4 clocking - I will try it in my Marantz CD60 - which will probably end up as a non-oversampler.

I have one 14 bit and fopur 16 bit machines now, my wife says everytime another CDP gets in: ''what on earth do you want with all those''

I want to experiment with them, one Non-Os, second Os, another with 4 dacs 1541 parallel with shift registers.
Sometimes i like OS, and sometime Non-os, depends on music also.

I use a 8.4672 MHz crystal which is devided by 2 to be fed to the SAA7000. I'll post the photo's and schematics next wee

Midimaze: Have you tried to clock direct in the 2 1540's? 😉

 

[tubee]

Just putted the 74LS163 next to the XO, connected in "free running mode"
Must say, not bad, the spaces of recording-rooms or delay-times are better defined now with the /4 clock to dac. A test with a background eighties sampling CD: the stock-aitken-waterman (correct me if i am wrong) Richard Marks sings now in front of me, is better articulated. The Blue Nile (by Linn records) has a greater ambience now. Also Jack Johnson has better refined siseling sounds of hi-hats.

The LS can only work to 32 Mhz and a HC on 5V +/- 50Mhz, but 32 is sufficient imo.
With OS you can take /2 from pin 14, for Nonos /4 of pin 13, and even /4 (1.4Mhz) or /16 (705kHz) is possible. Pin 1, 7 , 9 and 10 are on +5V supply by a 1k resistor.
Scope testing: very small overshoot(measured without load!) but nice squarewaves seen, wave doubling in size from QA to QD's output.

Have another clock, next i will do is a -04 hex inverter between clocks output and -163 to cope with the small time delay induced by the divider. Will try it with a jumper: with/without -04.
Try then a HC163 also.

Amazing how a simple Philips CD640 CDP can sound with only 2 small extra PCB,s , the simple I/V and a clock with divider.

 

[Guido Tent]

Amazing how a simple Philips CD640 CDP can sound with only 2 small extra PCB,s , the simple I/V and a clock with divider.

Yes, amazing, not ?

I use HC logic, Philips and On-semi

best

 

[Fin]

I have one 14 bit and fopur 16 bit machines now, my wife says everytime another CDP gets in: ''what on earth do you want with all those''

Almost the same! I have one 14 bit and three 16 bit machines.

The LS can only work to 32 Mhz and a HC on 5V +/- 50Mhz, but 32 is sufficient imo.
With OS you can take /2 from pin 14, for Nonos /4 of pin 13, and even /4 (1.4Mhz) or /16 (705kHz) is possible. Pin 1, 7 , 9 and 10 are on +5V supply by a 1k resistor.

This is very interesting - well, for those like me who didn't know this.

Amazing how a simple Philips CD640 CDP can sound with only 2 small extra PCB,s , the simple I/V and a clock with divider.

CD640 must be similar to Marantz CD60 - so I can try this too.

I use HC logic, Philips and On-semi

Are other brands not as good? I have some from TI.............


Now for a question to do with the next stage after direct /2 clocking from a Tentlabs XO2 - converted to a XO2.5. The Arcam's existing 74HC74 and hex inverter will be redundant. What do you think of replacing the 74/04 with another 74HC74 - and using one half of each FF to reclock WS and Data?

The attached picture shows the logic chips partially hidden by the wide, grey cable. This is where the signals enter the board and diverting them through the FFs would not increase the signal path by very much. The bottom left corner shows where the signals enter the TDA1541A via resistors. Is this a reasonable idea and should those 390R metal film resistors stay?

Regards

 

[tubee]

Is this a reasonable idea and should those 390R metal film resistors stay?

I have beed reading Guido's article (nice work Guido) about HF, so mounted 47ohm stopping-resistors there. Original was zero ohm, but i don't know 390 is too heavvy.

This is very interesting - well, for those like me who didn't know this.

Googled everything, all on the net.

CD640 must be similar to Marantz CD60 - so I can try this too.

I am experimenting further, wait til this is done. Last night i discovered bass was less contoured with the 163. But with 7220 (OS) it will always enhance sound quality.

 

[tubee]

I use HC logic, Philips and On-semi

Yes if possible choose Philips, Signetics is not bad either. (Have read that in an German article about a preamp. Have few others makes tried.
This morning i got a 74HCT163, for a lot: 1.95 Euro!! and not even Philips.

Almost the same! I have one 14 bit and three 16 bit machines.

I own a Philips CD300, its a pitty i havn't the Marantz CD73 equivalent, for me this CDP is a very nice looking vintage player:

http://www.audiogold.co.uk/archive/CDPlayers_Marantz_CD73_59.htm

or better: http://galerie.hifi-forum.de/details.php?image_id=614

But the CD300 looks not bad either.

and using one half of each FF to reclock WS and Data?

Search here on diy, someone has tried demreclocking:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=11949&highlight=

XO2 - converted to a XO2.5.

How is the change to .5 done, with a choke supply?

 

[Fin]

I have beed reading Guido's article (nice work Guido) about HF, so mounted 47ohm stopping-resistors there. Original was zero ohm, but i don't know 390 is too heavvy.

Yes - Guido mentions 47ohm for this application - so I will try that value.

I am experimenting further, wait til this is done. Last night i discovered bass was less contoured with the 163. But with 7220 (OS) it will always enhance sound quality.

I am happy to wait 🙂

I own a Philips CD300, its a pitty i havn't the Marantz CD73 equivalent, for me this CDP is a very nice looking vintage player:

Mine is a CD303 - with some problems.

Search here on diy, someone has tried demreclocking:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=11949&highlight=.

Is this similar to reclocking the signals going into the DAC - but doing it inside the TDA?

How is the change to .5 done, with a choke supply?

It is best to get the details directly from Guido. That way there can be no confusion.


Thanks for the help.

 

[tubee]

Is this similar to reclocking the signals going into the DAC - but doing it inside the TDA?

Dem reclock has something to do with internal clock to pin16-17 of 1541 i think, can someone enlighten this?

In that thread is stated that BCK reclock helps most.

 

[Pedja]

The “Dynamic Element Matching” is about the time averaging between the four current sources before their currents are sent further downstream (see the page 7 of the Philips’ Digital Audio Designer’s Guide). This averaging results in the four tightly matched currents two of which are used for the MSB, one of which is used for the MSB-1 and one of which is further divided the same way. After three such averaging the seventh output is used for 10 bit (10 LSBs) passive divider. (BTW the purpose of those 14 TDA1541(A)’s capacitors, seven for each channel, is to decouple these seven currents.)

Henk ten Pierick’s scheme suggests external (and synchronous) clocking of the DEM circuit instead of usage of the embedded oscillator (the capacitor at pins 16/17 is its external part). A DEM “reclocking” has nothing to do with what is conventionally accepted as a reclocking in the DACs.

Pedja

 

[tubee]

Thanks Pedja for your enlightment in this matter. 🙂

 

[tubee]

Arcamental: wat do you want to do with the Arcam:
If you want to keep SAA7220 and so stay wit OS, my advice is: (though i'm not aware of rest of your gear)

*Build a separate clock as good as possible. This is the most important
step imo
*Good I/V stage, several to be found here.
*divide the separate clock /2 with 74HC74 (simple but effective) and
feed this into 1541.
*Use a pre-amp as buffer to poweramp, (I/V's often doesn't like low
impedances to drive, a secondhand MF X-can could improve allready)

But this are guidelines only, have done them (and others) and i'm not regretting it.

Then maybe try Nonos with another player, again clock is most effective, as I/V, and some lesser effective(in nonos) is the divider (/4 to 1541) because the 7210 has a relative clean clock allready. 😉

 

[Fin]

Arcamental: wat do you want to do with the Arcam:
If you want to keep SAA7220 and so stay wit OS, my advice is: (though i'm not aware of rest of your gear)

Hi Tubee

Thanks for your advice - and it seems like you are a mind reader and may have been a design consultant for Arcam in a previous life. You will see what I mean as you read my reply............

Initially I would like to keep the SAA7220 and stay with OS - so that I can experience the best that TDA1541A with OS can offer. If I am converted to non-OS later, the SAA7220 can easily be removed as it is in a socket. An adapter could be plugged in to replace it.

*Build a separate clock as good as possible. This is the most important step imo

Arcam originally put a "better than standard" clock in this CDP. It is on the audioboard near the DAC and is based on a 74HC04 inverter. However, I have one of Guido Tent's XO2 clocks (canned oscillator on his pcb with low noise power supply) and a dedicated XO PSU (transformer, schottky diodes, choke.......), waiting to be fitted in the Arcam. I don't think I can get better than that.

*Good I/V stage, several to be found here.

Arcam have included a discrete I/V stage - but I will try one of the diy ones in my Marantz CD60 and compare.

*divide the separate clock /2 with 74HC74 (simple but effective) and feed this into 1541.

Arcam did it the other way - they reclocked BCK. However, as we have learned, direct /2 is better - so I will be obtaining /2 clock from the XO2 board.

*Use a pre-amp as buffer to poweramp, (I/V's often doesn't like low impedances to drive, a secondhand MF X-can could improve allready)

I will have to look into this. It is not something that I have thought of.

But this are guidelines only, have done them (and others) and i'm not regretting it.

It is good to know that we are on the right path.

Then maybe try Nonos with another player, again clock is most effective, as I/V, and some lesser effective(in nonos) is the divider (/4 to 1541) because the 7210 has a relative clean clock allready. 😉

Our thinking is very similar! I am happy to try this in my Marantz CD60. I will improve it with OS first - then compare it to the Arcam - then change it to non-OS and see if I like it.

 

[tubee]

I am happy to try this in my Marantz CD60.

Hi Arcamania,
Yes, try it first on this player before the Arcam.

An adapter could be plugged in to replace it.

I use the CD640 without 7220, there is a blank spot now. Connected SDAB to data, and WSAB to WS of 1541. So no adapter is used in Nonos.

Another suggestion: get a little grip on the pollution from 7220 on PS with a Oscon & bead or better: separate regulator (78L05?) with bead and two oscons.

I will try Schottky's in PS also.

Good luck!

 

[tubee]

Oh.. and i reminded:

How are the 14 decoupling caps at 1541: NP0 SMD-ones?
Or decent MKT-MKP's? I fitted twelve 220N MKT and two 0.47uF for MSB.

There is only one (i call it a) IF->THEN when you go in the Nonos world:

You hear the laserbeam tracking everything, also the rubbisch between tracks! To avoid this you have to mute from uP to 7210, so track down where muting signal comes from on uP, it goes to pin 22 ATSB, the mute from the uP has to be connected on pin 11 (mute) of 7210.
The attenuation of searching will be gone then too, as with the 7220, but i never missed that function.

The Philips based 7210-7220-1541 players are very robust imo, (but some think different here), and are easy to work with. But i never consulted Arcam..... My 640 is a test player, i have to move drawer by hand, a MAB8461P is partly broken, but rest still works...

Btw what CDM has the Arcam? -4?

 

[MidiMaze]

Why are you all trying to reclock the 1541 directly and risc the change of clock delays? Philips has well thought of how to use a master clock and feed this to all the parts. (tough this is a simple resonator...) All the pcb timings are calculated for a single master clock... Just calculate how long it will take for the clock to propagate over the pcb, you'll be amazed how "long" that will take. Therefor i suggest in my humble oppinion to not directly clock the 1541...

btw, i have pictures of my cd104, i'll update my stite tomorrow.
http://midiserver.student.utwente.nl/sitestuff/apparatuur/philips_cd_104/cd_104.htm

 

[Guido Tent]

Why are you all trying to reclock the 1541 directly and risc the change of clock delays? Philips has well thought of how to use a master clock and feed this to all the parts. (tough this is a simple resonator...) All the pcb timings are calculated for a single master clock... Just calculate how long it will take for the clock to propagate over the pcb, you'll be amazed how "long" that will take. Therefor i suggest in my humble oppinion to not directly clock the 1541...

btw, i have pictures of my cd104, i'll update my stite tomorrow.
http://midiserver.student.utwente.nl/sitestuff/apparatuur/philips_cd_104/cd_104.htm

Well, it is your choice, but by getting the timing right and directly clocking the DAC, the player performs better.

I hope you understand the reason why people want to directly clock the 1541

Getting the timing rigth is not that difficult.

best

 

[tubee]

Well, it is your choice, but by getting the timing right and directly clocking the DAC, the player performs better.

Midimaze: I'm experimenting with a separate clock, then a HC04 inverter and 163 divider now. The inverter is to shift the delay a little.
And remember, its digital, 1 or 0, music or silence, when timing will be that wrong, there will be no music. But i'm still listening here.......😉

 

[Fin]

Hi Tubee

Hi Arcamania,
Yes, try it first on this player before the Arcam.

Indeed - the Arcam is not for experiments. Only proven upgrades will be done to it.

I use the CD640 without 7220, there is a blank spot now. Connected SDAB to data, and WSAB to WS of 1541. So no adapter is used in Nonos.

I was thinking of making a small adaptor to replace the 7220. It could plug into the empty socket and make the connections as you describe. Then it could easily be removed in order to allow the 7220 to be re-fitted if needed.

Another suggestion: get a little grip on the pollution from 7220 on PS with a Oscon & bead or better: separate regulator (78L05?) with bead and two oscons.

The audioboard/dac board has its own transformer and power supplies. The 7220 is on the decoder board and is powered from a regulator dedicated to the digital section. The mechanical section has it's own regulation. So - I wil definitely use an Oscon (100uF or 180uF - the biggest I have) and a bead. If I can find space - I will look into a seperate regulator for the 7220 - but it is tight in there.

Oh.. and i reminded:

How are the 14 decoupling caps at 1541: NP0 SMD-ones?
Or decent MKT-MKP's? I fitted twelve 220N MKT and two 0.47uF for MSB.

The Marantz CD60 has 14 SMD caps - but the Arcam has ten 220nF MKS and four 0.47uF Tantalums.

There is only one (i call it a) IF->THEN when you go in the Nonos world:

You hear the laserbeam tracking everything, also the rubbisch between tracks! To avoid this you have to mute from uP to 7210, so track down where muting signal comes from on uP, it goes to pin 22 ATSB, the mute from the uP has to be connected on pin 11 (mute) of 7210.
The attenuation of searching will be gone then too, as with the 7220, but i never missed that function.

OK - thanks.

Btw what CDM has the Arcam? -4?

It has a CDM4/11 - and can take a CDM1mkii, CDM2/10/29.....and CDM4/31 plus some others.

Why are you all trying to reclock the 1541 directly and risc the change of clock delays? Philips has well thought of how to use a master clock and feed this to all the parts. (tough this is a simple resonator...)

Philips chose the worst possible place to generate the clock and from which to distribute it - the filter.

All the pcb timings are calculated for a single master clock... Just calculate how long it will take for the clock to propagate over the pcb, you'll be amazed how "long" that will take. Therefor i suggest in my humble oppinion to not directly clock the 1541...

We are still discussing the use of a single master clock. The suggestion is to replace the original one with a better one - and to improve its distribution.

 

[happycat]

Hi all,

I just tried the direct clock method, and this is what I have done so far:

1) keep 4x oversampling
2) cut CLBD trace between SAA7220 and TDA1541A; WSBD and DABO still attached to TDA1541A
3) divide 11.2896 MHz system clock by a 74hc163
4) pin 14 of the 74hc163 goes directly to BCK of TDA1541A. As I am experimenting with "1/2 clocking", I used a piece of 11" stranded wire to interconnect the two chips.
5) pin 4 of DAC still connected to BCK

And I got no sound at all. So probably I have done something wrong here? Wires are too long? and should I also cut the trace between pin 2 and pin 4 of the DAC and connect pin 4 to ground?

Cheers,
honinbou

 

[tubee]

Hi Honinbou

Check 1/2 Clock on pin 14 of 163 with scope, if it is 5.6Mhz this part works correct. Connect by a 47ohm resistor to dac.

Pin 4 must be on Gnd with the "A" 1541

Good luck

Btw 74HC74 can do /2 also, see somewhere else here on diyaudio. The /163 has more dividing-options though if you want to try nonos too.