Have tried it yesterday, connected the wires of all PCB's on my desk, powered it, cd in pushed play on my new made control PCB(better knob layout)....It works,so now 30 cm of extensions are gone...
But maybe there is still a need for a little cap between two flex wires, because the extension PCB has 1 smd cap somewhere, will find that out.
philips cd service manuals & schematics
Here is a website of a dutch guy who is doing a lot of old scanned philips cd manuals on cd for 20 euro's,seems a goood deal,some are in english and other languages including dutch and the shematics are all in english,hope this helps.
http://ultranalog.com/sale/
Here is a website of a dutch guy who is doing a lot of old scanned philips cd manuals on cd for 20 euro's,seems a goood deal,some are in english and other languages including dutch and the shematics are all in english,hope this helps.
http://ultranalog.com/sale/
OK.
I have CD304mk2 , it doesn´t work.
I put cd in drawer , push play , cd is turning , laser read information , but no no number of songs or countdown on display , and no music.
CD stops , and nothing is happening.
I resolder all solder points , replaced all caps on transport board and other boards.
What you all think problem is?
I start to be nervous about this , and nothing comes on my mind solve this problem.
When I bought it , it was working , back home dead.
I have CD304mk2 , it doesn´t work.
I put cd in drawer , push play , cd is turning , laser read information , but no no number of songs or countdown on display , and no music.
CD stops , and nothing is happening.
I resolder all solder points , replaced all caps on transport board and other boards.
What you all think problem is?
I start to be nervous about this , and nothing comes on my mind solve this problem.
When I bought it , it was working , back home dead.
Hi, I am really glad I found this thread as right now I am trying to tweak my recently acquired Philips CD300 too. My wife loves the design which must be taken seriously and properly explored (certainly not an everyday situation ).
The original CD300 sounds rather harsh to me. I changed all electrolytics in the PSU and in the signal route, swapped ceramics around DACs for proper foils, bridged the output electrolytics with foils. After all the changes, it is still way behind my sweet sounding FineArts CD9000 - TDA1541 with disabled oversampling (bypassed SAA7220). As you see, I am an avid junk collector
I hooked the two players on the scope and saw the major difference - for 1kHz square signal from the CD, the NOS 1541 + filters outputs clean rectangles, whereas the OS 1540 produces ugly overshoots, typical sign of oversampling.
It is not easy to simply bypass the 7030 as it takes 16-bit samples in and outputs 14-bit chunks for 1540.
Thus, I decided to try to link the 16bit 1541 DAC to the Philips. The old player does not use SAA7000; it has a whole board with several SONY chips instead. Following datasheets for SAA7030 and TDA1541, I found a common data format supported by both the chips, the separated left/right channels. I hooked up the DAC using shielded quality CAT5, 4 pairs twisted with digital ground, to corresponding INPUTS of the oversampling filter SAA7030. However, the resultant sound sucked. No details, no highs, dark, boring. The scope showed that for sine signals the output from 1541 was seriously distorted, a lot of noise (the scope could hardly synchronize).
I checked the data signals going into 1541 (and SAA7030) and found out they differ from the specifications - instead of 16 bits for each channel, there are two identical groups of 16 bits (i.e. 32 bits in total). The scope shows that the 50cm of CAT5 produce no serious distortion of the signal shape. Therefore, I suspect the 1541 does not like the nonstandard format and perhaps its input serial-to-parallel registers overflow and cause the chaos.
This is where I am today. I am going to try to remove the excessive 16bits from the clock signal using binary counter and some NAND gates - I will report if it fixes the problem.
).The original CD300 sounds rather harsh to me. I changed all electrolytics in the PSU and in the signal route, swapped ceramics around DACs for proper foils, bridged the output electrolytics with foils. After all the changes, it is still way behind my sweet sounding FineArts CD9000 - TDA1541 with disabled oversampling (bypassed SAA7220). As you see, I am an avid junk collector
I hooked the two players on the scope and saw the major difference - for 1kHz square signal from the CD, the NOS 1541 + filters outputs clean rectangles, whereas the OS 1540 produces ugly overshoots, typical sign of oversampling.
It is not easy to simply bypass the 7030 as it takes 16-bit samples in and outputs 14-bit chunks for 1540.
Thus, I decided to try to link the 16bit 1541 DAC to the Philips. The old player does not use SAA7000; it has a whole board with several SONY chips instead. Following datasheets for SAA7030 and TDA1541, I found a common data format supported by both the chips, the separated left/right channels. I hooked up the DAC using shielded quality CAT5, 4 pairs twisted with digital ground, to corresponding INPUTS of the oversampling filter SAA7030. However, the resultant sound sucked. No details, no highs, dark, boring. The scope showed that for sine signals the output from 1541 was seriously distorted, a lot of noise (the scope could hardly synchronize).
I checked the data signals going into 1541 (and SAA7030) and found out they differ from the specifications - instead of 16 bits for each channel, there are two identical groups of 16 bits (i.e. 32 bits in total). The scope shows that the 50cm of CAT5 produce no serious distortion of the signal shape. Therefore, I suspect the 1541 does not like the nonstandard format and perhaps its input serial-to-parallel registers overflow and cause the chaos.
This is where I am today. I am going to try to remove the excessive 16bits from the clock signal using binary counter and some NAND gates - I will report if it fixes the problem.
I think it is possible to get a very nice sounding cd300 player, and this with keeping TDA1540 in it. Even unmodified its sounding not bad at all. But my pre- and hybrid power-amp are tubed ones and thus can deal better with eventual HF rubbisch from a player. Solid state amps cannot always have the rubbisch and the sound is harsch with some CDP's, especially oversampled ones.
A new clock with divider to the dacs, and a better I/V would be sufficient. And also some new caps in PS ofcourse.
But the cd104 is 14 bit also, has a somewhat better CDM(1), newer (more stable) servo board, and a shielded decoder pcb. Its better to mod this cdp afaik, if you can get one.
I know someone has a 104, but is unfunctional.....
A new clock with divider to the dacs, and a better I/V would be sufficient. And also some new caps in PS ofcourse.
But the cd104 is 14 bit also, has a somewhat better CDM(1), newer (more stable) servo board, and a shielded decoder pcb. Its better to mod this cdp afaik, if you can get one.
I know someone has a 104, but is unfunctional.....
rfbrw said:
The 44.1kHz latch enable signal. In order to see the clock and data signals I have to use 10x timebase magnification and delayed dual timebase.
This assumption would hold, if the input format conformed the specifications, i.e. 16 bits between each latch enable signal. However, the SONY replacement board provides 32 bits. Tomorrow I will hopefully finish the circuit and report on the results.
phofman said:
The specifications are very flexible where multibit devices are concerned.
The SAA7000 also provides 16 bits and some 'undefined' data, as Philips put it. It is the 16 bit burst clock that ensures the right bits are loaded.
Is the clock on pin2, BCK of the TDA1541 burst or continuous ?
Bypassing the SAA7030 - NOS 1540
Back here to report as promised.
I cut the data clock signal to first 16 pulses, effectively changing the format to 16 bits as in the 1541 datasheet, using a 4bit counter and a few more logic gates. Surprise - no change at all. Still the same rather blurred sound. Clearly, the 1541 takes only the first 16 bits.
After playing for a few evenings with the CD300 -> TDA1541 link, I managed to "discomfort" the 1541. More on that here
My 1541 seems to work for now and I did not want to mess with it anymore. I decided to use the existing 1540 DACs, but bypassing the oversampling filter SAA7030 which causes the ugly overshoots.
I traced back input signals from 1540s. The data signals and data clock go directly from oversampled outputs of SAA7030, while the Latch Enable clock output is reclocked with master clock CLOX first (pin 19 SAA7030), by one of the 7408 AND gates on the player's PCB. BTW, this is probably the reason for the blurred sound from 1541 - I did not reclock the LE signal with the precise clock - I guess the original LE signal has a significant jitter, causing the loss of details.
I tried to feed the 2 x 16 bits signal to 1540s, hoping they were as fool-proof as the 1541. Well, wrong assumption, the resultant signal was perfect for rectangles but seriously messed up for sine. That clearly indicated the input registers disliked the excessive bits. In order to use just first 14 bits, I designed a simple circuit which passes just the first 14 pulses of the data clock.
The NOS LE pulses entering SAA7030 are too wide, almost overlapping the first data clock pulse. I solved that by ANDing the signal with the precise oversampled reclocked LE signal going originally to 1540s - it just happens there is one short oversampled pulse within duration of the NOS wide (jittery) pulse (the remaining three oversampled pulses fall into the NOS pulse gap).
The circuit below is straightforward. The final reclocked (V1/1) (inverted) LE signal presets the counter to 1 (as defined by combination of preset inputs). The overflow output is high (no overflow yet), allowing the input data clock to shift the counter upward. Upon the 14th clock pulse the overflow output (active low) disables the clock (using the NAND gate V1/3). This state lasts until new LE signal presets the counter to 1 again and the counting starts again. I used 74HC125 drivers to buffer the outputs - these could be probably omitted.
I put the three ICs with decoupling capacitors on a small trial PCB and soldered in on the bottom side of the upper PCB.
Now the NOS DAC produces perfectly clean 1kHz squares on the CD outputs. I find the sound much softer and sweeter, the harshness is gone. In fact, it's very close to my Grundig with 16bit TDA1541. Honestly, I cannot tell the 2 missing bits difference.
Back here to report as promised.
I cut the data clock signal to first 16 pulses, effectively changing the format to 16 bits as in the 1541 datasheet, using a 4bit counter and a few more logic gates. Surprise - no change at all. Still the same rather blurred sound. Clearly, the 1541 takes only the first 16 bits.
After playing for a few evenings with the CD300 -> TDA1541 link, I managed to "discomfort" the 1541. More on that here
My 1541 seems to work for now and I did not want to mess with it anymore. I decided to use the existing 1540 DACs, but bypassing the oversampling filter SAA7030 which causes the ugly overshoots.
I traced back input signals from 1540s. The data signals and data clock go directly from oversampled outputs of SAA7030, while the Latch Enable clock output is reclocked with master clock CLOX first (pin 19 SAA7030), by one of the 7408 AND gates on the player's PCB. BTW, this is probably the reason for the blurred sound from 1541 - I did not reclock the LE signal with the precise clock - I guess the original LE signal has a significant jitter, causing the loss of details.
I tried to feed the 2 x 16 bits signal to 1540s, hoping they were as fool-proof as the 1541. Well, wrong assumption, the resultant signal was perfect for rectangles but seriously messed up for sine. That clearly indicated the input registers disliked the excessive bits. In order to use just first 14 bits, I designed a simple circuit which passes just the first 14 pulses of the data clock.
The NOS LE pulses entering SAA7030 are too wide, almost overlapping the first data clock pulse. I solved that by ANDing the signal with the precise oversampled reclocked LE signal going originally to 1540s - it just happens there is one short oversampled pulse within duration of the NOS wide (jittery) pulse (the remaining three oversampled pulses fall into the NOS pulse gap).
The circuit below is straightforward. The final reclocked (V1/1) (inverted) LE signal presets the counter to 1 (as defined by combination of preset inputs). The overflow output is high (no overflow yet), allowing the input data clock to shift the counter upward. Upon the 14th clock pulse the overflow output (active low) disables the clock (using the NAND gate V1/3). This state lasts until new LE signal presets the counter to 1 again and the counting starts again. I used 74HC125 drivers to buffer the outputs - these could be probably omitted.
I put the three ICs with decoupling capacitors on a small trial PCB and soldered in on the bottom side of the upper PCB.
Now the NOS DAC produces perfectly clean 1kHz squares on the CD outputs. I find the sound much softer and sweeter, the harshness is gone. In fact, it's very close to my Grundig with 16bit TDA1541. Honestly, I cannot tell the 2 missing bits difference.
Attachments
rfbrw said:
Well, that was the case for 1541 which accepted the 2x16bits input. The 1540 was clearly confused until I cut the data clock pulses to 14 only. There could be another reason - perhaps the input registers are cyclical - 1541 cycled twice with the same data (2x16bits), whereas 1540 ended up with messed up information. That would explain why the 0dB square signal was correct - there were only zeros/ones in the input data chunk and the loops did not matter.
phofman said:
DACs don't get confused. They get subjected to an incorrect relationship between data, clock and trigger. The SAA7030 is a 16 bit device. When used with a continuous clock the position of LE will reflect this. When used with a burst clock the burst will be 16 bits long and the position of LE is less important.
The TDA1540 is a 14 bit device. Send it data intended for a 16 bit device with either sort of clock and the sign bit and the MSB of the data will be clocked out of the input register leaving the last 14 of the 16 bits.
rfbrw said:
You are right, the word "confused" was not chosen well. Now I know there are standard (continuous, passthrough) shift registers at the input of both 1540 and 1541. When 1541 was fed with 2 groups of 16 bits, it took the identical value of the second group and output correct signal. 16 bits fed into 1540 caused the DAC to convert only the last 14 bits, distorting the signal (the lower 14bit sine fragments were clearly visible on the scope). That is why I had to cut the data clock to first 14 bits only.
The reason I was trying to feed the 16 bits into 14 bits DAC is the faint hope, that perhaps the input register takes first 14 bits only and stops shifting afterwards, waiting for reset by the LE signal. Clearly that is not the case, as expected
phofman said:
The reason I was trying to feed the 16 bits into 14 bits DAC is the faint hope, that perhaps the input register takes first 14 bits only and stops shifting afterwards, waiting for reset by the LE signal. Clearly that is not the case, as expected
The solution is simple. Delay the data by two sck cycles.
How will the schematic look like then?
But if you don't want waste (a long) time on that schematic to draw i don't mind, and will figure it out myself, because i want to finish some other projects first (cdp with 4 dacs, speakers, turntable) so i have no time left to change this 14 bit machine yet.
The 4 paralleled 1541 CDP with shift registers in NOS is allmost listenable in a few weeks i guess.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.