![]() ![]() |
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
I'm puzzled !
I've read about adding extra capacitance to the MSB bits for better bass response, but some claim the MSB bits to be on pins 7&24 while others claim 13&18 ?
Looking at a schematic of a cdp with TDA1541, I noticed the voltages to be measured are -3.5Vdc on 7&24 and going down to -7.4Vdc on 13&18.Actual measurement on a TDA1541 confirmed this, but I noticed the voltage on pins 12&19 to be -7.5Vdc and on pins 13&18 to be -7.3Vdc.
So I'm wondering, is there a relation between significance of these 2* 7bits and the voltage you're reading, so for more significant bits the value on the decouple pin becomes more negative ?
If this is true, I would conclude the order of bits is (from LSB to MSB):
7&24, 8&23, 9&22, 10&21, 11&20, 13&18, 12&19
So note MSB would then be on 12&19.I tried on 7&24, 13&18, 12&19 adding 220uF ELNA Cerafine's to the 100nF WIMA MKS, but did not hear a change (+ to ground, - to decouple pins). Maybe the Oscon's 15uF would make a difference though, but I don't have'm.
Any final clarification on this ???
BTW I also tried connecting SAA7220 pin 9 (clock output) to TDA1541 pin 4 (cut connection with pin 2) but got no sound. Crystal on SAA7220 is 11Mhz. This should work I read in the archives.
Anybody ?
Follow Ups:
Hi,
According to the schematic of the Philips CD160 the pins # 7 and 24 are LSB and the pins # 13 and 18 are MSB. This old Philips player uses the TDA1541P.
The Pooge Chronicles give the same information.
I hope this helps to sort things out.
ELSO
Hi Rbroer
The Walt Jung/ Hampton Childress modification involves the change of all bit-filter caps to 1 mufarad high quality film type. They use polycarbonate caps and claim improved linearity. (From the Audio Amateur Magazine Pooge Chronicles). Will try it out!The DAC expects a clock of 2822.2927 kHz measured (bitclock and SCK) or exact 44.01x64=2816.64 kHz and 44.01 kHz for the WS (measured 44.0983 kHz).
Did you receive my 'snailmail'?
Best regards,
ELSO
Well, supposedly the system clock input at pin 4 of the TDA1541(A) could be a multiple of the bitclock at pin 2.
So in a cd player, one could feed it with a very stable 11.289Mhz clock. This same low jitter clock would also be used to feed the SAA7220 and SAA7210.
But apperently this feature is not available for all types.
My Philips CD360 has the TDA1541. I do have a TDA1541A lying around, so I might have to try this one.
As for the decoupling caps, I've read an article about mixing 3 types of film caps for all pins; 10 or 1nF, 100nF and 1UF. Currently I have 100nF MKS/63V on all pins and 220uF Elna Silmic on the MSB pins.
I will add 1u MKS/63V later, but might have to alternate between upper side and lower side connecting, because they are too big to be placed next to each other.
Hi Rbroer,
I am sorry but the frequency's I mentioned are for my non oversampling dac(output of the CS8412). When using a digital filter you get other values. Hopefully this is not too confusing.
ELSO
Impedance matching of ALL data lines in a CD player will make a monumental improvement in sonics. This will take some digging to make exact matches for the lines as they vary with line length. However, 25- 50 ohms works well for lines over 4-5 inches, and 10-15 ohms for shorter lines. This is accomplished by cutting the data trace at the output of the concerned IC and installing a chip R to bridge the gap from the IC pin to the trace. Also, try grounding all IC Gd. pins directly from the IC pin to the top GP.Happy listening !
I would suggest that the important spec to be observed when adding decoupling capacitance is leakage, I have heard improvements when just upping all of the bit bypass lines to 1 uF film caps, would be cautious with an electrolytic because leakage currents could disturb the levels associated with each bit/transition. My understanding of the topology is that the bypass caps reduce noise primarilly around transitions, cleaning up and improving settling characteristics, with secondary benefits being reduced sensitivity to induced noise on each bit line. But I could be smoking crack [grin].There will be process related variations in the measured voltage at the bypass pins, which should be of no concern. One of the dirty secrets of chip design is that absolute "matching" sucks, but relative matching can be much better, and the prudent choice of topologies can produce seemingly magical results with only 10 or 20 percent absolute value tolerance primitives, ie Rs and Cs.
There are [at least] two versions: 1541 and 1541A; one has provisions for feeding the mclk clock [not bclk] into pin 4, presumably for internal clocking/reclocking/registering. You will be put on hold without muzak when trying to mix the two...
Share and enjoy...
This post is made possible by the generous support of people like you and our sponsors: