Another view: All those chips covered with kapton/mylar tape seemed original to me. I've to correct myself, obviously silver ink probably was used in re-labeling too:
It is quite clear that those ICs were moulded in that distinctive Japanese cases, with that standing-out dimple in the left side. Something that can not be replicated anymore as to my knowledge all other DIP28 cases since the later 90s didn't use that form anymore.
It is quite clear that those ICs were moulded in that distinctive Japanese cases, with that standing-out dimple in the left side. Something that can not be replicated anymore as to my knowledge all other DIP28 cases since the later 90s didn't use that form anymore.
Attachments
Last edited:
Absolutely, still a bargain compared to PCM63 and AD1862! Those two are well 'in fashion' at the present time, I wonder if what I'm doing here will create upward pressure on PCM58's market price?
As regards to 'life out' I'm not completely clear on your meaning, I'm guessing you're wanting to say 'PCM58 offers a lifetime's worth of tweaking opportunities' (or something similar) ? I haven't played at all yet with all those bit weight trims, is that something you've explored?
Indeed many many quantums (quanta?) ahead of 1387, seems to me the desired metric is SNR/Watt and 1387 falls down on that even though its under 10% of the power of PCM58. I was looking at TDA1545 yesterday as that's more frugal in power than 1387, runs down to 6mW but SNR suffers too at its lowest supply rail (3V).
As regards to 'life out' I'm not completely clear on your meaning, I'm guessing you're wanting to say 'PCM58 offers a lifetime's worth of tweaking opportunities' (or something similar) ? I haven't played at all yet with all those bit weight trims, is that something you've explored?
Indeed many many quantums (quanta?) ahead of 1387, seems to me the desired metric is SNR/Watt and 1387 falls down on that even though its under 10% of the power of PCM58. I was looking at TDA1545 yesterday as that's more frugal in power than 1387, runs down to 6mW but SNR suffers too at its lowest supply rail (3V).
Richard - you've absolutely hit the nail in what I was referring to.
Almost forgot about the 4x times MSB adjustment the '58 provides.
Regarding the EU/Philips side - I've sticked to the 1541, so I've never really dived into it.
Replaced 2 CD Players that came with the TDA1543 by the TDA1387, that's it - OK, one got the SAA7220 bypassed with an CS8421 ASRC but that didn't blow my mind either.
Nevertheless - how about an universal 1mA DAC out passive filter board?
Or did I miss something in this thread?
(recently got a Yamaha CD8-AD with 8x PCM1702 for €20 that also wants to be played with it)
OK, that is now going way beyond the topic of that thread, I'll shut up
Almost forgot about the 4x times MSB adjustment the '58 provides.
Regarding the EU/Philips side - I've sticked to the 1541, so I've never really dived into it.
Replaced 2 CD Players that came with the TDA1543 by the TDA1387, that's it - OK, one got the SAA7220 bypassed with an CS8421 ASRC but that didn't blow my mind either.
Nevertheless - how about an universal 1mA DAC out passive filter board?
Or did I miss something in this thread?
(recently got a Yamaha CD8-AD with 8x PCM1702 for €20 that also wants to be played with it)
OK, that is now going way beyond the topic of that thread, I'll shut up
Last edited:
TDA1541A does have a fairly decent SNR number at 110dB, its not one I've spent any time playing with though. Replacing it with TDA1387 would need quite a few (32?) to gain an extra 15dB and even then I'd expect LF noise to be better on 1541A as its bipolar.
I've been meaning to create a 'soft' substitute for SAA7220 for many years now but other things have taken up my time. I figured STM32G431 is the MCU to do the job and even improve on 7220 it has an in-built random number generator for dithering purpose. I got an eval module with it last week.
A 1mA universal DAC passive filter board - you mean something like this? - https://www.diyaudio.com/community/...r-i-v-stage-for-nos-dacs.395208/#post-7292110. I can confirm it works very well with paralleled PCM58s but gives lowest noise with a mod to obviate the need for the BPO pin(s).
8 * genuine PCM1702 for 20euro seems definitely a bargain!
I've been meaning to create a 'soft' substitute for SAA7220 for many years now but other things have taken up my time. I figured STM32G431 is the MCU to do the job and even improve on 7220 it has an in-built random number generator for dithering purpose. I got an eval module with it last week.
A 1mA universal DAC passive filter board - you mean something like this? - https://www.diyaudio.com/community/...r-i-v-stage-for-nos-dacs.395208/#post-7292110. I can confirm it works very well with paralleled PCM58s but gives lowest noise with a mod to obviate the need for the BPO pin(s).
8 * genuine PCM1702 for 20euro seems definitely a bargain!
The re-engineered 'Stack DAC' PCB has just arrived. We went up a size, now the PCB is the same size as Celibidache and sports almost as many DAC chips (16, but overclocked in this instance so equivalent in output current to having 24). The filter's not a separate module and can be built with P14 or TDK 7mm inductors. With so much DAC output current on tap there comes a point where output buffers may not be needed as the Zout is low enough for signal driving purposes so there's a tap-off point prior to the buffers for creating a more minimalist DAC. This connector can also be used to 'stack' (electrically rather than physically) DACs together.
These two boards are the fruit of being hunched over my iron for a good part of the weekend just gone. The vertical filter boards saved me a fair chunk of time though - turned out I already had suitable 5th order filters built for the 'Stack DAC' so why not repurpose them?
Against the balanced PCM58 which previously occupied my bench they of course don't stand a chance, there's noise which confuses the ambience retrieval, particularly in the lower freqs but in isolation they're still an enjoyable listen with a very good sense of scale on orchestral works.
The next pair of boards I plan to build using TDA1545s to see if there's an appreciable SQ difference between the two chips.
One TDA1545 variant playing music, the other awaiting its filter and last few TH caps and connectors. Then we can listen to a balanced pair. More current is coming out of the TDA1545s than TDA1387s for the same power input which makes the 1545 a more efficient (and potentially lower noise) DAC.
For myself I don't think so but everyone has their own priorities. I don't go for the graded ones because I'm much more interested in noise than distortion and the premium ones only are spec'd on distortion. Besides that, the temptation for a seller to add their own 'K' to an unmarked device must be high when few are going to be running the necessary THD tests to verify.
nopants,
Just to validate abrax's comments. I've only purchased 4 PCM58 but all work well, I purchase my DACs from the same seller and I have purchased 6 AD1862 from him in the past, all without issue.
I think it is critical to find a good seller and not just go for the lowest price. Cheap parts are often either fake or poorly de-soldered chips that either have heat damage or bent up pins. Pay a little more and ask the right questions, maybe even get pictures of the chips before they are sent can help you be more satified.
Regarding K series, I also don't buy these. Less because of improved disortion vs. noise, and more because of the high potential for remarking. Even if you have a good vendor he might not know that he bought remarked chips... and I don't have a valid way to test them. The PCM58 has the ability to adjust 4 MSBs but I don't see many people taking advantage of this feature. Why spend money on a K and then not adjust the MSB?
Just my two cents.
G
Just to validate abrax's comments. I've only purchased 4 PCM58 but all work well, I purchase my DACs from the same seller and I have purchased 6 AD1862 from him in the past, all without issue.
I think it is critical to find a good seller and not just go for the lowest price. Cheap parts are often either fake or poorly de-soldered chips that either have heat damage or bent up pins. Pay a little more and ask the right questions, maybe even get pictures of the chips before they are sent can help you be more satified.
Regarding K series, I also don't buy these. Less because of improved disortion vs. noise, and more because of the high potential for remarking. Even if you have a good vendor he might not know that he bought remarked chips... and I don't have a valid way to test them. The PCM58 has the ability to adjust 4 MSBs but I don't see many people taking advantage of this feature. Why spend money on a K and then not adjust the MSB?
Just my two cents.
G
DAC8811 is a traditional R2R DAC designed only to give full DS performance looking into a precision opamp with very low offset voltage. Thus any discrete I/V stage would need a precision opamp servo arrangement hooked around it to get the advertised numbers. A complication I'd much prefer to avoid.
LTC1595's DS says they've reduced the sensitivity to offset compared to earlier generations. I wonder what solution they went with?
Certainly interesting, thanks.
LTC1595's DS says they've reduced the sensitivity to offset compared to earlier generations. I wonder what solution they went with?
Certainly interesting, thanks.
PCBs coming together for Abbado II rev2 using TDA1545A. Seeing as its not an I2S DAC, there's additional digital interfacing needed to allow it to run from an I2S feed. I've also added a multiturn trimmer which allows the output level to be varied - the chips have quite a wide tolerance on output current. The output level trim comes into its own when running balanced, to get the channels much more closely matched. The 5th order filter has been optimized for being built from standard cap values as far as possible, making it easier to turn into a kit.
Having listened to this revised DAC for a few days and enjoyed the sound, I did a very quick comparison with Celibidache and what struck me was Celi's bass was so much better. Wifey listened - just to Abbado - and also pointed out the bass was rather lacking. This made me pay more attention to the 1545's power rails, which for some reason I decided to feed from CJ431 shunt regs. I'm not sure how I decided on that originally but I swapped them out for TL431 (TO92 package, TI brand) figuring these would be lower noise. Now the bass is more tuneful and the overall acoustic picture warmer. Wifey concurs. So I am wondering about a discrete reg for the chips next, similar to what I did on Celi....
For the TDA1545A, how are you converting the data format to that the 1545A accepts ( EIAJ ??) ?
I assume what the TDA1545A accepts is: MSB first, Left-Justified. Correct?
I shared the schematic for interfacing I2S to 1545 a little way back on the thread : https://www.diyaudio.com/community/...-rbcd-multibit-dac-design.324933/post-7005633
I think its right-justified, MSB first. The interface delays the data by 15 bit periods and inverts the WS.
I think its right-justified, MSB first. The interface delays the data by 15 bit periods and inverts the WS.