What about the Crystek 957 @ 49.152 Mhz?
60EUR less expensive than the Kali.
Does it work with Clive's 384 kernel?
Crystek is too large to fit directly on the PCB. NDK would work.
Also, as ES9023 is being used with asynchronous MCLK then I would have thought an 80Mhz or 100Mhz would be appropriate. Vasilis can advise I'm sure.
Maybe try Epson SAW as mentioned here?
Not sure if decoupling caps would need changing?
Last edited:
On the decoupling caps, you would have to refer to the datasheet. OR possibly ask Vasilis, it seems he has tried many options with ES9023 DACs.
ES9023 Datasheet spec is for 192mHz input max, but they seem to be working ok most of the time at 384mHz.
The clock spec on the ES9023 is 50mHz max. OTOH, there was a report in a thread here that the ES9018 worked ok with SAW clocks above the 100mHz max, so that MIGHT translate to the ES9023 too.
I'll be very curious of the results if someone tried a clock higher than 50mHz on the ES9023. It could be very interesting.
Me, I'm REALLY happy with the ES9023 in sync mode. That produced a different quality in the music, removing what I've heard referred to as the 'ESS' haze and made the sound both more direct, trebles more delicate and detailed, overall more real-sounding, and at the same time, a bit more like many other DACs, yet still retaining many of the strengths attributed to ESS DACs
LOL... and I'm going to respond here as the whole HFBD+P / Mamboberry (LS or not) / Kali / Piano thing is so intertwined that it almost should be just one giant thread
I said earlier I was not sure running the clock from the Kali over to the SCK pin (20) on the PCM5122 in the HBFD+P running in slave mode would be worthwhile. I thought about it more after reading your comments and now believe it would. While I thought that it might have been sufficient for the SCK of the PCM5122 to be fed via the on-board clocks on the HFBD+P, the issue is that they won't be aligned with the leading edges of the BCK, LRCK and DIN being fed into the PCM5122 from the Kali. DACs have a spec for this and being outside the spec will not mean good things for the sound (possibly more jitter at the input or more noise being produced as the chip tries to deal with the offset, not sure).
This MAY correlate with what I heard with my highly modified HFBD+P in slave mode on a Kali... clearly generally better, but something was not quite right. I attributed that to how I fed power to the Kali, but my first try at improving that did not fix the sound. I now have one more thing to try on that front, but I suspect the sound will still be disappointing... and I'll next try cutting the trace from the on-board clocks and running a connection from the Kali MCLK output.
The other possibility is that it was on a Kali192, I need to try it with the Kali384 I have where the output waveforms were fixed.
That's a good one. I ran my Mamboberry on a Kali192 quite a bit at first with no issues... was in ASYNC mode, didn't try anything but CD rips, but had NO issues at all. Maybe Ioan can chime in here?
10inches - jumper cable I2S.
And this MIGHT be the reason the Kali192 does not work. Remember the odd waveform anomalies out of the first Kali's? That MIGHT be made even worse by the long leads.
Greg in Mississippi
Guys ,, I'm not sure if i have try the ES9023P with bigger clock than 50 mhz ( 90% i have ...) but if I'm right in my mind ,,,the result its worst than smaller freq , but maybe its the clock ,, * i think we have try a really special clock with SUPERIUS speq from the local university ( big rack device ...) but the small poor clock have better result in sound ..
for sure the best result taken from crystal , and from anagram clock ( truly expensive .. )
* doe to one dozen pcb,s for test this period ...
for sure the best result taken from crystal , and from anagram clock ( truly expensive .. )
* doe to one dozen pcb,s for test this period ...
Last edited:
10inches - jumper cable I2S.
LOL, bet @soundcheck didn't read a word I said past the long i2s pic... think I mentioned it's just test several times, no critical listening was done like this.
I'll try short connections etc. on the buggy kali192-- no big deal. Thanks!
Last edited:
Yeah, I'll stop trying to redirect... all interrelated here and other forums.
Some news... I had gotten two more Uptone Audio LPS-1's yesterday. I spent most of the afternoon/evening setting up energizing supplies (2x 12V/1.2A constant minimum). I re-worked an old (and pretty massive) dual linear supply I used to use to provide ATX24 (via a PicoPSU) and P4 to run a motherboard-based audio PC (and not to be confused with my fully-linear ATX24/P4 supply). I know, TMI! But what do you expect from me?
Anyway, after testing them last night and giving them some soak time, I tried them on my modified Mamboberry and modified HFBD+P today.
On the Mamboberry, I had been running it with 2 LPS-1s, 1 on the main DC in to the board and the other on the Kali. Using those with it in sync mode has been working very well. I tried the 2 additional ones on the Pi and the FMC (Fiber Media Converter) on the Pi-side of my network cable. Interestingly, for as much of a difference 2 LPS-1s made on the MB and Kali, adding these two more were not large improvements. The one on the Pi made more of a difference, calming down what Vasilis described as: "The RPI have some emphasis in hi freq , special in very expensive systems sound a little bit hard". With the LPS-1 on the Pi, I didn't feel any need to change out the MCLK series R for something higher than 62R and everything was a bit more distinct. BUT the sound was a bit more exciting with the linear supply I have to power the Pi.
Not quite a wash, but not the overwhelming improvement I got when I put the first 2 LPS-1s on the MB & Kali.
On the FMC, it was just a little cleaner with the LPS-1, not enough to justify using one there.
But then I tried adding a 3rd LPS-1 to my modified HFBD+P setup. Before I had fed the two add-on regulators with their own LPS-1 while the Kali was fed either from the raw DC from the R-Pi's linear supply (just ok) or from a separate linear (slightly better). As I said above, I thought the HFBD+P in slave mode on a Kali showed some improvements, but even with the separate linear supply on the Kali did not sound right.
With a 3rd LPS-1 powering the Kali, it sounded right! Very good in every way, placing it above the Mamboberry (in sync mode) in the same way it is better than the MB when both are powered from my linear supplies.
I'm sure I'll get a lot of cr*p on using $1,200 USD of add-on supplies to power a modified $45 USD DAC (not to mention the roughly $200-$300 equivalent in the materials and time of the mods to the DAC board). I'm ok with that.... it sounded VERY good, very exciting, engaging, and did a great job of communicating the sense of the music.
My systems do not have to make anyone else happy but me. And both of these do!
At this point I now question my 2nd thoughts on feeding the Kali MCLK to the PCM5122 SCK (pin 20) input. All of the 'it didn't sound right' is now gone).
Of course this is still a bit early, I will revisit over the next couple of days AND also add the 4th to power the Pi while using the HFBD+P and see what it sounds like on that DAC.
My takeaway from this (and also based on the comments from John Swenson and Alex Crespi of Uptone AND those of the users powering their gear with the LPS-1) is that the interface just upstream of the DAC AND the DAC are where you'll get the most benefit from the best supplies you can manage. That doesn't mean skimp on the other supplies... but make sure you have your best in those two locations.
Flame suit on!
Greg in Mississippi
Anyway, after testing them last night and giving them some soak time, I tried them on my modified Mamboberry and modified HFBD+P today.
On the Mamboberry, I had been running it with 2 LPS-1s, 1 on the main DC in to the board and the other on the Kali. Using those with it in sync mode has been working very well. I tried the 2 additional ones on the Pi and the FMC (Fiber Media Converter) on the Pi-side of my network cable. Interestingly, for as much of a difference 2 LPS-1s made on the MB and Kali, adding these two more were not large improvements. The one on the Pi made more of a difference, calming down what Vasilis described as: "The RPI have some emphasis in hi freq , special in very expensive systems sound a little bit hard". With the LPS-1 on the Pi, I didn't feel any need to change out the MCLK series R for something higher than 62R and everything was a bit more distinct. BUT the sound was a bit more exciting with the linear supply I have to power the Pi.
Not quite a wash, but not the overwhelming improvement I got when I put the first 2 LPS-1s on the MB & Kali.
On the FMC, it was just a little cleaner with the LPS-1, not enough to justify using one there.
But then I tried adding a 3rd LPS-1 to my modified HFBD+P setup. Before I had fed the two add-on regulators with their own LPS-1 while the Kali was fed either from the raw DC from the R-Pi's linear supply (just ok) or from a separate linear (slightly better). As I said above, I thought the HFBD+P in slave mode on a Kali showed some improvements, but even with the separate linear supply on the Kali did not sound right.
With a 3rd LPS-1 powering the Kali, it sounded right! Very good in every way, placing it above the Mamboberry (in sync mode) in the same way it is better than the MB when both are powered from my linear supplies.
I'm sure I'll get a lot of cr*p on using $1,200 USD of add-on supplies to power a modified $45 USD DAC (not to mention the roughly $200-$300 equivalent in the materials and time of the mods to the DAC board). I'm ok with that.... it sounded VERY good, very exciting, engaging, and did a great job of communicating the sense of the music.
My systems do not have to make anyone else happy but me. And both of these do!
At this point I now question my 2nd thoughts on feeding the Kali MCLK to the PCM5122 SCK (pin 20) input. All of the 'it didn't sound right' is now gone).
Of course this is still a bit early, I will revisit over the next couple of days AND also add the 4th to power the Pi while using the HFBD+P and see what it sounds like on that DAC.
My takeaway from this (and also based on the comments from John Swenson and Alex Crespi of Uptone AND those of the users powering their gear with the LPS-1) is that the interface just upstream of the DAC AND the DAC are where you'll get the most benefit from the best supplies you can manage. That doesn't mean skimp on the other supplies... but make sure you have your best in those two locations.
Flame suit on!
Greg in Mississippi
I looked at the Subbu thread and also had a chat with inmate Eldam (he's been quite active on the Subbu thread) . And... ....I reread Gregs prose.
After that I tried some small HW tweaks over the weekend. Took me 15minutes.
(The reading took me two hours)
The "NEG" (pin 11) and the "AVDD" (pin 5) seem to belong into the "Tier1" modification targets.
Vasilis uses 4.7uf on NEG and 2*4.7uf electrolytics on AVDD.
The Subbu folks pretty much all vote for not using electrolytics on NEG. I think Greg followed that route.
And most comments I read commonly agree to provide AVDD with plenty of quality
energy up 2 570uF --- after the regulator.
I wanted to use parts that I have around. And I found some high potentials.
1. "NEG"
Inmate Eldam told me about his recommendation to try a Wima MKS2 (<4.7uF ) as 2nd or 3rd best choice after using his No1 champion - a 1uF Blackgate N.
Luckily I found -- beside numerous MKPs -- a 3.3uF Wima MKS2 in my junk box.
(Guess what: I today found some 0,47 BG-Ns. I might try these later )
2. "AVDD"
I do have some 470uF/100uF BG-Ns around. But these are pretty bold.
Usually with BGs you also get along with smaller values quite well.
Beside that ESS recommends just 1uf.
I then mounted 2*33uf BG-Ns (N = non-polars) in "anti-parallel" config and also attached a 0.1 BG-N to it.
So far so good.
Preliminary impression.
I'd pretty much support calling these mods "Tier1" mods.
The laid back, rather boring, character in comparison to before disappeared.
I earlier criticized rather flat/low-impact percussion instruments. That's much better now.
Reverberation of bells and drums increased pretty obviously. Low level stuff such as textures increased.
Now you'll be able to distinguish drumheads by brand. ( Ok. Let's not overdo it - )
Bottom line. Things absolutely went into the right direction.
Thx Greg for inspiring me to heat up the soldering iron.
(And this proves again - to me a at least - that you can't fix from the outside ( e.g. by using overpriced powersupplies) what's already compromised from the inside.
Enjoy.
After that I tried some small HW tweaks over the weekend. Took me 15minutes.
(The reading took me two hours)
The "NEG" (pin 11) and the "AVDD" (pin 5) seem to belong into the "Tier1" modification targets.
Vasilis uses 4.7uf on NEG and 2*4.7uf electrolytics on AVDD.
The Subbu folks pretty much all vote for not using electrolytics on NEG. I think Greg followed that route.
And most comments I read commonly agree to provide AVDD with plenty of quality
energy up 2 570uF --- after the regulator.
I wanted to use parts that I have around. And I found some high potentials.
1. "NEG"
Inmate Eldam told me about his recommendation to try a Wima MKS2 (<4.7uF ) as 2nd or 3rd best choice after using his No1 champion - a 1uF Blackgate N.
Luckily I found -- beside numerous MKPs -- a 3.3uF Wima MKS2 in my junk box.
(Guess what: I today found some 0,47 BG-Ns. I might try these later )
2. "AVDD"
I do have some 470uF/100uF BG-Ns around. But these are pretty bold.
Usually with BGs you also get along with smaller values quite well.
Beside that ESS recommends just 1uf.
I then mounted 2*33uf BG-Ns (N = non-polars) in "anti-parallel" config and also attached a 0.1 BG-N to it.
So far so good.
Preliminary impression.
I'd pretty much support calling these mods "Tier1" mods.
The laid back, rather boring, character in comparison to before disappeared.
I earlier criticized rather flat/low-impact percussion instruments. That's much better now.
Reverberation of bells and drums increased pretty obviously. Low level stuff such as textures increased.
Now you'll be able to distinguish drumheads by brand. ( Ok. Let's not overdo it -
)Bottom line. Things absolutely went into the right direction.
Thx Greg for inspiring me to heat up the soldering iron.
(And this proves again - to me a at least - that you can't fix from the outside ( e.g. by using overpriced powersupplies) what's already compromised from the inside.
Enjoy.
A friend of mine donated 3*4.7uF BG-N caps.
They arrived yesterday. And they were soldered in yesterday.
On NEG I attached 2*4.7 in Super-E. (replacing the Wima MKS2 3.3uF)
On VREG I attached 1*4.7
Inmate Eldam told me that his favorite would be Blackgate on NEG.
I do have the impression - even after just 1 day - that this change
lifts the DAC one class up.
The MKS2 was not that bad. First it sounded a bit metallic/edgy.
This became better after a number of hours break-in.
The Blackgates though are a different story. The level of details increased even further.
Timbres are even more present. Really nice.
The Subbu V3 folks like the Vishay SAL-RPM type caps in these regions. (Blackgates are just not available ) Never tried them myself. No clue how close these get.
The only cap I havn't touched yet is the CN-CP (PIN 9-10). I'll look into that next.
The sync-clock mod must wait.
Enjoy.
They arrived yesterday. And they were soldered in yesterday.
On NEG I attached 2*4.7 in Super-E. (replacing the Wima MKS2 3.3uF)
On VREG I attached 1*4.7
Inmate Eldam told me that his favorite would be Blackgate on NEG.
I do have the impression - even after just 1 day - that this change
lifts the DAC one class up.
The MKS2 was not that bad. First it sounded a bit metallic/edgy.
This became better after a number of hours break-in.
The Blackgates though are a different story. The level of details increased even further.
Timbres are even more present. Really nice.
The Subbu V3 folks like the Vishay SAL-RPM type caps in these regions. (Blackgates are just not available ) Never tried them myself. No clue how close these get.
The only cap I havn't touched yet is the CN-CP (PIN 9-10). I'll look into that next.
The sync-clock mod must wait.
Enjoy.
Does anybody know which onboard LDOs are used on the Mambo?
#################################
Update
Yesterday I heated up the soldering iron again.
What was bugging me, was the "common" Subbu project opinion/experience that "NEG" should not be fed with >4.7uF. Things would get worse at values larger then that.
Then I also read a comment stating 10uF on VREG wouldn't be a bad idea.
I then swapped the 2*4.7uF NEG with the 1*4.7 I had on VREG. A 5 minutes exercise.
Guess what!?!? Sound changed again.
Meanwhile I'm having a hard time to judge what's wrong or right.
Now it sounds somewhat less powerful (lower NEG capacity?) . A little less natural.
However. It doesn't sound wrong either.
Dimensions seem to be more balanced and steady now. Still a lot of details and timbres as before.
Hmmh. I'll give it some more days to break-in. 20-30h playback should do.
This cap swapping exercise is exhausting and kind of driving me crazy. (It's well worth it though)
##
Then there was that CN-CP question? What is this pin ?
I guess it's sometimes also called Cfly/Cpump!?!? E.g AKM also has these pins on DACs.
1uF is the typical value everybody suggests. Low ESR types seem to be a must.
The TI recommendations seem to be in line what the Subbu folks use.
Something like X7R 0805 (according to Eldam) are recommended.
I suppose this cap should have an effect on sound.
If I understood correctly: The quality (e.g. ripple) of the negative voltage used for the output opamps might depend on it. And the CN/CP cap is directly involved - I guess.
Nobody mentions electrolytics at this spot. And than neither mentions it at a size of 4.7uF as found on the Mambo!?!? I guess Vasillis has a reason for that choice.
Not sure if a BG-N 1uF with 0.1COG ( still have these around) would be the right choice here. Might be better then the Mambo default.
Any other recommendations are appreciated.
Thx.
#################################
Update
Yesterday I heated up the soldering iron again.
What was bugging me, was the "common" Subbu project opinion/experience that "NEG" should not be fed with >4.7uF. Things would get worse at values larger then that.
Then I also read a comment stating 10uF on VREG wouldn't be a bad idea.
I then swapped the 2*4.7uF NEG with the 1*4.7 I had on VREG. A 5 minutes exercise.
Guess what!?!? Sound changed again.
Meanwhile I'm having a hard time to judge what's wrong or right.
Now it sounds somewhat less powerful (lower NEG capacity?) . A little less natural.
However. It doesn't sound wrong either.
Dimensions seem to be more balanced and steady now. Still a lot of details and timbres as before.
Hmmh. I'll give it some more days to break-in. 20-30h playback should do.
This cap swapping exercise is exhausting and kind of driving me crazy. (It's well worth it though)
##
Then there was that CN-CP question? What is this pin ?
I guess it's sometimes also called Cfly/Cpump!?!? E.g AKM also has these pins on DACs.
1uF is the typical value everybody suggests. Low ESR types seem to be a must.
The TI recommendations seem to be in line what the Subbu folks use.
Something like X7R 0805 (according to Eldam) are recommended.
I suppose this cap should have an effect on sound.
If I understood correctly: The quality (e.g. ripple) of the negative voltage used for the output opamps might depend on it. And the CN/CP cap is directly involved - I guess.
Nobody mentions electrolytics at this spot. And than neither mentions it at a size of 4.7uF as found on the Mambo!?!? I guess Vasillis has a reason for that choice.
Not sure if a BG-N 1uF with 0.1COG ( still have these around) would be the right choice here. Might be better then the Mambo default.
Any other recommendations are appreciated.
Thx.
Last edited:
MBLS regs are the L5907. Already a step up from the TPS7A49 final reg and TPS7A900 initial reg in the original MB.
Good you've gotten your iron hot again and are getting good results. BG's should be very good here, one reason I didn't dig into my stock of BG caps to use them on my MB mods was to use parts currently available so anyone could copy (and also to use ones whose current street prices are less than the DAC card... not that I am opposed to using higher-cost components, but this IS an ES9023-based DAC HAT and there's only so much potential there).
For good low ESR (and low lead inductance), try some of the SMD film caps. I have had very good results with the Panasonic PPS caps. The 1uf/16v unit in the ECP-U(A) line is very economical @ about $1 USD per cap in small quantities. I use these combined with a 33uf/16v Black Gate N as my go-to low-voltage coupling cap where space is at a premium.
I used 1 of the 1uf/16v Panasonic PPS across the CP-CN in my mod. Not that my mod choices are designed to be definitive, but IMHO that is a good choice there.
Greg in Mississippi
Good you've gotten your iron hot again and are getting good results. BG's should be very good here, one reason I didn't dig into my stock of BG caps to use them on my MB mods was to use parts currently available so anyone could copy (and also to use ones whose current street prices are less than the DAC card... not that I am opposed to using higher-cost components, but this IS an ES9023-based DAC HAT and there's only so much potential there).
For good low ESR (and low lead inductance), try some of the SMD film caps. I have had very good results with the Panasonic PPS caps. The 1uf/16v unit in the ECP-U(A) line is very economical @ about $1 USD per cap in small quantities. I use these combined with a 33uf/16v Black Gate N as my go-to low-voltage coupling cap where space is at a premium.
I used 1 of the 1uf/16v Panasonic PPS across the CP-CN in my mod. Not that my mod choices are designed to be definitive, but IMHO that is a good choice there.
Greg in Mississippi
Thx Greg. I'm gonna buy some of these Pana PPS soon.
###################################################
Next thing I'll try is the MCLK sync mod.
Plan is to do it as below.
BTW. I2S damping Rs are supposed to be located as close as possible to the transmitter. That's why I plan to move the 51R.
Any comments?
(Not sure why Vassilis puts the 3 I2S Rs into the center of the board instead of putting them closer to the pin-header. )
I think the clock LDO can be turned off by desoldering just the "EN" pin - correct?
There's also a 3.7V version of that LDO. Might work as drop-in replacement for the DACs 3.3V LDO!?!?
###################################################
Next thing I'll try is the MCLK sync mod.
Plan is to do it as below.
BTW. I2S damping Rs are supposed to be located as close as possible to the transmitter. That's why I plan to move the 51R.
Any comments?
(Not sure why Vassilis puts the 3 I2S Rs into the center of the board instead of putting them closer to the pin-header. )
I think the clock LDO can be turned off by desoldering just the "EN" pin - correct?
There's also a 3.7V version of that LDO. Might work as drop-in replacement for the DACs 3.3V LDO!?!?
Attachments
Last edited:
)Hi,
Try to try this particular 470 uF one (this ref only) at the feet of the AVCC pin : http://www.mouser.fr/ProductDetail/Nichicon/PLE0J471MDO1/?qs=sGAEpiMZZMsIwzbKW1rlgXFJKKS9VEOcfjcwhsHbqxk%3d
stunning in the mid-treb clarity without harchness ! while being a ittle thin in the bass but change with the main supply you have before ! Beats the red caned Nichicon rr : far better !
An other one less exceptional but often very neutral is the 100 uf: http://www.mouser.fr/ProductDetail/...GAEpiMZZMsIwzbKW1rlgbY%2bFZWf3p8Ji2/MLwOkXg0=: : more balanced (les thin)
olds oscon SP 820 uF in 4.5V is also stunning here (not the very less good actual SEPF and the one to be avoided here at avcc : the SEPC )
YMMV, and the results in relation tot the main PS as said Greg.
the MKS2 Wima 1 uF to try for NEG is the 2.5 mm pitch leg, which gave better result than the normal 5 mm pitch legs as if it was two different cap ! (the littliest is smoother) : this MKS2 2.5 mm pitch 1 uF finding deserves to Jean-Paul member: one of the two drivers of the Subbu DAC project (with Subbu member himself) !
I go back in my white room
cheers
Eldam
Try to try this particular 470 uF one (this ref only) at the feet of the AVCC pin : http://www.mouser.fr/ProductDetail/Nichicon/PLE0J471MDO1/?qs=sGAEpiMZZMsIwzbKW1rlgXFJKKS9VEOcfjcwhsHbqxk%3d
stunning in the mid-treb clarity without harchness ! while being a ittle thin in the bass but change with the main supply you have before ! Beats the red caned Nichicon rr : far better !
An other one less exceptional but often very neutral is the 100 uf: http://www.mouser.fr/ProductDetail/...GAEpiMZZMsIwzbKW1rlgbY%2bFZWf3p8Ji2/MLwOkXg0=: : more balanced (les thin)
olds oscon SP 820 uF in 4.5V is also stunning here (not the very less good actual SEPF and the one to be avoided here at avcc : the SEPC )
YMMV, and the results in relation tot the main PS as said Greg.
the MKS2 Wima 1 uF to try for NEG is the 2.5 mm pitch leg, which gave better result than the normal 5 mm pitch legs as if it was two different cap ! (the littliest is smoother) : this MKS2 2.5 mm pitch 1 uF finding deserves to Jean-Paul member: one of the two drivers of the Subbu DAC project (with Subbu member himself) !
I go back in my white room
cheers
Eldam
Last edited:
Sorry for my typo, regulators on MBLS are LP5907. I looked but didn't see 3.7v in SOT-23 package. Please share source.
Disconnecting the enable pin from power will disable the reg. My choice would be to cut the trace just outside of the lead, then one can scrap off the solder mask past the cut and do a quick solder bridge to re-enable.
On SMD film caps, I know other manufacturers make them too with different trade names for the acrylic film.'PPS' is a film made by Toray, notice that current Panasonics don't always specify PPS... when you order do confirm they are not mylar, which are not as good.
I have also seen 'PML' film specified for Rubycon's versions, which can be had in up to 22uf for a good sum! Also have some from Wima and CDE and have seen announcements of similar from Illinois Capacitor. In the spirit of the comments from Eldam on recommendations from the Subbu thread, it could be fun trying all the varieties! I guess I need to read that thread sometime soon.
Will be very interested in your comments on sync mode.
Greg in Mississippi
Greg.
Thx for pointing to the SOT23-5 issue. I missed that one.
There's a SOT-23 3.6V LP2985 but that one has 30uV noise.
Compared to the 10uV of the LP5907 it's probably a bit on the high side.
Probably a better choice:
The Subbu V3 folks seem to use Microchip MIC5205-3.6YM5-TR (3.6V). It's SOT23-5.
But that one seems to work ultra low noise only with a 470p Cbypass on the gnd pin to gnd.
Not 100% a 1:1 drop-in. But I think I'd try that one.
What 130k R as replacement for the 220k would you guys go for?
######
I got the funny feeling that nothing will be left from the Mamboberry DAC when I'm finished
with all this.
Probably we should start building our own HAT. A HAT that is just clocked by the Kali MCLK .
Perhaps we should check if Jean-Paul is interested in a Subbu-HAT project !?!?
Thx for pointing to the SOT23-5 issue. I missed that one.
There's a SOT-23 3.6V LP2985 but that one has 30uV noise.
Compared to the 10uV of the LP5907 it's probably a bit on the high side.
Probably a better choice:
The Subbu V3 folks seem to use Microchip MIC5205-3.6YM5-TR (3.6V). It's SOT23-5.
But that one seems to work ultra low noise only with a 470p Cbypass on the gnd pin to gnd.
Not 100% a 1:1 drop-in. But I think I'd try that one.
What 130k R as replacement for the 220k would you guys go for?
######
I got the funny feeling that nothing will be left from the Mamboberry DAC when I'm finished
with all this.
Probably we should start building our own HAT. A HAT that is just clocked by the Kali MCLK .
Perhaps we should check if Jean-Paul is interested in a Subbu-HAT project !?!?
Reread this Subbu thread discussion.
It seems that the MIC LDO is not as low noise as the TI in 3.3.
Another inmate brought up that this rather high noise would lead to 16 bit resolution max. Interesting angle
to look at this.
It seems though that the 3.6V + higher noise MIC would still be the prefered choice over 3.3V and lower noise on the Subbu V3.
My mravcla TPS7A4700 comes with around 4.17uV. It's better then everything else discussed for now (on paper) . And this regulator also has an inductor on the secondary side. That's what jean-Paul suggests and adds the Subbu anyhow. As I thought earlier I might try to use that reg feeding 3.6V straight to the DAC instead of using it as pre-reg.
It seems that the MIC LDO is not as low noise as the TI in 3.3.
Another inmate brought up that this rather high noise would lead to 16 bit resolution max. Interesting angle
to look at this.
It seems though that the 3.6V + higher noise MIC would still be the prefered choice over 3.3V and lower noise on the Subbu V3.
My mravcla TPS7A4700 comes with around 4.17uV. It's better then everything else discussed for now (on paper) . And this regulator also has an inductor on the secondary side. That's what jean-Paul suggests and adds the Subbu anyhow. As I thought earlier I might try to use that reg feeding 3.6V straight to the DAC instead of using it as pre-reg.
It seems you need a Subbu-HAT my friend Will need to know HAT dimensions etc. For now I think you only need ES9023, its 3.6V PSU and pads for a 3.3V PSU and an XO (if people don't use the XO like with Kali).
Will check with my partner in audio crime if he joins in for USA/Canadian members and a possible GB.
Will need to know HAT dimensions etc. For now I think you only need ES9023, its 3.6V PSU and pads for a 3.3V PSU and an XO (if people don't use the XO like with Kali).Will check with my partner in audio crime if he joins in for USA/Canadian members and a possible GB.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.