Terry,
Understood. I have though about using the lowest frequency audio clocks in the audio frequency families, and then multiplying them. Unfortunately, there is usually more than just multiplier jitter that has to be contended with. Still, it might come out a little better, practicalities considered. Maybe.
Wenzel in particular does not seem to make audio frequency clocks at all. So, if one wants 100MHz, fine, but if one wants audio frequencies no luck. They appear to be mostly aiming for military and aerospace markets. As always, you are welcome to buy what you want and share your results here. I don't think it is reasonable to try to talk mostly newbies into going off trying new experiments. For one complication, the Wenzel clocks appear to run on +-15v and put out high-ish levels into 50-ohms. Nice way to blow up a dac chip or some other chip if someone doesn't know how to interface them.
EDIT: Also, when people think they can start getting away with relying on nearly unobtainium clocks, it can cause a lot of trouble. Crane Song has had delivery problems from its 45fs clock supplier. Harder to make reliably than one might want to believe.
Understood. I have though about using the lowest frequency audio clocks in the audio frequency families, and then multiplying them. Unfortunately, there is usually more than just multiplier jitter that has to be contended with. Still, it might come out a little better, practicalities considered. Maybe.
Wenzel in particular does not seem to make audio frequency clocks at all. So, if one wants 100MHz, fine, but if one wants audio frequencies no luck. They appear to be mostly aiming for military and aerospace markets. As always, you are welcome to buy what you want and share your results here. I don't think it is reasonable to try to talk mostly newbies into going off trying new experiments. For one complication, the Wenzel clocks appear to run on +-15v and put out high-ish levels into 50-ohms. Nice way to blow up a dac chip or some other chip if someone doesn't know how to interface them.
EDIT: Also, when people think they can start getting away with relying on nearly unobtainium clocks, it can cause a lot of trouble. Crane Song has had delivery problems from its 45fs clock supplier. Harder to make reliably than one might want to believe.
Last edited:
Mark, I also want to replace the clock on my board. I found on Ebay CCHD-575X-25-100.000M, which will be better than CCHD-575-50. Cost $ 38. The Mouser will be more expensive ($ 40 for the regular version and $ 45 for 575X). I use ru.mouser.com for my country.
Tomorrow I will make measurements with an oscilloscope to find out the reason for LME heating.
I also do not know where to order LTC6655. There is Ali, but I'm afraid to get a fake (6.5 $ LTC6655 LTC6655BHMS8 3.3 LTC6655CHMS8 3.3 LTC6655BHMS8 4.096 LTC6655CHMS8 4.096 0.25ppm Noise, Low Drift Precision References-in Replacement Parts & Accessories from Consumer Electronics on Aliexpress.com | Alibaba Group). Ebay offers the same for $ 16 (1PCS LTC6655BHMS8-3.3#PBF IC VREF SERIES PREC 3.3V 8-MSOP LTC6655 6655 LTC6655B | eBay)
Tomorrow I will make measurements with an oscilloscope to find out the reason for LME heating.
I also do not know where to order LTC6655. There is Ali, but I'm afraid to get a fake (6.5 $ LTC6655 LTC6655BHMS8 3.3 LTC6655CHMS8 3.3 LTC6655BHMS8 4.096 LTC6655CHMS8 4.096 0.25ppm Noise, Low Drift Precision References-in Replacement Parts & Accessories from Consumer Electronics on Aliexpress.com | Alibaba Group). Ebay offers the same for $ 16 (1PCS LTC6655BHMS8-3.3#PBF IC VREF SERIES PREC 3.3V 8-MSOP LTC6655 6655 LTC6655B | eBay)
@pixel28
as for the clock, this one will be good enough: NDK NZ2520SD 3.3V 80Mhz 49.152Mhz 45.1584Mhz 24.576Mhz 22.5792Mhz 12Mhz Ultra low phase noise oscillator - DIYINHK
of course with a dedicated reg like LT3042 or LP5907 ...
for AVCC i would suggest something simple, just google for "ИОН-ФШ-ОУ"
as for the clock, this one will be good enough: NDK NZ2520SD 3.3V 80Mhz 49.152Mhz 45.1584Mhz 24.576Mhz 22.5792Mhz 12Mhz Ultra low phase noise oscillator - DIYINHK
of course with a dedicated reg like LT3042 or LP5907 ...
for AVCC i would suggest something simple, just google for "ИОН-ФШ-ОУ"
I tried one of those 80MHz NDK clocks from diyinhk and didn't like it. Used a Crystek 100MHz instead.
In more recent times, NDK has come out with NZ2520SDA audio clocks (A - suffix added) which they claim are an improved type for audio, but 49MHz is the highest frequency one they make in that grade. https://www.ndk.com/images/products/catalog/c_NZ2520SDA_e.pdf
https://www.ndk.com/jp/news/2016/1190948_j.pdf
In more recent times, NDK has come out with NZ2520SDA audio clocks (A - suffix added) which they claim are an improved type for audio, but 49MHz is the highest frequency one they make in that grade. https://www.ndk.com/images/products/catalog/c_NZ2520SDA_e.pdf
https://www.ndk.com/jp/news/2016/1190948_j.pdf
It was on one of the modded dac boards with a dedicated LT1763 (with C0G noise cap, per data sheet) regulator. Don't recall the rest in great detail other than I recall I didn't like the sound as much as Crystek so I removed it (which I don't like doing) and put in a Crystek. Problem fixed.
OTOH, the NDK clocks in Allo Katana (which is a synchronous mode dac) sound fine. That being said, there is more than one way to get good sound quality. Also, I have heard of other people who prefer the 49MHz and 45MHz NDK A-suffix clocks over Crystek at those frequencies. Not sure we know everything that is going on sometimes, such as with different voltage regulators, circuit layouts, bypass caps, etc.
well, that is exactly that we do not know everything that is going on sometimes, such as with different voltage regulators, circuit layouts, bypass caps, etc, so is there anything we can do about this? just to make it clear where we are heading to?
btw, thanks Mark for the tip, I listen to Steely Dan "Black Cow" every time I mod something, but measurements still rule, nevertheless
btw, thanks Mark for the tip, I listen to Steely Dan "Black Cow" every time I mod something, but measurements still rule, nevertheless
just to make it clear where we are heading to?
btw, thanks Mark for the tip, I listen to Steely Dan "Black Cow" every time I mod something, but measurements still rule, nevertheless
I would agree that measurements combined with careful listening tests are appropriate and necessary for high end audio design (we have listened to Black Cow way more times than I can recall
). Unfortunately, component manufacturers do not completely specify or control for certain properties of what they make. It would cost too much. Neither Crystek or NDK tells us what kind of power quality their products need to perform their best. I do know that noise isn't the only factor, but a lot of people have been so impressed with noise numbers, it has become all they think matters.
It seems to me there are various ways to make really good dacs, but it is a lot of work to test out every different option, so once people find a working combination they tend to stick with it for as long as possible. It's not like making a billion dollar satellite where lots of engineers keep busy analyzing every little detail. Nobody can afford that level of analysis for dacs.
Even in our little groups like here nobody tests every power supply option somebody else says they like. Its too much work, too costly, and there are too many variables.
Last edited:
There is some new ''DC Link'' KEMET PP caps with lots of capacitance for cheap on mouser that could be good for the DAC PS filtering.
They provide ESR, ripple current and inductance... WIMA doesnt state any these specs on there normal MKP range, but they do give ESR on their own ''DC Link'' caps and these KEMETs are lower ESR and much cheaper.
C4AQCBW5400A3LJ KEMET | Mouser Ireland
75uf version, about 20% better value
C4AQCBW5750A3NJ KEMET | Mouser Ireland
They provide ESR, ripple current and inductance... WIMA doesnt state any these specs on there normal MKP range, but they do give ESR on their own ''DC Link'' caps and these KEMETs are lower ESR and much cheaper.
C4AQCBW5400A3LJ KEMET | Mouser Ireland
75uf version, about 20% better value
C4AQCBW5750A3NJ KEMET | Mouser Ireland
Last edited:
yes, only 14uf of PP caps for op amp rails after you suggested the idea, which still had very good results.
It was a pair of 10uf Jantzen Standard caps that were previously for amp input coupling but removed with the idea that ''the best cap is no cap'', plus cheap 2.2uf and 2 pairs 1uf.
I have a better use for those Jantzen now so those KEMETs will be the replacements, with more capacitance this time and some for AVCC and headphone amp supply too.
It was a pair of 10uf Jantzen Standard caps that were previously for amp input coupling but removed with the idea that ''the best cap is no cap'', plus cheap 2.2uf and 2 pairs 1uf.
I have a better use for those Jantzen now so those KEMETs will be the replacements, with more capacitance this time and some for AVCC and headphone amp supply too.
Please let us know how you like the sound after the change. Others may be interested in the caps too if they turn out to work well.
Hi Mark,
I changed the opas to 5532s and got the same distorted sound. I then went back to my pcb design comparing it to the Output schematic and Voila! Stupid mistake there. I forgot to put ground between R12 & C20, and R10 & C18 on this forum's schematic. And worse, I did not connect C19 to opamp pin 1 output(is it feedback?) & C17 to output pin 7 on the other channel.
I connect those points with wires and now I got very good sound indeed albeit on one channel only. I also tried the opa1612s back, yay they seem to still work!!
The only work left now is to fix the other channel. Do you have any troubleshooting tips for that?
Thanks & Cheers.
Kay.
I changed the opas to 5532s and got the same distorted sound. I then went back to my pcb design comparing it to the Output schematic and Voila! Stupid mistake there. I forgot to put ground between R12 & C20, and R10 & C18 on this forum's schematic. And worse, I did not connect C19 to opamp pin 1 output(is it feedback?) & C17 to output pin 7 on the other channel.
I connect those points with wires and now I got very good sound indeed albeit on one channel only. I also tried the opa1612s back, yay they seem to still work!!
The only work left now is to fix the other channel. Do you have any troubleshooting tips for that?
Thanks & Cheers.
Kay.
It even works on bigger (+3A) streamer power supply and Khadas tone board dac.
3x 40uF big mkp caps on diy LT1084 regulated supply
Attachments
More of what you did for the first channel might help. The only other thing I can think of would be use you spare amplifier to check for clean sound at points in the circuit to see if the circuitry is working okay up to that point. For example, the dac ouputs, if not grounded by I/V stages should have some fairly undistorted sound on them. If the I/V opamps are working as they should then the dac output will be shorted and there won't be any sound there. But, there should be some sound at the output of each I/V opamp. You would need to use a DC blocking cap in series with the input of your amp, and start with the volume low so you don't overdrive it. Its sort of the same way we would use a scope to trace a signal flow, but using your ears instead of your eyes. It can't necessarily tell you exactly what the problem is, but it can help you narrow down where to look.
Also, don't forget to check AVCC for the channel that still has problems, since a problem could be there rather than the output stage.
Last edited:
democles,
Its very interesting you put such big, very low ESR caps so close to the voltage regulator. With high gain regulators than can cause some instability such as ringing on transients or maybe oscillation. Having some wire between the regulators and caps may help since the wire can add a bit of ESR which can help stability. It may also be better to have the caps closer to the load rather than having long wires there. Always nice to have a scope of some kind to check for such issues. For most audio work, a 20MHz scope is good enough, although for the digital and clock parts of dacs it would be better to have at least 100MHz, and higher if possible. I know they are expensive things, but one can last for many years, in some cases almost a lifetime. Viewed in that way maybe the investment can make sense.
Other than that, happy listening.
Its very interesting you put such big, very low ESR caps so close to the voltage regulator. With high gain regulators than can cause some instability such as ringing on transients or maybe oscillation. Having some wire between the regulators and caps may help since the wire can add a bit of ESR which can help stability. It may also be better to have the caps closer to the load rather than having long wires there. Always nice to have a scope of some kind to check for such issues. For most audio work, a 20MHz scope is good enough, although for the digital and clock parts of dacs it would be better to have at least 100MHz, and higher if possible. I know they are expensive things, but one can last for many years, in some cases almost a lifetime. Viewed in that way maybe the investment can make sense.
Other than that, happy listening.
democles,
Its very interesting you put such big, very low ESR caps so close to the voltage regulator. With high gain regulators than can cause some instability such as ringing on transients or maybe oscillation. Having some wire between the regulators and caps may help since the wire can add a bit of ESR which can help stability. It may also be better to have the caps closer to the load rather than having long wires there. Always nice to have a scope of some kind to check for such issues. For most audio work, a 20MHz scope is good enough, although for the digital and clock parts of dacs it would be better to have at least 100MHz, and higher if possible. I know they are expensive things, but one can last for many years, in some cases almost a lifetime. Viewed in that way maybe the investment can make sense.
Other than that, happy listening.
I mesured with 100Mhz scope,no ringing ,no oscillation.(under load).
No instability,no hiccups on playing music.
Just curious if this would work with a bigger supply of 2/3 Amps.
And it works.
I didn,t place the MKP,s right after the regulator and took care of the ESR demands of the regulator.
Sinds there are no regulators giving more than 1 ore 1,5 Amps this is a nice solution for people playing with raspberry,s and other little computers.
And keeping it on the cheap solution.
Just adding a 100uF+ MKP after the regulator,that's it.
And some other little things
