looks like you are using too much heat, youve started to burn the flux
the idea of C1 is for it to be as small physically as you can manage, it looks like you have the makings of a resonant peak there. you guys who may not have done much SMD should really grab yourself some separate solder flux.
the idea of C1 is for it to be as small physically as you can manage, it looks like you have the makings of a resonant peak there. you guys who may not have done much SMD should really grab yourself some separate solder flux.
Last edited:
I don't have any way to control this - basic irons only - so it is what it is I'm afraid.
I understand the size affects the properties so it makes sense that it needs to be as physically small as possible, hence the 0603 underneath the 0805, but is an 0805 really going to cause a problem ? Or is it the values I've used ?
Thanks,
Tom
For extermer only, (try at your own risk, es9018 can die immediately and 99 lost)
I just accidently feed 5v to to the ES9018 DVCC and AVCC
it's after testing the new AK4399 PCB and reconnect the power supply directly to the next version ES9018 board and forget to change the output voltage back to 3.3.V (AK4399 uses 5V DVCC and AVCC...)
Originally, I think it's the new PCB version boost the sound quality much much more, yes, it noticeable, when first power on!, without notice I feed 5V to the ES9018, hopefully no smoking. after change back to 3.3V, the sound restore back to original.
According to the datasheet,
WARNING: stresses beyond those listed under "absolute maximum ratings" may case permanent damage to the device. These are stress ratings only and functional opteration of the device at these or any other conditions beyond those indicated under "recommended operation conditions" is not implied. Exposure to absolute-maximum-rated conditions for extended periods may affect device reliability
I just accidently feed 5v to to the ES9018 DVCC and AVCC
it's after testing the new AK4399 PCB and reconnect the power supply directly to the next version ES9018 board and forget to change the output voltage back to 3.3.V (AK4399 uses 5V DVCC and AVCC...)
Originally, I think it's the new PCB version boost the sound quality much much more, yes, it noticeable, when first power on!, without notice I feed 5V to the ES9018
, hopefully no smoking. after change back to 3.3V, the sound restore back to original.According to the datasheet,
WARNING: stresses beyond those listed under "absolute maximum ratings" may case permanent damage to the device. These are stress ratings only and functional opteration of the device at these or any other conditions beyond those indicated under "recommended operation conditions" is not implied. Exposure to absolute-maximum-rated conditions for extended periods may affect device reliability
you cant afford $60 temperature controlled iron? proper solder joints on a high speed (preferably) impedance controlled circuit, will improve performance much more than any tweaks of clocks and caps. it is burning because you are taking too long to make the joint and reflowing, adjusting the part position, reflowing again etc, if you add flux and reflow the joint quickly it will help.
as for the caps, arbitrarily choosing the cap values and stacking them roughly on top of each other is worse than just having a single well chosen size. parallel caps like this cause resonance and without doing the math/measurements of the trace impedance and other components its unlikely to make any improvement and likely to make things worse. the clock you have there has well chosen internal ceramic decoupling caps, by adding arbitrary (actually any) additional caps you will lower the effectiveness of the internal caps, especially the 15pf; possibly (more like probably) causing a resonance
this position in particular is not the place to be experimenting blind, its one of the more critical decoupling caps in the build, but if any, just put a single 10-100nf np0 0603 cap or none at all.
yes adding on top of one another of course one effects the other, it puts them in parallel. the point of small caps is lowest inductance/impedance and shortest loop area, the large cap will ruin all of these specs. the 15pf is too small (too close to internal cap value but not the same)and the 1µF is too large, neither should really be there IMO.
as for the reg bypass caps, they'll be fine with the ceramics. they re designed for that as most modern ldos are. the lm317/337 have a resonant peak in the output impedance that can be excited by the use of very low impedance caps, the ones on the board are designed to avoid this, but make them all the same value, dont choose random stepped values thinking it will cover all frequencies when you dont know the interactions.
diyinhk:
was that a recommendation to try 5v AVCC as long as its at their own risk? then post what is effectively a disclaimer?
as for the caps, arbitrarily choosing the cap values and stacking them roughly on top of each other is worse than just having a single well chosen size. parallel caps like this cause resonance and without doing the math/measurements of the trace impedance and other components its unlikely to make any improvement and likely to make things worse. the clock you have there has well chosen internal ceramic decoupling caps, by adding arbitrary (actually any) additional caps you will lower the effectiveness of the internal caps, especially the 15pf; possibly (more like probably) causing a resonance
this position in particular is not the place to be experimenting blind, its one of the more critical decoupling caps in the build, but if any, just put a single 10-100nf np0 0603 cap or none at all.
yes adding on top of one another of course one effects the other, it puts them in parallel. the point of small caps is lowest inductance/impedance and shortest loop area, the large cap will ruin all of these specs. the 15pf is too small (too close to internal cap value but not the same)and the 1µF is too large, neither should really be there IMO.
as for the reg bypass caps, they'll be fine with the ceramics. they re designed for that as most modern ldos are. the lm317/337 have a resonant peak in the output impedance that can be excited by the use of very low impedance caps, the ones on the board are designed to avoid this, but make them all the same value, dont choose random stepped values thinking it will cover all frequencies when you dont know the interactions.
diyinhk:
was that a recommendation to try 5v AVCC as long as its at their own risk? then post what is effectively a disclaimer?
Last edited:
Electrical circuit design is engineering not high school trial and error science experiment ... learn then build not the other way.
I don't want to sound like a grumpy but I will but I think this has to be said. There is much that can be learned before building these boards and you will not get close to the datasheet specifications of these DAC chips without very close attention to detail and electrical circuit knowledge. There is a reason that these boards are much cheaper than alternatives, they require particular thought on the builders behalf.
Last edited:
Impedance controlled irons? Why haven't I seen those yet? You mean there's a special iron for 50R, 75R and 110R balanced?
you obviously caught the post quickly before I caught the typo, see the edit, it looks as if you are the one speaking nonsense now
you should have quoted my post while it was there lolThere's never only one way to learn and there's not a soul alive who knows it all.
If you have a specific piece of assistance or advice you'd like to offer about building this board, the components, etc, everyone here would like you to share it.
General criticism such as the above lower the tone and do no-one any favours. I hope that's just a bad day you're having. It's Friday. Lighten up mate.
Frankly, I have had a great day and beer is not in short supply so nothing to complain about for a Friday night!
My advice is clear and not directed at you personally. It is intended as general advice to anyone in the diyaudio community that is considering undertaking a relatively expensive project to construct an ES9012/8 DAC with these PCBs with the misconception that they will save money.
Learn about decoupling and circuit design; read the datasheet(s); learn to use proper SMA (surface mount assembly) techniques and equipment.
I personally work with power systems for my day job, I am far more comfortable dealing with M.VA (capital M) many kV (kilovolts) and I readily admit the gaps in my knowledge at design of these circuits. This might as well be civil engineering for the amount of similarity between my professional qualifications and this as my hobby.
This stuff is not easy and not something that can be designed by simple and easily articulated 'rules of thumb'. Expecting hand holding by the community to provide support for every component selection is arrogent and presumptuous.
There are reasons why the ackodac/TPA/musicaltech/qnktc pcbs and designs (that target these same DAC chips) have an initial cost that is more than what you're using here. I cannot sugar coat the downside of having a PCB that is designed quickly for cheap manufacture and that is provided without schematic/BOM. IMO the target audience is not the my-first-sma-project crowd. I simply do not have the time to start from the begining and provide the education service. I also don't believe I should apologise for that. I do help where I can. I believe there is as much value (or more) in my (or anyone else) posting that this approach will not reach a level of performance that will be commensurate with your expectations of the performance of the end result, if you are constructing this DAC based on the subjective reviews of the aforementioned designer's work. If you are simply constructing this DAC as a learning experience and fully intend to dispose of the result and move on to something better then sure, though I think that is an expensive way to learn.
Here is the essence of my advice, if your vendor is recommending exceeding datasheet specifications by 50%-66%, in contrast to dire warnings from the chip manufacturer, you should ask him what the warranty that he offers if you follow this suggestion!
Chris
yes, hopefully at the same time =) hochopeper may seem a bit harsh here, but he does have a point and I do not believe he intended to single you out.
What is the point of starting with a blank board and choosing the parts if you dont use that opportunity to make educated choices before applying the parts?. what I mean by that is thinking about why a particular part value/type may be chosen and try to apply that knowledge. just throwing stuff at the board to see what sticks; particularly as you have no way to measure the result with any accuracy; is not an effective use of your hobby time and money. You will find yourself in the same position next time, just a little bit poorer and wearier.
he did not suggest knowing it all, quite the opposite, he suggested you start at the beginning as we all did.... but with the recommended level of research before picking up the soldering iron. That is what I suggest, do not get ahead of yourself and try to reinvent the wheel before you know where its wobbles are (lol ad-lib platitudes) grasshopper
trying to out-design the ESS designers when you are just learning? really?the application note for the es9008 is publicly available, just about everything applies here, not the specific IV resistor values, use the ones here, but the general advice is good. some you will not be able to achieve with this PCB, but it will give you some ideas I hope.
someone just telling you what part to put where, as in a specific suggestion, does not really teach you anything, I suspect that is why you are still making these mistakes, everyone makes them for sure, its part of learning, but thats completely different to crossing fingers and hoping for the best.
be prepared, build stock and then improve on it IMO, modifying it to take into account some of the suggestions in the 9008 app note may be a good place to start after that.
Some recommendations
now this is where I must bring up a probable mistake that would be worth modding if you can (wait till youve got it up and running first).
All of the 1.2v locations are treated as one. correct me if i'm wrong, but the 1.2v regulator connects to pin 1, pin 16, pin 20, pin 33, pin 48, pin 61? there is no other 1.2v reg on the board correct? this is where I fear the design has taken way too literally from the 9008 app note as if it applies directly to es9018....
9008 does not have 1.2v analogue supplies, 9018 does. DVDD_L and DVDD_R are ANALOGUE supplies, they power the same circuit as analogue references AVCC_L/R respectively, perhaps some bias current.. they are not simply VDD which is a plain vanilla digital supply. the fact that there is 8 bypass caps for AVCC and its brethren have 2 and hardly any room for those...could use some attention.
the digital pins will have much higher slewrates, they need bypassing at least as much if they only have basic regulators. actually they will be more demanding.
so find a way to power DVDD_L/R pins separately if you can, youve got the datasheet glt yeah? this will actually make a significant improvement if i'm reading this correctly that there is only 1 x 1.2v supply
Last edited:
Hi,
No worries - maybe he's on the grog eh ? ;-)
I've never read about oscillator decoupling so it's pure ignorance on my part, so fair enough.
Yes, all the 1.2V are shared from the one reg and I was wondering what caps to choose because of this. The caps that sound best for analogue don't suit digital, and vice versa. So I'll look at separating the analogue and digital and creating another 1.2V supply for the analogue.
Do you know any good-for-analogue 1.2V regs ?
Thanks,
EDIT - I don't think it can be done. The routing is under the IC.
No worries - maybe he's on the grog eh ? ;-)
I've never read about oscillator decoupling so it's pure ignorance on my part, so fair enough.Yes, all the 1.2V are shared from the one reg and I was wondering what caps to choose because of this. The caps that sound best for analogue don't suit digital, and vice versa. So I'll look at separating the analogue and digital and creating another 1.2V supply for the analogue.
Do you know any good-for-analogue 1.2V regs ?
Thanks,
EDIT - I don't think it can be done. The routing is under the IC.
Last edited:
I figured as much.... something to put in your pipe to smoke later perhaps. sorry I also do not want to set a precedent, there are many designs for analogue regs here in the forum (hint, its not very dynamic or high current, you want lowest noise) and without knowing the regulator type, you cannot really make a capacitor choice.
I just accidently feed 5v to to the ES9018 DVCC and AVCC
Originally, I think it's the new PCB version boost the sound quality much much more, yes, it noticeable, when first power on!, without notice I feed 5V to the ES9018
Hi,
How long did you leave it at 5V for ? Did you check if the chip had got very hot after you switched off ? I'm very interested in finding the soa limit but not with my DAC ! Maybe you can try with an old V1 ? Try 4V for a week .... then 4.2V....etc... ? We'd all love to know at what voltage / duration it really is damaged.
Heatsinking is very important and monitoring temperatures too. There is often a threshold where performance does not increase but heat does, and then beyond that, it shuts down / fails (at least with CPUs and memory etc).
People have reported that 125Mhz is better than 100Mhz but 150Mhz doesn't work, so it seems that trial and error has found the (subjective) limit for clocking. So maybe you can find it for power supply ?
subjective? that would seem to be a pretty literal error mode to me, not working that is... OK so you want to push it to the highest voltage at the highest speed, so it can produce the highest protest and highest noise?
the ESS epoxy package is going to have a lot of thermal impedance, heatsinking will not be very effective
Lowest noise I know is the super Teddy reg. But seeing as I'm gonna have to toke on a bad bong, I'll decouple the adp151 1.2V with something like Panasonic FM.
For the voltage / clock limit, lots of people say higher voltage is better, and a few who've tried say a higher freq clock is better. I know manuf. spec safe limits for reliability, but part of that is sample variance; they factor for the worst part that still comes within spec. and discard the rest.
Anyway, at the end of the day, I want to find out for myself. I value experience as much as theory, so long as it doesn't go smokey and the bandit on me ;-)
Last edited:
Hi,
No worries - maybe he's on the grog eh ? ;-)
Yes, all the 1.2V are shared from the one reg and I was wondering what caps to choose because of this. The caps that sound best for analogue don't suit digital, and vice versa. So I'll look at separating the analogue and digital and creating another 1.2V supply for the analogue.
Do you know any good-for-analogue 1.2V regs ?
Thanks,
EDIT - I don't think it can be done. The routing is under the IC.
The Buffalo DAC also has a single 1.2v regulator. Elsewhere it was explained that the 1.2V is a "holding" voltage. I wouldn't worry about separating the 1.2V supply.
I plan to use .1 uF bypass everywhere and parallel them with 10 uF on the digital/analog 1.2V and the digital 3.3V. On the Analog 3.3V I would just use 0.1uF bypass (and maybe a smaller one in parallel).
TPS7A47 seems a good candidate for the 3.3V digital (which also feeds the ADP 1.2v)
and Salas LV/Placid/other shunt seems a good choice for the analog 3.3 which benefits from shunt regulation
Last edited:
Still waiting for parts, so in the meantime I fitted a socket for the oscillator. Positioning is tricky but it's doable. I made it low-profile by cutting off the legs, soldering bridges, and filing down so it sat flat on the board. Put solder on one pad and positioned the socket.
The angle of this photo is not so good - looks like C1 connects to the socket with solder but that's just the angle.
The other angle :
Before filing down, with a Crystek that's grown legs :
For the xo decoupling cap C1, I soldered on a 0603 15pF to the board's pads and then a 1uF on top of it.
There's quite a lot of low esr ceramic caps on the outputs of the adp's. The datasheet says match the input and output caps, which I will do, but my experience with LM317s oscillating still makes me a little uncomfortable about this. What do you guys think ?
If using Crystek CCHD-950/957, no bypass is necessary.
I am thinking of fitting something that can accommodate Ian's oscillator carrier boards. Haven't figured out how. Maybe make a frame with a piece of protoboard...
Chips are too expensive for this trial-error thing
. Since certain inputs are 5V tolerant, is possible that IC technology used can accommodate 5V. But I wouldn't go pass 4V.The speed test is reversible -either it works or the chip can't catch up, but if you kill a single gate or junction with overvoltage, you'll have to throw away the chip.
Last edited:
- Status
- Not open for further replies.