 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
In Reply to: Re: PLL jitter reduction and lock question for Dave I . or anyone else posted by tenor39 on July 23, 2000 at 20:05:54:
The ERF pin will be high for an error condition - a parity error or biphase coding violatiom or out of lock PLL. Check it with a DMM - it should be 4-5 V when not locked and close to zero when in lock. I would normally route it through a driver to a LED but you could just connect it directly to a high efficiency LED with a series resistor to ground set for about 1 mA of current or whatever brightness you want. It would be lit when out of lock (error condition). Or wire the LED and resistor to 5V so it is lit when ERF is low to indicate lock condition.I have used good epoxy molded tantalums and low leakage aluminum electrolytics but I think the biggest payoff is in the first few mfds. With the big caps you should use two filters, one with a small resistor and small capacitor and the other with a big resistor and big capacitor. Both networks will be in parallel. With a big enough resistor in series with the electrolytic (about 1K) you should still lock OK and the small value network helps the voltage on the PLL to build up quicker. The voltage on the FILT pin 20 is always positive so the cap should be oriented with + to pin 20.
So you still didn't answer whether you found it to sound better with the revised filter. This is a way to add some jitter rejection and I have found it effective in the early DACs I worked on. There are much more effective means to remove jitter but they are much more complex also. If you find this mod effective, you would probably also benefit from more care with the power distribution to the CS8414 as much of the incoming jitter is propogated to subsequent stages because of ground and power noise on this chip. An additional PLL in an isolated section using a good quality VCXO is the next step...:-) Keep me posted on your progress.
Dave
Dave-Yes, I do find a subjective improvement in sound quality with the larger caps. Would the CS8414 benefit from its own regulator? Or would choke filtering of its power work? The DAC section is on its own daughter board with its own ground plane. I've also covered the DAC's and receiver chips with lead tape connected to ground for shielding. An additional PLL? I have lots of room in the chassis. How would I build one up? I'm going to try the LED with 5V and pin 25. When it goes low is it close to ground? Thanks for my continuing ed. in digital world. What do you think of jitter reduction devices like the Monarchy DIP? They have a new one coming out that will pass 96/24 and the DD/DTS bitstream.
Mike
The analog ground (AGND pin 21) and digital ground (DGND pin 8) still need to be connected to the same ground plane but I have used separate regulators for each power input (VA pin 22 and VD pin 7). The power for VA should be as clean as possible with choke isolation and good capacitors if possible. Use care with any power or digital returns around the CS8414. Return the current from the FILT circuit directly to the AGND pin (or as close as possible) instead of the ground plane to reduce the amount of noise it sees.The lead is a good idea but not for shielding. Use about .062" thick lead (or multiple lead tape layers) coupled to the top of the CS8414 with something like Blu-tack for a very effective damping/mass loading. Works great on any analog op amps or other circuits too. No ground connection is necessary and should probably be avoided for the CS8414. I have tried it both ways and can't say the ground connection is of any use. Works great on capacitors too. I have talked about this in previous posts such as THIS ONE as well as others.
When pin 25 goes low it will be very close to ground depending on how much current it is sinking. With only one mA it should get very low.
You can build a very effective PLL using a simple IC like the 74HC4046 and a good VCXO but that is way too complex to talk about here. If you are serious, read and digest the data sheet and application notes for the 4046 and we can talk about it more later here or by e-mail.
The most effective jitter reduction is done in the DAC but external ones can be effective too. Use caution with external devices that use a sample rate convertor because much of the jitter will be encoded in the data and you can't remove it in later operations. The same type of jitter reduction circuitry used in an external device would be much more effective if built into the DAC which is what added PLL circuits do.
Dave
Dave-OK, I used a 3mfd film/100 ohm in parallel with a 820mfd/1 Kohm, and it takes about a second to lock. The odd thing is I swear the highs are now rolled off, and it seems I need to push the volume contol higher to achieve the same level as before. Am I going nuts? Separating the Vd from the Va may be next to impossible, as the traces are under the surface mount chip. The LED lock indicator worked perfectly. What's next?
Mike
I don't know Mike. That is kind of up to you. I don't know anything about your processor other than it uses the CS8414. What kind of digital input do you use, wire or Toslink? There are many methods to assure whichever one you are using is optimized to reduce jitter. But after that you would probably have to start substituting parts or redesigning circuits. Bob Fitzgerald has taken on some similar reworking of commercial DACs and some of the work was published in past issues of Audio Electronics. Many posters here believe you can make significant improvements with liberal use of highly miniaturized electrolytics so you could try that. Or you could take a different approach and ....Dave
Dave-Had you experienced the change in freq. response that I heard? My wife heard it, too, and then it seemed to disappear after about 10p.m. I have extensive power line filtering, but could this be the case of "night is right" phenomenon? I have done major power supply upgrades, including FRED's, low ESR caps everywhere with PP bypasses, chokes in the PS, LT regulators, etc. The analog stages have WIMA fim & foil caps for filters, AD 827 op-amps. The unit, A Golden Theater GTX-1, came with REAL RCA's, not those cheap PC mount pieces of tin. I replaced the DVD dig. input with a 75 ohm BNC. A TNT DIY power cord completes the set. My DVD is a heavily modified Pioneer DV-05 Elite, with the mods from the Soundstage article, plus most of the things from the AA threads. If I used a relay on a short time delay how big a cap could I use on the PLL, and would I still need the 1K resistor to slow down its charging? What should I be hearing as I increase cap size? What's the lowest freq. I can remove jitter from? Audio Alchemy used to claim jitter reduction to 10Hz after lock. Thanks again for my crash course in digital. I'm usually an analog kind of guy.
Mike
I don't ever recall experiencing a loss of highs or reduced volume due to the PLL filter change. Was it set up so you could easily install and remove the additional cap? Or was it based on memory? My experience is just that the sound becomes more open with better body and depth and bass when the capacitance is increased. I think it would be better to stay with film caps or solid tantalum though and avoid the large aluminum electrolytic cap as you will probably need to implement a time delay for consistent operation from power off and other modes. Some aluminum electrolytics are rather leaky and not very good until they have "formed" to the working voltage which is only a couple volts in this case. Maybe this is what was happening to you. Sometimes subjecting them to a higher voltage first and then discharging improves their leakage performance.Changing the receiver PLL external filter components does not provide much jitter reduction and certainly not to the level of the old AA dual PLL setup. I have made jitter measurements on the MCLK from the receiver with both the stock filter and high capacitance filter but have not been able to verify the reduction in jitter that the large sound improvement would seem to indicate. There does seem to be a change in the jitter spectrum which may account for it but I don't have the necessary equipment to analyze the entire bandwidth and my measuring setup is optimized for data correlated jitter.
I use a different jitter reduction system now so have not played with the input receiver PLL as you are doing for a couple years. At that time I had a relay with a time delay from the lock signal that either reduced the value of the series resistor or shorted it out after a few seconds. The last one used a 1000uF capacitor I think but I am not sure on the resistor. That was on my own dac. At one point I changed the values on a friend's Parts Connection DAC to 100 ohm and 100 uF tantalum with .1uF in parallel (again, values are from memory) and no relay and it had no problem. I will try to look up the schematic on my design that used the relay later and check the values but I don't think there is much to be gained by increasing it more.
It sounds like you have made some extensive changes. My own designs follow a little different philosophy so I won't comment much except to say that I would experiment with the termination of the digital input if I was you, removing all the components that may be there now and directly connecting to the receiver. Then try it with the series capacitors that are probably there now. And then experiment with the resistor termination value. Sometimes it works better with a higher than 75 ohm termination value and sometimes a small cap helps and sometimes changing the input grounding (if present) helps. I have seen some DACs with a large 60 or 120 Hz jitter component due to the way ground was connected to the digital input cable shield.
Dave
Dave-My problem right now is that I'm working without a schematic. The parent company of Golden Theater, Solo Electronics, went under about two months ago, and I can't contact anyone to get a schematic. I know the dig. input looks to be transformer coupled, with a 74 series buffer chip. I didn't know you you could input right to the receiver. I'd like to see your circuit for the relay option. I was thinking of keeping the 3mfd film caps in place, and then having a larger(1000mfd?) cap come in after lock was established. Would I keep the 1K resistor on the larger cap? I'm using Panasonic HFQ's, which have low leakage/low ESR. If you'd like to share the slightly different direction you've taken please feel free to correspond by e-mail.
Mike
Mike
A good transformer input will work well but some cheaper implementations do not. I would never buffer the S/PDIF input either because all the timing variation on the data edges which becomes jitter on the PLL extracted clock will be introduced into the system at this point. Some designs do it to square up the edges for the receiver but it seems redundant to me as the CS8412/14 has a built in receiver circuit with hysteresis. It also has some internal input termination so one can experiment with different (or even no) external terminations without problem.The relay I used simply shorted out the resistor which was in series with the big cap leaving the big cap and little cap in parallel with no series resistor (or small value resistor depending on implementation). The CS8414 will lock much quicker due to its wider frequency coverage (which also leads to less jitter rejection). If you have installed pins in the board (highly recommended) you can experiment by just plugging in different combinations of cap and resistor to decide what the best values are for you.
Panasonic HFQs are one of the worst sounding electrolytic capacitors I have ever tested and I wouldn't use them anywhere close to an analog circuit or power supply but opinions do vary :-) Many seem to like them but in direct plug-in testing against many other older and less minituarized types they sound very hard and brittle with very pronounced glare. Wrapping them with lead helps immensely but other types respond much better to treatments. Just my opinion though so if you love them don't be offended.
We have probably gone on enough here and it doesn't appear anyone else is interested so why don't you e-mail me and we can talk more if desired.
Dave
Hi GuysFirst, Dave, what electrolytic caps do you like that are still available to buy somewhere? I always thought the bigger Panasonic HF caps were better myself, but never took it too far. Some swear by low ESR, but I would think dielectric absortion would be a pretty good way to judge these. I was looking at the Jung - Curl test method in TAA from '85 this week and wondered if anyone else ever uses that?
Suggestion for tenor39 next steps - upgrade the transport clock. Don't know what you have done there, but if possible, upgrade it to a separate clock with it's own regulator. Then use a flip flop (74hc74) to reclock the spdif output with this new clock. From my testing, these steps remove quite a bit of data correlated jitter that originates in the transport. I can't give you a schematic or more details, but can tell you G&D transforms sells both clocks and output reclocking boards for most transports.
More effective, but possibly more difficult to achieve, as Dave mentioned is to add a 2nd stage PLL to the DAC. I don't know anyone selling kits to do this, and am not aware of any published designs on the web either. This is very DAC dependent, and to do easily will only work with one frequency(i.e. 11.2896MHz), so if you need both DVD-A and normal 16 bit CDs, this won't work.
Another possibility is to install the transport clock in the DAC, and feed it back to the transport. Or easier yet, run the transport clock (the new high quality one) to the DAC
The idea then is to use this transport clock to reclock the critical timing line right before the DAC. Your DAC is now nonstandard, however. I think this is a pretty cool solution, though, and it is a shame transport/DAC designers have not offered this as an option (clock in / clock out ports). We see this all the time in the world of test and measurement, with counters and so on...
Bob
The HF are a little better but I was never too impressed with anything Panasonic offered although most of the testing I performed was a few years ago.By far the best easily available and inexpensive electrolytics I tested and subsequently used a lot of were the old Sprague 672 and 673 series developed originally for switchmode power supplies. They have thick aluminum cases, thick aluminum plates, internal damping material, copper leads and an epoxy end seal - all things that the more modern miniature caps lack. They are MUCH bigger than those modern caps though. Sprague has gone through a lot of changes since then and I haven't ordered any for a few years as I don't use them anymore but they might still be available.
I was fairly impressed with some of the audio grade caps in the Nichicon line at the time and obtained many, many samples for testing. Their top of the line version was very good sounding but a little soft I believe. I listened to literally hundreds of different brand and value aluminum electrolytics after inadvertently "discovering" how bad the HFQs were. At the time I didn't believe that electrolytics made much difference, instead thinking they were all equally bad so used the commonly recommended HFQ almost exclusively. I had prototyped an amp and it was all set to get boards made and I needed to finalize the transformer winding spec. At that time I realized that I would have to increase the gain stage supply voltage a few volts to meet my power spec under all line conditions and it would be close to or above the 63V cap value I was using at the time. I decided to go with 100V caps but to fit in the same footprint, the value would be somewhat smaller so I rigged up the PC board with pins to allow me to easily swap in and out different value caps to see how much it would affect the sound. I substituted some older caps with a lower capacitance value that I had on hand to simulate what the 100V caps would be and listened expecting the sound to be (hopefully) as good or maybe a little worse. But I was astounded to find they sounded much better than the HFQs. So much smoother. Back and forth I tested and always the same conclusion. I had others listen and even my girlfriend could easily tell. This led to much further investigation and buying of many surplus caps to test and cut open and examine and on and on ...
The moral of the story is they all sound awful compared to even bad film capacitors but the miniaturized ones even sound worse. Get the biggest, heaviest caps you can find and conduct your own test with pins installed in the board so you can make quick changes. You may be surprised how far we have "advanced" in the last 25 years! I have a bunch of 1000uF/15VDC Sprague 672D capacitors if someone else would like to do some testing and report their results. I have some wrapped in lead also that are in a different league. If someone thinks they have the capability to conduct an impartial test maybe we could set something up. What about you Bob? Or I could mail a set to a couple people and we could get some feedback. Limited time offer...:-)
Now days I try to design audio gear without electrolytics, at least not anywhere near the analog sections so I haven't been doing any tests on them for a few years. I have been using epoxy molded tantalums in the digital sections.
Dave
Those Panasonic HFQ caps (and FA/FC) have not the best DF/Q specs, just low ESR compared to other electrolytics. For this application, look for a cap with a low DF, often suggested for coupling use.A good film cap really does blow away electrolytics in most areas except capacitance per unit volume. ESR down to as little as .003 ohms while most electrolytics are .040 along with Tantalums, with better ones getting into the .015 area. These values are for modestly sized values. Small film caps and small electrolytics can have an ohm or two!
Bravo conrad johnson for having tube products with all film caps, even at the expense of full-power bass as in the MV45/50 amps.
Have you guys measured/tried Black Gate electrolytics? They don't give any specs in their literature, just lots of verbiage. I tried measuring them one on a bridge and got good numbers for an electrolytic. Since they take so long to break in, I'm sure I was not getting the best readings and would need to form them and leave them on the meter for a week or two and see if the numbers got better...
BTW, Sprague (now a Vishay company) makes solid aluminum electrolytics (OS-CON) too and may be a little easier to buy than Sanyo's. I just find that the selection of values limited with large values nonexistant so low impedence can't be gotten at low frequencies.
Dave -
Really interesting post. I think I heard you say film caps, perhaps even of a substantially lesser capacitance value, are far ahead of electrolytics. This jives very much with opinions from the single ended tube guys, who are using "puny" power supply caps (say, 100-200uF total) compared to typical SS PS(say,10-50,000uF). But those small value caps are BIG, film or paper in oil caps in many cases.
BTW, here are some numbers from the Jung Curl article (TAA 4/85)
I mentioned. A differential amp is set up with two caps on the two inputs, one a high quality reference, the other the cap under test. It is balanced for signal level and ESR, and a band limited pulse is used as the test signal. The "residue" is the distortion to the waveform the lower quality cap creates.
here are a few examples of the results, using a 200V polypropylene cap as the "reference".
10uF 10V aluminum electrolytic - 12% residue
10uF 25V nonpolar aluminum electtolytic - 3% residue
10uF 35V tanatalum electrolytic .9 % residue
.1uF 50V ceramic 2.4% residue
.1uF 100V polyester (mylar) .11% residue
"The residue from the test units is dominated mostly by DA errors...Not only is the basic waveform of the input pulse distorted in shape,but also note that that appreciable output is still occuring for several tens of milliseconds after the input is complete."I would love to dive into another project, but I have way too many open now. I may see If I can duplicate the test above using an FFT analyzer with an intrumentation quality differential inputs (HP3562A)
If it is super easy I'll test what is on hand, report back, and expand it..Very interesting discussion, has perhaps changed my perspective on caps a bit...
Bob
The film caps in my case and in the single ended tube guys case are preceded by large value chokes which is more work and expense but enables the building of a component such as preamp, D/A convertor, turntable PS or even low power amp with NO electrolytic capacitors. I can tell you from my own experimentation that introducing even the highest quality electrolytic into such an electrolytic-free system is like turning off the magic switch. The gains are sometimes staggering from eliminating such a large source of irritation.I think a lot of the problem is due to the delayed resonance problem that the Jung test hinted at. Energy is lost to the movement of the plates inside the capacitor setting up radial waves from the center to the outer case and then resulting reflections back into the core. The highly etched aluminum plates, which are like ceramic in the more miniaturized ones, can resonate at high and not very damped frequencies and this energy in turn causes changes in the capacitance and thus affects the voltage across the capacitor. The capacitor value versus time will be a very complex function related to the exciting voltage and all the individual properties of the parts it is constructed from. And when used in the typical capacitor input power supply, they are subjected to very high current pulses which probably excites many internal resonance modes. All conjecture on my part and no evidence to back it up except based on these ideas, capacitors wrapped in a layer of lead viscoelastically coupled to the case have proven to myself and others to be a dramatic improvement when used in power supplies and most other areas that are not "directly" handling an audio signal (power supply caps of course do supply all the audio current and so I think should be considered directly in the signal path).
Dave
Hi Dave,Like Bob and others, I've found your recent posts to be fascinating.
Have you ever tried Sanyo Oscon electrolytics? I've not seen them mentioned here, which I find surprising as they are excellent. Unbeatable impedence values, plus they have a solid electrolyte which must substantially reduce microphony. I always use them now in low-level circuits (max voltage is 25V).
You are a fan of chokes then. I'd love to hear a bit about your experiences with these PSUs. Are they superior to gyrator/regulated PSUs in your opinion?
Dan
Hi DanielHope you got the recent e-mail I sent. We are about ready to fall off the edge of my screen so better end it here or start some other thread...:-)
I haven't tried the Oscons but they do seem to solve some of the problems with coiled-plate type caps. Philips makes a comparable solid aluminum electrolyytic I believe. The sound of the solid epoxy molded tantalum is pretty good as well when compared to typical electrolytics but values are somewhat low and they get expensive fast. Not in the same league as a well made film cap though, especially my favorite type which is a solid block of epoxy impregnated paper film.
By gyrator regulated PSU do you mean capacitor multiplying type circuits using a pass transistor? I have used similar schemes in the past but in recent times I have been directly regulating across the choke resistance using shunt regulators. This is the best I have tried out of many different designs including all manner of series type regulators. Nothing but copper from the rectifiers to the circuit load.
Dave
Dave,This discussion has been VERY interesting, in case you are wondering.
Thanks for the information.
Kevin.
Dave -
Thank for the tips. Can you suggest a source for the .062 lead?
This is not a local hardware store item, is it?
I know the blue tac is easier to find.
Bob
Hi BobI was using a lot of it on a past project so bought a big sheet at a supply house. They do carry .031" lead flashing that is sold by the pound at normal hardware stores such as Orchard Supply in the SF Bay Area and it works just as well. Just use as many thicknesses as you need. I was originally using it to wrap electrolytic capacitors for which it is extremely effective because the mass working in conjuction with the visco-elastic coupling helps to remove the mechanical energy that is internally generated and really smooths out the sound of a good electrolytic such as the old Sprague 672 series.
As a fluke we tried it on some op amps in a CD player and were startled at the difference and this in turn lead (pun intended) to many other uses. I saw mention later in Audio Amateur of similar results obtained by others. The uses are really endless. I have even built (and still use) amazing sounding cables and interconnects using high lead content solder as the conductor that are the best I have heard but that is another story.
Dave
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
