Filburt
-
Posts
843 -
Joined
-
Last visited
-
Days Won
1
Content Type
Profiles
Forums
Events
Posts posted by Filburt
-
-
So, let me get this straight, for $1500 you get the sloppy recovered clock of the PCM2707 hooked into a low end, obsolete d/a converter pushed through what is probably a basic common-base i/v all housed in a generic off-the-shelf hammond?
Sign me up!
-
-
You guys are all wrong. The hybrid brown/white noise run through caps with 75 years' burn in is the proper method. I found out from head-fi.
-
I don't really care for the idea. It just means more clutter and I'm not sure it would serve much purpose other than to create tangential discussion based on the prior edits that may contain irrelevant or inaccurate content to begin with.
-
I'd take the Silmic II over the BG, personally. On top of that, it costs a small fraction of the price (check digikey for Elna RFS).
grenert - the cheap polystyrenes are not of the same performance as something like an RTE, RTX, or (so I've heard; never tried) some of the good Philips/Siemens KS box caps. The construction of them is such that series inductance is apparently higher, I guess because it's essentially they shove leads into the film stack inside the plastic, so precision is lower and the path is longer.
edit - I see luvdunhill mentions the P/S caps too; maybe there are worthwhile then
-
So, I got a bit better look at the circuit...
One of the other problems lies in the I/V. I'm not sure if it's just that this phenomenon was not well understood at the time, but they have essentially nothing damping the current step at the inverting input, nor do they have anything damping the non-inverting input. Both of these things can end up causing the i/v to barf on transient response when using something like the PCM1702, depending on the op-amp. They do, at least, have a feedback capacitor which helps _a lot_. If you're feeling particularly adventurous, you may want to cut the traces to the inputs and wire in a 100 ohm resistor in series between the non-inverting input and ground, and something along the lines of 100-500 ohms, probably, in series with i-out to inv. I'm not sure what specific value is optimal.
I don't know if the THS4031 is especially suceptible to this problem since my DAC already has these measures in place. The AD797, though, performs very poorly without it, and in fact sometimes needs a capacitor to ground as well. The OPA627, interestingly enough, is less suceptible to this problem, and this fact makes me wonder if it is not the reason some people have reported favourable results with the OPA627 compared to some other op-amps for I/V, as I didn't find the performance particularly superlative compared to some other options.
-
Way to let the children down. Had to be that guy didn't you.
Well, I told you the same thing yesterday!
-
Good chance I won't be able to attend on that date. Sorry
-
I was suggesting soldering the adaptor directly into the board, though socketed probably would work. I have I/V and LPF both socketed on my DAC. I don't see much gain in soldering the chip directly to the dip socket like that versus using a good adaptor like the ones Aries makes (A724-ND or A744-ND from digikey). In fact, I'm not sure it would work as well, especially when we factor in the bypassing stuff that needs to be done. Input impedance at the inverting input goes up dramatically at high frequency, so milliohms in the high frequency region from the adaptor is a drop in the bucket. I don't know how much stray capacitance is going to affect it or not; some of that will depend on how good a job you do on the bypassing as well as what the rest of the circuit layout looks like (which I, obviously, can't see from these photos).
-
Okay, looks like 4570 for I/V and 5218 for LPF. The bypassing arrangement pretty much stinks; I don't see bypassing from both rails, and certainly not across the rails, so substantial alteration in that respect is probably in order. Definitely so if you have any interest in using the THS4032.
So, gameplan, I guess....
Rel-Cap RTE can probably fit and sub in for whatever they're using for the filter caps at the moment. Looks like it's probably MKS (mylar) or MKP (metallised polypropylene) based on the package shape and size. Polystyrene (aka KS) is considerably better for filter applications.
For bypass, you're probably looking at, after soldering in an op-amp (or an op-amp on an adaptor), soldering caps directly from pins 4 and 8 to ground, observing polarity where applicable. Although, inductance could become a problem, so if that does happen, you can put around a 1-4 ohm resistor between the larger rail capacitor (e.g. the electrolytic) and the rail. I'd say probably start with somewhere around 100-330uF for rail cap, and have 0.1uF in parallel. Then, put 2.2uF across the rails. I'm basing this off of some recent results I've been getting using a spectrum analyser and an oscilloscope. With certain op-amps, sometimes you really will need that resistor in series with the electrolytic, so don't count it out as a possible remedy to problems, especially since the placement of the rail caps could be such that you really have to consider the effects of lead inductance. When you bypass across the rails, do it right on the pins of the op-amp on the adaptor.
For the I/V, again I'd say use THS4032, although you could also try thr AD8022 as the rails are +/- 12V. For LPF, I've recently been getting some pretty good results with the AD797. You could try the AD8599 dual which, as far as I understand, is compensated differently but it is a similar topology. At least, it behaves like it is. It's a bit slower, the distortion is higher, but it's less cranky about running in unity gain and it's a dual, making your life easier. It is SOIC, though, so it will need an adaptor, just like the THS4032.
What's left? I guess you could change the regs if they're noisy. You'll probably need a scope to really see what's going on and make sure you don't make things worse. You could also switch out the resistors, if they're carbon film, for metal film. Looking at how small those are, you'd probably have to use something like RN50/CMF50.
I noticed it uses a YM3623. This receiver seems to be fairly infamous for jitter problems...maybe try looking into improving the loop filter if anything can even be done about that, like on the CS series chips. That won't really do much about low frequency but might help with high. I've seen adaptors for it to be swapped with CS8414, but I have no idea how well those work.
Other than that? I dunno, could swap out for PCM1702-K. I got measureable and audible results out of doing that with the Transdac.
-
I use Multicap RTX and Relcap RTE (e.g. polystyrene). KS works great for audio, but it can't take high temperatures, so it _must_ be hand-soldered, and carefully so. Still, even something like the Multicap is a lot less expensive than something like PTFE, which I'm not entirely convinced is superior.
-
If you're getting excessive noise on the rails for the op-amps, you can cut the traces to pins 4 and 8 (-/+ rails) and solder in something like a 10-100 ohm resistor in series with a wire to supply power, then give it fairly beefy bypassing to keep the impedance down (I've used 330uF + 0.1uF to ground, 2.2-10uF across the rails to good effect thus far, including checking on a scope and spec analyser). I've found this to be pretty much mandatory for getting optimal results (or, in some configurations, just being able to make the thing work) with fussy chips like the AD797 or THS4031 on a breadboard when using something other than a battery for power. If using a battery, it can sometimes be helpful to put in a small resistor in series and have good bypassing in order to try to damp inductance from the battery or its leads, if that is causing problems for you.
-
Hi, Filburt! Which do you think would be preferable?: Using a 4032, or converting a dual I/V to 4031s using a single-to-dual adapter? I know there are a lot of folks who don't care for Browndog adapters, but they have a pretty nifty dual DIP -> two single SOIC adapter that I've had good luck with before.
Also, is there any problem with increasing output coupling cap size from, say, 22uF to 47uF? This would be on a DA-500. Stock is 16V 22uF, and I have some 16V 47uF Silmics I could use there.
BTW, if anyone has technical questions about the DA-500, I have a service manual I can check.
Definitely use the 4032 instead. The problems I'm talking about that crop up with the 4032 vs the 4031 crop up in greater degree on something like the browndog dual-to-single SOIC adaptor. The 4032 is definitely a usable IC; you just have to be careful with it, just like the 4031. It's mostly just a matter of doing what is needed for the 4031, but with just that little bit of extra care on top of that. Well worth the effort, though.
I would appreciate if Filburt could post the mods for the DA-500 somewhere, so it could be documented and available publicly.Honestly, I can't remember what the DA-500 looks like on the inside. If you could give me some good pictures of the inside of it, that would be really helpful in terms of my being able to give advice on mods. A schematic of the output stage would also help, perhaps from grenert
-
Thanks for the insight on the THS4031 vs. 4032 Filburt. Had I not gone to a discrete output stage, those were where I was going next, and my player's stock I/V was setup to use duals. Of course I knew if I went to these, I would have to do alot of rework/bypassing to the stock stage to begin with.
Regarding the NPO ceramics, are COG available in this size? I forget how big they get (0.1uf? or 0.01uf?) but it would be interesting to see if you noted similar problems with these.
C0G is the same dielectric as NP0. It's just two ways of designating the same temperature coefficient. The biggest I've personally seen is 0.1uF. After some recent results using a scope and an analyser, I've started using PPS in place of NP0/C0G more often than I used to, although sometimes the ceramic cap ends up being a better choice, depending on the circuit and the components being used.
The 4032 is certainly usable; you just have to be extra careful about bypass because you want to keep it especially happy at high frequency or you could end up with garbage floating down to the audio band from the dac being unhappy (especially if it's something that works well into the MHz range, like a PCM1704).
yes, some red lights kinda went off in my head when he compared it to the bel canto3. But I'll just look on the bright side of things for now: Its a very compact dac/amp design which will be perfect for my small college dorm. I will also not have a problem reselling it for the same as I bought it because the DAC1's are very popular and in good demand on places like head-fi.Yes, I imagine it'll work fine for you. It does seem modded units somtimes have trouble selling, for whatever reason, at the same price. I'm not sure why, but I guess because people don't trust the mods.
-
I settled on the modded Benchmark DAC1 (I know you're all going to kill me!) but I have a few reasons. First, it has a decent integrated amp, so I'm not going to buy an amp until I have a lot more money to get a really nice one.
Also the guy I bought it from said that before he upgraded to the modded dac1 he actually had the perpetual tech p3a/p1a/monolithic ps setup and found the dac1 to be a clear step up. He said that compared to the stock dac1, the modded one had a clearer and wider sound. He even said that he compared it with a bel canto dac3 and found it relatively close in sound.
Lastly, there are a lot people out there who have modded (or paid to have modified) their dac1's, and if I ever choose to, it will not be that difficult to find things to do to improve the sound.
Hey, it's your money, man. I don't think the DAC1 is complete garbage; I just don't think it's a good value. It's decent stock I suppose, but it's not exactly a great platform for upgrading. There are some things that can be done, but the design overall closes the doors on some of the most effective modifications that don't involve simply replacing the whole output stage and at least part of the power supply.
By this I mean, for example, you can't do much about stuff like the analogue bandpass filter, other than changing the shape (which is not recommended unless you have a scope and a spec analyser), so no opportunity to put in something good like Polystyrene caps. It looks like they use NP0 ceramic filter caps, which work...OK. When I've checked them in filters on an analyser, they show pretty decent THD+N (maybe about the same as mylar), but I suspect the DIM products are poor. Polystyrene and Polypropylene are considerably better performing materials, but there really is no way to use them in this device as those caps are through-hole. You could maybe try PPS since that comes in SMT package, and it does an OK job (better than NP0, worse than the other two). I got a slight glimpse of some of the problems with NP0 recently but I'm still not sure of the specific problem involved. This is an unfortunate limitation of the design, as the filter design can have a substantial bearing on final performance (though not of the magnitude of the op-amps used).
If you really want to go the mod route, I'd suggest focusing on the output stage almost exclusively, other than making adjustments to the bypassing or other power related stuff to keep the devices stable. Meaning if you're not wanting to throw out the output stage design, replace the op-amps, and maybe increase the size of the bypass caps to ~100uF and use 0.1uF to ground if it isn't that already, and then put somewhere between 0.1 and 10uF across the rails, depending on the needs of the device. Still, a nagging problem with this design is options. THS4031 is probably of the best (if not the best) for i/v, but you can't use it because of the DAC1 using all duals. There is a dual (THS4032), but that does suffer from crosstalk in the MHz range and it's harder to get stable IME. The layout on the DAC1 looks like it may be suboptimal for implementing this chip, but maybe with some shrewd use of extra or modified bypassing (perhaps across the rails), it could be made stable. You may have to swap those X7R bypass caps out for PPS or NP0 as the microphonic characteristics of X7R seem to cause problems with the 4031. However, the LPF is still going to be a problem area, since most of the worthwhile chips for this are single-channel only. For buffer I guess you could try the AD8022 (if your supply voltage is not above 24V; it can be used for I/V also) or AD8599, but either may require some adjustment to the bypassing scheme.
It may be worth noting that I've seen a decent amount of mods that are probably not worth pursuing from the standpoint of value, such as messing with the clocking or replacing all the diodes or something like that. You could try to improve the regulation, but a real effective attack on that problem would involve a pretty invasive process; pretty much the same deal with anything worth doing with regard to clocking. Try not to get caught up in the mod hype and bleed out several hundred dollars into the thing; you're better off just getting a better platform at that point. This may include, even among the options you proposed, the 840c, although I still think a Parts Connection dac would deliver higher price/performance ratio.
Also, not to put too fine a point on it, but did you expect the seller wasn't going to talk up the thing he was selling?
-
I got some ATH-CK7 (earphone) for use with my iPod Touch and I'm quite happy with this combination. Good isolation, strong low end (but controlled, unlike the Sony EX series), and very good detail.
-
Forgot to mention, upgrading to K-grade on the dacs (e.g. PCM1702P-K) did improve things as well (that's what I had done in mine, now jp's
). It's about 55 bucks in parts, so you may or may not want to include that as well with respect to keeping costs down. -
There is a transdac on ebay here: SONIC FRONTIERS Trans DAC MIB - eBay (item 250267149272 end time Jul-12-08 16:12:16 PDT)
for $500
I usually see them show up for less, including on AGon. However, if you're in a hurry this might work. You could try using the "Make Offer" option. If you do decide to get one, there are some basic mods you should probably have someone do for you (e.g. op-amps, some bypassing stuff on the rails) if you can't do it yourself. THS4031 for I/V, AD825 for LPF/Buffer. Swap the 10uF bypass for the buffer for either 10uF tantalum or 100uF Panasonic FM, and parallel some 0.1uF X7R with them (e.g. soldered across the electrolytic cap's pins on the bottom). You can ask jp11801 what he thinks of my old Transdac (which now resides with him since I picked up an Assemblage DAC-2.6).
-
whoa how about the Stello DA100 Signature?
I thought the DA100 was OK but nothing especially amazing. If I recall correctly, they use kind of cheap aluminum electrolytics coupling caps on the output of the AK4395, which is already limited by its switched-capacitor output. Then it goes to a very aggressive lowpass filter using cheap JRC NE5532s, then finally to the discrete buffer stage...but at that point it's sort of like "why bother?"
Still, it sounds pretty decent. I don't think I'd shell out what it costs for one, though.
-
One of my concerns is that I will be living in a dorm starting a little over a month from now and I won't really have room for a setup like the Perpetual Technology stuff. It looks really neat though.
I also need an amplifier because I just sold of the only one I have.
I am leaning toward the modded DAC1. Although I have read many reviews claiming the DA10 is superior, that would give me an extra ~$150 vs. the Lavry to upgrade my cables, possibly the internals of the DAC1, or put $ toward an actual amplifier and not just the one onboard the DAC1.
Several of the Parts Connection dacs are as small or smaller than the DAC1. They show up on Audiogon and eBay fairly often. Look for Sonic Frontiers Transdac and Assemblage DAC-1, DAC-2, or DAC-2.6.
-
I personally would pass on the DA10. I know it's popular on head-fi, but I was seriously unimpressed with the DA10 whenever I've tried it. It has this character of being simultaneously stale yet weirdly aggressive at times; I think possibly due to the use of OP275s for I/V and buffer, which tend to have this characteristic. I don't find the 275 to have a particularly good design overall. The loop gain seems to fall off too quickly, so distortion rises pretty quickly with frequency, it has a decent amount of input capacitance which ends up inducing distortion at the input, and the VAS doesn't seem particularly linear to begin with.
I liked the Benchmark DAC1 better, but that isn't saying much. It's got kind of a bland sound to it as well. I don't know why people call this neutral. I wonder if Benchmark tried measuring IMD or DIM (TIM for instance) at >6KHz or something of that sort, or if the 'transparency' claim was based just on running some basic THD+N plots.
If value is really a concern, I'd say you're much better off looking at a good quality budget DAC from the mid-late 90s like something from Parts Connection (Assemblage, Sonic Frontiers) or Parasound (such as the D/AC-1100 or 1600) that uses one of the high-end R-2R dacs and has a solid analog stage design. Modernising the parts in the output stage can be very helpful as there has been a lot of advancement in op-amps since then (though modern dacs rarely take advantage of this) and substantial gains can be made for low cost (e.g. ~$30-50). D/A converters like the PCM1704, PCM1702, and PCM63, in my opinion and experience, offer substantially higher audio performance than the delta sigma chips used nowadays, especially versus the voltage out type (e.g. CS*, AK*, WM*), though the AD1955 is good as far as delta sigma chips go (though not impervious to poor use...e.g. DA10).
-
Sorry, the DDE was what I was talking about. I wouldn't bother with modding the DDS.
Was the DDS positioned by AA as a transport, or as a CDP? Or both? Just curious. The DDE interests me...
DDE could definitely use it. I heard that it uses a passive i/v scheme stock, which is fail. So, replacing that would probably be a good thing to prioritise, if so. Then again, probably a good thing to prioritise, anyway.
-
More like its screaming for a decent (discrete) output stage. Why fuck around with opamps?
Discrete would make more sense if it were i-out; plus it's more complicated, and far more costly, and there is more of a risk of permanent damage to the device due to its design. If I can find out what the voltage is at the rails for the op-amps, though, I can probably make a reasonably good suggestion on what to replace what look to be NJM4556s on there. Plus, the money, time, etc. involved in a discrete stage is better spent on the DDE.
grenert - No idea what their costs were or what the MSRP was. Only reason I recognised it was that I've seen the CDP-291 around a few times. You can find them at junk/thrift stores sometimes, otherwise eBay will probably have them fairly reliably.
-
Looks like a Sony CDP-291/391 put into a different chassis. For example, SONY CDP-291 COMPACT DISC CD PLAYER - eBay (item 110269440424 end time Jul-14-08 18:46:50 PDT)
The text isn't clear but it says 'High Precision D/A System' and under that it says 'Servo Stabilizer Circuit'. As you can see, put behind a red plastic window, the display would look just like that of the DDS III.
Diy SS amp with dip opamps
in Do It Yourself
Posted
Andrea?....
Anyway, you can just build a PIMETA with the LMH6321, which is a better buffer than the BUF634 anyway. Just observe the differences in pin-out appropriately.