X-24K DAC Mods

[matt7941]

Following on from the thread posted elsewhere in the forum, I have actually bought the X-24K DAC from canjunkie, yes the one in the pictures, and can report some big improvements.

I have done a full recap on the power supply side with Panasonic FCs and replaced the three tantalums down there also. The ICs and opamaps have all had their caps replaced with WIMAs or Epcos Polyester film caps, and the output caps replaced with 470uf Panasonic FCs bridged as per Mike's procedure on the X10-Dv3.

The diffference is immense. The soundstage has broadened and become more 3D, bass is deeper, tighter and faster, treble is much sweeter and separation and clarity has increased tremendously.

I had been using a Cyrus CD7Q and PSX-R until recently but I am now using this with a Micromega Stage 1 CD player of some vintage as transport, mainly for the Philips CDM 12.4 mechanism. I have also constructed a power supply based on the toroidal in the diy section of this site. I can happily say that the X-24K combo now easily beats the CD7Q on all-round ability. The tweaks are certainly worthwhile.

However, can anyone tell me if it is worth replacing the tantalum capacitors on the audio board for some Panasonic audio grade aluminium caps as I have heard the tantalums are not as accurate and have worse tolerances? Should I leave them be or is there any alternative for replacement. They are 47uf 16v rated if this cannot be gleaned from canjunkies pictures.

Thanks to Mike for a great site and to all those threads from which I have gleaned information, especially SandyK.

Thanks javascript:add("%20:D")
Cheesy

[matt7941]

Sorry, I forgot to add that the opamps have been replaced with Burr-Brown opa2134pas as opposed to the stock NE5534s.

Are there any better suited opamps anyone knows of, the AD843ANZ looks interesting.

Thanks

[Deleted]

matt7941
You could try the DIL version (not SMD, unless you put it on a converter board) of the LM4562, but it can make for a fairly hot top end in some situations.
SandyK

[matt7941]

Thanks Sandy.

Would I need to bypass the LM4562s, presumably with 0.1uf poly caps from +v to ground, -v to ground and + to - under the opamp? Also, would it be beneficial to change all 4 opamps or just the final pair?

Having said that the rest of my set up is a MF A3 integrated and Epos M15.2 speakers. This is probably on the slightly bright side of neutral as it stands, although I have tried to temper this with Audioquest interconnects and speaker cable.

Any ideas on the tantalums or is it not worth the effort?

Sorry for all the questions, its just I've recently got the soldering iron back in my hand and I'm too sad to put it down again!

Any other opinions most welcome.

Thanks, Matt.

[Deleted]

Matt
If the OPA2134 have no problems, I don't think that the LM4562 SHOULD have problems using existing bypasses.It depends on the actual PCB layout. I would try doing just the final pair initially, and see how that sounds. I would leave the tantalums as is for now. I must admit to not really liking tantalums, as I have had a few go short circuit in power supplies, even though they were well within their voltage ratings. BTW, I got both versions of the LM4562 from Digi-Key
SandyK

[matt7941]

Thanks again Sandy.

I think I might just have a go at the last two op amps. I can get them in the UK after sniffing around on t'interweb from Mark Grant Cables at a reasonable £5.50 each. (Sorry, not trying to advertise, I have no affiliation with them!). I suppose it's a case of suck it and see, if they start to oscillate I can look at bypassing.

I might also replace those tantalums and see what happens, you never know it might be the final tweak that leads to hifi nirvana.

Just waiting for the bypass film caps for the power supply board to come in to stock at my local supplier and I'll get busy again. I'll check in once all the mods are done, someone else may find them useful. The X-24K has already seen a huge leap in performance so for the cost of a few (!) caps and diodes might save someone a few quid in reviving an old X-24K rather than buying a newer replacement. Plus, they might just derive as much enjoyment as me from the exercise!

Cheers, Matt.

[XTRProf]

Hi matt7941,

The OPAs will depend on the sound you like. 4562 has a clearer dilvery as compared to 2134. So depends on your musical genre you always frequent. There is no right or wrong or poorly designed OPAs here.

For me I started off trying the 2134 as recommended by the Sandyk thread here, although I had heard about that locally, as I must just start off by having something as a reference over the stock 5532 in my X-Dac V3. I found that OPA very very good for vocal, some jazz and classical music. But when it comes to rock and those pratty music, it somehow slowed down the music and I swing and tapped less of my feet and snaped less my fingers.

For 4562, as I'm presently using, the rock and pratty muzak goes back to about correct. Although not as good for especially vocal, I found that overall for all kind of muzak, it's about righter at least for my musical taste. So do experiment with the various combo and mix. As usual, YMMV.

The bypassing 0.1 uf cap is just for filtering off the high frequency and is not enough to improve the transient of the OPA as when you have a 10 uf or more there as recommended in application notes. I have experienced a relevation there when I got extreme 3 D performance from my X-Dac V3 when I used super e-caps pairs of BG and EC to boost up the local PS near the OPAs as detailed in my thread here.

As for tantalum caps, it's not my cup of tea as well. John Curl also doesn't like them. But again this depends on where the tantalum are. If for local PSs, you better changed them if it's not too costly. If for further upstream PSs, improvement will be little. However, if it's not too costly to change, then just do it as in Nike ad!

Cheers and have fun - the most important of all in diys and modes.

[matt7941]

Thanks Prof.

I certainly am enjoying this modifying lark, it does get a little addictive. However, it just gives simpletons like me a feeling that I've at least contributed something of my own to the listening experience, rather than just listening to shop bought "stock" components.

I think I will replace the 47uf 16v tantalums with the same rated Panasonic AD or AM (?) series aluminium electrolytics as they claim to be for audio use and I can get them locally for peanuts. Has anyone tried these and would they work well? They are located in the audio stages, power supply thereof presumably, of the DAC (pictured on the upper board in canjunkies' photos in the previous X-24k thread).

Thanks, Matt.

[Deleted]

For me I started off trying the 2134 as recommended by the Sandyk thread here,

To put that into context, the NE5532 were replaced in all but the output I.C. with OPA2134. The Output device was changed to an AD8066. Using a LM4562 for the last position would most likely have given a fairly similar result. It is usually best to mix the devices, rather than end up with the I.C. manufacturers "house signature" sound.
SandyK

[matt7941]

Just replaced the tantalums with some "audio grade" Panasonic Aluminium Electrolytics. Initial impressions are favourable; there appears to be more bass all across the range, better separation and bras and strings have more "bite". I'm going to use a cliche here but it does sound like a veil has been lifted. I can see why some of you don't like tantalums, they just seem to get in the way of the music.

Local supplier still hasn't got my 1uf WIMAs yet for cap bypassing! Still, its an excuse to get the soldering iron out again at a later date.

If anyone else has one of these DACs I cannot recommend modifying it in a similar way highly enough. "Stock" it is ok but not special, modified it is excellent. I dare say the X-DACv3 is probably better but I'm very happy.

It begs the question, which I know has been asked before on here, why do MF put such cheap components in a design that obviously has well thought out basics re circuits, but out of the box comes nowhere near realising its potential?

Thanks again to all who have responded, SandyK, Mike and everyone whose ideas I have incorporated.

Matt.

[Deleted]

why do MF put such cheap components in a design that obviously has well thought out basics re circuits, but out of the box comes nowhere near realising its potential

matt7941
It's all about getting the gear into the market at an affordable price. That, and the bloody beancounters that go through the designer's parts lists like a dose of Epsom Salts.
SandyK

P.S. Did canjunkie contact you ?

[leo]

May 17, 2008 22:13:02 GMT matt7941 said:

Just replaced the tantalums with some "audio grade" Panasonic Aluminium Electrolytics. Initial impressions are favourable; there appears to be more bass all across the range, better separation and bras and strings have more "bite". I'm going to use a cliche here but it does sound like a veil has been lifted. I can see why some of you don't like tantalums, they just seem to get in the way of the music.

Local supplier still hasn't got my 1uf WIMAs yet for cap bypassing! Still, its an excuse to get the soldering iron out again at a later date.

If anyone else has one of these DACs I cannot recommend modifying it in a similar way highly enough. "Stock" it is ok but not special, modified it is excellent. I dare say the X-DACv3 is probably better but I'm very happy.

It begs the question, which I know has been asked before on here, why do MF put such cheap components in a design that obviously has well thought out basics re circuits, but out of the box comes nowhere near realising its potential?

Thanks again to all who have responded, SandyK, Mike and everyone whose ideas I have incorporated.

Matt.

Tants do the job well but only if used in moderation, some of their electrical properties is better than electrolytics, use too many though and their signature dominates things especially if values is over 10uf's are used so replacing the 47uf's with electrolytics was a good move

I presume your using the blue cased panasonic ECA series? you'll probably notice the sound will change a bit in a few days once they have had chance to burn in

Does this MF dac use the PCM1716?

[matt7941]

Good evening Sandy.

Yes thanks, canjunkie did get in touch. He's a good guy, it's he who put me on to Mike and the forum. Good to hear from him.

I'm thinking pairs of opa627s (with adaptor boards) for the final opamps as a finishing touch. I was briefly thinking about fitting a new clock module as the icing on the cake but a) I don't think I have the skill or knowledge to implement it, b) I'm not sure there's the room in the X-24K to fit it in and c) I'm not sure the rewards would justify the outlay, they seem to be quite expensive.

Help, I'm becoming addicted to modding! I may need counselling.

Cheers, Matt.

[matt7941]

Good morning Leo.

Yes, they were the blue cased Panasonics. I live quite near CPC so its easy for me to pop in and pick them up. At less than 15p per cap they've made a definite positive difference. Good job they're not too expensive as there were 12 to replace. I'd already replaced the 3 on the lower supply board with FCs. It certainly looks different in there now compared to the photos canjunkie posted on the thread elsewhere in the forum. There's hardly anything left untouched apart from some of the ICs and the resistors!

I'm not sure re the PCM1716 question. I'll open her up again either later tonight or tomorrow.

Any ideas where to get some decent wire locally, and whcih wire to use, to make a new power supply cable? Maplins have some 10a rated equipment wire which I was considering using in a twisted pair. I've got the 2.1mm plugs and have made a PSU but the lead is pants, just put together to get it up and running.

Cheers, Matt.

[leo]

Good morning

Those Panny ECA's are decent caps, especially for the price

If it uses a PCM1716 theres two capacitors , one goes to pin15 and the other goes to pin20, value should be around 10uf.
I'd suggest use the best you can here, the quality has quite an effect on the performance of this chip, its right in the signal path too

Probably not worth changing the clock in this one tbh, if it uses a 4 pin crystal oscillator one thing you should do that will help is feed it a nice clean 5v supply, an oscon here would be worth trying, those are hard to beat in clock circuits

I'm afraid I'm not a cable guru, I tend to use heavy gauge copper wire for all power leads with good results, the 10a stuff from Maplin should be more than adequate

[matt7941]

Leo,

I've opened up the wee beastie and it appears the dac is a PCM1728e (BurrBrown), with an accompanying CS8414(?). The crystal is a 2 pin, not 4, at 6.144Mhz marked as SPK E8 with 2 accompanying 22pf ceramics. I don't know whether this alters anything re the clock etc. or whether it would still be money better spent elsewhere.

Matt.

[XTRProf]

May 16, 2008 20:05:44 GMT matt7941 said:

Thanks Prof.

I certainly am enjoying this modifying lark, it does get a little addictive. However, it just gives simpletons like me a feeling that I've at least contributed something of my own to the listening experience, rather than just listening to shop bought "stock" components.

I think I will replace the 47uf 16v tantalums with the same rated Panasonic AD or AM (?) series aluminium electrolytics as they claim to be for audio use and I can get them locally for peanuts. Has anyone tried these and would they work well? They are located in the audio stages, power supply thereof presumably, of the DAC (pictured on the upper board in canjunkies' photos in the previous X-24k thread).

Thanks, Matt.

Don't worry, in some time u will reach expert level and advising all brothers and sisters what to do.

Ya, those shoukd be local smoothing PS caps. Great to see that u have changed them.

[XTRProf]

May 17, 2008 22:13:02 GMT matt7941 said:

It begs the question, which I know has been asked before on here, why do MF put such cheap components in a design that obviously has well thought out basics re circuits, but out of the box comes nowhere near realising its potential?

Ya, I concur with what Alex had written in this area. I once went to a meeting many moons ago when my company was making some pcbas for a new model of product coming up. I was one of those from the manufacturing side. There were those bloody accountants around when the engineers were proposing the bom and likely cost. It was not pleasant to be interrupted by those bloody accountants now and then during the presentation.

[XTRProf]

May 18, 2008 3:26:43 GMT matt7941 said:

Leo,

I've opened up the wee beastie and it appears the dac is a PCM1728e (BurrBrown), with an accompanying CS8414(?). The crystal is a 2 pin, not 4, at 6.144Mhz marked as SPK E8 with 2 accompanying 22pf ceramics. I don't know whether this alters anything re the clock etc. or whether it would still be money better spent elsewhere.

Matt.

Err, the clock will be a worthwhile upgrade. So is changing the opas to discrete ones. So just do the rest first and enjoy the music and leave that to the later. In this way, u learn better about the improvements after the burn in period.

[matt7941]

Thanks for the advice guys, I think I will leave it alone for now barring the cap bypassing on the power supply side which is in the pipeline. Time to sit back and enjoy the music.

Now, where did I put that Celine Dion CD? (Only joking!)

Cheers, Matt.

[matt7941]

Sorry, it's a useless struggle, I just can't leave the soldering iron to go cold in my tool box. The urge is too strong.

I've been checking other forums and sources too see if the reclocking idea would be worthwhile. The balance of opinion seems to be 95%/5% in favour of replacing the simple low quality crsytal with a separate circuit. The space constraints within the X-24K seems to exclude most of the units available, however the "Micro-clock" from NetAudio seems to be promising. Its dimensions seem feasible for shoehorning in to the little spare space available. Its also reasonably priced at just under £40. Reviews seem favourable, but limited in number.

The problem I have is will it be compatible? The crystal in the X-24K is marked as 6.144Mhz. The PCM1728e seems to have virtually identical spec to the PCM1716, but the data sheet suggests a clock of higher Mhz should be used. The datasheet states a clock of 256/384/512/768fs (where fs is sampling frequency) should be used. As it will accept 32, 44.1, 48 or 96 khz frequency, should the lowest clock speed actually be 8.192Mhz? If utilising 44.1khz should this be 11.2896Mhz?

The manual from MF for the X-24K suggests it is an upsampling dac, not just able to handle 24bit/96khz but actively interpolating the signals to this bit length and thus frequency. Is therefore the upsampling done before the signal reaches the dac and thus the PCM1728e only ever being fed 24bit/96khz signals; or is this done internally within the PCM1728e and reactive to whatever sampling frequency reaches it? If it is the former, using my limited electronic logic and knowledge, it would seem ideal to have the clock set at one of the frequencies recommended in the PCM1728e datasheet as being ideal for 96khz. In this case 24.576Mhz (256fs, fs96khz) as the datasheet informs us that the crystal oscillator should have no higher actual frequency than 24.576Mhz.

Thus, is the clock in my unit a replacement? At 6.144Mhz it seems to be too low. Although, it is a factor of 24.576Mhz (exactly 1/4). Is this therefore something to do with the oversampling, although if it was I'd expect it to show in the datasheet (?).

If then it is only receiving a 24bit/96khz signal it would seem logical to use a clock frequency for this, on the other hand if it is receiving 44.1khz/16bit signals it would seem logical to use a 11.2896Mhz clock as this would be a clock rate for this frequency. The dac is only used for CD and fed by a Micromega Stage 1 Cd player being used as a transport.

Any advice would be welcomed, I hope my cobbled-together logic makes sense and can be understood. Us mere mortals have a long way to go before we can claim to be on a par with the knowledge of many of the respected forum members. I also trust the torrent of questions is not irritating or fatiguing.

Thanks, Matt.

[matt7941]

Good evening folks.

Could anyone help me out with a query?

I've ordered some AD8066s (with SOIC to DIP adaptors) and LM4562s. I have fitted some dip sockets so I can do a spot of op-amp rolling and have bypassed the 15v+ and 15v- supply lines. It is this that I could do with confirmation of.

I have used a 10uf 16v electrolytic and 1uf 63v polyester cap to bypass each rail. They have been connected to the supply lines as close as possible to the sockets by scratching a little of the insulation off to reveal bare copper and soldering. The negative side of the caps have been connected, individually paired, to ground on top of the board via a short piece of insulated wire after again scratching a little of the insulation off and soldering. I trust that this is correct. Also, is it the case that the wire to ground length is largely irrelevant (within reason) as long as the positive side of the caps are attached as close as possible to the supply rail socket pins?

Thanks and hope someone can help out.

Cheers, Matt.

[matt7941]

Sorry, that should be a 10uf 16v electrolytic and 0.1uf polyester.

Cheers.

[canjunkie]

May 19, 2008 16:44:46 GMT matt7941 said:

Sorry, it's a useless struggle, I just can't leave the soldering iron to go cold in my tool box. The urge is too strong.

Gotcha

I'd love to help but you'd be better waiting for Alex, Leo or the Rickster to take a look at your query - they love a challenge

[matt7941]

Cheers, no worries.

You wouldn't believe how different things look inside compared to the photos you took. If I can get my digital camera working when it's finished I'll post some shots. I can't claim it to be a masterpiece akin to the XdacV3 posted elsewhere, but it seems to be improving the sound.

Matt.