Jump to content

PDX DAC project


Guest Benjet

Recommended Posts

2 hours ago, rmpfyf said:

......  if you want a SOTA outcome you’ll run those puppies at 768kHz or as fast as you reasonably can.

Thats not certain. Extreme rates might sound worse.  Higher sample rates can result in more glitch noise and impair full resetting. 

Link to comment
Share on other sites



Hey guys this is getting me all motivated with the PCM1704k again.

I have a JLsounds I2S over USB board lying around & it certainly looks dead easy to couple it to a pair of PCM1704K’s.

CAD up a PCB with lots of ultra low noise regs for everything - disable the clocks on the JLsounds & clock it from pair of Crystek Femto clocks I also happen to have lying around. Now just the IV stage & output to decide on. Maybe just the Burr Brown data sheet circuit with a OPA627/OPA2134, also lying around ha ha - lots of stuff lying around in my parts hoard that’s for sure.

On second thoughts maybe not the OPA2134 - must be something better to replace that with now.

Forgot to mention that I also have about ten brand spanking new PCM1704K’s as well. Open to offers!

Edited by Gieseler Audio
  • Like 1
Link to comment
Share on other sites

1 hour ago, Nada said:

Thats not certain. Extreme rates might sound worse.  Higher sample rates can result in more glitch noise and impair full resetting. 

Solvable problems. From an upsampling/filtering perspective it's ideal, if a diminishing returns argument. Min phase (or nearly so) is effing glorious in the audible band when at 384kHz... just ask the team at Meridian.

 

27 minutes ago, Gieseler Audio said:

Hey guys this is getting me all motivated with the PCM1704k again.

I have a JLsounds I2S over USB board lying around & it certainly looks dead easy to couple it to a pair of PCM1704K’s.

CAD up a PCB with lots of ultra low noise regs for everything - disable the clocks on the JLsounds & clock it from pair of Crystek Femto clocks I also happen to have lying around. Now just the IV stage & output to decide on. Maybe just the Burr Brown data sheet circuit with a OPA627/OPA2134, also lying around ha ha - lots of stuff lying around in my parts hoard that’s for sure.

On second thoughts maybe not the OPA2134 - must be something better to replace that with now.

Forgot to mention that I also have about ten brand spanking new PCM1704K’s as well. Open to offers!

 

I understand there are a few PDX's in circulations sporting JLsounds I2S boards - there's life after Hiface! As you're aware JLsounds has a separate clock board for this sort of hijinks... a SOTA implementation was never intended to use $2 oscillators. 

 

Various opinions on the Crystek clocks. A step up, there's other stuff too.

 

Clay, rather than doing something off-the-bat new with your new stock, what about an official upgrade program for the PDX's current in circulation?

 

They're unlikely to turn into a Phasure overnight but I'd think a NOS upgrade kit could get pretty close, and the DAC ICs are good. It'd end up as something that could play DxD no problem and that could do DSD with an input DSP doing the conversion - there's a few boxes in market that can do that, or it'd be just as easy to setup a cheap PC to do it.

 

Clay, I'm updating my Killer to the same spec. Old DACs for the new age. Update the PDX, let the old battle begin anew :D 

 

 

Link to comment
Share on other sites

Hey guys this is getting me all motivated with the PCM1704k again.
I have a JLsounds I2S over USB board lying around & it certainly looks dead easy to couple it to a pair of PCM1704K’s.
CAD up a PCB with lots of ultra low noise regs for everything - disable the clocks on the JLsounds & clock it from pair of Crystek Femto clocks I also happen to have lying around. Now just the IV stage & output to decide on. Maybe just the Burr Brown data sheet circuit with a OPA627/OPA2134, also lying around ha ha - lots of stuff lying around in my parts hoard that’s for sure.
On second thoughts maybe not the OPA2134 - must be something better to replace that with now.
Forgot to mention that I also have about ten brand spanking new PCM1704K’s as well. Open to offers!

You forgot the price ;)
Link to comment
Share on other sites

Richardo - PDX upgrade - actually that is quite a good idea considering the first few veroboard PDX’s (very embarrassing)

& the first generation PCB’s all used PCM1704’s on plug in adaptor boards. Very easy to remove & plug into a new dedicated design. After that they were mounted directly on the PCB in a densely populated layout so difficult to remove without damage.

Unfortunately (or fortunately) my resources are taxed to the max these days with new products I’m about to release so I’m not keen on upgrade or repair work. Certainly nothing to stop someone else taking it on though - zenelectro??

 

Ben unfortunately NOS PCM1704K’s from reliable sources are commanding big dollars these days ($250 - $300 each)

That would make any new design with them not particularly good value for money.

However if I’m totally blown away with my little prototype I may consider doing a small production run with the few I have left.

I still have a soft spot  for the PCM1704 so I’d much rather build something with them rather than just sell the chips.

Edited by Gieseler Audio
Link to comment
Share on other sites



2 hours ago, Gieseler Audio said:

Richardo - PDX upgrade - actually that is quite a good idea considering the first few veroboard PDX’s (very embarrassing)

& the first generation PCB’s all used PCM1704’s on plug in adaptor boards. Very easy to remove & plug into a new dedicated design. After that they were mounted directly on the PCB in a densely populated layout so difficult to remove without damage.

Unfortunately (or fortunately) my resources are taxed to the max these days with new products I’m about to release so I’m not keen on upgrade or repair work. Certainly nothing to stop someone else taking it on though - zenelectro??

 

Ben unfortunately NOS PCM1704K’s from reliable sources are commanding big dollars these days ($250 - $300 each)

That would make any new design with them not particularly good value for money.

However if I’m totally blown away with my little prototype I may consider doing a small production run with the few I have left.

I still have a soft spot  for the PCM1704 so I’d much rather build something with them rather than just sell the chips.

 

Clay, I reckon you make a new upsampling/NOS input stage, something that you could re-use in other DAC designs and offer to PDX owners too. 

 

Then you do a limited run of 5 PDX's with the new front end. 

 

I enjoyed my PDX for the time I had it; I did find the input stage the most limited bit. Admittedly mine was early and at the time it was build a HiFace was about as good as it got, though now that things have moved on there's a lot that can be done. 

 

This, and the 1704 is a very very good DAC IC. Last of breed FWIW.

 

(This, or you go all-out on a multi-chip-per-channel one-off design IMHO). 

Link to comment
Share on other sites

2 hours ago, Gieseler Audio said:

Richardo - PDX upgrade - actually that is quite a good idea considering the first few veroboard PDX’s (very embarrassing)

& the first generation PCB’s all used PCM1704’s on plug in adaptor boards. Very easy to remove & plug into a new dedicated design. After that they were mounted directly on the PCB in a densely populated layout so difficult to remove without damage.

Unfortunately (or fortunately) my resources are taxed to the max these days with new products I’m about to release so I’m not keen on upgrade or repair work. Certainly nothing to stop someone else taking it on though - zenelectro??

 

Ben unfortunately NOS PCM1704K’s from reliable sources are commanding big dollars these days ($250 - $300 each)

That would make any new design with them not particularly good value for money.

However if I’m totally blown away with my little prototype I may consider doing a small production run with the few I have left.

I still have a soft spot  for the PCM1704 so I’d much rather build something with them rather than just sell the chips.

Clay! What have you done!!!! Will this new DAC have a volume control?? 

Link to comment
Share on other sites

30 minutes ago, Nada said:

Are you claiming you have a way to lower glitches and settling time for the PCM1704 running at extreme speeds?

Me? Nup.

 

Evidence suggests a few NOS 1704 projects at 384kHz and 768kHz are doing fine.

Link to comment
Share on other sites

Looking at the specs of the JLsounds I2SolverUSB board and it look is really straightforward. What are the advantages/disadvantages to this approach over the iancanada. Price looks very attractive, but if the iancanada yields greater performance, I would prefer to spend the extra.

Link to comment
Share on other sites



Just a quick shout out to Clay. I heard quite a few DACs in my time and the PDX, even in it’s current condition, is a darn good unit. My biggest concern is that with all my planned “upgrades” I don’t lose , but enhance it’s magic. What I’m attempting should be referred to as an update not an upgrade. All of the implementations I plan were not available at the time of the manufacturer of this unit (correct me if I’m wrong). [emoji106][emoji106][emoji106][emoji122][emoji122][emoji122]

Link to comment
Share on other sites

18 minutes ago, Benjet said:

Looking at the specs of the JLsounds I2SolverUSB board and it look is really straightforward. What are the advantages/disadvantages to this approach over the iancanada. Price looks very attractive, but if the iancanada yields greater performance, I would prefer to spend the extra.

 

Bigger buffer on Ian’s. More complexity tho. Not sure max freq in JL. Both take better clocks, which is def where to spend.

 

Ian’s allows you to slot something between  USB and I2S. Which is powerful, and can affect clock choice.

 

13 minutes ago, Benjet said:

Just a quick shout out to Clay. I heard quite a few DACs in my time and the PDX, even in it’s current condition, is a darn good unit. My biggest concern is that with all my planned “upgrades” I don’t lose , but enhance it’s magic. What I’m attempting should be referred to as an update not an upgrade. All of the implementations I plan were not available at the time of the manufacturer of this unit (correct me if I’m wrong). emoji106.pngemoji106.pngemoji106.pngemoji122.pngemoji122.pngemoji122.png

 

You will not degrade it. Tested this much audibly. Zenelectro has attested similarly, and his sample size isn’t small.

Link to comment
Share on other sites

Hey just another angle on this.

The PDX has around 200v across the SPPP output stage - 100v across each tube section. (ECC88 is a dual triode)

Now rawl99 (who builds the killer dac) assures me that around 75v is the sweet spot for this tube so Ben why don’t you try reducing the HT a bit & see if it sounds even better.

From memory there is a 1k resistor is series with the anode of the top tube section so just try increasing this a bit until you get about 150v total across both tubes. Say try 1.5k etc.

Please be careful in the HT area - leave it switched off for for at least ten minutes & check the voltage across the filter caps before making  any changes.

Apparently the ECC88 sounds a lot sweeter & more relalaxed at 75v vs 100v.

 

Edited by Gieseler Audio
Link to comment
Share on other sites

On 16/09/2018 at 10:01 AM, rmpfyf said:

Yes, some of the answer is in the valves chosen, though really there's quite a lot else to do here to make it a super beautiful sounding DAC - however much of my application here is others' learnings and hard work, it's not for me to post. 

 

When I wrote the above, I meant this:

 

55 minutes ago, Gieseler Audio said:

Now rawl99 (who builds the killer dac) assures me that around 75v is the sweet spot for this tube so Ben why don’t you try reducing the HT a bit & see if it sounds even better.

 

This would be the start of a journey on the tube side of things. Sky's the limit, about as deep as your pockets :P 

Link to comment
Share on other sites

16 hours ago, Nada said:

Are you claiming you have a way to lower glitches and settling time for the PCM1704 running at extreme speeds?

These are not extreme speeds, 1704 is specced to run at 96k x 8 (digital filter oversampling)  = 768kHz.

Quote from top of page 2 of data sheet:

 

All specifications at TA = +25°C, ±VCC = ±VDD = ±5V, fS = 768kHz (96kHz • 8), and 24-bit data, unless otherwise noted. 

 

So as always - read the instructions :)

 

Now - let's take what you are suggesting and apply it in reverse - maybe running 1704 at 384kHz or even 192kHz actual DAC speed will improve the specs.

You are a genius - haha :)

 

Terry 

  • Like 1
Link to comment
Share on other sites



OK - I've had a quick going over this thread.

 

I was the one who first implemented JLsounds on the PDX DAC.

That was a pretty standard implementation and by no means close to what is ultimately possible in a really optimized (but time consuming) approach -  it made a big improvement.

A lot has changed since that job, there are more options available for USB bridge and the JLsounds has in itself gone through an upgrade which is questionable. 

There are more clock options now but all the usual clocks out there have a large variance - which they don't tend to tell people.... :)   The ones I prefer to use are individually measured

so you get guaranteed performance - but obviously, this adds cost.

 

I'll think this one through for a few days - I've got a lot on the plate ATM so a diversion in my game plan has to be carefully considered.

 

cheers

 

Terry

 

 

 

 

 

 

 

  • Like 2
Link to comment
Share on other sites

30 minutes ago, zenelectro said:

OK - I've had a quick going over this thread.

 

I was the one who first implemented JLsounds on the PDX DAC.

That was a pretty standard implementation and by no means close to what is ultimately possible in a really optimized (but time consuming) approach -  it made a big improvement.

A lot has changed since that job, there are more options available for USB bridge and the JLsounds has in itself gone through an upgrade which is questionable. 

There are more clock options now but all the usual clocks out there have a large variance - which they don't tend to tell people.... :)   The ones I prefer to use are individually measured

so you get guaranteed performance - but obviously, this adds cost.

 

I'll think this one through for a few days - I've got a lot on the plate ATM so a diversion in my game plan has to be carefully considered.

 

cheers

 

Terry

 

 

 

 

 

 

 

I’ll back this up.

 

I was on a similar journey with my DAC (TDA1541A) which started with my playing around with my PDX (since sold). I was able to improve the performance of it a ton just by updating the input stage - and my attempts were pretty marginal. I had purchased an Acko board to start with - good but a long way to go - and thought best to enlist a professional. @rawl99 was kind enough to introduce me to @zenelectro, who had done a ton of research into oscillators for earlier reference-grade projects. It’s been a journey since and a lot of learning! 

 

For various reasons we settled on 384kHz as an ideal target (NOS).

 

@zenelectro is presently in the final stages of building me two reclockers, one using an Audial kit and the other with iancanada’s kit. Particularly with the latter there is more to it than just the boards as ordered. I’ll leave it to @zenelectro to post pics and talk to performance here.

 

I’ll quite openly state that dealing with him has been an absolute pleasure at very fair rates. (And that as an engineer I’m an absolute PITA of a customer). He’s got a few things in development which go beyond reclocking too.

 

There’s quite a few PDX’s in circulation that  have great chips and sound amazing with a well-clocked input the likes of which just wasn’t available when the DAC was conceived. They’re cheap to acquire and, properly upgraded, play well above their pay grade.

 

If there are others beyond @Benjet wanting to take their PDX to the next level I’d suggest getting together for a group buy on a kit developed and fitted by @zenelectro. You won’t look back.

Link to comment
Share on other sites

Just about to break ground on this project. Just got a headphone dac/amp on the bench just about completed and the PDX is next. Tubes will be here next week and once their in and I’ve swapped out some resistors, I’ll be looking into sorting out some new voltage regs. I’ll probably also order the JLsounds interface as well. Would be great starting point. I will go down the NOS path to see how that sounds. Will look into the clock side of input as well. I haven’t had much experience with these, so It’ll be a steep learning curve. I’ll post up some pics when I have them.

 

Link to comment
Share on other sites

Finished the headphone/DAC. Threw every trick in the book I know at the DAC section and the PDX is still lightyears ahead. Fortunately I included a RCA input on the amp so I can bypass the DAC section if I want. 

 

Anyhow, that’s enough practise, PDX next!!!!

2421A54E-9F35-42A5-A16E-3B66CD579F44.jpeg

83EC0AAF-3211-4C22-8E9A-C13CD3C9A842.jpeg

E429C70C-F648-4F43-A3E7-88B2C7D8E22C.jpeg

9D04C5F2-E6D9-4607-AA3F-99A9AA25D7B1.jpeg

Link to comment
Share on other sites

I’m about to order the bits for the input stage. I need some advice so I make sure I get the right bits.

The shopping list is:
1. JLSounds i2soverUSB VIII
2. Oscillator board
3. Crystek oscillator. Need help with which one to order. I’m aiming for 384kHz output direct to the DAC chips. I think the 49.152MHz one is the right one, but I really don’t know much about this side of the interface.

With the power supply side of the DAC, I would like to get rid of the plug pack and move all the supply to inside the case.

From what I can see is the supplies I’ll need are
1. +5v for PCM1704’s
2. +5v USB side of i2soverUSB VIII
3. +5v for Oscillator board
4. +6.2v for tube heaters
5. +218v for B+

All voltages are DC. I could upgrade the voltage regs on the board, but would like to build a seperate power supply board and keep AC voltages off the main board. Looking at getting a r core transformer with 230v and 9v output taps and locating it inside the chassis in a shielded section. I also looking for a new chassis that would better suit the new configuration.

That should bring the power supply and input stages to a new level. Output modification will be the change in the plate voltage to 70v and also the new Siemens tubes.

Link to comment
Share on other sites



Hi Ben,

hey just to keep things simple (in the clock dept) to start with I would just try the JLSounds i2soverUSB VIII

This new version uses NZ2520SDA clocks - these are excellent & very low phase noise - zenelectro may care to comment.

Also on the PSU side of things the PCM1704 also needs -5v

Cheers  Clay

  • Like 1
Link to comment
Share on other sites

Hey just another angle on this.
The PDX has around 200v across the SPPP output stage - 100v across each tube section. (ECC88 is a dual triode)
Now rawl99 (who builds the killer dac) assures me that around 75v is the sweet spot for this tube so Ben why don’t you try reducing the HT a bit & see if it sounds even better.
From memory there is a 1k resistor is series with the anode of the top tube section so just try increasing this a bit until you get about 150v total across both tubes. Say try 1.5k etc.
Please be careful in the HT area - leave it switched off for for at least ten minutes & check the voltage across the filter caps before making  any changes.
Apparently the ECC88 sounds a lot sweeter & more relalaxed at 75v vs 100v.
 

Just had a quick look at the plate resistors. They are 4.7k and have a voltage drop of 14.5v. I’m looking for a voltage drop of 40ish volts and by my calculations I will need around 12k (I think). Also with the heaters, you have them floating around 75v. So the change in plate voltages should bring their potential differences closer to that of the plates.

Let me know how far off the mark I am.

Thanks,

Ben
Link to comment
Share on other sites

Ben, are you familiar with the john kenny spdif converter. It used lipo battery banks with automatic switching and charging to supply an inbuilt hiface device. Might be a bit cumbersome compared to a r-core though. 

http://www.johnkenny.biz/home-1/mk3-hiface

Obviously the hiface is not the point of interest here. I had one of the jkspdif and liked the simplicity of operation in the later units. 

Link to comment
Share on other sites

3 hours ago, Benjet said:

I’m about to order the bits for the input stage. I need some advice so I make sure I get the right bits.

The shopping list is:
1. JLSounds i2soverUSB VIII
2. Oscillator board
3. Crystek oscillator. Need help with which one to order. I’m aiming for 384kHz output direct to the DAC chips. I think the 49.152MHz one is the right one, but I really don’t know much about this side of the interface.

With the power supply side of the DAC, I would like to get rid of the plug pack and move all the supply to inside the case.

From what I can see is the supplies I’ll need are
1. +5v for PCM1704’s
2. +5v USB side of i2soverUSB VIII
3. +5v for Oscillator board
4. +6.2v for tube heaters
5. +218v for B+

All voltages are DC. I could upgrade the voltage regs on the board, but would like to build a seperate power supply board and keep AC voltages off the main board. Looking at getting a r core transformer with 230v and 9v output taps and locating it inside the chassis in a shielded section. I also looking for a new chassis that would better suit the new configuration.

That should bring the power supply and input stages to a new level. Output modification will be the change in the plate voltage to 70v and also the new Siemens tubes.

 

You may have missed a quote from @zenelectro - the new i2soverUSB is FPGA-based (like an Amanero) and that's going to limit clocking accuracy irrespective of the oscillator used. 

 

You'll probably find better phase performance for 22/24 (or so) Mhz pairs in the NDK NZ2520SDA range; the question then is can you find a solution that'll give you genuine simultaneous performance out of lower-speed clocks (i.e. 384kHz) - this isn't possible with I2S - look at the bitstream. 

 

Pedja Rogic's USB board is one such contender, if he's got stock I'd ask. From memory it can be configured to spit out data for PCM DAC ICs, I could be wrong. Otherwise you're stuck at the aforementioned (double) clock speeds for kits  that are I2S native, and iancanada's kit is about as good as it gets - though there's a few boards involved and it ain't cheap. 

 

I would move on from the Crystek's. 

 

Seriously - you want this one https://www.audialonline.com/topics/usb-to-simultaneous-data-or-i2s-converter-series-2/ - get in touch with Pedja.

Link to comment
Share on other sites

7 minutes ago, rmpfyf said:

 

You may have missed a quote from @zenelectro - the new i2soverUSB is FPGA-based (like an Amanero) and that's going to limit clocking accuracy irrespective of the oscillator used. 

 

You'll probably find better phase performance for 22/24 (or so) Mhz pairs in the NDK NZ2520SDA range; the question then is can you find a solution that'll give you genuine simultaneous performance out of lower-speed clocks (i.e. 384kHz) - this isn't possible with I2S - look at the bitstream. 

 

Pedja Rogic's USB board is one such contender, if he's got stock I'd ask. From memory it can be configured to spit out data for PCM DAC ICs, I could be wrong. Otherwise you're stuck at the aforementioned (double) clock speeds for kits  that are I2S native, and iancanada's kit is about as good as it gets - though there's a few boards involved and it ain't cheap. 

 

I would move on from the Crystek's. 

 

Seriously - you want this one https://www.audialonline.com/topics/usb-to-simultaneous-data-or-i2s-converter-series-2/ - get in touch with Pedja.

One wee problem, under the purchase section it reads - "EDIT: The offer is closed, and payment buttons are removed.""

Link to comment
Share on other sites



  • Recently Browsing   0 members

    • No registered users viewing this page.




×
×
  • Create New...
To Top