[Sapientiam]

I would suggest you don't go for a toroidal in this design - reason being toroids have higher interwinding capacitance. What this means in practice is more mains-borne noise gets in, and this will suck out the dynamics. Since one of the reasons for going multibit is dynamics, this would be a pity. Besides unless you get a custom wound toroid (always possible but not very cheap) you'd need to use three of them. That's 3X the noise. An R-core is a pretty good choice for this DAC so if I were you I'd stick with it. There are plenty of other ways to enhance it.
 
Good idea about the larger chassis - saves cursing how tricky it is to get the S/PDIF connector back into position.
 
I modded the power supplies sure, but when I was really into these DACs I wasn't fully aware of how big a difference PSUs make (hint - huge). So I did stuff like snubbering the bridge rectifiers to reduce RF emissions and I put small value series Rs to provide better attenuation of HF ripple components on the reg inputs. One of the drawbacks of IC regulators is they suck rather badly at rejecting high frequency noise, hence passive filters are needed on their inputs, not just the usual shunt caps. I'd suggest you do the series R (or if you're even more advanced, series inductors) mods on the feeds to the 3 regulators. Also I swapped the regs from fixed voltage (7806 I think) to LM317s which are variable and have the facility for including another capacitor to give much better noise and ripple rejection. They're the same package size but different pin-outs, so its a bit fiddly, I think I turned them around 180o.
 
Another major mod I did which I've not seen anywhere else is cutting out all the TDA1543s and fitting either TDA1545s (in DIL packages) or TDA1387 (another fiddly mod with small wires as these are only SMT packaged DACs). You can see the results of that on my blog - http://www.diyaudio.com/forums/blogs/abraxalito/871-dacalito-transversal-modification-dac-ah-lite.html

 

[Bronco2011]

You're right about the toroidal! I will stick with the R-core. 
 
Just one question about the bridge rectifier R2. Some people changed the rectifier to a fast/soft recovery bridge rectifier. Is this total nonsense or is this a improvement? And why only change R2 and not R1 en R3 also?
 
Plus is it useful to change the caps C25, C26, C29 and C30? I saw that some use Panasonic TC 4700uf/25V caps. My Lite Dac came with Sanyo 2200uf/35V caps. 
 
Your blog about TDA1545 looks very cool 

 ! 

 

[Sapientiam]

Its not total nonsense, no - because normal diodes (a bridge inside is 4 diodes) generate a burst of RF energy when they switch off. I use snubbers myself rather than changing the whole rectifier, but some like to gild the lily. You're right that it doesn't really make sense to do this to only one, at the very least to the two which power analog circuits (BR2 & BR3) and preferably to all three. There's a pic of how I added some snubbers here - http://www.whatsbestforum.com/showthread.php?9121-The-Ozone-layer-modding-the-Lite-DAC-AH&p=178969&viewfull=1#post178969
 
I can't think of a reason to change the 2200uFs - I measured mine (Sanyo, just like yours) and they have fairly impressive (i.e. low) ESRs, Do though improve the HF rejection with a series resistor or choke. A CRC combination would perform better than a single C with larger value. So by all means add a second 2200uF after a lowish value resistor. This isn't really worth doing on the digital supply (BR1) but a step up for the other two supplies..
 
P.S. Glad you enjoyed my blog!

 

[Bronco2011]

What you done with the snubbers does look good! You've really done quite a lot of modding on the Lite Dac. That's for sure 

 
I shall read your post about the snubbers. Did you ever hear about the website NewClassD? They make discrete opamps and other stuff like UWB Voltages Regulators that can be used to replace the 7806 (actually not the 7806 but the 7808). Only thing is....NewClassD stuff is not cheap.
 
I'm still not certain if I want to let the TDA1543 on 8V. Because than I need to get heatsinks. Yesterday, I think, I finally found suitable PP High frequency caps from Vishay on a dutch website (and also sells all the other stuff that I need).
 
But first things first....Let's start reading your post about snubbers. Thanks!

 

[Sapientiam]

I hadn't heard of NewClassD before, I went over to their website. The PCM1704 DAC (DexaDAC S) interested me as I'm a devoted multibit DAC fan. Its over 500 Euros and they say 'this is a high end DAC system'.
 
They're kind enough to supply schematics so I went to investigate just how high end the design really is and I found it doesn't live up to that claim. In particular the power supplies aren't high-end, and that's the main thing that distinguishes high-end from mainstream stuff. They more or less just follow the datasheet example circuits for decoupling. They do though use series resistors (R16,17) before the IC regulators which is a step-up from stuff with run of the mill PSUs like those in DAC-AH so let's give them some credit

 

[Bronco2011]

Saying "high end" is mostly not the same as being actually "high end"
But nice of them to send you schematics. I only use the discrete opamps of newclassD. I'm still very pleased with them. And 500 euro for a DAC! That pricey

By the way, when the lite DAC runs at 6v. How high should the vref voltage be? I found that 3,3V bias voltage the sound is somewhat harsh. I also tried 2,35V (I read on a website that 2,35 would "the best" setting) but I hear some distortion. Which bias setting do you prefer? And do you use the opamps or the discrete output?

 

[Sapientiam]

They didn't send them - the link is there on their website. Its rather ironic that in the blurb for another of their products (a reg PCB I think) they say that the IC regs used are 'cheap' ones. But they still charge 500 euros, I can't disagree with your price analysis (but PCM1704 are expensive chips).
 
As for Vref, I assume you're talking about pin7. I and others on DIYA reckon the chip sounds best without anything connected to this pin, but then that requires the I/V resistors to be connected to a bias voltage - on my blog I have a post about this and show a filter for the bias voltage. I use a very cheap shunt reg (TL431) to generate it.
 
I tend to avoid opamps when it comes to I/V - I do passive I/V then filter and after that use an AFA ('active feedback amplifier). This is a special kind of opamp which addresses some of the drawbacks of normal opamps in terms of SQ. The AFA topology was developed by Barrie Gilbert at Analog Devices and has been incorporated into several of their chips - for example AD603, AD605, AD830 and AD8129. I've tried all of these in my various DACs and they sound great.

 

[Bronco2011]

I still haven't found the time to research and read about your suggestions. All work and no play 

 
 
But I'm wondering about an idea I have. Probably a wild idea. As you maybe know I already got one Lite Dac and the second one is under way. Is it possible to connect the 2 Dac's together? Make it into a 16 paralleled Non oversampling Dac! The chassis is big enough for it. The question, for me as a newby, is where to tap the signal off on the first board and where does it needs to be "plugged" into on the second board. 
 
I got the schematic of lite Dac, but not everything is clear to me. Did you ever connect the two boards together? 

 

[Sapientiam]

No, I didn't connect two boards together - but in respect of the number of paralleled DACs that I tried, I went further than 16 - I got up to 40, of which 32 were in parallel. Not with TDA1543 as that's too power hungry, but I did it with TDA1545s. Having listened to that, I realized that there's no value in terms of SQ from simply adding more and more DAC chips.
 
What distinguishes a 'true audiophile' DAC from a run of the mill one is noise - or rather, lack of noise. Hence I'd not recommend connecting two boards together as this will likely increase noise (with all the wires needed between them, two transformers etc.) whereas the direction of clearer sound requires lower noise. Of course if you could increase signal whilst making no increase to the noise, that would be another way to go. But putting two boards together does not increase the signal level at all. So I reckon two boards together would sound worse (i.e. less dynamic) than one.

 

[Bronco2011]

Darn, I forget about the noise again. There are several complaints about noise. People with high sensitive speakers, above 99db, say that the TDA's are quite noisy. Anyhow, it was a wild idea that just kept sticking in my head. Better to ask if it possible or just a bad idea. Conclusion, it's just a bad idea.

 

[Sapientiam]

Yes - much better to ask. The way to get good ideas is to have lots of ideas, discarding the ones that don't work.

 

[Bronco2011]

@sapientiam. Finally I finished my first upgrades on the Lite Dac.
 
I did the following:
01. Swapped the R35 en R36 resistors for 154 ohm 
02. Changed blue caps to Vishay (type:416) 0,1u caps (except for C38 and C39. They are still original)
03. Desoldered the opamps and put in IC feet for the Dexa opamps
04. Swapped the sanyo elco's 2200uf en 220uf for Panasonic FR serie elco's (they have more low ESR en higer Ripple)
05. Soldered wires nearby U5 and U6 for the passive output. Put in Soniccraft 3,0uf sonicaps and connected them to the RCA outputs (between earth and signal I soldered 47Kohm resistors)
06. Swapped the Elco's C35 and C36 also for Soniccraft 3,0uf Sonicaps
 
Desoldering was not easy task! But anyhow with lot's of patience I managed to upgrade it. The sound of the passive output is really great. It sound so much better. More details and depth. I really like the passive output. Before the upgrades I liked the active output. But now the active output sounds just loud and harsh. I also made some pic's of the result.
 

 

 
Only the back of the chassis looks like s**t. I didn't had the right tools to make it better. 
 
I only still want to upgrade the rectifier bridges. I read something about it. Is it possible to change it with some diodes? or to put in a IXYS (type: FBE 22-06N1) Fast single Phase Rectifier Bridge? But I read that the pin out on the IXYS bridge is somewhat different. Plus the IXYS hase 5 pins instead of 4. What is the function of the pin in the middle? 
 
I also came across a PDF File on changing bridges into diodes (http://www.hificollective.co.uk/pdf/diode.pdf). It looks simple but is it really that simple? Or do I need to make more changes before I can put some diodes?
 
Did you ever change the bridge for more fast and soft diodes? And thanks for your advises before. 

 

[Sapientiam]

I changed the bridges to discrete schottky diodes yes, on at least one of my mod binges. No other changes required as a consequence of this. That IXYS brdge looks to be massive overkill - pin3 looks to be connected to the back side metal tab of the package.
 
Have you removed the grounds from the groundplane and used star earthing? That makes a significant difference.

 

[Bronco2011]

Because I was looking into what you wrote about star earthing a came across a site about building a gain clone (http://www.decdun.me.uk/gainclone_psu.html#gl3). I found a great explanation for making a rectifier bridge with MUR diodes. Because the IXYS bridge is overkill and very expensive!

So I bought a strip board and MUR860 diodes and made my first diode rectifier and it works well! I think that I'm going to strip the 3 bridges. Take a larger piece strip board and mount the 3 bridges on it to feed the DAC.

I took a pic of the experiment


And try to star earth the DAC.

 

[polaron]

Well, I've made a few more changes. I removed the 1000uf Blackgate caps that were between the DAC heatsinks and the LM7808 heatsink because they were getting too hot and replaced them with 220uf/16V Panasonic FC caps (C28,C37) near their original locations. I haven't noticed any more noise or reduced bass. I also added another pair of RCA jacks connected to the opamp outputs through bipolar Elna caps (included in the socketed opamp option from diykits). I added 680 ohm resistors in parallel with the 390 ohm opamp feedback resistors to lower the buffer stage gain to 1.48 from the original 1.76. This results in 2V RMS max output voltage from the opamps. I'm also using opa627 opamps. Finally, I twisted the wires going from the transformer to the board and moved them out from under the board. This totally eliminated the slight 60Hz hum I had at max volume.

Now I can directly compare the sound of the passive output and the buffered output. I was wrong in my earlier assessment. The passive output sounds much more transparent and "real" to me. It's sounds like the singer and musicians are performing right in front of me. Amazing! I definitely prefer the passive bypass to the opamp outputs.

I've tried increasing the 2.2uf Auricaps to 6.9uf (2.2uf and 4.7uf in parallel) and didn't hear much difference. I must admit that I haven't yet given the buffered output a fair chance yet since it still uses an electrolytic cap in the output. I may try installing an Auricap in the opamp output just to try it, but I'm so happy with the sound with the passive bypass that I'm not very motivated to do that. By the way, I'm listening through a World Audio Design HD83 tube headphone amp and HD600s.

I do plan to add more ventilation to the top panel since there's is still a lot of heat generated.

Anyway, I highly recommend trying the passive bypass. You don't necessarily have to do the 6V-to-8V mod. Changing to 8V does result in more heat and the necessity of DAC heatsinks and case ventilation. If you stay with 6V and change RLoad to 150 ohms, then the max RMS output will be about 0.9V. That's about 3db less than the standard line out spec of 2V. If you switch to 8V and change RLoad to 220 ohms, then the max RMS output will be 1.35V. On my second DAC-AH I'm thinking of staying at 6V and living with the slightly reduced output voltage.

If you don't want to get into extensive mods like this, then I strongly recommend at least correcting the resistor error and adjusting the trimpot. That alone will give you significantly better sound.

I hope this info is useful to you DAC-AH owners or those of you considering purchasing a DAC-AH.

Hi Kim!
Sorry I know that this is a very old thread, but I am trying to make a few modds to a Lite Ah.
I decided that at least I will bipass the opamps and install a direct output. I will use Hovland (black ones), because I used it in a Paradisea. The Paradisea also has a TDA dac.
Te Hovlands with the TDA make such an improvement! More, much more deep and large stage, but in an incredible level. It sounded magical with tones pf detail, like sabre, but with TDA performance.
Anyway, I want your help in order to understand how to "the resistor error and adjusting the trimpot".
How to identify these resistors?
What value and brand?
What is the trimpot?
If the explanation is to long at least please explain about the resistors.
Tank you very much!