IanCanada's Latest RPi GB Goodies Impressions... and your tweaks, mods and hints...

I do plan to use the 3042 regulator boards that Ian generously gave us. I expect that they will lead to even more improvements.
One thing that may not have been clearer in my post on the clock power supply... I used 3 Regulators in series. I can't attest to the Improvement that multiple Regulators make but after reading many posts about the science behind it I felt that it was well worth putting them in place. I can say that the sound is quite good.

My next experiment will be with Andre's well-tempered master clock. I look forward to seeing those results although that'll be quite a piece down the road. The experiment with reg on the NDK was really just a step in that direction.

Also pleased to hear about the long break in. I suspect the entire thing will continue to get better like fine wine. That's why I keep powered 7x24. A separate switch for digital and tube sections proved very valuable.

Again thanks to Ian. I was attracted to his device based partly on the flexibility to safely try different clocks and PS mods. It certainly delivers.
 
Member
Joined 2003
Paid Member
@jacklee,

Sorry for being so slow to respond to your request on transformer outputs. Honestly I have little experience in this realm and have only used 3 setups and only with Ian's DACs... the GB LL1544a, the Onetics 600:600 units I got for this type of use a few year back, and a pair of Lundahl LL1570XL.

The LL1570XL trail behind the other two. I can't recommend them.

The LL1544a are behind the Onetics, but not by a huge amount. I can listen to either. AND since getting more of the Onetics tranformers seems to be difficult nowadays (I reached out to them and know some others who did too, with no responses), I would not set my heart on getting a pair of them.

I asked a friend with MUCH more experience with output transformers and he suggested Cinemag, Jensen, and Sowter as good sounding alternatives at reasonable prices. AND if one has deep pockets, Tango and Finemet. AND maybe some from Silk that will be between these two extremes in cost.

Anyone else with some experience in this area?

Greg in Mississippi

P.S. I should add I have LL1684 here to try too at some point.
 
Last edited:
@Greg,

Thanks for sharing your experiences with transformers. I really like the LL1544a already and it's not even a premium model in the Lundahl catalog so I thought there must be something even better.

I'm not entirely sure about what transformer specifications are required for this I/V conversion duty so have a hard time finding alternatives. Do you have any suggestions?

I found in the Lundahl catalog a premium input transformer: LL1948
https://www.lundahltransformers.com/wp-content/uploads/datasheets/1948.pdf

Do you think this will work for I/V duty?

Thanks again!

Jack
 
More Clock Experiments

Just want to share some initial impressions from installing the newclassD Neutron Star 2 (45Mhz) clock into my fifopi stack.

A couple things for the installation:

  • The NS2 is powered using a separate 13.2V battery rail.
  • The NS2 has two outputs. I used the one with coaxial cable output (which has a max length of 50cm. The other one is limited to 10cm.)
  • I cut the coaxial cable to length (about 10cm) and then soldered the wires to a clock socket with center wire to the signal pin and shield wire to the GND pin.
  • The NS2 can support enable/disable via the ENABLE pin connected to the HOST connection point in the NS2 board but I ended up not using that (more on that later).
Some observations:

  • I initially connected the ENABLE/HOST via a third wire (the brown one in the photo) but that didn't work - the fifopi won't boot (stay on displaying running dots in the ESS Controller). I then realized that for the ENABLE signal to work the NS2 board must have its GND connected to the fifopi GND as the clock output is galvanic isolated.
  • So I connected their GNDs and it worked as intended. BUT then I found the sound to be not as good as before. Puzzled for a while and then I went back to remove the GND connection (and disabled the ENABLE/HOST function via a solder joint) and voila the sound improved!
  • I also found that the fifopi can work correctly w/o the ENABLE/HOST function as I can play both 44.1khz and 48khz files normally with a NDK clock (49Mhz) in the second socket. The fifopi must have a "mute" switch for each clock for when one is not in use.
  • However, at this point I still found it slightly worst then when I first tested it and the only difference was I have installed a second NDK clock (49Mhz) into the second socket so off it went, and voila again the full sound quality was restored!

What it means is that to get the best sound out of the NS2 it must be used alone and completely isolated from the fifopi GND. I'm not sure if this also applies to other clocks so could be something simple to experiment.

So how does it sound? I hesitate to draw any conclusion this early as it has just been installed but suffice to say that with it in place I was drawn into the music rather than focusing on doing comparisons. I'll do the usual run-in and report back in a couple weeks.
 

Attachments

  • IMG_0438.jpg
    IMG_0438.jpg
    984.5 KB · Views: 821
Last edited:
Regarding Neutron Star 2 - 45MHz
Specifications at: NewClassD NeutronStar2 45.1584
shows jitter as: "1.0pS RMS (10 Hz - 50 kHz range)"

Seems kind of high, at least compared to some of the better audio clocks.

Indeed. I thought Phase noise was the key spec. NDK and Crystek have been compared, but I fail to find any information on phase noise for this device, unless I am missing something.
 
For those using line power supplies rather than batteries, I just tried a couple of different voltage regulators for AVCC on one of Ian's FIFO_Pi dacs. One regulator sounded a lot better than the other. Details posted in the ES9038Q2M thread (since people over there might be interested too): https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-433.html#post5773733

Also, noticed a couple of other issues that might be worth mentioning. I am using OPA1612 opamps on Brown Dog adapters for Ian's I/V board. They are running from a +-15v supply with film cap filitering. The sound pretty good, but a couple of concerns: (1) they are running much warmer than they should. I noticed the same effect on my 1st modded dac, but not the 2nd one. I believe the heating is due to RF getting into the opamps from other sections of system. My second modded dac was laid out and wired differently, which I believe accounts for why no RF heating with its opamps. And, (2) the dac sound quality suffers, IMHO, from not having harmonic distortion compensation programming supported in Ian's controller. I use the distortion compensation in my modded dacs, and Allo uses it with Katana. It does make a difference and would likely help this dac sound at least a little bit better. Even if it only needs a very few dB of compensation, the effects still make for an audible improvement, IMHO.
 
@Greg,

Thanks for sharing your experiences with transformers. I really like the LL1544a already and it's not even a premium model in the Lundahl catalog so I thought there must be something even better.

I'm not entirely sure about what transformer specifications are required for this I/V conversion duty so have a hard time finding alternatives. Do you have any suggestions?

I found in the Lundahl catalog a premium input transformer: LL1948
https://www.lundahltransformers.com/wp-content/uploads/datasheets/1948.pdf

Do you think this will work for I/V duty?

Thanks again!

Jack

I've been looking into iron lately, and reviewed Ian's board its is set up for 1:4 transformer ratio, if I read the schematic and data sheet correctly. There is a guy here who winds iv tx's, and he recommends 1:4 as a good compromise for the windings ratio.

Your tx would not support 1:4, but no idea how it would compare.

This is the thread from the guy here, with info on the ratios and why 1:4 was picked
Output transformers for DACs

I'm thinking about trying his stuff when I actually start working on this.

Greg
I read somewhere Bud from Onetics retired, so unlikely to get anymore there.
 
More Upgrades....

Hi all,

Since my last post, I've added a few more upgrades to the 'DAC system'. First I replaced the standard clocks with the Crystek equivalents using Ian's supplied adapters. Haven't done a lot of SMD soldering, but managed okay to get everything soldered and working correctly.

I next upgraded all 3 of the output opamps to OPA1622s using Brown Dog adapters. At least I'm getting more exposure to SMD soldering! They went well.

Lastly I replaced the I/V STD PSU with a DIYINHK +- 15VDC regulator board using the same R-CORE xformer.

Now for the sound... At least to my ears I can't say there has been a dramatic, or even significant change in the sonics. What I can discern is a cleaner, less clinical sounding with highs and certain vocals. The soundstage also seems to be more open, with less focus on each speaker. IMO that's a good thing. The lows and mids seem unchanged, yet still have the punch I prefer. I find that well recorded acoustic music a good test for any system. I have a couple of tracks I always default to, to test various DACs and amp I swap in and out occasionally.

I still have a couple more changes to make like a stand-alone AVCC PSU, and a separate supply for the RPi 5V using the isolated connector. Once I finalize the enclosure I'll add that to list. :)

For now, Ian's 'DAC system' is my preferred roon playback choice. I'm more than happy with the upgrade in sound I'm enjoying.
 
Q

...replaced the I/V STD PSU with a DIYINHK +- 15VDC regulator board using the same R-CORE xformer.

LDOs for +-15v? Probably enough film cap power rail filtering will help improve SQ.

OPA1622 for all opamps? Probably OPA1612 would have been a better choice, except for maybe the differential summing stage (only if headphones are to be driven directly). Besides, what about that ground pin they designed into OPA1622? That pin left floating if using an 8-pin adapter, or something else?

Turns out running each of the dac 3.3v supplies on separate regulators is a good idea. LDOs work okay for the two digital sections. Suggest some other approach for AVCC. I am kind of impressed so far with New Class D regulators for AVCC, which are not LDOs. They could be powered from the output stage +15v. NewClassD UWB2 LC78033

Could be with the big Crystek clocks, it might help to put the clocks on their own voltage regulator (for FIFO_Pi). Haven't tried Crystek yet for this dac, but NDK "A" suffix clocks seem to work reasonably well as is. The Crystek's would seem to come so close to the dac board that some capacitive coupling between clock cases and dac board signals/ground seems possible, all the more so since the Crystek's are socketed which means their grounds have some ESL going back to the clock board ground.

Regarding RF heating of the output stage opamps, could be separating the output stage from the dac board with a little distance could help. Might also help to twist together the dac +- output wires for each channel between the dac board and the output stage board. That will add a few pf of capacitance which might help with the RF heating. Or, a small-ish cap could be tried instead. Might not take much, have to try and see.
 
Last edited:
....OPA1622 for all opamps? Probably OPA1612 would have been a better choice, except for maybe the differential summing stage (only if headphones are to be driven directly). Besides, what about that ground pin they designed into OPA1622? That pin left floating if using an 8-pin adapter, or something else?

I forgot the OPA1622s I used came pre-installed on 8-pin adapters. Visually, it appears the GND is floating. I'll have to check the next time it's on the bench. I assumed they were drop-in replacements for the stock opamps - according to the docs/recommendations for the the I/V STD board. I assume they are working correctly since there was no magic smoke after replacing them. :) I do plan on trying the OPA1612 as well though. What's the audible improvement with using the 1612 over the 1622?

Plan to use the TP Trident-SR for AVCC.
 
What's the audible improvement with using the 1612 over the 1622?

Plan to use the TP Trident-SR for AVCC.

The difference between the two types of opamps is likely small-ish compared to the benefits you will get from adding enough film caps from your +-15v rails to ground, either than or get some kind of non-LDO, good audio opamp power supply to use instead. There are at least a few choices, but they all cost money and or involve varying amounts of construction. At this point, film caps are probably the easiest thing to try. The lowest cost option would probably be to build Nazar regulators as per eziitis recommendations. Once the +-15 power quality is suitable for opamps to sound really good as they should, then the OPA1612 should probably sound better, although the differential summing one won't sound as good as OPA1622 if directly driving low impedance headphones (since that's what OPA1622 is better at).

Also, unfortunately the TP Trident-SR uses ADM7150, so it may be kind of grainy and distorted for use with ES9038Q2M. At least that's what I found, since Trident-SR was exactly the ADM7150 regulator I was talking about in another post (although they are fine for DVCC and VCCA). Suggest you try a New Class D Mk2 3.3v regulator for AVCC instead of Trident-SR if at all possible (*much* better sound quality, at least in the case of ES9038Q2M AVCC). Either that, or build an opamp type AVCC supply as recommended by ESS.

Another thing you might try is adding a resistor from Trident-SR output to ground to see if that helps improve sound quality (Sometimes audio voltage regulators sound better with an added resistor in parallel with the load, it may take some experimentation to find the best value).
 
Last edited:
Another options you can try are:
Paul Hynes 3.3V shunt regulator.
Belleson 3.3V - they regs comparsion in their website (including new class D).
Salas Reflector - D for 3.3v
Salas BiB 1.3 +/-15V - this one really rocks in terms of bandwith, Z-out and low noise.
Check also ldovr website. The guy makes nice LT3045 base small form factor regs.
Those mentioned are on my top list.

I tried also a bunch of others:
ADM7150, Old Trident shunts from twistead pear, PSU Shunts 2013 -/+15V from Audio Gd, Kubota design regs, Diyinhk, SuperTeddyReg, Audio GD PSU-A Shunt - all of them are also, but I found the mentioned ones at the begining of this post to be the best.
 
Last edited:
A Cautionary Tale

So.... I was tinkering with various values of additional clock decoupling capacitance on my FiFoPi. It is absolutely possible to mislead yourself with unintended consequences here. One addition caused me to wonder if I'd killed the player completely. Only a factory default reset brought back replay. TrueSYNC was no sync...:D

I learned that monitoring the DPLL ratio through the controller is a powerful tool to tell you if you are on the right or wrong track. Remember that this is RF here and parasitic resonances can appear in unpredictable fashion and shape your clock signal integrity.