And loading the transformer secondary at 10k?
Also, sometimes MKP bypass on electrolytic blocking caps is a better fit than FKP. The latter can be too bright (maybe too low ESR) to match up well with the sound of the electrolytic. Another electrolytic to possibly try would be Nichicon UKZ 100uf, but they are EOL. Still some around though. Positive end goes to the dac, right?
Only point is that when using a passive solution like a transformer, associated passive elements may need to be adjusted to give a balanced sound.
Also when using a transformer SE that way, the dac inverting output is not loaded the same as the non-inverting output. That increases ripple on the 22uf X5R bypass caps on the shift registers. The effect may be to increase distortion, so if you are hearing some HD/IMD with the transformers then maybe you would want to try to tweak that imbalance out.
Also, sometimes MKP bypass on electrolytic blocking caps is a better fit than FKP. The latter can be too bright (maybe too low ESR) to match up well with the sound of the electrolytic. Another electrolytic to possibly try would be Nichicon UKZ 100uf, but they are EOL. Still some around though. Positive end goes to the dac, right?
Only point is that when using a passive solution like a transformer, associated passive elements may need to be adjusted to give a balanced sound.
Also when using a transformer SE that way, the dac inverting output is not loaded the same as the non-inverting output. That increases ripple on the 22uf X5R bypass caps on the shift registers. The effect may be to increase distortion, so if you are hearing some HD/IMD with the transformers then maybe you would want to try to tweak that imbalance out.
Last edited:
One other thing just occurred to me. When using electrolytics for DC blocking, you might want to give them a few days at least with music playing through the dac before you even bother to listen. The sound of electrolytics can change pretty fast at first. Easiest way not to get confused/distracted by it, is just don't listen to them at all until the dac has been running continuously with the caps for 12 or 24 hours. Even that may be a bit soon.
It's better to leave the negative output open when it is not in use. The filter output behaves more or less like two voltage sources to ground in antiphase, rather than as a floating voltage source.
I did wonder about that, however at the time it was just simpler to ground it. I will do as you suggest Marcel and pull those wires from the connector. Thanks for the advice.
Well, I hesitate to share my initial findings - but...
The change to the Beyschlag MELF resistors is not subtle. My ears are not great anymore but I would defy anyone not to hear the difference.
Now, a caveat is that I am not necessarily comparing exactly like for like, just very close. However, I can do another test which is to replace the Susumo resistors on the 2nd DAC board and if the 2 DACs then sound identical then it would confirm that the change is purely down to the resistor change and nothing else. (Although these 2 DACs sounded identical to my ears previously) I just want to use due diligence in the experiment.
Now to the difference. I would say that the HF is somehow clearer and more defined in the sound-stage. This DAC was superb at sound-stage to begin with but somehow this change opens and deepens the image. Almost like your brain gets a rest as it doesn't have to work so hard, if that makes sense? Switching the A/B knob back to the DAC with the Susumo the image compresses and a sense of mild disappointment ensues, even though if I started listening with that DAC I was perfectly happy. Considering the 150 resistors I bought only cost less than £6 I consider this a very worthwhile exercise. I did take me a few hours with a bridge to match them up but I believe I got 32 that are within 0.05% of each other, possibly closer. Now whether it is the accuracy of the matching or the construction of these particular resistors I have no idea but I am convinced they do make a positive difference.
For anyone wanting to try the same though I will report back once I have made the double test of modifying the other DAC to ensure I get the same result.
The change to the Beyschlag MELF resistors is not subtle. My ears are not great anymore but I would defy anyone not to hear the difference.
Now, a caveat is that I am not necessarily comparing exactly like for like, just very close. However, I can do another test which is to replace the Susumo resistors on the 2nd DAC board and if the 2 DACs then sound identical then it would confirm that the change is purely down to the resistor change and nothing else. (Although these 2 DACs sounded identical to my ears previously) I just want to use due diligence in the experiment.
Now to the difference. I would say that the HF is somehow clearer and more defined in the sound-stage. This DAC was superb at sound-stage to begin with but somehow this change opens and deepens the image. Almost like your brain gets a rest as it doesn't have to work so hard, if that makes sense? Switching the A/B knob back to the DAC with the Susumo the image compresses and a sense of mild disappointment ensues, even though if I started listening with that DAC I was perfectly happy. Considering the 150 resistors I bought only cost less than £6 I consider this a very worthwhile exercise. I did take me a few hours with a bridge to match them up but I believe I got 32 that are within 0.05% of each other, possibly closer. Now whether it is the accuracy of the matching or the construction of these particular resistors I have no idea but I am convinced they do make a positive difference.
For anyone wanting to try the same though I will report back once I have made the double test of modifying the other DAC to ensure I get the same result.
I had back luck with Susumu RG resistors in a FIRDAC similar to this (it was one of Andrea's FIRDACs). Poor sound apparently due to resistor excess noise. OTOH the resistors that came with Akco's Marcel RTZ dac which I tired here were fine.
I have warned about the Susumu resistors before, but usually people who don't listen like to insist I must be imagining something.
BTW, everything said helped the sound of Marcel's dac did help it. I think you will keep being surprised if you get around to trying those things for yourself.
EDIT: In fact I mentioned the Susumu resistors in this very thread more than once:
[/QUOTE]
I have warned about the Susumu resistors before, but usually people who don't listen like to insist I must be imagining something.
BTW, everything said helped the sound of Marcel's dac did help it. I think you will keep being surprised if you get around to trying those things for yourself.
EDIT: In fact I mentioned the Susumu resistors in this very thread more than once:
[/QUOTE]
Last edited:
When I said "not subtle" I guess that's all relative. Of course it's subtle at this level but I stand by hearing an improvement! I seriously wasn't expecting to. Plus I wasn't criticizing the Susumo resistors, as such, just that these Beyschlag offer an alternative that may be better suited for very little outlay. They even fit the pads well, drop-in replacement. I found the top use for double-sided tape however! ;-) Tiny little creatures run away. (I only lost 1 in the whole process, which I thought was an achievement!)
There may be better than Beyschlag too. I tested a few different resistor types and they all sounded different. Whatever were on Acko's RTZ dac are ones I would recommend. They were pretty close to as good as much more expensive metal foil. IIUC, @nautibuoy assembled Acko's dac. Maybe he can tell you what they were and you could try them on your other dac?
EDIT: IIUC resistor design involves tradeoffs. MELF resistors are designed for rugged conditions such as high humidity and or automotive use. That doesn't mean they are the most optimized for Excess noise. Maybe some brands are and other brands aren't. Since this is for audio, I would suggest to listen to some different resistors.
EDIT: IIUC resistor design involves tradeoffs. MELF resistors are designed for rugged conditions such as high humidity and or automotive use. That doesn't mean they are the most optimized for Excess noise. Maybe some brands are and other brands aren't. Since this is for audio, I would suggest to listen to some different resistors.
Last edited:
I would assume those are the ones I used as they were specified in the BOM. @nautibuoy could confirm as he built it. (Susumu RR1220P-3011-D-M)
Last edited:
I have used MELFs in audio path for many years so I'm happy with your results. But the difference between Susumu and Beyschlag could be at least partly due to better matching (i.e. lower distortions).
MELFs are easy to work with if using a reflow oven or heater plate. They stick well to the solder paste and are chunkier than 0805 thin films so easier to handle with tweezers.
MELFs are easy to work with if using a reflow oven or heater plate. They stick well to the solder paste and are chunkier than 0805 thin films so easier to handle with tweezers.
I completely agree with you on the handling of them. They sat firm in the solder paste and even using hot air they didn't move much and got dragged into place very nicely with the surface tension. I was apprehensive but they turned out to be a dream to fit.
I am sat here delighting in the sound. Another big thanks to Marcel and of course @olo111 and @PJotr25 for the PCM2DSD and all the help and advice from the users of this forum. It's a journey and a hobby but ultimately I just want to hear nice music that makes me happy. I wouldn't have been able to do that for any sensible amount of money if it weren't for DIY and generous individuals who devote time and effort to bringing their designs within reach. I salute you.
I am sat here delighting in the sound. Another big thanks to Marcel and of course @olo111 and @PJotr25 for the PCM2DSD and all the help and advice from the users of this forum. It's a journey and a hobby but ultimately I just want to hear nice music that makes me happy. I wouldn't have been able to do that for any sensible amount of money if it weren't for DIY and generous individuals who devote time and effort to bringing their designs within reach. I salute you.
In principle poor matching of the resistors in a single-bit FIRDAC doesn't cause distortion, like in a multibit DAC, but only poor stop band suppression.
Then again, poor stop band suppression also means poor suppression of the frequency-modulated idle tones around half the sample rate, if any, and hence more chance of those getting mixed down to the audio band if there is something around an odd multiple of half the sample rate ending up on the clock or the voltage reference.
How about the vulnerability to overheating when you hand-solder them with a too hot soldering iron? I know ordinary thin-film resistors are quite sensitive to that.
I have never had issues with MELFs even when hand-soldering or desoldering. The termination sometimes breaks in thin-films (and ceramic caps) when desoldering.
Well, I have made the same alteration in my 2nd DAC unit and compared with the A/B switch and I can now say with certainty that (to my ears) the 2 DACs are indistinguishable. This is good since it proves that the effect the resistor change had was reproducible and consistent.
I now have a test-bed to test further alterations on one and compare again.
What I don't know is whether the change in spacial definition and tone was due to the resistors themselves or the fact that they are now far more closely matched. The obvious way to test that would be to buy an excess of the Susumo RR types and match them up in the same way and swap them out in one DAC. I am just not sure I care enough though to put myself through that. lol.
I will try Mark's quick test of the resistors shunting the shift reg decoupling caps since that is easy to implement and remove without too much trouble. I did try it previously but not having two DACs that matched up sound-wise at the time I couldn't tell where I was. (At the time if there was a difference it was swamped by the change that the resistors made)
I now have a test-bed to test further alterations on one and compare again.
What I don't know is whether the change in spacial definition and tone was due to the resistors themselves or the fact that they are now far more closely matched. The obvious way to test that would be to buy an excess of the Susumo RR types and match them up in the same way and swap them out in one DAC. I am just not sure I care enough though to put myself through that. lol.
I will try Mark's quick test of the resistors shunting the shift reg decoupling caps since that is easy to implement and remove without too much trouble. I did try it previously but not having two DACs that matched up sound-wise at the time I couldn't tell where I was. (At the time if there was a difference it was swamped by the change that the resistors made)
I had chance this afternoon to try Mark's mod bypassing the shift-reg decoupling caps with resistors.
I can report that it does make a difference and I believe that the difference is positive. Subtle but positive. I have not yet worked out what the acoustic differences are so I will continue to listen to both during the rest of the working days this week.
I can report that it does make a difference and I believe that the difference is positive. Subtle but positive. I have not yet worked out what the acoustic differences are so I will continue to listen to both during the rest of the working days this week.
With a little practice and careful setup, these tweezer tools can remove smd resistors, ceramic caps, and even smd film caps quickly and without damage. I usually add a little past flux to the component ends first. https://www.amazon.com/Portable-Tweezers-Soldering-Parallel-Repairing/dp/B09L4BM33P/ref=sr_1_6?crid=3KSD2Q7KQYN3N&keywords=soldering+tweezer+smd&qid=1701196305&sprefix=soldering+tweezer+smd,aps,210&sr=8-6
I decided to integrate my STM32F7 USB-I2S board with PCM2DSD to have RTZ dac with PCM input.
RTZ board and the USB-DSD board have same form factor and can be connected directly with pin header or if needed with MMCX for DSD + MCK. The USB-DSD board is quite versatile as in addition to isolated DSD it has isolated I2C and GPIOs for connecting to e.g. AKM or ESS dacs. It has NZ2520SDA clocks but can also use external clocks instead via MMCX. It also supports displays and encoders.
Here with a 128x160 IPS display:
Admittedly encoder or display is not very useful for RTZ dac as it does not have volume control or any settings. But with AKM or ESS dacs encoder and display are definitely needed.
I had some issues running the "latest" pcm2dsd firmware as it did not work properly for 44k1 sample rate (or multiples). I need to check that further. I'm currently using the firmware by @xx3stksm (https://www.diyaudio.com/community/threads/simple-dsd-modulator-for-dsc2.370177/post-6671447) as it runs without any hickups.
Here is 1k -4bFS @88k2:
My plan is to fit the dac, display and PSU ("SilentSwitcher") in a 160x100mm Hammond 1455 box. I just need to make a front panel board for the display and encoder.
RTZ board and the USB-DSD board have same form factor and can be connected directly with pin header or if needed with MMCX for DSD + MCK. The USB-DSD board is quite versatile as in addition to isolated DSD it has isolated I2C and GPIOs for connecting to e.g. AKM or ESS dacs. It has NZ2520SDA clocks but can also use external clocks instead via MMCX. It also supports displays and encoders.
Here with a 128x160 IPS display:
Admittedly encoder or display is not very useful for RTZ dac as it does not have volume control or any settings. But with AKM or ESS dacs encoder and display are definitely needed.
I had some issues running the "latest" pcm2dsd firmware as it did not work properly for 44k1 sample rate (or multiples). I need to check that further. I'm currently using the firmware by @xx3stksm (https://www.diyaudio.com/community/threads/simple-dsd-modulator-for-dsc2.370177/post-6671447) as it runs without any hickups.
Here is 1k -4bFS @88k2:
My plan is to fit the dac, display and PSU ("SilentSwitcher") in a 160x100mm Hammond 1455 box. I just need to make a front panel board for the display and encoder.