I did an interesting set of comparisons last night last night. To help another Ian GB FiFoPi / DM DAC user troubleshoot his setup, I did some comparison listening.
I powered my Ian GB stack and let it warm up still running a modified IVStd with 2 NewClassD Ultimate discrete opamps for a balanced output. After a short ear-calibration listen, I first put in the LL1544A board, then the Bisesik transformer output board.
What I heard surprised me. The LL1544A board sounded MUCH better than I remembered. THEN I remembered that since I last put it in this setup, (sometime early-middle last summer), I made the following changes:
* Replaced 2 3.3V LiFePO4 rails powering the FiFoPi and DM 9038Q2M DAC to 2 Uptone-Audio LPS-1.2 float-charging 2 350F UC pairs.
* Moved from a stock RPi 2B to one modified with linear regulators replacing the onboard DC-DC converters for the 3.3V & 1.8V rails to an Allo USBBridge Signature.
* Replaced a modified K&K Low Voltage supply powering the RPi/Modified RPi/USBBridge Sig with a modified Allo Shanti and then enhancing it by adding a 350F UC pair output buffer on the Shanti.
* Replaced my LPS-1.2-powered ZyXEL switch, modified Jameco 5V linear-regulated wall-wart-float-charged 5V battery-powered Trendnet sending FMC, and LPS-1.2-powered modified Delock receiving FMC network with a 2 LPS-1.2 paralleled into an MPAudio dual section 3||LT3045 board-powered Uptone Audio EtherRegen.
* Upgraded the 3 pairs of 350F Maxwell UCs to their newer 325F units.
Ok, that was a lot of changes in that 6 month timeframe. AND none were small improvements, with the UC power, the USB Bridge Signature, and the EtherRegen each being a very significant improvement.
So while I would not have called it a bad output stage or a slouch before, NOW the LL1544A setup was very listenable. Smooth, no digitalitus, fairly detailed.
Then I put in the Bisesik transformers. Better with more delicate and detailed highs, better separation of instruments/voices, wider soundstage. I heard a small thing in one recording I’d never noticed before… an electric bass note being held at end the of the vocal 1st part of a song as it went into an instrumental section… and then clearly hearing when it ended after the initial sax solo started. I more easily heard/noticed little things like that with the Bisesik transformers compared to the LL1544A outputs.
I also swapped back to powering the FiFoPi and DM DAC from the 2 3.3V rails on Ian's LiFePO4 PS. This was from my now-standard Ultracap supplies where I use a pair of 325F/2.7V Maxwell UCs in the configuration of Ian's UCPi circuit and float charge each pair with an Uptone Audio LPS-1.2 Ultracap supply which is a bank-switching UC setup where the bank supplying the output is not connected to the charging circuits, so off-the-grid (and therefore free of any AC line noise AND AC-DC conversion noise such as transformer and diode ringing.)
The LiFePO4 3.3V rails powering the FiFoPi and DM DAC were definitely a step back, less palpable, less delineated, less dynamic and lifelike. Also less natural.
Finally I put my modified IVStd with the NewClassD discretes back in and heard what I liked with it as it got warmed up (of course, not an issue with the transformers… after break-in, they just start playing!). AN interesting alternative to the transformers, I put that setup roughly on par with the Biseseks with some differences in strengths/weaknesses.
So in my setups and to my ears, I could live with the LL1544A. I heard no obvious issues except they were not as good overall as my two better output setups.
Greg in Mississippi
P.S. Clocks in the FiFoPi were a pair of 45/49 NDK SDA's and the DM DAC is run in True Sync mode.
I powered my Ian GB stack and let it warm up still running a modified IVStd with 2 NewClassD Ultimate discrete opamps for a balanced output. After a short ear-calibration listen, I first put in the LL1544A board, then the Bisesik transformer output board.
What I heard surprised me. The LL1544A board sounded MUCH better than I remembered. THEN I remembered that since I last put it in this setup, (sometime early-middle last summer), I made the following changes:
* Replaced 2 3.3V LiFePO4 rails powering the FiFoPi and DM 9038Q2M DAC to 2 Uptone-Audio LPS-1.2 float-charging 2 350F UC pairs.
* Moved from a stock RPi 2B to one modified with linear regulators replacing the onboard DC-DC converters for the 3.3V & 1.8V rails to an Allo USBBridge Signature.
* Replaced a modified K&K Low Voltage supply powering the RPi/Modified RPi/USBBridge Sig with a modified Allo Shanti and then enhancing it by adding a 350F UC pair output buffer on the Shanti.
* Replaced my LPS-1.2-powered ZyXEL switch, modified Jameco 5V linear-regulated wall-wart-float-charged 5V battery-powered Trendnet sending FMC, and LPS-1.2-powered modified Delock receiving FMC network with a 2 LPS-1.2 paralleled into an MPAudio dual section 3||LT3045 board-powered Uptone Audio EtherRegen.
* Upgraded the 3 pairs of 350F Maxwell UCs to their newer 325F units.
Ok, that was a lot of changes in that 6 month timeframe. AND none were small improvements, with the UC power, the USB Bridge Signature, and the EtherRegen each being a very significant improvement.
So while I would not have called it a bad output stage or a slouch before, NOW the LL1544A setup was very listenable. Smooth, no digitalitus, fairly detailed.
Then I put in the Bisesik transformers. Better with more delicate and detailed highs, better separation of instruments/voices, wider soundstage. I heard a small thing in one recording I’d never noticed before… an electric bass note being held at end the of the vocal 1st part of a song as it went into an instrumental section… and then clearly hearing when it ended after the initial sax solo started. I more easily heard/noticed little things like that with the Bisesik transformers compared to the LL1544A outputs.
I also swapped back to powering the FiFoPi and DM DAC from the 2 3.3V rails on Ian's LiFePO4 PS. This was from my now-standard Ultracap supplies where I use a pair of 325F/2.7V Maxwell UCs in the configuration of Ian's UCPi circuit and float charge each pair with an Uptone Audio LPS-1.2 Ultracap supply which is a bank-switching UC setup where the bank supplying the output is not connected to the charging circuits, so off-the-grid (and therefore free of any AC line noise AND AC-DC conversion noise such as transformer and diode ringing.)
The LiFePO4 3.3V rails powering the FiFoPi and DM DAC were definitely a step back, less palpable, less delineated, less dynamic and lifelike. Also less natural.
Finally I put my modified IVStd with the NewClassD discretes back in and heard what I liked with it as it got warmed up (of course, not an issue with the transformers… after break-in, they just start playing!). AN interesting alternative to the transformers, I put that setup roughly on par with the Biseseks with some differences in strengths/weaknesses.
So in my setups and to my ears, I could live with the LL1544A. I heard no obvious issues except they were not as good overall as my two better output setups.
Greg in Mississippi
P.S. Clocks in the FiFoPi were a pair of 45/49 NDK SDA's and the DM DAC is run in True Sync mode.
Last edited:
Hi Greg,
That’s correct. It’s 5v from a Reflektor D into the isolator board. I have a Pi Model 3B so the mezzanine board should be okay. Also got a Kali here (running NDK SDA clocks) and an Allo isolator on the way.....should make for an interesting comparison.
James
Hi Greg,
Not surprising to me that you observed such findings. Happy to hear of your experiments.
I don't use an isolator pi, but by chance last week I noticed further isolation benefits the FIFO. To my surprise.
I have upgraded to an Allo USB signature (well worth it - if your using USB as an input! but I haven't tried it as an I2S source and doubt I will) and I was experimenting with different powering options for the USB input board (JL Sounds).
When I powered the isolator from the same source, IE grounds shared on both sides of the USB board and therefore eliminating the benefit of the isolator. The SQ became less natural, less delineation/blackness between individual components etc.
I then decided to use two different supplies (two 5v's from a Shanti) and tested by combining both grounds and it was easy to replicate the loss in SQ.
Thanks, Greg, really good insight. I would be interested to know if you tried any other discretes, where you hear differences between the two different output approaches and how you power them.
I am quite interested to have a play with the IVstd.
What does it 'typically' mean when the 'ESS Controller' stays on the display, and the periods (.) move from left to right and then repeat. After many months of my Ian stack working perfectly, it now does not. I will have more to post later, but was just curious of typical symptoms when the display acts like above.
It means the controller has no serial comms with the DAC. Check all your power connections and voltages, and if you've unplugged anything from the stack make sure it was lined up properly when you plugged it back in...(I've been guilty of that!)
After measuring voltages before and after regulators, all voltages going to their respective boards were spot on, including DVCC, VCCA, and AVCC.
I first noticed the DAC not working when it did not show up as one of my roon endpoints a couple days ago. As it was not working, I unplugged it and left it in my rack. When I did take it out and put it on my bench, it powered up fine, roon recognized it, and it played and sounded beautiful - on my bench. Took it back out to my rack and no go. Nothing. Urgh!
During this whole time, since it went into service, it stayed powered on and was not opened up, nor any changes made to the inside, or DAC stack. I will say, my enclosure does not have any ventilation of the inside. So, possibly something got 'cooked' the slow way despite minimal heat being generated by the PSUs and other supporting subsystems.
My son flashed the latest version of the roon bridge software and we were able to get the DAC to be recognized by roon, but still no audio at all. I've tried different ethernet cables, different ports, even a USB based DAC plugged into the RPi, still nothing. I'm using the standard I/V output stage with OPA1612 op amps with little heatsinks on top. Even swapped out the RPi.
A good reference for handling U.fl connections. Three Quick Tips About Using U.FL - learn.sparkfun.com
If you are using sync mode, I’d turn it off and pull the cable to see if you do get a connection to the controller.
I don’t quite understand one point. You plugged in a usb dac to the Rpi with the stack on top and don’t get music from the USB dac? Could there be a larger underlying issue?
If you are using sync mode, I’d turn it off and pull the cable to see if you do get a connection to the controller.
I don’t quite understand one point. You plugged in a usb dac to the Rpi with the stack on top and don’t get music from the USB dac? Could there be a larger underlying issue?
@ozcal, See post #335 of this thread:
Shanti Dual LPS 5V/3A , 5V/1.5A
@Brumjam, Looking forward to your results. When I first received a FifoPi, I did compare it against an Isolator / Kali combo. I preferred the FiFoPi, but it had the better SDA version NDKs which were not, at that time, fully broken in. It was not a huge difference.
Related, back in the day I did compare Ian’s Isolator to the original Allo one. I preferred the Allo one (also slightly). I assume due to the slightly better regulation.
Based on that and that I haven’t upgraded any Kali to use the SDA version NDK’s, I haven't repeated that test. Also note Ian’s current version of the FiFoPi has some small improvements over the original ones shipped, as Ian noted in one of his main-line threads.
@sebbyp, Interesting result. On other opamps in the IVStd, I have tried (in order of preference):
- The stock opamps Ian included with the ready-to-use boards.
- OPA1642A (very nice mids, slightly trashy highs, MUCH better and quite listenable with revised filtering on the IVStd.)
- LKS discretes
- Burson V5 Vivid
- OPA1612 (very even performance top to bottom, very detailed, many would prefer it to my preference over it, the Sparkos discretes.)
- Sparkos discretes (a tiny bit more lifelike to my ears than the OPA1612, but a tiny bit less detail. An interesting contrast to the OPA1612.)
- NewClassD Ultimate discretes (more natural sounding and detailed than the Sparkos. The Sparkos beat every other top output setup on Ian’s DM DAC for me except the Bisesik transformers. I’d put these discretes roughly even with the Bisesiks with a different set of strengths/weaknesses. NO, I haven’t spent a lot of time trying to characterize the differences… my setups have gotten to the point I’d rather just listen!)
My mods to the IVStd are fairly simple:
- Replace the 4 270pF IV stage feedback filtering caps with 1500pF Panasonic PPS caps.
- Add a “Joe Rasmussen pre-IV filter” consisting of a 1uF Rubycon PML across the + & - DAC outputs per channel with a 3.3R Susumu resistor in series in each signal line after the cap.
Note that you may have issues with noise pickup with the pre-filter. I didn't, Ian did when he tried it. BUT he uses the WiFi on his RPi 3B & I did not in any of my setups.
My experience is that with these changes all of the opamps sounded somewhat better afterwards. I assume that the filtering lowered the amount of out-of-band high frequencies that the opamps had to handle. The biggest gain was with the OPA1642A, which went from ‘interesting with a great midrange’ to an overall very good opamp. OTOH, the OPA1612 is overall better and not THAT much more expensive.
I have a pair of 2 OPA627’s on dual adapters here to try at some point.
@markw4, an interesting 6 months. With the changes I did during that time, I’m finding it harder and harder to listen to the gear and just want to enjoy the music, both on this setup and my very similar Allo Katana setup.
Greg in Mississippi
Shanti Dual LPS 5V/3A , 5V/1.5A
@Brumjam, Looking forward to your results. When I first received a FifoPi, I did compare it against an Isolator / Kali combo. I preferred the FiFoPi, but it had the better SDA version NDKs which were not, at that time, fully broken in. It was not a huge difference.
Related, back in the day I did compare Ian’s Isolator to the original Allo one. I preferred the Allo one (also slightly). I assume due to the slightly better regulation.
Based on that and that I haven’t upgraded any Kali to use the SDA version NDK’s, I haven't repeated that test. Also note Ian’s current version of the FiFoPi has some small improvements over the original ones shipped, as Ian noted in one of his main-line threads.
@sebbyp, Interesting result. On other opamps in the IVStd, I have tried (in order of preference):
- The stock opamps Ian included with the ready-to-use boards.
- OPA1642A (very nice mids, slightly trashy highs, MUCH better and quite listenable with revised filtering on the IVStd.)
- LKS discretes
- Burson V5 Vivid
- OPA1612 (very even performance top to bottom, very detailed, many would prefer it to my preference over it, the Sparkos discretes.)
- Sparkos discretes (a tiny bit more lifelike to my ears than the OPA1612, but a tiny bit less detail. An interesting contrast to the OPA1612.)
- NewClassD Ultimate discretes (more natural sounding and detailed than the Sparkos. The Sparkos beat every other top output setup on Ian’s DM DAC for me except the Bisesik transformers. I’d put these discretes roughly even with the Bisesiks with a different set of strengths/weaknesses. NO, I haven’t spent a lot of time trying to characterize the differences… my setups have gotten to the point I’d rather just listen!)
My mods to the IVStd are fairly simple:
- Replace the 4 270pF IV stage feedback filtering caps with 1500pF Panasonic PPS caps.
- Add a “Joe Rasmussen pre-IV filter” consisting of a 1uF Rubycon PML across the + & - DAC outputs per channel with a 3.3R Susumu resistor in series in each signal line after the cap.
Note that you may have issues with noise pickup with the pre-filter. I didn't, Ian did when he tried it. BUT he uses the WiFi on his RPi 3B & I did not in any of my setups.
My experience is that with these changes all of the opamps sounded somewhat better afterwards. I assume that the filtering lowered the amount of out-of-band high frequencies that the opamps had to handle. The biggest gain was with the OPA1642A, which went from ‘interesting with a great midrange’ to an overall very good opamp. OTOH, the OPA1612 is overall better and not THAT much more expensive.
I have a pair of 2 OPA627’s on dual adapters here to try at some point.
@markw4, an interesting 6 months. With the changes I did during that time, I’m finding it harder and harder to listen to the gear and just want to enjoy the music, both on this setup and my very similar Allo Katana setup.
Greg in Mississippi
Last edited:
Finally got my DAC/Stack working again! I believe there were a couple of different issues that caused to stop working out of the blue. The first thing I did (actually my son) did was to reflash the microSD card with roonBridge. I believe it may have gotten corrupted during a push update, but not sure.
After opening it up and undoing the stack for possible issue and finding nothing, I put everything back together. One thing I did notice was the small control display board, was not seated as tight as it should have been. Since it aces the back of the cabinet, and being inside I never see or use it. I reseated all the boards and double-checked all power connectors. Fired it up and music was once again playing! I did loop into the power amp trigger so it's not on all the time. Only when using the amp.
There are still a couple more mods I want to try; (1) using sync mode, and (2) I still need to add my film caps at the I/V power input connector.
It sounds so good, I can't top listening to it!
It's on most of the day while I'm in my office. It continues to sound better and better and finally without the fatiguing characteristic of the sound.I got a new Parasound Amp over the holidays, and it sound wonderful to my ears. I feel as if I finally got some gear that approaches reference level quality
Edit: I also was able to get my hands on 2 of the First Once mosfet amp modules. Now I can repair my first flagship amp that sounded soooo beautiful and articulate.
Last edited:
Hello Ian and his customers,
I have been trying to collect the information i need by reading threads regarding your circuits but somehow my mind gets mixed up all the time.
As you might have noticed there are people in the dddac thread on this site that are interested in trying your creations. I am one of them.
Right now i am using a pimped dddac which is connected by wave io to an Aurender mediaplayer.
I wanna improve this set up by adding some parts made by you. Maybe in the end i will replace the Aurender by another mediaplayer that allows me to use an I2S connection.
Right now i use choke input power supply for the wave io that needs a max current below 500 mA.
What do i need to improve my present set up.?
Your lifepo4 power supply with supercaps added. That is one 3,3 for the accusilon clocks. The other supply on the fifopi board is also 3,3 but it takes quiet some current when there is an RP board connected to it. I could add some 3,3 volts in parallel to take care of the extra current or is this not necessary?
There is a 5 volt power supply that can be improved by adding a little circuit board would that be an improvement over my choke input that is used now for the wave IO?
Normally my dddac is switched on all the time. Both the 12 and 5 volt power supply. Of course your supply needs charging.
Is it possible to use your board and just let it stay in charging mode when i am sleeping so eight hours a day or just stop the charging mode when i am listening?
The next upgrade on your board will make the installing of supercaps easier because it will have some current protection.
Thank you for explaining.
I will copy and paste your info to the dddac thread because there is a rising interest because the ddddac designer has made some test up with your fifopi.
Sincere greetings, eduard
P.s when will the updated board be on the market?
P.s as you can see in the attachments i work with metal so it will be easy to make the positioning of the circuit close to each and still be flexible for modifications
I have been trying to collect the information i need by reading threads regarding your circuits but somehow my mind gets mixed up all the time.
As you might have noticed there are people in the dddac thread on this site that are interested in trying your creations. I am one of them.
Right now i am using a pimped dddac which is connected by wave io to an Aurender mediaplayer.
I wanna improve this set up by adding some parts made by you. Maybe in the end i will replace the Aurender by another mediaplayer that allows me to use an I2S connection.
Right now i use choke input power supply for the wave io that needs a max current below 500 mA.
What do i need to improve my present set up.?
Your lifepo4 power supply with supercaps added. That is one 3,3 for the accusilon clocks. The other supply on the fifopi board is also 3,3 but it takes quiet some current when there is an RP board connected to it. I could add some 3,3 volts in parallel to take care of the extra current or is this not necessary?
There is a 5 volt power supply that can be improved by adding a little circuit board would that be an improvement over my choke input that is used now for the wave IO?
Normally my dddac is switched on all the time. Both the 12 and 5 volt power supply. Of course your supply needs charging.
Is it possible to use your board and just let it stay in charging mode when i am sleeping so eight hours a day or just stop the charging mode when i am listening?
The next upgrade on your board will make the installing of supercaps easier because it will have some current protection.
Thank you for explaining.
I will copy and paste your info to the dddac thread because there is a rising interest because the ddddac designer has made some test up with your fifopi.
Sincere greetings, eduard
P.s when will the updated board be on the market?
P.s as you can see in the attachments i work with metal so it will be easy to make the positioning of the circuit close to each and still be flexible for modifications
