ProtoDAC TDA1387 X8 project

[hifinet]

Yes, it says the Q7II comes with a higher grade FemtoMck clocks. No FIFO memory, but MPD already has a buffer. See the comparison chart . No programmable delay, so you need a manual delay (see @michaelagiles post).

Power supply would be the 2S LifePO4/LT1963. Hopeful about this. 

When you have another board like the IsolatorPi, you have an extra EMI shield.

[wolf7250]

When you say the MA has no programmable delay and that a manual delay is needed, are you referring to @michaelagiles post referring to adding a switch to the power line?

I'm really interested in your 2S LifePO4/LT1963. I saw the links to AliExpress that you shared before, but I confess I can't picture how it would all come together. Are you able to share a bit more detail about your idea for how this would look as a finished product?

[hifinet]

I think a delayed turn on is needed based on @michaelagiles post.

The BMS is needed to provide balanced charge to the two cells and to prevent overcharging and overdischarge. Connections are per the schematic. You need to use the one for lifepo4 2S 6.4V. Details below. Need a series type two battery holder for the two 18650 cells. Need a wire attached to the connection plate between the two batteries and connected to BM on the board. Charger and load connect to the P+ and P-. During use, it's connected to the 5V LT1963 regulator and then to the 5V stack through Isolator. Since there is no connection to a power supply, there is no noise. Disconnect and charge overnight. It says 7.4V charging voltage is recommended. It might need a cheap DC DC converter, if I can't find an appropriate voltage SMPS. I think Ian is using Hakadi (same part number). Trying to verify ESR. I might order these instead, with ESR < 15, since I need to order more parts for ProtoDAC.

I always keep my DIY stuff on ceramic flooring tiles, just in case of fire. Would recommend keeping the batteries on these during charging.

When I used the 2S LiPO, I went cheap and just got the balanced 2S charger. At first, I had to watch the voltage during use to make sure I wasn't overdischarging. Once I had an idea how long it would operate on a charge, I didn't need to monitor voltage. It could run the stack almost 8 hours, which would be a very long listening session. The BMS device combines a balanced charger with overdischarge protection and are now very inexpensive.

Lifepo4 2S 6.4V 15A/20A/25A BMS
Applicable 3.2V lifepo4 battery(Lithium Iron Phosphate)
Connection method:Common Port
Over-discharge protection voltage :2.0V±0.1V
Continuous working current :15A/20A/25A
Instantaneous protection current:65A/80A/90A

Overcurrent recovery conditions :charge recovery or disconnect load                          

Short-circuit recovery conditions :charge recovery or disconnect load
Overcharge protection voltage :3.7V±0.1V
Continuous charging current :15A/20A/25A
Charger voltage :7.4V (recommended)
Activation method : (1) B- and P- wires can be short-circuited once
(2)Recharge and activate
Work power consumption:≤10uA
Main circuit conduction internal resistance:≤5mΩ
Temperature range :-40℃ ~ 80℃
Appearance size:60*25*3.5 MM
other instructions:15A BMS=6MOS  20A BMS=8MOS  30A BMS=10MOS
Due to different production batches, we will randomly send blue or green or red circuit boards.

[hifinet]

Nelson Pass often uses SMPS for his preamp designs. He triple filters with simple RC filters. I have also seen him use LM317 (60dB ripple rejection) linear regulators at the end. The SMPS DC Filter P089ZB Kit and the other filter posted by @wolf7250 specifically filter noise typically generated by SMPS. These techniques may be enough to adequtely filter noise, and provide continuous power.

I found the LT1963 regulator in this article of listening tests of various regulators. The LT1963 was the highest rated for an IC regulator. I think it sounds very similar to the Chinese Jung Super regulator I tried. It is an LDO with fast transient response and needs the proper cap on the output to prevent ringing (see the datasheet).

Transient response determines a lot of the sound of regulators, ultracaps and batteries. In my tube amps, I use the Maida LM317 regulator. This uses a 2R7 and 1uF mylar cap snubber at the output, which affects the sound quite a bit. It dampens the transient response. Similar to the 10uF ESR up to 3R with the LT1963. Ultracaps have super fast and hard transient response, with no overshoot or ringing, because they can supply huge amounts of instantaneous current. See here and here . LifePO4 vs UC . There are still the sharp vertical lines (barely visible) in the Uc transient response images, probably from stray capacitance/inductance. Sometimes I do hear some harshness (= ringing) with the PurePi 5V UC supply. I wonder if a snubber would help this?

[hifinet]

Started with the snubber from the Maida regulator (2R7 and 1uF mylar cap), and so far I think it helps the UC. There are snubber calculators, but it's hard to determine the inputs. Just fiddle with the values. They generally won't do anything bad. They are basically just shunting (dampening, tuned RC) high frequency to ground.

Type of music? Operatic soprano is usually a good one to trigger harshness. Reminds me of a Renee Fleming concert. Sitting in the first balcony. Her high notes were so powerful. No harshness, just a very loud fundamental frequency, even from the balcony.

Snubbers can also be useful across the secondary winding of audio transformers (see the W5M and the 47R 0.1uF snubber). Also, across standard diodes in power supplies. The sharp on/off behavior can induce transients in the inductance of the power transformer.

[wolf7250]

Nelson Pass often uses SMPS for his preamp designs. He triple filters with simple RC filters. I have also seen him use LM317 (60dB ripple rejection) linear regulators at the end. The SMPS DC Filter P089ZB Kit and the other filter posted by @wolf7250 specifically filter noise typically generated by SMPS. These techniques may be enough to adequtely filter noise, and provide continuous power.

I'm going to give the diyAudio SMPS DC Filter P089ZB Kit a go. Have ordered it and will give it a crack.
Haven't done much soldering... between this kit and the ProtoDAC I'll probably have a bit more experience. Looking forward to learning. I will be watching YouTube tutorials etc before beginning   .

Thanks @hifinet for all the information and advice!

[hifinet]

The difference between the two is the one you posted uses a common mode choke, which is connected to both the positive and ground. The P089ZB just has a choke on the positive. But if you can't get the parts, you can't build it. There is a thread in diyA about SMPS and using common mode chokes, and I don't think they could fully remove the noise with the common mode choke either. You just need to lower it enough that you can't hear it.

[wolf7250]

Interesting... so it's better to have the common mode choke connected to both the positive and ground?
I had already ordered the PCB's for the filter I posted. So if the parts become available in the future I might order them and compare the diyAudio filter with it.
Yeah, in my mind, if the noise is lowered enough that I can't hear it then I'm not fussed about whether it's fully removed or not.

[hifinet]

Looking at obtaining the parts for CMC filter. Everything is available except for the CMC. Parts cost would be $6.40. Mouser has the part on reorder with a long reorder time. The product is discontinued by the manufacturer. It might be available on the surplus market. 

The main problem with this design is that it is SMD, which locks you into specific size parts. This is critical with the SMD CMC. These parts have similar specs, but the footprints are different. Substituting through hole parts would make it easier. It could be constructed on a perf board. Soldering SMD is also very difficult with an iron.

UPDATE: I actually did find the CMC at LCSC (LCSC Part # C710586), which is the parts supplier for JLCPCB. They have 5 in stock. They are missing 4 parts from the BOM.

CMCs are also available on Ali. They are so cheap, they usually sell in multiples. A CMC is nothing more than two pieces of enameled wire wound around a core. The dot in the schematic indicates the top of the winding, and the lead numbers in the part diagram need to be followed. In this circuit, both wires start winding at the same end of the core. Common mode signals such as external EMI, RFI, etc. (identical signals in both + and -) cancel, and differential signals (5V) pass unimpeded. The core material and shape can affect the performance, as well as the windings, type of wire, etc. You just need to find one with similar specs that will fit the SMD pads on your PCB.

The inline choke (L2) helps to filter switching noise from the SMPS on the + line, along with the ceramic and tantalum caps on the output, which comprise an LC low pass filter.

It can help to wrap the SMPS lead with a clip-on ferrite bead or even better, wind it several times through a ferrite toroid. The long wire acts as an antenna, and it can feedback RFI into the SMPS error amplifier.

[hifinet]

Received data on the Ian Canada LifePO4 batteries from Hakadi. The HNGN038ELC measured internal resistance 23.2 mOhm (1.8Ah version) and 21.25 mOhm (1.5Ah version) for two samples. Same part number, but lower capacity than the IC batteries, which are 2.0 Ah. Maybe a special order, since they are not available in the Hakadi store.