I have used the THS without those feedback caps. Now I use NE5532 with the caps
I will adjust MSB too this time can do this with oscilloscope
I will adjust MSB too this time can do this with oscilloscope
RPI has quite high jitter with internal clocks (much higher than PCM2706)
Raspberry Pi I2S output working | Crazy Audio
Using external USB-I2S device (like xmos) could improve the sound (I have not tested, @Ripster tested it here: https://www.diyaudio.com/forums/dig...862-tht-i2s-input-nos-2r-138.html#post6490317 )
I think the hat which is reclocker is ok and in pair with external usb about the jitter as both are asynchronous. The hat is even better as the signal is partially buferred (fifo ). What I wanted to highligth is the GPIO is not reclocked imo just has a galvanic isolation. So as you say, the clock is comming from the unique 15 Mhz XO reclocked by the intrnal FGPA of the Rpi. This should not be seen at this level of DAC quality 🙂 , cause it's big waste imo. Hence I said it could be enhanced by taking the clock from the uf-l which are the only reclocked I2S outputs on that hat ! And also Ernst had a problem with one of his Rpi clock as well : no sound !
But also the pic was just a POC proto as well.
@ Tubee, so what tubes will you use please , ecc88 ? 6N23P ? Maybe at the end of the year I would use tubes, a mate of mine said one is not needing expensive trafo for the plate, Moglia's pcb or KRK ones are much cheaper and close as good than a good step down traffo when noise is a problem...
Well I do know nothing about tubes 🙁.
Last edited:
I am planning ef86 in triode (low distorsion) and maybe 6n16 mini tube srpp.
But think of transformer output too as they can sound delightful
But think of transformer output too as they can sound delightful
I am planning ef86 in triode (low distorsion) and maybe 6n16 mini tube srpp.
But think of transformer output too as they can sound delightful
Too much expensive for me (OT), but I can afford a good tin foil from SCR... Thorsten liked it a lot when it comes to save from the cost of a good OT (of course better, but it's expensive to dance the Tango nowadays), if I remmeber the TDA1541 designs of him...
Sorry if this question has been covered, I couldn't find an answer above.
I've built a couple of Power Supply 2s, which work nicely, and also an 1865 DAC which I haven't fired up yet. The 1865 board needs VData +5 VDC and VAudio/analog +/-5 VDC. The schematic doesn't indicate that VD ground and VA ground are independent and the meters showed these pins to be connected together, so as part of the pre-chip-insertion tryout I fed VD+ and VDground from VA+ and VAground, respectively.
The power supply definitely didn't like this, acting as if there was a short. Meter says no connection between VD+ and VA+. There's a little conduction between + and ground on VA and VD, and between - and ground on VA, I think as the filter capacitors charge?
Any ideas why this isn't working? I'm happy to use two separate power supplies if that's the answer but also want to learn what I'm missing here.
I've built a couple of Power Supply 2s, which work nicely, and also an 1865 DAC which I haven't fired up yet. The 1865 board needs VData +5 VDC and VAudio/analog +/-5 VDC. The schematic doesn't indicate that VD ground and VA ground are independent and the meters showed these pins to be connected together, so as part of the pre-chip-insertion tryout I fed VD+ and VDground from VA+ and VAground, respectively.
The power supply definitely didn't like this, acting as if there was a short. Meter says no connection between VD+ and VA+. There's a little conduction between + and ground on VA and VD, and between - and ground on VA, I think as the filter capacitors charge?
Any ideas why this isn't working? I'm happy to use two separate power supplies if that's the answer but also want to learn what I'm missing here.
@Paddy ... connect each side of the power supply (5VD, +5VA, -5VA) through 3R3 resistors (except GND which is connected directly) and see what happens. Measure voltage drops across resistors 🙂
Power Supply 2s is the "PSU 2: LT1963A/LT3015"?
Power Supply 2s is the "PSU 2: LT1963A/LT3015"?
Paddy,
Yes, you need two separate supplies.
Not good to “borrow” the +5v from the analog psu rail for both 5v dac chip requirements.
Yes, you need two separate supplies.
Not good to “borrow” the +5v from the analog psu rail for both 5v dac chip requirements.
Yes, the LT3015/LT1963 power supply. With 3R3 resistors in series on +5 and -5, there is no voltage drop across the resistor but there is also no apparent problem for the power supply.
I assume they're acting to limit current?
I assume they're acting to limit current?
@Paddy
Most LDO regulators don't like a high value of low-ESR capacitance on the output. When two rails were connected together, capacitor C11 increased total output capacitance and lowered the ESR for the LDO power supply, which apparently created the problem
It is hard to estimate the maximum "limit" for capacitors, and if such problematic situation happens, jumpers on the PCB (on power rails) come in the play - resistors are used there (low value like 2R2, 3R3). Thanks to this resistance, the power supply is then stable for a high capacitance.
Regulators 78xx/79xx have an emitter folower stage with no gain, so the big output low-esr capacitance makes no difference to the phase shift, thus stays stable for high capacitive low-esr loads, but the drawback may be higher noise
I added this information about PSU 2 on the main page 😱
Most LDO regulators don't like a high value of low-ESR capacitance on the output. When two rails were connected together, capacitor C11 increased total output capacitance and lowered the ESR for the LDO power supply, which apparently created the problem

It is hard to estimate the maximum "limit" for capacitors, and if such problematic situation happens, jumpers on the PCB (on power rails) come in the play - resistors are used there (low value like 2R2, 3R3). Thanks to this resistance, the power supply is then stable for a high capacitance.
Regulators 78xx/79xx have an emitter folower stage with no gain, so the big output low-esr capacitance makes no difference to the phase shift, thus stays stable for high capacitive low-esr loads, but the drawback may be higher noise

I added this information about PSU 2 on the main page 😱
Last edited:
And today I learned something! Thanks very much.
PS Vunce - excellent point, it'll have its own dedicated supply in its forever home...
PS Vunce - excellent point, it'll have its own dedicated supply in its forever home...
Last edited:
Interesting reading on the PSUs there. I have built a PSU 1 hot rod using Aliexpress regulators that use the Ti TPS7A chips. Have had good results with them in my Cambridge Audio Dacmagic.
I was thinking just recently if the 4700uf on the outlets would suit them . Have read the datasheets.
I guess I will try the PSU with them and maybe with some lower values too.
I was thinking just recently if the 4700uf on the outlets would suit them . Have read the datasheets.
I guess I will try the PSU with them and maybe with some lower values too.
Attachments
Nice looking psu board Jim!
Be careful with the capacitor size and type downstream of the LDO regulators. They could become unstable. Refer to the Ti TPS7Axxxx data sheet.
Be careful with the capacitor size and type downstream of the LDO regulators. They could become unstable. Refer to the Ti TPS7Axxxx data sheet.
Interesting reading on the PSUs there. I have built a PSU 1 hot rod using Aliexpress regulators that use the Ti TPS7A chips. Have had good results with them in my Cambridge Audio Dacmagic.
I was thinking just recently if the 4700uf on the outlets would suit them . Have read the datasheets.
I guess I will try the PSU with them and maybe with some lower values too.
I don't know why but I always found them better than the more often seen LT304x... Go figure why ?!😕
Thanks Vunce. Ds doesn't allude to much so I shall suck and see. They were cool with no load.
Diyiggy...I believe they have lower impedance than LT304x despite being noisier.
Diyiggy...I believe they have lower impedance than LT304x despite being noisier.
@jimk04 ... you did very nice mod on the PSU 1 
There can be instability with TPS7A regulator, because there is not output resistor in the functional block diagram. If you find any instability (check with scope - during start-ups, during different loads), just scratch the top soldermask, break traces and bridge the traces with 2R2 or 3R3 resistors 😉

There can be instability with TPS7A regulator, because there is not output resistor in the functional block diagram. If you find any instability (check with scope - during start-ups, during different loads), just scratch the top soldermask, break traces and bridge the traces with 2R2 or 3R3 resistors 😉
Attachments
Diyiggy...I believe they have lower impedance than LT304x despite being noisier.
You mean the same enhancement occurs when putting some Z or very bad esr cap at the output of a 78xx/79xx reg ? 😉
I can contribute an update about some experiments with output stages. Myself and a few pals have all made this DAC on the same base - the PSU2 and DAC with various inputs (SD card player, USB, SPDIF) but we all wanted to experiment with output stages. So we built a test bed DAC with SD card input, level matched the room and stages as close as possible and built:
EUVL current mirror
Pedja Rogic diamond buffer
6n2 SRPP from aliexpress (lightly modded)
heavily modded 3a5 DHT stage
SEN stage also from EUVL
standard opamps (rolled a few different kinds, thought OP27 were the best)
All with Andrea Mori clocks on the sources except the SPDIF
In short, all give excellent results, but with different levels of difficulty of assembly. By far the most complicated is the 3a5 DHT stage. This has Rod Coleman filament regs, filament bias, CCS on the plate and all that needs to be done to make it quiet. The SQ though is stunning at the end.
Sticking with valves, the 6n2 stage is a dream - all you need is the board from aliexpress and a transformer. Works perfectly, but replacing the original valves with russian EV spec ones improved things quite a bit, and adding an LED in the cathode also improved things a bit. A very easy way to implement a valve output stage and you get a very punchy dynamic sound, but not harsh at all.
The Pedja Rogic stage is a beauty, and combines some of the values of the 6n2 SRPP and the next one, the current mirror. Very full sound, lots of detail, tint of warmth there. It needs +/- 20V of good clean power to shine. There are gerbers available as well.
The SEN needs no introduction, and is just beautiful as well. It sounds very "right", very evenly balanced. The one drawback is that it needs the floating power supply, and I used 2 20V drill battery packs to give me that. There is a floting supply referenced in the thread, but I didn't have the parts needed to make it, so I didn't make it and used the batteries instead. Sound here was lovely, and the only reason it was put to one side was because the guys here reckoned charging would be a bit of a pain. We might end up revisiting that though. The 2sk170s are still available although getting dearer all the time.
Which brings me to EUVLs current mirror. This is a lovely build, really neat, low current, works excellently from the PSU2, and gives great performance. It has become our "stock" build and the guys here have been living with it now for a while with high praise for it.
The standard op-amps are not to be sneezed at, but I do think the discrete stages bring the DAC to a new level, and bluntly, the DAC deserves it. I have not tried any of the boutique op-amps as I don't have any, and wasn't going to invest either.
If you've read this far, you'll see that I am not recommending one over another. That is because we have all come to the conclusion that it is very subjective - one prefers the current mirror, another Rogic, another the DHT. Personally for me, I will have one current mirror and a DHT stage. I am gathering parts to make a 01a DHT stage at present, waiting on sockets and my plan is to make a stage that can act as IV output but also voltage input as a regular preamp (if I can!).
It has been a lot of assembly and a lot of listening which has been difficult to organise because of the restrictions here over the past while. But it has also been enormously enjoyable too!
There are more experiments to come yet - power supplies have made a difference in every build I've ever done, and I expect this one will be no different, so it seems obvious to look at that. LiFePO4 batteries or supercaps are likely to feature there, echoing what sligolad has reported earlier here.
Thanks again to Miro, EUVL, all the combined wisdom from here and the tubes forum who helped me with the DHT stage, and all who have contributed.
EUVL current mirror
Pedja Rogic diamond buffer
6n2 SRPP from aliexpress (lightly modded)
heavily modded 3a5 DHT stage
SEN stage also from EUVL
standard opamps (rolled a few different kinds, thought OP27 were the best)
All with Andrea Mori clocks on the sources except the SPDIF
In short, all give excellent results, but with different levels of difficulty of assembly. By far the most complicated is the 3a5 DHT stage. This has Rod Coleman filament regs, filament bias, CCS on the plate and all that needs to be done to make it quiet. The SQ though is stunning at the end.
Sticking with valves, the 6n2 stage is a dream - all you need is the board from aliexpress and a transformer. Works perfectly, but replacing the original valves with russian EV spec ones improved things quite a bit, and adding an LED in the cathode also improved things a bit. A very easy way to implement a valve output stage and you get a very punchy dynamic sound, but not harsh at all.
The Pedja Rogic stage is a beauty, and combines some of the values of the 6n2 SRPP and the next one, the current mirror. Very full sound, lots of detail, tint of warmth there. It needs +/- 20V of good clean power to shine. There are gerbers available as well.
The SEN needs no introduction, and is just beautiful as well. It sounds very "right", very evenly balanced. The one drawback is that it needs the floating power supply, and I used 2 20V drill battery packs to give me that. There is a floting supply referenced in the thread, but I didn't have the parts needed to make it, so I didn't make it and used the batteries instead. Sound here was lovely, and the only reason it was put to one side was because the guys here reckoned charging would be a bit of a pain. We might end up revisiting that though. The 2sk170s are still available although getting dearer all the time.
Which brings me to EUVLs current mirror. This is a lovely build, really neat, low current, works excellently from the PSU2, and gives great performance. It has become our "stock" build and the guys here have been living with it now for a while with high praise for it.
The standard op-amps are not to be sneezed at, but I do think the discrete stages bring the DAC to a new level, and bluntly, the DAC deserves it. I have not tried any of the boutique op-amps as I don't have any, and wasn't going to invest either.
If you've read this far, you'll see that I am not recommending one over another. That is because we have all come to the conclusion that it is very subjective - one prefers the current mirror, another Rogic, another the DHT. Personally for me, I will have one current mirror and a DHT stage. I am gathering parts to make a 01a DHT stage at present, waiting on sockets and my plan is to make a stage that can act as IV output but also voltage input as a regular preamp (if I can!).
It has been a lot of assembly and a lot of listening which has been difficult to organise because of the restrictions here over the past while. But it has also been enormously enjoyable too!
There are more experiments to come yet - power supplies have made a difference in every build I've ever done, and I expect this one will be no different, so it seems obvious to look at that. LiFePO4 batteries or supercaps are likely to feature there, echoing what sligolad has reported earlier here.
Thanks again to Miro, EUVL, all the combined wisdom from here and the tubes forum who helped me with the DHT stage, and all who have contributed.
Simply fantastic Fran! You and your fearless band of brothers paired Miro’s AD1862 Dac with some excellent I/V output stages. I have come to the same overall opinion, opamps can get you very good/excellent sound, but discrete I/V output stages gets you a notch higher performance. I too would like to give EUVL’s current mirror a listen.
Excellent work by ALL!!
Cheers,
Vunce
Excellent work by ALL!!
Cheers,
Vunce
- Home
- Source & Line
- Digital Line Level
- DAC AD1862: Almost THT, I2S input, NOS, R-2R