Page 6 of 11

#76 Re: Millett LR + TP Opus + AMB α24 reboot

Posted: Fri Nov 19, 2021 11:15 am
by Thermionic Idler
Thanks Nick - well, I've decided now to just make the DAC single-source (USB only) for the music server, so unless I decide to go for the upgraded oscillator board, it'll just be the 2 PCB's plus the Silentswitcher. There'll be a separate box handling the 2 channel signal from the AV rig, either SPDIF or HDMI breakout.

That way I can return the linestage to 'normal' analogue source switching, doing that reduces the dependency on the LCDuino infrastructure, giving me more options if I want to re-do the linestage in the future, say as an Aikido.

#77 Re: Millett LR + TP Opus + AMB α24 reboot

Posted: Mon Nov 22, 2021 9:33 am
by Thermionic Idler
Got back from the frozen north to find two gremlins have appeared under this roof in my absence - one, the clock on the central heating system has started flickering and advancing one minute every four seconds (anyone seen that before?), secondly, I went to cue up an LP last night, reached for the volume control to adjust the level and suddenly - no sound.

Every time I adjust the volume pot manually now, the volume resets back to minimum (thank god it didn't decide to set itself to maximum). It does continue to function properly with the remote, though. Cannot think of any valid reason why it's started doing this.

Given the DAC source control is now redundant, I think I'm just going to re-flash the current version of the AMB factory software back onto it. Very odd - it's not using any more memory than the default software, in fact it's using less because I disabled some functionality that I didn't need, so it can't be that. It worked perfectly up until the point where I starting mucking about with it, I'm starting to regret that now. Still unclear on whether the Silentswitcher conversion has actually done the sound any favours. Last night everything just seemed to sound a little flat, although this could just be after several hours of listening to Dom's rig and cartridges.

By the way, on that subject, short review of the DH3s - so on 'reference quality' vinyl, such as any Speakers Corner issue, the improvements over something like the VdH Black Beauty are definitely there, but more subtle - you can hear more layers, more melodic lines, and overall the music is presented in a way that makes sense - to be honest you just forget about the system and get completely immersed.

What completely floored me was when we got out my ropey old copy of "Who's Next". I love The Who but I've never heard a studio recording of theirs that sounded any good - the digital transfers all sound awful, and my pressing of "Who's Next" has never failed to sound congested, worn, crackly, some of the congestion I'd put down to the source tape. We use it as a test to see whether the system just highlights all the faults or can still make it musically enjoyable, so I tossed that into the bag to see what Dom's cartridge would make of it.

We played it all the way to end of side (by the time it gets to the last track it's normally unlistenable) and I kid you not, it sounded like a new record, in fact it sounded like Speakers Corner had mastered it. The same effect was apparent on the few other older and previously less-good sounding LPs that I'd brought with me. So that, for me, is its key strength - what it does for older, non-reference quality vinyl - just make sure it's clean. If the information is there, somewhere, it'll dig it out.

#78 Re: Millett LR + TP Opus + AMB α24 reboot

Posted: Sun Nov 28, 2021 1:44 pm
by Thermionic Idler
Paging RayP, Paging RayP (and anyone else who might be able to spot any schoolboy errors).

I've drawn out the below schematic for a backplane PCB to connect the I2SoverUSB to the TP Opus. I'm still unsure what the "Controller" block interface on the Opus is for, oddly the pin marked MUTE doesn't actually route to the MUTEB pin on the DAC chip, the one marked ML does (on the Opus schematic and confirmed when I inspected the PCB itself).

The other bit I wasn't 100% sure of was which MCLK output to use from the I2SoverUSB - I think I've got it right though, and I think everything else is as it should be.

The analogue power and signal outputs will be hard wired, so they aren't shown. My understanding is that the GND from H1 should be kept separate from the main ground plane, in order to preserve the galvanic isolation of the USB input. I've added the tell-tale LEDs there just to help me check the sample rate is being interpreted correctly when testing.

If anything doesn't look right, please shout! Many thanks.

Image

#79 Re: Millett LR + TP Opus + AMB α24 reboot

Posted: Sun Nov 28, 2021 5:16 pm
by jack
Can't see any 100n caps decoupling each chip's Vcc/Vdd - well, there are no chips as such, but decoupling pins 1 & 2 of J1 would be a good plan.

Crowbar (diode) between pins 1 & 2 of J1 as well, plus possible tiny power LEDs for both usb & iso sides.

#80 Re: Millett LR + TP Opus + AMB α24 reboot

Posted: Sun Nov 28, 2021 9:21 pm
by Ray P
jack wrote: Sun Nov 28, 2021 5:16 pm Can't see any 100n caps decoupling each chip's Vcc/Vdd - well, there are no chips as such, but decoupling pins 1 & 2 of J1 would be a good plan.

Crowbar (diode) between pins 1 & 2 of J1 as well, plus possible tiny power LEDs for both usb & iso sides.
No need for decoupling Nick, that'll all be taken care of on the TPA and JLsounds boards - we're just looking at ameans of connecting the two boards together.

#81 Re: Millett LR + TP Opus + AMB α24 reboot

Posted: Sun Nov 28, 2021 9:31 pm
by Ray P
Thermionic Idler wrote: Sun Nov 28, 2021 1:44 pm Paging RayP,
Dave, I think your data/clock lines look OK. though I'll have another look tomorrow as I've been a bit short of time this evening and feeling a bit tired so don't want to trust myself. One thing to check out is the mute signal and whether the Opus DAC expects a mute signal that is hi or lo?

Personally, as well as eyeballing it, I would be tempted to hook it up with some header patch cables to verify it is working correctly before ordering a PCB. You can get a pack of 10cm header patch cables on ebay for peanuts.

Have you managed to try the TV DAC yet?

#82 Re: Millett LR + TP Opus + AMB α24 reboot

Posted: Sun Nov 28, 2021 9:36 pm
by Nick
Ray P wrote: Sun Nov 28, 2021 9:21 pm
jack wrote: Sun Nov 28, 2021 5:16 pm Can't see any 100n caps decoupling each chip's Vcc/Vdd - well, there are no chips as such, but decoupling pins 1 & 2 of J1 would be a good plan.

Crowbar (diode) between pins 1 & 2 of J1 as well, plus possible tiny power LEDs for both usb & iso sides.
No need for decoupling Nick, that'll all be taken care of on the TPA and JLsounds boards - we're just looking at ameans of connecting the two boards together.
Yes, but I would agree with Jack, you still have the inductance of the supply tracks and connectors between the power supply and load, so never hurts to decoupling in and out of the board.

#83 Re: Millett LR + TP Opus + AMB α24 reboot

Posted: Sun Nov 28, 2021 10:29 pm
by jack
Ray P wrote: Sun Nov 28, 2021 9:21 pm
jack wrote: Sun Nov 28, 2021 5:16 pm Can't see any 100n caps decoupling each chip's Vcc/Vdd - well, there are no chips as such, but decoupling pins 1 & 2 of J1 would be a good plan.

Crowbar (diode) between pins 1 & 2 of J1 as well, plus possible tiny power LEDs for both usb & iso sides.
No need for decoupling Nick, that'll all be taken care of on the TPA and JLsounds boards - we're just looking at ameans of connecting the two boards together.
Not so. Decoupling, as is implicit in the other Nick's comments, is subtle and sometimes far from obvious.

EMI, track inductance, differential noise on supply lines, capacitive coupling between the daughter and motherboard etc. Lots of ways you can add noise in unexpected ways with a simple board like this.

Adding decoupling is cheap and easy. Likewise, board layout is a subtle art and important to get right. Supply design in mixed-mode designs is also important, especially WRT ground returns.

If in doubt, a few loops of wire on a scope (with decent bandwidth) probe can show all sorts of stuff. Stuff may work, but getting stuff to work well is subtle - all that digital signalling has fast rising and falling edges with all the harmonics and noise that creates.

Plus, I haven't looked at the design of the JL board in detail, i.e. the ground planes and decoupling. Galvanic isolation is one thing, how it's implemented is another. It may be absolutely fine...

#84 Re: Millett LR + TP Opus + AMB α24 reboot

Posted: Sun Nov 28, 2021 11:03 pm
by Nick
Also ceramic surface mount caps are better at decoupling that through hole caps, less lead inductance.

#85 Re: Millett LR + TP Opus + AMB α24 reboot

Posted: Sun Nov 28, 2021 11:03 pm
by Ray P
jack wrote: Sun Nov 28, 2021 10:29 pm Not so.
Fair enough.

Dave, have you looked at the sample rate matrix in the JLSounds documentation? If you incorporated some logic you could arrange for an LED specific to the incoming data rate to be lit.

#86 Re: Millett LR + TP Opus + AMB α24 reboot

Posted: Sun Nov 28, 2021 11:15 pm
by Thermionic Idler
Ray P wrote: Sun Nov 28, 2021 11:03 pm
jack wrote: Sun Nov 28, 2021 10:29 pm Not so.
Fair enough.

Dave, have you looked at the sample rate matrix in the JLSounds documentation? If you incorporated some logic you could arrange for an LED specific to the incoming data rate to be lit.
I have - I thought about it and, to be honest I'm not bothered about having a front panel display for the sample rate, I just want to be able to test with the top off that 44.1 send = 44.1 receive and so on, and I can do that with LEDs in the A0/1/2 lines, and figure it out from which ones are lit. Once I know it works I can forget about it. The only thing I want on the front panel is a soft orange on/off light. Nice and minimalist.

#87 Re: Millett LR + TP Opus + AMB α24 reboot

Posted: Mon Nov 29, 2021 11:14 am
by Thermionic Idler
Thanks everyone for the input - I'll redo the schematic tonight and check I've interpreted all your suggestions correctly.

#88 Re: Millett LR + TP Opus + AMB α24 reboot

Posted: Sat Dec 04, 2021 7:13 pm
by Thermionic Idler
Nick wrote: Sun Nov 28, 2021 9:36 pm
Ray P wrote: Sun Nov 28, 2021 9:21 pm
jack wrote: Sun Nov 28, 2021 5:16 pm Can't see any 100n caps decoupling each chip's Vcc/Vdd - well, there are no chips as such, but decoupling pins 1 & 2 of J1 would be a good plan.

Crowbar (diode) between pins 1 & 2 of J1 as well, plus possible tiny power LEDs for both usb & iso sides.
No need for decoupling Nick, that'll all be taken care of on the TPA and JLsounds boards - we're just looking at ameans of connecting the two boards together.
Yes, but I would agree with Jack, you still have the inductance of the supply tracks and connectors between the power supply and load, so never hurts to decoupling in and out of the board.
So something like this? (I've also added a tell-tale LED for the mute signal). Presumably the caps go as close to the board inputs as poss.

Image

#89 Re: Millett LR + TP Opus + AMB α24 reboot

Posted: Sat Dec 04, 2021 7:41 pm
by Nick
Presumably the caps go as close to the board inputs as poss.
Yes

#90 Re: Millett LR + TP Opus + AMB α24 reboot

Posted: Mon Dec 06, 2021 2:10 am
by Thermionic Idler
Thanks Nick.

OK, sit-rep. The DAC project is now on the back burner until next year as there are still some fiddly design details I need to work out. I want to focus on getting the Millett LR phonostage working first, I've decided to make that the main Christmas project, as I realised yesterday that I just didn't have the bandwidth to do both by year end. There are a number of reasons I want to prioritise the phonostage, which hopefully will become clear in due course if all goes to plan.

Last night I completed the design on the PCB that will hold the Lundahl step-up transformers, and this morning ordered it from JLPCB. Also placed an order from Mouser for a few remaining bits and pieces, and Accu for some fixings etc. Assuming everything makes it here ahead of the holidays, I'll have a nice little project to tackle over the Christmas break. It'll be interesting to hear how it compares to the Whest, which is the only piece of commercial gear left in my vinyl-to-speaker signal chain. The Whest is kind of a tough act to follow.