Millett LR + TP Opus + AMB α24 reboot

What people are working on at the moment
User avatar
Thermionic Idler
Old Hand
Posts: 732
Joined: Sun Apr 22, 2012 8:22 pm
Location: Southsea

#1 Millett LR + TP Opus + AMB α24 reboot

Post by Thermionic Idler »

With the advent of the SilentSwitcher, I realised that this has given me quite a good opportunity to re-think the power scheme of everything in the system that isn't a power amplifier or a turntable.

The Millett phono stage and Twisted Pear Opus DAC projects are planned to go forward pretty much in tandem with each other, and it occurred to me that I could also convert my existing linestage to mains-free operation as well. So I've decided to combine the design and build log for all three projects into one thread, as they are kind of related to each other, and just basically add to it as I make (typically slow) progress.

So here's the idea so far. Reading up on the SilentSwitcher, it seems that feeding it with as clean power as possible is beneficial. So I want to use lithium battery packs, 2 x 3.7v connected in series to give 7.4V (the SilentSwitcher will work with any input voltage between 3V and 10V).

These can power three Silentswitchers - one feeds the linestage circuit in my existing AMB preamp (a.k.a. the Dieselpunk Preamp of Doom), and the other two feed the Opus DAC and Millett phono stage respectively. The rack will then have nothing with a mains transformer in it at all - in fact the only thing running on mains voltage will be the Garrard 301, and that takes its power from a balanced regenerative power supply anyway so it won't be throwing out hash everywhere.

To recap, at the moment the AMB preamp has three linear power supplies - a 5V always-on supply which runs the Arduino control system for power on-off, switching relay volume, source select etc. There's a 12V supply which runs the buffer amp for the VU meters, their illumination bulbs, and also provides the 12V trigger to turn on the power relays in the 300Bs. Then there's a ±10V dual-rail supply for the linestage circuit itself. Removing all of that will leave plenty of room for the lithium battery packs and a Silentswitcher.

The plan for the AMB is to use the 6.5V Silentswitcher supply for the VU meter buffer (although I may need to replace the TL072 op-amp with something like an LM358 so it'll work on the lower voltage). Then the dual-rail supply can be set up for ±12V to feed the linestage circuit (this will work on voltages up to 15V). The battery voltage will also be routed to a double USB socket which can then be used to power the DAC and phono stage.

There will be an always-on wall-wart 5V 3A supply into the AMB, which, when the system is running, will power the stuff that doesn't need an ultra-clean supply. So the Arduino control system, the meter bulbs, and the power relays in the 300B's (which I'll need to swap out with 5V coil voltage units). When the system is powered down, this will automatically switch over to feed two lithium charging boards (the common 5V in, 4.2V 1A out ones that are everywhere) which will charge the two 3.2V battery packs in parallel.

I haven't run the numbers but I'm hoping the battery packs will have enough capacity to power the system for many hours at a time, but in the event they do run flat whilst the rig is powered up, the supply to the Silentswitchers will be automatically flipped over to the wall-wart. This is the bit I'm trying to figure out at the moment, I have a simple voltage monitoring circuit drawn up in Spice which uses a Zener and an NPN transistor to switch 'on' the battery voltage and trigger a solid state relay, which in turn should trigger a latching circuit powered by the wall wart, if the series battery voltage falls below 6V. But I'm a bit hopeless with transistors (valves, I'm fine with).

The good thing about all this low-voltage stuff though, is that I can test it on the bench without fearing for my life.
Deck: Garrard 301 - Audio Origami PU7 - Van Den Hul Black Beauty
Phono: Whest PS30RDT
Linestage: AMB α24/LCDuino
Power amps: 'Newton' 300B PP based on a design by Lynn Olson
Speakers: FF225WK 'Big Met' Metronomes c/o Cressy Snr
User avatar
Ray P
Eternally single
Posts: 4966
Joined: Thu Nov 22, 2007 5:18 pm
Location: Somerset

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

Post by Ray P »

I'm using a SilentSwitcher in my buffered passive preamp - they're excellent little packages.
User avatar
Thermionic Idler
Old Hand
Posts: 732
Joined: Sun Apr 22, 2012 8:22 pm
Location: Southsea

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

Post by Thermionic Idler »

Thanks Ray - good to know, I've heard a lot of positive things about them.

Today, I have been mostly trying to fathom out my Li battery protection mechanism, and I think I've finally got a circuit that works:

Image

Why use anything more complicated than a zener diode as a voltage reference. The 4.7V zener is padded out with a 1N4148, and activates one half of the Panasonic SSR (two relays on one chip) when the battery voltage falls below about 5.9V across the two battery packs (according to the Spice sim anyway).

This then causes the latch circuit on the right, powered by the wall wart, to engage and activate the other half of the SSR. At this point it doesn't matter what the batteries do. This closes a circuit to a mechanical dual-throw relay, which will switch over the power from the batteries to the wall wart, and stay that way until power-down. Ideally at the same time I need to disconnect the batteries from that monitoring circuit as it will still be drawing current for the first SSR - there's a whole wider infrastructure outside what I've drawn here.
Deck: Garrard 301 - Audio Origami PU7 - Van Den Hul Black Beauty
Phono: Whest PS30RDT
Linestage: AMB α24/LCDuino
Power amps: 'Newton' 300B PP based on a design by Lynn Olson
Speakers: FF225WK 'Big Met' Metronomes c/o Cressy Snr
User avatar
ed
retired
Posts: 4711
Joined: Thu Jun 21, 2007 4:01 pm
Location: yorkshire
Contact:

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

Post by ed »

Thermionic Idler wrote: Fri Aug 27, 2021 7:15 pm The 4.7V zener is padded out with a 1N4148, and activates one half of the Panasonic SSR (two relays on one chip) when the battery voltage falls below about 5.9V across the two battery packs (according to the Spice sim anyway).

This then causes the latch circuit on the right, powered by the wall wart, to engage and activate the other half of the SSR. At this point it doesn't matter what the batteries do.
Hi Dave

I'm curious how you arrived at the figure for 5.9v. Dropping a 2 cell lipo to this point might not be good for longevity.
if you're using Li-ion or life cells things might be slightly different.
If you haven't seen this it might be of use: https://blog.ampow.com/lipo-voltage-chart/
There's nowhere you can be that isn't where you're meant to be
User avatar
Thermionic Idler
Old Hand
Posts: 732
Joined: Sun Apr 22, 2012 8:22 pm
Location: Southsea

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

Post by Thermionic Idler »

ed wrote: Sat Aug 28, 2021 8:34 am Hi Dave

I'm curious how you arrived at the figure for 5.9v. Dropping a 2 cell lipo to this point might not be good for longevity.
if you're using Li-ion or life cells things might be slightly different.
If you haven't seen this it might be of use: https://blog.ampow.com/lipo-voltage-chart/
Hi Ed,

It was based on an article I found where someone had pointed out that the protection chip commonly used in a lot of these 3.7V packs was set to 2.5V, and that a better point to start charging was about 2.9 - 3.0V.

Thanks for the heads-up on that table though. Looks like I actually need to be targeting a shade below 7.5V - interesting as that's actually about the same as the nominal voltage, and much higher than the internal protection level. Should be easy enough, just change out the zener for a higher voltage one.
Deck: Garrard 301 - Audio Origami PU7 - Van Den Hul Black Beauty
Phono: Whest PS30RDT
Linestage: AMB α24/LCDuino
Power amps: 'Newton' 300B PP based on a design by Lynn Olson
Speakers: FF225WK 'Big Met' Metronomes c/o Cressy Snr
User avatar
Thermionic Idler
Old Hand
Posts: 732
Joined: Sun Apr 22, 2012 8:22 pm
Location: Southsea

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

Post by Thermionic Idler »

I've now extended the model in Kicad to incorporate the various relays and connections needed to make this work. Today was mostly spent working out power budgets to make sure I am not about to weld relay contacts together, fry power supplies or have lithium batteries reduce my living room to a smoking ruin.

The three SilentSwitchers will together consume 3.5W worst case. With the 10.4Ah battery packs I've specified, I worked out that they should give me 18.85 hours of continuous operation using 80% of that capacity (I assumed that 10.4Ah was for 100% to zero, but the table that Ed linked to had the cutoff happen at 20%, and that's what I'm designing in). Then 8.32 hours to recharge.

I'm wondering about including a trim pot where R1 is in the circuit posted previously - that resistor value seems to affect the actual cutoff voltage, and it might provide a useful way to fine tune it within the zener diode tolerances. Either that or just get a range of resistors and prototype the thing before committing it to a PCB.

Getting shot of all these linear power supplies of course makes the whole thing a lot more efficient in terms of electricity usage. It offsets, at least by a little bit, the mahoosive inefficiencies that come from running 300B Class A valve amps! :oops:
Deck: Garrard 301 - Audio Origami PU7 - Van Den Hul Black Beauty
Phono: Whest PS30RDT
Linestage: AMB α24/LCDuino
Power amps: 'Newton' 300B PP based on a design by Lynn Olson
Speakers: FF225WK 'Big Met' Metronomes c/o Cressy Snr
izzy wizzy
Old Hand
Posts: 1367
Joined: Fri Nov 02, 2007 7:02 pm
Location: Auckland NZ
Contact:

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

Post by izzy wizzy »

Really enjoy following your threads. As an owner of hugely inefficient class A power amps, I can see why you wouldn't want to double down on that.

I'd still like to ask, given the silent switcher can be powered from the mains why the move to batteries given regulators usually remove the need for heavy bulky passive filtering. Is it not swapping one for another?

I can see that batteries remove the need for LF filtering but as the silentswitcher chops all that up at HF then doesn't that make the power feed to it less bulky?

I have no horse in this race. Just like hearing about people's design decisions.
User avatar
Thermionic Idler
Old Hand
Posts: 732
Joined: Sun Apr 22, 2012 8:22 pm
Location: Southsea

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

Post by Thermionic Idler »

izzy wizzy wrote: Mon Aug 30, 2021 9:22 pm I'd still like to ask, given the silent switcher can be powered from the mains why the move to batteries given regulators usually remove the need for heavy bulky passive filtering. Is it not swapping one for another?
Well it was prompted by this post on DIYaudio. The wall warts seemed to be a little bit variable as to how much 'hash' got through. Battery power can be seen as the one option that's guaranteed to be the most hash-free.

The 5V supply into the preamp also powers an Arduino processor that controls the attenuator relays amongst other things, and I suspect that would add to the hash on the supply to the silentswitchers too.
Deck: Garrard 301 - Audio Origami PU7 - Van Den Hul Black Beauty
Phono: Whest PS30RDT
Linestage: AMB α24/LCDuino
Power amps: 'Newton' 300B PP based on a design by Lynn Olson
Speakers: FF225WK 'Big Met' Metronomes c/o Cressy Snr
User avatar
Nick
Site Admin
Posts: 13656
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

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

Post by Nick »

There are a lot of options between a battery and a wall wart. But I think the most important point is made by Jan later on in that thread:
If you see it check the grounding because 9 out of 10 cases the hash comes in through some surprise ground connection.
Thats why the filtering you think will do some good doesn't as the actual noise is going via the supply to the test kit via a shared earth and the actual equipment that is being powered does not see the noise you think you are trying to filter out.
Little known fact, coherent thought can destructively interfere with itself leaving no thought at all, that’s why I prefer incoherent thought.
User avatar
Nick
Site Admin
Posts: 13656
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

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

Post by Nick »

Battery power can be seen as the one option that's guaranteed to be the most hash-free.
If you are actually powering it directly from a battery (as Jan also points out). A lot of usb battery packs have a switching reg on the output to maintain a constant non battery friendly voltage.
Little known fact, coherent thought can destructively interfere with itself leaving no thought at all, that’s why I prefer incoherent thought.
User avatar
Thermionic Idler
Old Hand
Posts: 732
Joined: Sun Apr 22, 2012 8:22 pm
Location: Southsea

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

Post by Thermionic Idler »

Nick wrote: Tue Aug 31, 2021 11:47 am
Battery power can be seen as the one option that's guaranteed to be the most hash-free.
If you are actually powering it directly from a battery (as Jan also points out). A lot of usb battery packs have a switching reg on the output to maintain a constant non battery friendly voltage.
Indeed, I also noticed that, and discounted the devices that turn 3.7V into 5V as a result. My cunning plan uses two standard Li-Po 3.7V packs, series connected.

Here's the draft schematic for what will hopefully become my first PCB design, to slot into the preamp in place of the power supply. Some values may still require fine tuning but I think it's largely there.
Deck: Garrard 301 - Audio Origami PU7 - Van Den Hul Black Beauty
Phono: Whest PS30RDT
Linestage: AMB α24/LCDuino
Power amps: 'Newton' 300B PP based on a design by Lynn Olson
Speakers: FF225WK 'Big Met' Metronomes c/o Cressy Snr
User avatar
Thermionic Idler
Old Hand
Posts: 732
Joined: Sun Apr 22, 2012 8:22 pm
Location: Southsea

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

Post by Thermionic Idler »

I've made an improvement to the battery level monitoring. The Lipo voltage chart calls for the cutoff voltage to be really quite precise. A little too low, and the longevity of the battery pack is compromised, a little too high and available system up-time on battery power significantly decreases. I decided some sort of trim pot was necessary to dial in the correct value.

I couldn't figure out where in the existing circuit this could be effectively implemented so I decided to rethink it. It was a tricky one because most circuits assume a constant rail voltage, whereas in this case not only is it varying, the circuit has to be both powered by it, and change its state depending on its value.

I reasoned that if I at least had some kind of constant current source, then I could put that across the trim pot and end up with a stable voltage range. So I started with a simple example circuit I found online, which I then turned "upside down" to tie the zener to the moving voltage rail rather than 0V. The rest of the circuit was worked out by a lot of trial and error in Spice, but according to the modelling this should work.

The screenshot below shows the same circuit duplicated, except with the 50 ohm pot adjusted one way, and then the other.

Image

The model output shows the decreasing voltage rail (blue line) and the voltages across the SSR trigger (R17, R23). The red and cyan lines show the voltage between R17/R23 and Q5/Q9, indicating the range of adjustment where that transistor switches on. This range is between about 7.2V and 7.7V (target should be a shade below 7.45V). The green line shows the CCS current flowing through the trim pot, indicating that it remains quite stable within that range of adjustment.

Image

I'll make the necessary changes to the KiCad schematic in due course.
Deck: Garrard 301 - Audio Origami PU7 - Van Den Hul Black Beauty
Phono: Whest PS30RDT
Linestage: AMB α24/LCDuino
Power amps: 'Newton' 300B PP based on a design by Lynn Olson
Speakers: FF225WK 'Big Met' Metronomes c/o Cressy Snr
User avatar
Nick
Site Admin
Posts: 13656
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

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

Post by Nick »

FWIW (which is nothing) my tendency would be a voltage reference and a comparator, or more likely a ucontroller with a analogue comparator, then all the logic becomes simple.

Actually, my tendency would be a linear supply, but that’s a different conversation.
Little known fact, coherent thought can destructively interfere with itself leaving no thought at all, that’s why I prefer incoherent thought.
izzy wizzy
Old Hand
Posts: 1367
Joined: Fri Nov 02, 2007 7:02 pm
Location: Auckland NZ
Contact:

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

Post by izzy wizzy »

Nick, as there is an ever changing world of ucontrollers, what would be your choice for a simple cheap one, easy to program without expensive tools, with a few analogue/comparator inputs and maybe a couple of analog and some digital/switch outputs?

You can probably tell by my question I know very little but come from a programming background so hopefully can cope.
User avatar
Nick
Site Admin
Posts: 13656
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

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

Post by Nick »

Well, I am happy being out of date and use the Atmel 8/16 bit parts. Most of the surface mount parts are unobtainable now as the world has run out, but the PDIP ones are still out there, It hasn't got a analogue output, but the simple ATTiny24A fits that bill. Programming interface are readily available, and the dev tools are built around gcc and Visual Studio on windows and are freem likewise it can be driven from linux as well. If you need analogue outputs its easy to interface a i2c DAC chip. The atxmega32 is bigger and more powerful and has all sorts of nice toys, but its unobtainable ATM.

Jack (Nick) has more experience of newer parts especially low power ones, but you can do a lot with a £1.50 Tiny24 and I work on the if its not broken why look for a fix mindset.

I use an Atmel Dragon to program and in theory debug (though I do most debugging by waving an output about as required), but there are loads of cheap as chips programmers out there.

And of course there is the Arduino stuff which adds a USB interface and some simple libs to the basic chips and makes it more plug and play. Depends on how close to the metal you want to get. But an attiny24 wired up to a 6 pin AVR programming interface is simple and easy place to start.
Little known fact, coherent thought can destructively interfere with itself leaving no thought at all, that’s why I prefer incoherent thought.
Post Reply