advertisement


ALWSR for NAXO queries

nobeone said:
Cheers Arkless. I have done a simple sanity check and the ALWSR output is measurably quieter than my bench PSU just with a 1mV/div scope envelope trace so I can't have completely crapped it up!

Yes super reg may well be overkill on an active cross over but it seems separate regulation for each channel and each stage is beneficial, hence the Naim SNAXO with Supercap. So overkill with a bunch of ALWSR can't hurt and is an order of magnitude cheaper and needs no more rack space and may be better with regulation closer to the load, well that is my theory.

Ok, so I will stand on the shoulders of giants and assume if they give the correct output voltage as I vary the input voltage and remain stable on load and are quieter then my bench supply they are good to go. I will build up the rest then work out how to sneak them in and connect them up taking inspiration from timH.
Click to expand...

A reasonable assumption! :)

If you can see any noise whatsoever on a scope trace at 1mV/div then be worried! Gain/Bandwidth product is finite though of course so at frequencies way above the audio range (RF) impedance will rise and so will noise but by these frequencies even a fairly small capacitor has a very low impedance and output cap can take over from regulator (a few if's and but's to that but near enough)
 
Well the 1mV/div envelope does show noise, some is scope noise, if I short the probe tip to the ground connection I can still see something in the envelope. Perhaps I just need to turn the bandwidth right down, it is a wide bandwidth digital scope so perhaps not the best for looking at audio bandwidth, R&S 10 bit ADC, fancy 10" touch screen, you name it, for a softy doing various digital and comms stuff not an audio hardware guy! I can see on the FFT it is wide band noise, no hooters in the audio band.
 
nobeone said:
Well the 1mV/div envelope does show noise, some is scope noise, if I short the probe tip to the ground connection I can still see something in the envelope. Perhaps I just need to turn the bandwidth right down, it is a wide bandwidth digital scope so perhaps not the best for looking at audio bandwidth, R&S 10 bit ADC, fancy 10" touch screen, you name it, for a softy doing various digital and comms stuff not an audio hardware guy! I can see on the FFT it is wide band noise, no hooters in the audio band.
Click to expand...

I've always found digital scopes pretty unusable for very low level stuff due to the trace becoming several mm's thick anyway... I nearly sent back a batch of £30K Agilent scopes as faulty at one place I worked at!
If very wide bandwidth then maybe seeing RF noise from common mode sources?
I would guesstimate audio band noise from a good super-regulator which is optimised for low noise to be less than 1uV.. as I had to amplify it by 100x to use a 100uV FSD meter after the reg...
BUT. I did "cheat" by using A weighting for these measurements. Obviously when trying to measure noise this low all local noise sources such as hum from mains transformers, CFL lights. PC's etc becomes a nightmare and so to get decently usable readings I had to use A weighting. This with a balanced low noise amp to reject as much common mode noise as possible from the wiring, grounding etc. The true noise over say 10Hz to 50KHz bandwidth would no doubt have been triple or more what I was measuring.
This was a super-regulator of my own design (but not dissimilar to this one) very much optimised for ultra low noise, and would be considerably quieter than the ALWSR one in the form it is built here, mainly down to the use of AD797 as the error amp, which is a much quieter op amp than used here (and can be rather tetchy stability wise) A filtered reference and even low noise pass transistor were used.
I used it for powering MC head amps with a PSRR of only maybe 10dB
 
There is a build of ALWSR using the AD797 but not recommended for general purpose use for stability issues I believe. The guys at work know a few things about noise, we have a screened room like a full EMC test house in our offices where I work. I could set up in there, run from batteries and set up the scope from outside but I'm not that bothered!
 
nobeone said:
There is a build of ALWSR using the AD797 but not recommended for general purpose use for stability issues I believe. The guys at work know a few things about noise, we have a screened room like a full EMC test house in our offices where I work. I could set up in there, run from batteries and set up the scope from outside but I'm not that bothered!
Click to expand...

I was pretty sure there was a quieter version of the ALWSR with the AD797 hence "quieter than the ALWSR one in the form it is built here":)
 
Moving on to looking at the NAXO ...

It has 5 signal circuit sections: Channel 1 Low Frequency (LF), Channel 2 LF, Channel 1 High Frequency (HF), Channel 2 HF, and the output relays.

The NAXO takes two 24Vdc rails from a Hicap.

What interests me is, as I piece together Neil McBrides NAXO schematics and the PCB in front of me, is that given there are two rails, why did Naim decide the best thing to do was use one for nearly everything?

Each of the channel sections has an input stage and an output stage. The LF sections use a single 27R/47uF from a rail for both input and output sections. The HF sections split the power to the input and output sections, so the input section has a 27R/47uF from one rail, and the output section has a 27R/47uF from another rail.

It turns out one rail supplies the Channel 1 & 2 LF, Channel 1 & 2 HF output stages, and the output relays, while the other rail just does the Channel 1 & 2 HF input stages. I would have thought it best to split the load so channel 1 had one supply, channel 2 the other, or perhaps channel 1 and 2 LF on one, channel 1 and 2 HF on the other - a bit like the arguments for and against vertical or horizontal split on amplifiers.

I assume in the SNAXO they split them so, like the HF section in the NAXO, the LF section is split into input and output sections, can anybody share the insides of a SNAXO?

I don't really want to cut tracks on my NAXO, I prefer my modifications to be easily reversed so nobody would know I had visited. To this end my plan is to add ALWSRs where there are 27R/47uF from a rail. I had planned on 4 but it looks like it is already 6, that is OK, I have 8 ALWSR boards thanks to the group buy. I wasn't going to do anything with the relay supply though I note timH did.

So next question is, how do I wire up the ALWSR? Calling all ALWSR experts!

I assume I wire the ALWSR TRIN to a Hicap +V rail star point, and the ALWSR TRGND to the Hicap GND star point. Each ALWSR getting separate wires back to the star points.
I assume I remove the 27R and 47uF and wire ALWSR OP1 to where the 27R was.

But should I follow the wiring from the ALWSR manual for local or remote sense?

If I follow the local sense, I assume I wire the ALWSR SRGND to the GND on the removed 47uF?
If I follow the remote sense, I assume I wire the ALWSR +VSENSE to the 27R too, and I wire the ALWSR 0VSENSE to the GND on the removed 47uF?
 
So I looked at where the NADI was powered from and updated the image above, NADI is on the same 24v rail as most other stuff :)

I suppose to keep the brightness the same I will have to adjust the 180R that feeds the NADI from what is expected to be 24V but will be more like 28V to allow for the ALWSRs to output 24V. I will "worry" about that later.
 
I think I have worked out how to fit 6 ALWSRs in there, they are bigger than the Teddy Regs so space is tight
naxo 2-4 ALWSR positionshttps://www.flickr.com/photos/184147951@N07/ on Flickr

The aim will be to make a pair of aluminium brackets like timH did and fit 3 on each side supporting the PCBs on one screw hole and providing heat sinking for the TPR and the output device. I wanted to place the ALWSRs so I could still get at the trim pots, they did not get in the way of the 10uF tants, but they were as close as reasonable to each section they will power.

The reason I want to keep them away from the 10uF tants is I will be replacing them as part of the modification. Probably with WIMA MKS2 10uF 50V 10% parts unless anybody suggests something else. These are taller than the tants and so prevent placing the ALWSRs above them.
 
You do realise that leaving the 27R and 47uF in will negate any and all advantages of ANY voltage regulator? (other than to filter some noise from a spectacularly noisy one!)
 
If you’ve got enough boards it’s quite straightforward to also separately power both sections of the LF filters. You don’t have to but as I always say - why not just do it? ;)
 
timH said:
If you’ve got enough boards it’s quite straightforward to also separately power both sections of the LF filters. You don’t have to but as I always say - why not just do it? ;)
Click to expand...
I haven't had the PCB out yet but I can see the LF input stage has unpopulated space for another 27R/47uF but it isn't obviously linked out anywhere so I assume it is a track cut to separate the rails? As you say, simple enough but I was specifically trying to avoid track cuts, if it was just a link then I would! I also can't quite fit another ALWSR in there without covering the trim pot ... hmm, I will think about it! I can see the SNAXO does separately power each section so it would be nice to go there too but ... ah indecision!

EDIT: yes I randomly buy twice as much as I need for most projects JIC and so have 8 ALWSR PCBs, just enough as it happens ...
 
Last edited:
I had been planning to bend the ALWSR TPR and the output transistor over and heatsink on a flat piece of aluminium to which I could also support the ALWSR PCB, as per timH Teddy Reg project. But Tim has me thinking, I'll crack the NAXO open again and see if there is space for two U channels down between the PCB posts, I could then mount the ALWSR TPR and output transistor upright, I'll need some other way to support the ALWSR PCBs, but I can then fit 8 in and split the LF circuits and power all the sections from their own ALWSR. Suggestions for U channel suppliers appreciated!
 
Opted for ebay, 1m (!!!) of 12mm wide 20mm high U channel for about £3 and £3 delivery, not bad, you were right various DIY sheds have U channel but tended to be same width as height and I am tight for space on width so the ability to pick that was useful. Plus I live a decent drive from everywhere making mail order preferable! Component kits on order for the full 8 ALWSR, plus hardware to mount.

I had a look at splitting the rails on the NAXO PCB. To lift it up I had to remove the switch and NADI, I can see some ceramic caps tacked on the back (!!!) and the hole for the switch is somewhat agricultural, you can see the score marks from marking up before drilling and additional drilled holes, looks like a factory hack of a sled designed for something else ...

Anyway, easy enough to see how to cut the rails on the LF section but I was wrong about the additional 27R/47uF mounting positions just about the LF section, they are for powering the mids on a NAXO 3-6 and not an option on splitting the LF, I guess that was too good to be true!
 
I have also worked out how to support the ALWSR PCBs on the U channel, I found these nifty nylon PCB mounting blocks in RS that allow for various mountings. With a couple of M3 nylon washers of 0.8mm thickness I can make up the difference of a 1.6mm PCB and mount the TPR, drive transistor and nylon mounting block flush on the U channel.

EDIT: The nylon block can be seen on the ALWSR bottom right of the above photo of the 8 ALWSRs.
 
Last edited:
I'm going the remote sense wiring route, I will run a twisted pair from the 0V and +V sense down to the NAXO as the NAXO ground plane provides a return to the ground star point.
 
As if by a magic another thread appears regarding the simple modification of a Hicap to provide the 28V needed to power the ALWSRs, just linking the two to save me repeating myself here.
 
You must log in or register to reply here.


advertisement


Back
Top