Arcam Alpha 5 CDP upgrades

A complete aside, but I've just realised those are 'clickable' hires images and had a close look. Those Netaudio elements look very good value for the retail cost!

(I've used Block transformers too, and like them a lot)
 
I agree, they do seem good value.

I can't fault the service either. David (the man behind Net Audio) was very helpful.
 
Re-clocking the SAA7220

Hi everyone,

Since my last post on this thread things have developed a bit on the Alpha 5 front.

For a while now I've been pondering the relative merits of non-oversampling on TDA1541 based players versus retaining the oversampling and digital filtering performed by the SAA7220 chip.
Previously on this thread I've converted one of my two Arcam Alpha 5 CD players to NOS by fitting the NET Audio Super Non Oversampling kit plus a host of other tweaks and upgrades. The conversion was a great success and the modified player sounded far superior to my 'spare' standard player.

However I decided that I'd like to see how a upgraded player which retained oversampling/digital filtering would compare. The planned upgrades centred around providing a low jitter 'clean' clock signal in an effort to reduce jitter whilst retaining the digital filtering function of the SAA7220.

A plan was born:

The standard clock crystal would be replaced with an after-market high quality clock. I chose to use the NET Audio Rock Clock 3 for this as it is compact and seems to offer a great performance for the money.

The Rock Clock would be powered by its own dedicated regulated power supply. for this I used an Audiocom clock PSU which I had spare.

One clock feed would go to the SAA7220 chip and a separate clock feed would go to the decoder chip (SAA7310). This is beneficial as the SAA7220 normally distributes a very noisy clock feed to the decoder. By isolating the clock feed from the SAA7220 to the SAA7310 the decoder can be provided with its own clean clock signal.

A better explanation of the above can be found here:
http://www.fidelityaudio.co.uk/1541 clocking_03.2ver.pdf

Here are some pics:


photo sharing


image hosting tiny pic


capture software

Initial listening impressions were very promising and the player seemed to improve considerably over the next few weeks, perhaps as a result of the new clock burning in.

I liked what I heard so I made the decision to make the best player I could by combining all the best bits from both my Alpha 5 CDP's into one 'Frankenstein' super player. That meant stealing the main circuit board from the NOS player which I had already populated with upgraded capacitors and diodes as well as the discrete op-amps and upgraded capacitors off the audio board. I also swapped over the voltage regulator tweaks (zener diodes and LED's) which I'd used on the NOS player.

I'm super pleased with the result. To my ears the re-clocked and upgraded player sounds absolutely stunning and well worth the effort.

Next I plan to install a dedicated PSU for the SAA7220. Watch this space.
 
Very neat work.
As for watching this space - looks like you might well run out of space as it's getting quite full in there!
 
Yes I wish the case was a bit bigger!

I've read elsewhere that the SAA7220 is a very noisy chip and tends to propagate a lot of noise back into the power supply to the detriment of the rest of the components. For this reason it has been suggested that giving the SAA7220 its own dedicated power supply can be a very good upgrade on TDA1541/SAA7220 based players.

The designers of the Alpha 5 seem to have already given this some thought and the SAA7220 gets a +5v feed from the main board (shared with the decoder chip and other IC's on the mono board).

The audio board has a separate PSU section which supplies the DAC and op-amps (separate regulators and transformer winding etc).

Do you think it would still be worthwhile giving the SAA7220 its own PSU? How sensitive to PSU noise do you think the decoder/microcontroller IC's will be?

Alternatively I could give the TDA1541 DAC chip its own dedicated PSU (currently it shares +5/-5 and -15v supplies with the op-amps).
 
The thing with the SAA7220 is it handles the bit clock as well as the digital data, and because the 5v supply and 0v pins are on opposite corners it's difficult to decouple properly (large loop area). it translates directly into noise on the critical bitclock. I can recommend as a minimum using a 47uF oscon with the legs stretched to tie the supply and 0v pins as close to a straight line as you can (diagonally across the chip.) Unless you have a ground plane avaialble and are happy working with SMT caps that's about as good as you'll get easily. Forget ordinary electrolytics here, and don't use some boutique cap in the existing bypass cap mounting holes - too far away to do any good.

This filter does need its own +5v regulator capable of supplying 250mA - from well-sorted LM317T to as good as you like, things only get better. But a separate PSU is not required if you get the rest halfway-right.
 
Thanks Felix, that's very helpful advice.

The Oscon decoupling cap sounds like a 'no brainer' I'll get a suitable capacitor next time I'm placing an order for some parts.

In preparation for the SAA7220 PSU I have already picked up a 7VA 12v mini toroidal transformer and a pre-built adjustable LM317 based module from ebay (from the seller Audiowind). I also picked up one of the 'low noise' +5v regulator modules from the same seller. These were super cheap and I figured they'd come in handy one day even if I don't use them now.

The rough idea was to run one of these pre-made modules to power the SAA7220 or I could even run them in series with the adjustable module set at around 10v followed by the 5v reg module (hence the 12v transformer).

As standard on the Alpha 5 the SAA7220 gets its +5v supply from a 7805 reg which it shares with other IC's on the main board. An easy and neat option would be to replace this 7805 with a fancy super duper 3 pin regulator. Or would it really be better to have a dedicated PSU for the SAA7220? Or both?

Sorry for all the questions, I'm only slightly further up the learning curve nowadays.
 
Just a separate 5v reg for the '7220 (just try a 7805 first) sharing the same raw supply would be a very good start. Pre-filter that raw supply before your new 7805 with say a 1.0 up to 2.2ohm* resistor and a big cap of 100uF or more (from 7805 Vin pin to 0v pin) will help, too, by providing a decent filter for HF noise that such regs don't cope with very well.

Even better if you can reference the 7805's 0v pin to /or locate right-at the 0v pin of the '7220.

Such things don't have to be fancy, just well-considered :)


*SAA7220 draws about 200mA so 2.2ohm will have about 0.5v drop across it - you probably dont want to waste any more. Regular 0.6W metal film type is just fine, for ~0.1w disspation. The 7805 will disspate about a watt - it will be hot but OK. A small clip-on heatsink is a good idea if you have space.
 
I've just taken delivery of 25x 47uF 25v Oscon radial capacitors.

I couldn't find any in the UK as they appear to have been discontinued. In the end I got some from Poland on ebay.

I'll be soldering one directly across the underside of the SAA7220 from the power pin (pin 24) to the ground pin (pin 12). As recommended by Felix.

The +5v power supply comes in through a inductor/coil followed by 100uF electro and a 47nF ceramic cap paralleled to ground. Should I just leave these in place?

Is it worth upgrading any of the other decoupling caps to Oscons at the same time? For example on the +5v, -15v and -5v power supplies to the TDA1541? Or the +5v supply to the decoder chip?
 
Good-o, I suggest you try that as it is a few days and see what you think before considering other changes. I find oscons remain a bit 'leaky' electrically for a few hours after soldering - don't be surprised if the sound changes on day 2 ( - and I'm not a big believer in burn-in, but this seems repeatable)

Final nerdy suggestion - if there is enough clearance when installed, twist that oscon back up so the leads lie as close as you dare to the pcb. It keeps loop-area minimal. (Don't worry, the green solder mask is not conductive)

>100uF electro and a 47nF ceramic cap paralleled to ground. Should I just leave these in place?

Yes, the ESR of the electrolytic will help 'damp' the supply.

[I didn't know there is a ground plan under the 7220 in the Alpha 5. If I had - as a more invasive mod personally I'd probably choose to use a surface-mount ceramic cap between the +5v pin (on the bottom left in you pic) and that ground plane, as close as poss (scratch off the green mask and solder to the copper below). I'd then whip out the existing 47nF ceramic cap in parallel with the 100uF electrolytic. ]
 
I've done a bit of listening over the previous few days with the new Oscon decoupling capacitor in place and I'm very pleased with the results. An audible improvement for next to no money can't be bad! Thanks Felix.

Like you keep telling me:
"Such things don't have to be fancy, just well-considered"

As a side note when I looked at the PCB the in-situ decoupling caps arrangement for the SAA7220 doesn'teem to be the same as shown on the schematic. The schematic shows a 100uF electrolytic and a 47nF cap but in reality the PCB appears to have 3 decoupling caps, a 100uf electrolytic (now Rubycon ZLH), what looks like a film cap of some sort (the orange one) and also what looks to me like a small tantalum cap? (the blue one). If you look back at earlier pictures on this thread they are the three caps at the bottom right hand corner of the PCB. Perhaps this was a late improvement to the design?

I've got another 24 of those Oscons begging to be put to good use somewhere.
 
Well I finally got around to doing something about the power supply to the SAA7220.

In the end I've opted for a dedicated power supply consisting of a 7VA 12v mini toroid feeding a small ready made adjustable voltage power supply module which I purchased from ebay.

The module is based on an LM317 regulator and has a small trim pot which can be used to set the output voltage.

I swapped the rectifier diodes on the module to schottky 11DQ10 types in the hopr that they might be less noisy.

Amazingly everything works and it sounds good too. I'll need to do more listening before I can say if it's an worthwhile upgrade over the standard set-up.

The only problem is that the heatsink on the regulator is getting too hot for my liking so I may swap to a different transformer with a lower secondary voltage (9v?).

The transformer which I have just fitted is supposed to be 12v but measured at around 18v so perhaps that's why the heatsink is getting too hot.


image hosting facebook

I think I'm going to struggle to fit much more under the lid!


screen shot windows 7
 
12VAC small transformer giving 18vdc looks about right.

So is that warm or hot, hot,hot? c.13vdc dropped at 200mA is 2.6w wasted as heat. Looking at the heatsink, that's maybe 10c/w so say 26degC above ambient, and that means maybe 50degC running. Means you cant touch it for much more than 3seconds, I'd guess. Yes, that's hot, but still OK for the 317. Dropping to a 9VAC toroid will save a bit of waste though! (~c.12 or 13vdc once running = c. 1.4w dissipated)

Do check the LM317 board has a capacitor fitted from 317 pin Vadj to 0v (ie across the adjustable pot - and preferably '0v' near the output, not the input.)

If not do add your own - 10uf or more, up to as much as you like. It improves performance a good deal, for nearly free.
 
Hi Felix,

Thanks for the input.

The heatsink does get pretty hot and that was with the lid off, with the lid back on it's going to be an oven in there. I'm going to order a 9v toroid to replace the 12v currently fitted. Do you have any advice on what VA rating to go for? The readily available ones seem to come in 3.2VA, 5VA and 7VA versions.

The PSU module does have a 22uF capacitor fitted across the adjustable pot:)

I suppose I could substitute the adjustable pot for a zener or two green LED's for even better performance?

Whoaa....I just realised that there is fresh air between the LM317 and the heatsink! The screw is practically falling out it is so loose and there is no thermal compound or sil pad.

I'd better sort that then!
 
Good work.
I'd agree with Martin on 7va.
Regulation is usually better at higher VA, though with a 9v type that's not going to impact much here.
 
Hi Rob, thanks for the input.

I've already ordered a 9v 7VA toroid to replace the 12v one currently fitted. I hope that this stops the regulator from running so hot. At present it gets too hot to touch for more than a second or two.

I've done a bit more listening and the results are superb, the dedicated regulator/power supply for the SAA7220 makes a dramatic difference. I'd put it in the same category as the re-clocking.

I might swap out the adjustable pot on the PSU module for either a zener diode or a couple of LED's as a better way to fix the voltage at 5v (as previously discussed on this thread). I've no idea how the adjustable pot affects the LM317 in terms of performance in comparison to a standard fixed value resistor, perhaps Felix can provide some insight on this?
 
It's a very small matter, even less so with the capacitor across it. But if all you need is a fixed voltage, why not go for that?

Unless sealed multiturn types I think adjustable pots all tend to be horrible and ultimately a bit flaky. Not as reliable as a fixed resistor, and since the cheap ones are carbon film, they suffer a little from excess noise too. Not necesarily a big deal in a PSU, but since you are doing this to squeeze as much out as possible for DIY curiosity ...

Using a fixed resistor (or leds etc) allows another slight finesse - where, exactly, you connect it. It should run between the Vadj pin and the 0v at the load for best regulation - it's even mentioned in the datasheet:

r3route.PNG


- Obviously you can't do that with a variable pot on a PCB which has to tie to the board's 0v output pin ;)

Lastly - your 22uF cap formerly across the voltage setting pot should then be connected across the resistor directly, so it links the Vadj pin to the load's 0v pin. That gives best performance overall because it keeps the most direct AC feedback path to the regulator (along with the DC feedback). Have fun!
 
Last edited by a moderator:
I have now fitted the 9v transformer in place of the 12v item. I stuck with 7VA as suggested.

The heat sink on the voltage reg still gets pretty hot but significantly less hot than with the 12v transformer.

For now I have removed the adjustable pot/variable resistor and replaced it with two miniature green LED's. This gives me an output voltage of +5.2v. The data sheet for the SAA7220 states a nominal power supply voltage of +5v and a max of +7v so 5.2v will do nicely.

I also reused two of the old rectifier diodes as circuit protection diodes for the PSU circuit.


free screenshot tool


image url
 

Latest posts

Back
Top