Arcam Alpha 5 CDP upgrades

I have a couple of quick questions:

My first question is about the +5v supply to the SAA7220.

As standard the SAA7220 gets its +5v supply from the main board and this passes through a 33uH inductor just before the chip. When I fitted a dedicated PSU for the SAA7220 I removed this inductor and used it as a convenient way of separating the chip from the normal power supply. It also gave me a nice convenient point to solder the wire carrying my new +5v supply to. Would it be a good idea use the inductor again on the new +5v supply? A brief explanation would be helpful to help me understand the reasoning/implications.


My second question is about the decoupling arrangement for the SAA7220.

As standard the +5v supply is decoupled to ground with 3 different capacitors in parallel:
A 47nF mica,
A 100uf electrolytic (this is now a 220uF Rubycon ZLH as this was part of the Net Audio upgrade kit).
and finally a 33uF Tantalum

(strangely the tant does not appear on the schematic but it's definitely there).

I have also fitted a 47uF Oscon under the chip which is soldered from the power in pin (pin 24) to the ground plane.

My question is, should I just leave the other decoupling caps in situ now that I have the Oscon fitted?
 
mjp200581 said:
I have a couple of quick questions:

My first question is about the +5v supply to the SAA7220.

As standard the SAA7220 gets its +5v supply from the main board and this passes through a 33uH inductor just before the chip. When I fitted a dedicated PSU for the SAA7220 I removed this inductor and used it as a convenient way of separating the chip from the normal power supply. It also gave me a nice convenient point to solder the wire carrying my new +5v supply to. Would it be a good idea use the inductor again on the new +5v supply? A brief explanation would be helpful to help me understand the reasoning/implications.


My second question is about the decoupling arrangement for the SAA7220.

As standard the +5v supply is decoupled to ground with 3 different capacitors in parallel:
A 47nF mica,
A 100uf electrolytic (this is now a 220uF Rubycon ZLH as this was part of the Net Audio upgrade kit).
and finally a 33uF Tantalum

(strangely the tant does not appear on the schematic but it's definitely there).

I have also fitted a 47uF Oscon under the chip which is soldered from the power in pin (pin 24) to the ground plane.

My question is, should I just leave the other decoupling caps in situ now that I have the Oscon fitted?
Click to expand...

Personally I'd just use the Oscon since I think everything else is overkill for the job.
I seem to recall Martin posting some information about multiple parallel caps in power supplies causing unwanted ringing and that might apply here?
 
Thanks Rob, yes it does seem a bit overkill.

The Oscon is operating with tighter loop area than the others as I have one leg soldered directly to pin 24 and the other to the ground plane. I'm not sure what, if anything the other caps will be bringing to the party. Perhaps I need retain the tant or the electro though to keep things stable?

(Who am I kidding I have no idea really).

Let's see what Martin thinks.
 
mjp200581 said:
Thanks Rob, yes it does seem a bit overkill.

The Oscon is operating with tighter loop area than the others as I have one leg soldered directly to pin 24 and the other to the ground plane. I'm not sure what, if anything the other caps will be bringing to the party. Perhaps I need retain the tant or the electro though to keep things stable?

(Who am I kidding I have no idea really).

Let's see what Martin thinks.
Click to expand...

...and the Oscon has very low ESR, and being designed for SMPS uses it has excellent properties for HF filtering. I think its going to do a great job tucked up nice and snug with the chip.
 
I refer the gentleman to post #52 :)

I reckon you can bin most of the bits but I think I'd keep the 100uF electrolytic in there. it's lossy nature will help guarantee the new reg is nicely behaved when 'seeing' a 47uF oscon on the output.

Have a look at this, it's from a NatSemi design note on the LM317 by E Dietz:

dietz.gif


As the output cap increases in size, the tendency to peak drops both in magnitude and frequency. By the time you get around 22-47uf (and maybe 0.02ohm esr for the oscon) that puts the peak around 2-2.5khz, right in the sensitive audio midrange. I know the SAA7220 is a digital part but it directly impacts the bit and word-clock performance, and so...

Anyway - leaving the 'big' 100uF cap in there certainly does no harm. Even if no added damping were required, you can see that extra capacitance alone further suppresses the tendency to ring for any reason including load-realted ones, and drops the magnitude right down for free.


My inclination would be to ditch the inductor for similar arcane reasons - I'll have a look at the A5 chematic and have a think about that some more.


(PS note the results in that graph also depend slightly on the load current drawn from the 317, the set output voltage, and any passing on the adjust pin ..!

Second point - the very last line of Dietz notes is a nice giveaway I can confirm - don't ever rely on just the typical 0.1uF at the input of a regulator, a few tens of uF is helpful if you have a sensitive load)
 
Last edited by a moderator:
Hi Martin,

Perfect timing, the soldering iron is smoking away as I type this.

I realise I had already covered this in post 52 but back then I identified the capacitors incorrectly. The three capacitors are definitely: 47nF Mica, 33uF Tant and 220uF electrolytic.

I'll remove the mica cap and the tant but leave the electrolytic in place if that seems sensible.
 
I think this epic thread needs knocking into a pdf FAQ at some point :)
 
Ok, pics being worth more than words here's a quick and dirty model of that inductor vs. caps new and old.

The bits on the left are a rough approximation of an LM317 running at decent current with the ADJ pin bypassed if you believe the datasheet graphs. The bits on the right represent the inductor with a guessed third of an ohm resistance (I have no idea what it is really, but must be rather low since it carries over 200mA and should not drop more than about 0.2v from the 5v supply to the 7220). There's also 100uf cap with about half an ohm of ESR, and a 47uF cap with about 15milliohms of ESR, representing the existing cap and new Oscon respectively. The load is a constant 200mA, being the '7220:

alpha5vmodel.png


Then three sweeps run, as drawn (red) removing the inductor and its esr (green) andf inally removing the 100uF cap and its ESR (blue)

alpha5vmodelgraph.png


1. The difference between the blue and green curves shows the magnitude of the peaking increases, and goes up in frequency, the oscon and the regulator interact more strongly without the added damping of the 100uF cap - as we expected :)

2. The inductor makes for a larger peak at some lower frequency, no great surprise really since it adds to the output impedance of the regulator. Note a lower resistance for the inductor will make the peak larger.

The inductor also rolls-off incoming noise faster as a trade-off. But the major noise source in this circuit is the '7220, so on that basis I'd probably whip the inductor out, leave the 100uF cap in and hope for a curve like the green one. You could just short the inductor with a wire to bypass it and see how you go.
 
That's superb! Thank you for going to the trouble of doing this.

The Rubycon ZLH series is quite low ESR. I just measured a 220uF 16v like the one which I am currently using in this position and it has an ESR of 0.10 Ohm.

In my pile of bits I have some Nichcon Muse Bi-polar 47uF and some bog standard Rubycon 100uF (which came out of the Arcam) both with an ESR of 0.50 Ohm.

Is it worth swapping out the ZLH to a capacitor with a more ESR to better replicate your model?

Last night fitted some Mundorf M-Cap MKP 1.5uF in positions C13 and C113. These are the DC blocking caps near the RCA plugs. I figured they sounded great before so I would stick with them and use the same thing again.

I also fitted some standard red 5mm LED's on the adj pins of the LM317/337 +/- 5v regs (z203 and z204). This has now given me the following voltages:

+4.79v at the output of the LM317
-4.79v at the output of the LM337
+4.72v at the +5v pin of the TDA1541A (pin 27)
-4.80v at the -5v pin of the TDA1541A
9.68v measured from the -5v to -15v pins of the TDA1541A

I'm pretty happy with all of that. I'd rather run the DAC slightly cool for a longer lifespan rather than run it hot.

This is how it all looks now.


photo hosting sites

I was very excited to find some ERO 100nf boxed radial polypropylene capacitors with a 5mm lead pitch which I was hoping to use for bit decoupling on the DAC. Unfortunately there are just a bit too wide. I think I could get away with it if I fit them alternately one above the board then one below the board. I'm just not sure it is worth all the effort just to swap from polyester to polypropylene though.

If you had room on the PCB I think these caps would be the absolute business.


imgurl
 
'Is it worth swapping out the ZLH to a capacitor with a more ESR to better replicate your model?' Probably not, to be honest, the model is *rough*.

Oh well since it only takes 30s with the model as built, let's see:

alpha5zlh.png


Based on the same circuit as above but changing the 100uF /0R33 cap to 220uF/0R1: now Blue is the curve with the 100uF cap; green with the 220uF rubycon; red the effect of rubycon and leaving the 330uH inductor in-circuit. All rather small changes , but in the right sort of direction - slightly smaller peaks, and smoother/slower rates-of -change.
 
Excellent.

There is actually another 220uF Rubycon ZLH at the output of the PSU module after the reg. In post #64 you can see my sketch of the PSU schematic. The module came with a 0.1uF film cap but I swapped this first to a 100uF electrolytic (post #68) and then finally to a 220uF ZLH (just because).

Hopefully the extra Rubycon at the output of the PSU might actually improve things yet further than the line shown by the green trace on the last plot.

I have managed to get the other audio board working again now. It turned out to be a fault with one of the discrete op-amps. I've now swapped the first pair of op-amps nearest the DAC to op27 and the board is working again. This is especially good news as it means that the TDA1541A DAC chip which I recently purchased is a good. If you want to know where you can get a genuine TDA1541A from a U.K. supplier send me a PM.
 
felix said:
OPA 627 are excellent opamps but they are verywide-bandwidth (50MHz + at unity-gain) and that can cause other problems that degrades the effective result unless real care is taken in implementation - tight, specific local decoupling, care over gain structure - and preferably, a fast 'scope available to check that things are happy.

It should be obvious from the rest of this thread that I do trust Martin's advise but RS were doing some great deals on OPA627's and well, I just couldn't resist the temptation. I'm sure they'll come in handy some day even if they don't stay in the A5.

So I now have a pair of BB OPA627AP installed nearest the output (filter stage?) and some OPA604's nearest the DAC (IV stage?)



The OPA627's in this position have a bit more 'sparkle' that the 604's which by comparison sound a little more laid-back.

It sounds great but looking at the schematic is there anything which could be/should be done to optimise the decoupling of the op-amps?

I haven't yet experimented with swapping the 627's to the IV stage and the 604's to the filter stage. I also have the original OP 27's to throw into the mix.

I've also just ordered some fancy super-duper copper bodied gold plated RCA jacks (cheap off ebay).
Click to expand...
 
Can't leave well-enough a lone, eh? :D 627s are a very, very good choice for I/V duties IME. Your pic appears to show film caps for decoupling duties, which is a good match to the 627s too. They can respond very well to 4n7 of tiny film cap across the power pins (4 & 7) on the opamp if the decoupling is marginal, though thats something I'd prefer to try only with a scope to hand.

The fact you find them add a subtle 'sparkle' is good - IMO changes are (or should be) subtle.

While its no substitute for measurement, my subjective impressions that make me suspect a wildly oscillating or unhappy opamp is(a) if it runs hot (b) just not sound right in a way you can't describe but don't care for or (c) sound obviously dull. In other words, radical shifts in sound quality.

Leave them be and see how you feel about them in a week or so :)
 
Sorry, I can't help it, I think I'm addicted to 'fettling'!

I'll try swapping the op-amps around to put the 627's in the IV position and 604's in the filter position.

The decoupling caps around the op-amps are indeed all polypropylene film types.
 
Arcam Alpha 6 CD

Sorry to join this thread about modification, but I feel that someone reading it might have some advice for me. I have an Alpha 6 CD and I have recently replace the toothed wheel that opens the cd tray. It opens perfectly, but when it closes the motor seems to run on for a couple of seconds even after the tray is closed. Does anyone have any idea why this is happening please? Thanks
 
Hi,

I've got two Alpha 5 players and the motor runs on slightly on both when opening the tray but stops straight away when closing it.

I'm sure someone on here will be able to help.

I have the service manual if that's any help?

Mike
 
Hello everyone,

This thread has been quiet for a while which you can all take as a positive sign that the Alpha 5 is running well without trouble.

It is also sounding so good that I've not had the compulsion to fiddle with it either.......well not much anyway:)

Since my last posts it gained another pair of OPA627's so it's now running 627's throughout.

However, a thought popped into my head the other day about the power supply for the SAA7220 P/B filter chip.

You may recall that I fitted a dedicated PSU for this chip, it now has it's own regulated +5v power supply (separate transformer, rectifier, LM317 etc).

I also fitted an 470uf Oscon to tie the +5v Vin pin of the chip to the ground pin on Martin's recommendation which was a great little mod.

What I didn't do however is separate the ground of the SAA7220 from the rest of the player, it still is connected to the same ground plane as everything else on the PCB.

I'm wondering if the ground plane of the DAC board in general could be made quieter by separating the SAA7220's ground?

Probably a stupid question.
 
if you do that, the player won't work, the clock signals distributed around the player from the 7220 need that 0v return too. Don't worry about it :)

Keeping the power bypass loop-area as small as possible - as you already have - is the most direct and useful thing by far to making it 'quiet'.
 

Latest posts

Back
Top