Arcam Alpha 5 CDP upgrades

I recently ordered a TDA1541A chip from a UK supplier called Little Diode. I've used them before and their service is nice and quick.

The chip arrived today carefully mounted on a piece of antistatic foam and in an antistatic bag.

Unfortunately however what they've sent me is a TDA1541A R1 . The 'R' bit is significant as it stands for 'Relaxed' tolerance so it's not what I wanted.

The good news is that it looks to be a genuine part.



I've just fired off a very polite email asking if they can supply an exchange non R1 variant.
 
Last edited by a moderator:
This afternoon I spotted another Arcam Alpha 5 CD player looking lonely and forgotten in my local Oxfam shop. I certainly don't need another one but like a moth to a flame I was helpless to resist and well....it's for charity so everyone wins....right?

This one takes my collection up to three. I know this is all getting a bit confusing so just to recap I have this one which is standard, another with a modded DAC board (which I use to test out modifications) and finally the fully modded layer which is the main subject of this thread.



When I got home I fired it up and was pleased to find that it reads discs well and sounded pretty good. In fact I'd forgotten how decent the Alpha 5 is in standard trim.

I took the lid off and found that the player has been repaired at some point in the past. It's clear that at least two of the main smoothing capacitors on the main board have failed at some point and leaked all over the PCB.



Four of the electrolytics, at least one resistor and one ceramic cap have been all been replaced. It's nice to see that whoever did the repair used good quality Panasonic FM and Rubycon YXF electrolytics. I feel vindicated in my decision to replace all of these when I first started modding mine. If anyone reading this has an Alpha 5 CD I'd thoroughly recommend you do the same.

The DAC chip has also been replaced at some point (the solder joints are not original). The chip has some glue residue on it which suggests to me that it had a a heat sink glued to it at some point. The glue has taken off all of the lettering so I can't tell which variety of TDA1541 it is. So far two of the three Alpha 5's I've had have come with a replacement DAC chip which makes me think they maybe run a bit too warm in this player. On my fully modded player the voltage supplies are all slightly under spec now which ought to help and I've put a heat sink on top too.

 
A large portion of chips.

Last week Martin kindly lent me his spare TDA1541A S1 for testing.

I recently fitted an IC socket for the DAC chip in my 'reference' Alpha 5 so it seemed like a good opportunity to do some listening comparisons between different TDA1541A chips.

Over the last few years I've built up a small collection of chips comprising of: Three Dutch made TDA1541A's, one Taiwan made 'A' and also the TDA1541A R1 which I only recently purchased. Added to this mix is Martin's TDA1541A S1.

I did the initial listening tests on my 'development hack' player which has a heavily modified DAC board as is still a very good player. The best chips were then compared in my fully modded 'reference' player.



I fabricated a little tool out of some steel strip which I used to carefully lever the IC's out of the socket with less risk of damaging them.



Here are my listening impressions:

TDA1541A: HSH9349, HSH9423 and HSH8844
I found these chips to all have a very similar sound. From this group I think the HSH94 was my favourite by a very small margin and sounded a little clearer. The HSH88 was perhaps a shade 'warmer' with marginally softer bass and the HSH93 was somewhere in the middle. Essentially these three chips all sound the same, I doubt I could tell between them in a blind test.

TDA1541A R1 HSH8740:
I was fully expecting this chip to sound awful but very quickly my negative preconceptions were proven largely unjustified. The R1 chip sounded almost identical to the three Dutch made TDA1541A chips. If I'm being ultra critical I think I did find the soundstage a bit flatter, less '3D' if you know what I mean but basically it sounds like any TDA1541A does.

TDA1541A S1 HSH8736:
I was very excited to here the S1 and was expecting great things from it. I found the S1 to be a nice sounding chip with a pleasing 'crispness' and good detail. My impression is that channel balance and stereo imaging was perhaps a shade better than the three TDA1541A chips. Overall the S1 shared the same sonic character as the three TDA1541A chips but with a tiny pinch of added 'je ne sais quoi'.

TDA1541A Taiwan 1997:
For me this was the big surprise of the listening tests. I found this chip to have quite a noticeably different character to the other chips tested. The sound of this chip is more dynamic and vigorous than the Dutch chips. Detail is on a similar level to the S1 but the bass of the Taiwan chip is especially impressive and superior to any of the other chips tested.

Conclusions?:
TDA1541A - The three chips I tested sounded very very similar. Measurements may be able to differentiate between them but I doubt I could reliably identify each one in blind auditions.

TDA1541A vs TDA1541A R1 - The R1 chip has almost all of the same sonic virtues as the standard TDA1541A. If you have a choice between a standard 'A' and an R1 chip then go for the standard 'A' but if you already have an R1 fitted or if R1 chips are all you can find then rest assure the difference is small and not worth worrying about unless you are looking for perfection.

TDA1541A S1 vs TDA1541A - You have to remember that the S1 is 'just' a selected TDA1541A with guaranteed 'good' tolerances and that's exactly how it sounds. Listening to an S1 chip is like listening to any good TDA1541A, nothing more. To my mind the advantage of an S1 chip is that it comes with the reassurance that you're getting a 'good' chip. Personally I don't think that they justify the dramatically higher price which they tend to sell for nowadays. I haven't heard an S2 chip so I can't comment on those.

TDA1541A Taiwan - The final few years of TDA1541A production (1997-2000) were done in Taiwan. My Taiwanese made chip has a noticeably different sonic character to the earlier Dutch made chips. To me the Taiwanese chip has better dynamics, sounds more energetic and has more powerful bass. I will concede that some people may prefer the sound of the Dutch chips and they may work better with some systems. However for me the Taiwanese chip is the clear winner so far and this is the chip which has been put back into my best CD player.

The S1 may have the crown but is it still the King?
 
Very nice little test, and good news in the sense that if you need a replacement chip it's not probably worth agonising over exactly which variant you can find at a reasonable price - probably more mileage in either
(1)buying several non-blessed spares and sorting them to taste rather than getting bitten for a 'blessed' crown variant.
or
(2)At this age spending the difference on ensuring the machine is in good order in all other respects e.g. replacing old/suspect PSU caps and the like.
-or both.
 
felix said:
At this age spending the difference on ensuring the machine is in good order in all other respects e.g. replacing old/suspect PSU caps and the like.
-or both.
Click to expand...

I definitely agree with that. My first Alpha 5 sounded dreadful until I replaced the PSU caps on the main PCB.

Little Diode has offered to exchange the R1 chip they sent me. I've asked them to send me the chip with the latest possible production date if they have a selection to choose from. Let's see what turns up.
 
My replacement chip finally arrived today. This is what they've sent me:



I'll report back after testing it in my spare player.
 
A quick test in my spare player has shown that the chip is fully working and makes all the right noises.

Happy days! :)
 
I been waiting for a rainy day to post about the latest series of improvements to my Alpha 5 CD player.

These latest upgrades have proven to be spectacularly successful! I hope you approve.

First up, I decided to beef up the filter capacitors after the rectifiers on the main PCB:

C406 and C407 smooth the rectified 25vdc supply to the DAC board and these were swapped to Panasonic FM 1500uF 35v (up from 1000uF).

Similarly C403 and C405 are the post rectifier smoothing caps for the main PCB. These were swapped to 3900uF Panasonic FC and 1500uF Panasonic FM respectively (up from 3300uF and 1000uF).

Here are the new bigger caps in situ:



While I had the main PCB out the electrolytic cap on the SAA7310 decoder chip was swapped to a 47uF Oscon. It's probably too far away from the IC to have any real benefit but hey, it can't hurt right?



The 3300uF Rubycon ZLH which was taken from the main PCB was reused on the SAA7220 PSU to replace the nasty cheap (and old) capacitor which I had been using as a temporary measure.



After reading one of Martin's posts on another forum I also had a rethink about the layout and wiring of the SAA7220 power supply. Previously I had long AC carrying wires from the transformers secondary to the rectifier and then very short DC wiring from the PSU output to the decoupling cap(s) at the SAA7220 chip. I now realise that it's better to keep the transformer and rectifier close to each other and as long as you have a decoupling cap next to the power consuming device the longer wiring carrying smoothed DC isn't a problem.

Bear with me the exciting stuff is coming up in the next post....
 
Next job was to swap the Audiocom clock PSU module to the NetAudio one which I also have. There's nothing wrong with the Audiocom module but somehow it just seems 'right' to be powering the Net Audio clock from the PSU which it was designed to partner. I managed to get it tucked up nice and close to the clock.

Note this is the superior regulated Net Audio clock PSU.


I've been wanting to fit some really top class capacitors in the DC blocking role at the output for some time but I've been putting it off as I couldn't find a neat way of installing them. Well I've decided that sound quality has to take priority over aesthetics so in went these big Clarity Cap polypropylene film capacitors.



As you can see it's a very ugly installation but they sound absolutely wonderful so I don't really care. These are from the Clarity Cap ESA range and they are the 250v and 2.7uF variety.
To fit them I needed to drill the through holes out to 1.5mm to accommodate the thick copper leads.

More to follow...
 
Over the last year I've been doing lots of reading about the TDA1541 DAC and how to get the best results from it.

Something which keeps coming up is the subject of the 14 bit decoupling capacitors which surround the DAC. I already know from past experience how important these capacitors are how the type of capacitor(s) used dramatically impacts on the sound quality.

Slowly, I've been coming to the conclusion that surface mount polyphenylene Sulphide film capacitors are likely to be an excellent choice. Inspired by the positive comments from other people who have tried them I decided to give the PPS caps a try.
After searching through the usual parts suppliers I settled on Panasonic ECH-U1H104GX9 100nF. These are 4.8mm x 3.3mm x 2.1mm so will fit over the existing solder pads vacated by 5mm pitch radial boxed film caps nicely.

I chose to leave the 1uF polyester caps on the MSB pins 13 and 18 and also the 470nF polyesters on pins 12 and 19.

The remaining 220nF caps were removed and replaced with the PPS films.

Patience and a steady hand required!



The TDA looks odd without the familiar boxed film caps around it.



The small blue capacitor C227 is a 47nF multilayer ceramic. There are three of these around the DAC, two for decoupling of the +5 and -5v lines and a third which ties the analogue ground to the digital ground between pins 5 and 14. I think these could be replaced with something better, any suggestions? Surface mount COG ceramic perhaps?
 
Blimey, it sounds fantastic! :)

Much more of a noticeable difference than you'd expect.

Cymbal crashes in particular stand out as sounding far more lifelike than they did before.
 
Martin,

I found this post from you over on another forum. It's regarding the SAA7220 chip:

One more little thing - for chips designed to work together - they don't interface very well without a bit of work! Find the long 'TDA1541 thread in this forum of more details, but the key tweak is to use RC decoupling in the signal lines between these two chips to reduce HF noise injection into the dacs substrate (1K/10pF was suggested)*. Your Arcam already has some resistors in place in these lines, so definitely add small caps (10-22pF) to digital ground at the 1541 input pins. It helps a surprising amount.

*this is because the 7220 swings 5v pk-pk on its outputs, and the 1541 only needs a tiny current centred c1.4v to switch - it's a current-routing logic; the excess voltage swing just pushes HF currents into the dac.[/I

Is this something which you'd still recommend?
 
Yes! Use COG ceramics for the tiny cap values (the maplin caps at this value are COG/NPO, just fine) Ideally they should return to 0v plane at the SAA7220, but very shortest route from resistor on the dac side to common digital 0v is best.

In my CD2 I settled on <checks> 1K (1206 size smt) in series with the data lines both with 5.6pF to digital 0v; I just have 75 or 100R in series with the bitclock (5.6Mhz0signal from 7220 to the dac - no capacitance to 0v on this line.

Try it, costs pennies since the resistors are already there in the A5.
 
Great, I'll definitely give that a try!

Since I've been gaining confidence with surface mount components would it be worth looking at the power supply decoupling of the SAA7220 again?

Currently I have a 47uF Oscon soldered onto the underside of the PCB directly onto the +5v pin and to digital ground (see pic in post #68). There is also a 220uF electrolytic to 'damp' the supply.

I could add a surface mount ceramic directly onto the ground plane as you suggested in post (#51) if you think there would be any benefit?


I must say the last few mods have been something of a revelation. There has been a BIG jump in SQ which I find hard to explain.

Are the surface mount PPS caps really be THAT much better than boxed polyester?
Or are the ClarityCaps really THAT much better than the Mundorf MKPs?
Or perhaps the Net Audio clock wasn't very happy running from the Audiocom clock?
 
I've had some mixed results with the small NPO caps.....

I tried this mod first on my spare 'development' player. I used a 5.6pF NPO from the DAC side of R213 (390R) to digital ground and another 5.6pF from the DAC side of R215 (390R) to digital ground. I didn't do anything with R214 which carries the bit clock.

I very much liked the results so I then tried the same mods on my 'best' player. I repeated the mods exactly as far as I'm aware using the same solder points and the same capacitors but after fitting them there was no audio output from the player. All I would get is a brief FZZZTTT sound through the speakers and then silence. In the end I removed the two caps and the player worked again with no ill effects. I don't understand why it would work on one player and not the other :confused:. The only major difference between them is that the 'best' player uses an external clock and associated clock PSU.

As usual I couldn't resist doing some other fettling too while I was in there.

Now that I've 'seen the light' regarding surface mount components I swapped C222 and C220 for surface mount NPO ceramics of the same value (47nF), these are decoupling caps for the +5v/-5v supplies to the DAC.
I did the same for C227 which connects digital ground to analogue ground (joins pins 5 and 14 of the TDA1541A).

I also tried a neat trick which I read about on another TDA1541A thread:

It seems that the feedback resistors on the op amps of the I/V stage have a significant effect on sound quality and resistors with low inductance sound the best. Some people have reported great success with bulk 'Z' foil resistors such as those made by Charcroft, Vishay and Texas.

A neat little trick using less expensive metal film resistors is to use multiple resistors in parallel as a way to reduce the inductance. I therefore swapped the single 1K2 resistors (R4, R104) for a pair of 2K4 in parallel, one on the upper surface of the PCB and the second soldered across its legs on the underside. I used the PRP resistors from Hi Fi collective again as I find these to be great sounding and very reasonably priced (think of them as the Panasonic FC of resistors).

In this pic you can see the larger surface mount 0.1uF PPS bit decoupling caps under the TDA1541A. Behind them you can see the slightly smaller 46nF NPO caps and in the background the two tiny 5.6pf caps on R213 and R215 going to digital ground.

On the right you can see the red coloured paralleled op amp feedback resistors.

 
Last edited by a moderator:
I recently treated myself to a good quality multimeter which has a frequency test function.

This has allowed to do some much more in depth analysis of the DEM frequency on my Arcam Alpha 5.

I tested three different TDA1541A chips with a variety of DEM capacitor values. All measurements were taken with the CD player playing music. The results are shown in Fig. 1.



As you can see there is clearly some variation between different TDA1541A chips. Most likely due to the very slack tolerances of R Int (+/- 25% on the data sheet).

Previously I had gone to some lengths to try and calculate what the capacitor value would be needed to give me a DEM frequency of 176.4 kHz (see posts #118- #120).

Using the formula given in the Philips data sheet I arrived at a value of 533pF which is what I have been using recently. However the results shown in Fig. 1 show the 533pF capacitor was not getting me terribly close to 176.4kHz in practice.

I was able to achieve a DEM frequency closest to 176.4kHz when using a total capacitance of 580pF made up of several smaller values in parallel. I don't really like the idea of using several caps in parallel and the closest standard value you can get is 560pF.

So...... I purchased a bag of 10 Wima FKP2 560pF and tested each one in the circuit with my favourite TDA chip (the 1997 Taiwan). The results are shown in Fig.2

Using the small variations in capacitance within the sample of 10 Wima capacitors I was able to select the most optimum value. The 5th capacitor tested gave me a frequency of 175.2 kHz which is pretty close to ideal.

When selecting capacitors this way it is obviously a positive advantage to buy lower tolerance components.

Some important notes:

  1. During my testing I found that the DEM frequency started low and increased steadily as the TDA1541A warmed up. Once the TDA had reached thermal equilibrium the DEM frequency plateaued out. I found it was necessary to leave the CD playing for at least 3 minutes before the frequency stabilised.
  2. I found that the 6K8 resistors tying each pin of the capacitor to -15v had a significant stabilising effect on the DEM frequency. Without the resistors in place the DEM frequency would fluctuate up and down much more than with them fitted (typically around +/- 4kHz fluctuation).
  3. Fitting the resistors alters the DEM frequency. I found the DEM frequency was higher for a given capacitance with the resistors fitted. Without the resistors fitted the formula given in the Philips data sheet is quite acceptably accurate.
 
Now that's interesting.

I think all your points 1-3 can be explained by the warm-up of the transistors in the DEM oscillator and the consequent increase in hfe (current gain) and slight drop of Vbe you'd expect as a result IOW. I've got a bit of an internal schematic somewhere - will post what I have in mind later.

I think I need to drag an Alpha 5 out of storage and play... & compare to my longsuffering CD2...

ETA:
DEMosc.png

Vbe for a transistor drops -2mV/degC. That will mean the cross-coupled transistors - the pair with the emitters at about -9v - will turn on earlier/faster (and slight hfe enhancement means 'harder') as the die warms. That means the Cosc cap wil be charged a tad harder, so voltage acrosss it rises faster and so the timing period drops - as you've observed.

Then - loading them with more than the default 200uA (6K8 to -15V adds about 1ma, which is massive) will swamp this: so making the pair run much harder, i.e. at a level that is insensitive to the die temperature - principally by enhancing hfe. That matches your observation also.


And it makes me think that that using a current source for 'boosting' the DEM circuit thus would be the better idea than 6K8 or 10k resistances - owing to considerably better PSRR through higher dynamic impedance offered by CCS. Something like a 600uA CRD (J503 - J505 or similar), of which I have a couple spare. Must go play soon...
 
Last edited by a moderator:
felix said:
And it makes me think that that using a current source for 'boosting' the DEM circuit thus would be the better idea than 6K8 or 10k resistances - owing to considerably better PSRR through higher dynamic impedance offered by CCS. Something like a 600uA CRD (J503 - J505 or similar), of which I have a couple spare. Must go play soon...
Click to expand...

Very interesting, please keep me up to date with your thoughts on this!

I think I need to drag an Alpha 5 out of storage and play... & compare to my longsuffering CD2...
Click to expand...

I didn't know you had one!

It'd be interesting to audition my Alpha 5 alongside your CD2 sometime. You are very welcome to borrow my Alpha if you are curious to hear it.

I have another Farnell order arriving today (mostly bits for the Leak amp). I'm going to have another go at implementing the RC filtering on the data lines between the SAA and TDA. This time I'll be using 1K resistors and 5.6pF as you suggest.

I'll report back later.
 

Latest posts

Back
Top