NAIM CD5 - opamps

[Fidelity Audio]

Hi

Has anyone had any joy changing the opamps in one of these players. I changed the final 2 from OPA604 to AD8610 (soic) and there were nice gains. But what about the other 4. Is it worth the hassle and if so what would you go for.
I was thinking of the LM4562 equiv in single form (LM49710NA).

Thanks
Brent

 

[martin clark]

The 49710 sounds awful here. Dull, dull, dull. I don't think it works very well in filters anyway, but I can specifically tell you it doesn't work in Naim's 3rd order bessel filter.

Fast FET input types do however. Take a good look at datasheets for output impedance - you want ths to be low, open loop (its often not given in datasheets though)for best effect in sallen-key filters.

 

[Fidelity Audio]

Thanks Martin.

I have them on sample order and thought they maybe ok.

I tried the 627 this afternoon in place of the AD8610 and think these were much more suited. Better bass and rhythem.

I'll check through the data sheets of what I have or could order.
What do you have in your 3.5?

Brent

 

[martin clark]

I had 627s. in all three positions. The output pair made the biggest difference by far. If using faster opamps, do add 27-47R in series at the final pair's output to decouple the interconnect capacitance.

AD8065s might be worth a try - haven't used them myself.

The 3.5 and 5 are virtually identical circuits BTW, so you'll find many of the mods common. Specifically, if you haven't already, do #7 (here: http://www.acoustica.org.uk/t/naim/35mods2.html) - it will make a much bigger difference than dicking with these extra four opamps.

 

[Fidelity Audio]

I noticed the pcb said 3-5 on it. It did make me wonder.

So far I have fitted a clock, bypassed the inverter chip (i don't like these). Fitted low noise regs on the dac and opamps today.
I changed the cap on the dac for the ripple current but need to experiment more here.

The difference is amazing, much more open and detailed, also the listenability is much better.

I'll take a look on the site see what other tweeks there are.

Thanks

Brent

 

[lucpes]

All opamps sound better with additional decoupling on-situ (right on top of the opamp). I use Rubycon ZA 220uf/35V soldered directly on the rail legs with great results (sense of power reserve, openness and so on).

 

[vynuhl.addict]

If you want to get really fancy for the outputs try the Walt Jung composite op-amp, better results than you could ever realistically obtain from a single op-amp, excellent current drive and superlative ease of sonics. If used in the I/v stage I have found the OPA627 to be pretty bad sounding when compared to the opa604, Ad823 is my favorite in this role, however it is a dual op-amp. The Opa627 has its own sort of rounded sonic flavour, best when biased into class-a with an active CCS. not a resistor and the AD744/ad811 combo is very neutral, some may call it clinical, mechanical so it really becomes the best of both worlds in some ways.


Colin

 

[Fidelity Audio]

Thanks guys

Martin I did the Analogue bias decoupling hack and it worked very well indeed, it improved pretty much everything but most noticably was the stereo seperation. The layout was different on this player but was easy to sort.

Thanks

Brent

 

[oam]

All opamps sound better with additional decoupling on-situ (right on top of the opamp). I use Rubycon ZA 220uf/35V soldered directly on the rail legs with great results (sense of power reserve, openness and so on).

More details please Would I be mad to try this on my Sony CD-player which just had new op-amps? And how could one determine the suitable cap? Thanks!

 

[KP11520]

Hi Colin,

If you want to get really fancy for the outputs try the Walt Jung composite op-amp, better results than you could ever realistically obtain from a single op-amp, excellent current drive and superlative ease of sonics. If used in the I/v stage I have found the OPA627 to be pretty bad sounding when compared to the opa604, Ad823 is my favorite in this role, however it is a dual op-amp. The Opa627 has its own sort of rounded sonic flavour, best when biased into class-a with an active CCS. not a resistor and the AD744/ad811 combo is very neutral, some may call it clinical, mechanical so it really becomes the best of both worlds in some ways.

Is there any place you can point us to, to find the best way to make an active CCS and employ it? Also, when you biased the OPA627, how many mA did the CCS bias the 627 for best results or did that not matter with an active CCS?

Thanks!

Regards//Keith

 

[vynuhl.addict]

Hi Keith,

For the CCS I am using one BJT a 2SC2705 to be exact one 3mm standard green led and a resistor. The 2sc2705 is a very low collector capacitance part, many seem to use FET's, or current diodes but i have used this one with extremely positive results which makes a simple resistor load sound quite crude .I have the opa 627 biased at 2-3ma max. I really like the opa627 although not so much as a line driver but use it for a first stage preamp buffer so 2-3ma is much more than sufficient in that case.


Colin

WALT jUNG lINK
http://waltjung.org/Classic_Articles.html

 

[KP11520]

Hey Colin,

Thanks for the Link!

I replaced my 5534s in my CD Player with OPA627s (Analog Output stage) and started looking into biasing it after the fact so I am trying to learn about it beyond skin deep.

Some here have said they got the most positive results biasing it to around 8 mA (using various resistors depending on rail voltages). So now I am in the process of installing sockets with a place to easily change cascodes (and maybe resistors to find the best mA spot and then build a cascode at that bias) to find the best results. I have already built a JFET Cascode using 2N5484s (I have a thread on that) but wonder if it is my best option, hence sockets to experiment more until I find Nirvana.

Thanks for the enlightenment, we all hope to use this info to get the best from our systems!

Regards//Keith

 

[vynuhl.addict]

Hi Keith,


Used as an analog output 8ma sounds correct if being used as a final stage. I have found that biasing the opa627 into that high a bias tends to make it quite warm sounding at the expense of low level detail, dynamics also seem to audibly suffer. These are a couple of the reasons i decided to go with a fet input opamp and a current feedback buffer, the combination of excruciatingly low dc offset(great oppurtunity to remove nasty output electrolytics) and plenty of current drive meaning plenty of low distortion control and the ability to effortlessly work with different interconnects and input impedences. Datasheets are indispensable here.


Colin

 

[lucpes]

More details please Would I be mad to try this on my Sony CD-player which just had new op-amps? And how could one determine the suitable cap? Thanks!

- rail (cap) + rail. Well, I use a quality cap here of over 47uf making sure that the rating exceeds both rails combined (usually 30V hence my Rubycon ZA 220uf/35V). Just make sure you orient the - stripe to the neg rail...

See attached pic... and yes, that's a changed LM49720 on a sound blaster 24 bit card (sounds really decent with changed caps) (sorry but that's what I had at hand at the office).

I'd like to know what others think of this mod for OPAs, I find it VERY worthwile.

 

[jives11]

interesting

- rail (cap) + rail. Well, I use a quality cap here of over 47uf making sure that the rating exceeds both rails combined (usually 30V hence my Rubycon ZA 220uf/35V). Just make sure you orient the - stripe to the neg rail...


I'd like to know what others think of this mod for OPAs, I find it VERY worthwile.

Hi, I'm intrigued by this. You are not adding extra decoupling to each supply pin BUT bridging the 2 supply pins with a large polarised cap ? How does this work ? I have seen this topology on a Rotel schematic elsewhere in this forum, though it was a small plastic cap and hence bipolar. I'm wondering if this might work with a 5534 IV & output stage on my old CD104. Thanks

 

[JonHarrison]

You are not adding extra decoupling to each supply pin BUT bridging the 2 supply pins with a large polarised cap ? How does this work ?

I too am puzzled by this. Decoupling the rails I can understand but not with an electrolytic. For analog you'd normally decouple the rails with something like a 100nF film cap - you want something with good HF properties not a large electrolytic.

Putting a seperate electrolytic between each supply and ground risks coupling noise from the supply into the ground which you don't want to do although you do obvious need enough storage to support the rails so you have to reach a compromise on this.

 

[jives11]

I too am puzzled by this. Decoupling the rails I can understand but not with an electrolytic. For analog you'd normally decouple the rails with something like a 100nF film cap - you want something with good HF properties not a large electrolytic.

Putting a seperate electrolytic between each supply and ground risks coupling noise from the supply into the ground which you don't want to do although you do obvious need enough storage to support the rails so you have to reach a compromise on this.

but why couple the +ve and -ve supplies together at all ? Surely they should be independent

 

[lucpes]

Try it, it sounds really good. Dual opamps have no ground connection of their own and they can be operated very well using a single supply (in which case you should use signal coupling caps). So basically what I'm doing here is providing a power reserve and also making sure that we have the same noise in both rails (maybe common mode rejection gets better or???).

Can't really explain why it sounds REALLY good, but it does.

 

[greierasul]

very interesting.....