Philips CD 750: advice for output stage rebuild

[matose]

Hey guys,

I've been modding the hell out of this CD player in the last months and the last stage of the rebuild is the output stage and I need some advice from you guys.
The mods until now: power supply caps changed to Panasonic FC, all other caps upgraded to low ESR polymer Nichicon / OS-CON. Fitted 3 ultra low-noise 5V regulators in addition to the one already present to power the demodulator chip, digital filter chip and DAC.

Right now I've made the output straight to RCA after the first differential OPAMP (half of LM833-N) - where the green line is - and the sound it's very very good. I guess the second half of the LM833-N is a LPF filter or buffer.

I'm a big fan of LME49720HA and I have a couple on the way so the next stage is to build an output stage around those. Should I just swap them instead of the existing LM833-N and use only half of each OPAMP per channel or try to use the second half in slave mode like in the second picture attached? The distortion should go down by a significant margin, what do you think?

The DAC is SAA7350AGP if I didn't mention. What does the 3318 resistor and 2332 capacitor does in the below schematic? It's a filter at the OPAMP input I guess, can I remove them? The 10K resistors (3314 and 3316) must be for level attenuation at the OPAMP input and should be left in place. Am I right?

Thanks a lot for your time,
Matei

 

[rrrremus]

Salut Matei!

I would suggest that you search for a direct replacement op-amp so that you keep the modifications to a minimum. There is no need to fit the second op-amp as slave as you call it. This topology only helps reduce distortion while the op-amp is driving really low impedance loads. That is not the case here and will only cause more harm than good in my view.
After you have swapped the op-amps and have checked that they do no oscillate ( you will need a scope for this ) you can try to remove what appears to be a mutting transistor ( 7304 and 7382 BC818 type ). Ther eis a possibility that without these in the circuit you will hear some pops while changing tracks, stopping, pressing play etc. See how that goes.
Another step would be to change resistors to thin film type 0.1% and film caps.

Best of luck.

 

[matose]

Thanks for the quick answer Remus!
The mutting transistors are long gone 🙂. I already bypass the rest of the circuit after the first half the LM833N opamp (from the green line straight to RCA, no need for caps because DC offset is very small).

So you are saying to fit the LM49720HA straight in the actual circuit and use it as such. Should I use one LME49720HA per channel (one half not connected) or use both its sections (one for left channel and one for right channel)?

Cheers

 

[DF96]

I guess the second half of the LM833-N is a LPF filter or buffer.

Yes, it is a modified Sallen-Key second-order low pass filter. I'm sure they only included it to avoid wasting a spare opamp. The passive components around it must have been going spare in their Stores Department at the time.

What does the 3318 resistor and 2332 capacitor does in the below schematic? It's a filter at the OPAMP input I guess, can I remove them? The 10K resistors (3314 and 3316) must be for level attenuation at the OPAMP input and should be left in place. Am I right?

Maybe these too were going spare? Or perhaps the designer intended a low pass filter to prevent digital edges from causing slew rate limiting and hence distortion in the opamp?

I am puzzled how you can think it possible to improve a circuit which you clearly don't understand.

 

[matose]

There's always room to improve and simplicity is the way to go. Thanks for the input but nobody was born a doctor in electronics, so we must learn 🙂

So I can get away with half of OPAMP per channel, the question is if I will get a better sound by not using the second half at all or should I use one OPAMP for both channels?

If I understand correctly the 3318, 2332, 3314 and 3316 can go? 🙂

 

[poynton]

If I understand correctly the 3318, 2332, 3314 and 3316 can go? 🙂

Not really !!

They form the required feedback components for the unity gain buffer.
They are not to attenuate the signal.


Andy

 

[matose]

You're reffering to 3314 and 3316 or all of them?

Thanks

 

[DF96]

There's always room to improve and simplicity is the way to go.

Simplicity is only the way to go if either you value simplicity over sound quality or the circuit is too complex for the task it performs i.e. you understand the circuit and its task better than the original designer.

If I understand correctly the 3318, 2332, 3314 and 3316 can go?

All the components around the first opamp are needed if you wish to retain its functions, which are:
1. low pass filter
2. differential mode buffer
3. common mode rejector
Which of these functions do you feel you can discard?

 

[matose]

Thanks for the reply!

1) From what I read the LPF is not really needed so I guess it can be discarded

2) Differential mode buffer should be kept if I want single-ended output so it stays

3) I know that the LME49720HA that I will use has a good CMMR so I don't know if I need extra common mode rejector

Sorry to put a lot of questions but I'm trying to better understand the circuit and make the best choice. Thanks!

 

[DF96]

1) don't believe everything you read, instead work out what the LPF does and decide whether you need it or not
3) don't confuse chip CMRR with circuit CMRR

 

[korpberget]

KISS.
Check out Lampizator.
CD_ALPHABETICAL
I am working on a Cambridge Azur 640c now. I have two of them for comparison after putting a tube in one of them.
Exciting!

 

[matose]

Yes, I know the Lampizator and read a lot of stuff from him. He is saying the same thing: remove everything you don't actually need. I just want to know what I can remove safely 🙂

 

[DF96]

Much of what Lampizator says can be regarded as eccentric. I am being polite.

 

[korpberget]

I am not a believer yet, but I just fired up the lampized Cambridge and had sound!
Need some fine adjustments and stuff.
Gonna test them this week I hope.
Excited if I can hear any difference or improvement. Like the theory tho.

 

[korpberget]

BTW removed six opamps and all sound output links from both dac's. They are not connected to anything else but the tube circuit. (signal, that is)

 

[matose]

Attached is the reference SAA7350 output circuit which looks pretty similar, the values are a little difference but the topology is identical. Any opinions?

 

[DF96]

It is reasonable to assume that the chip manufacturer's application engineers know how to get the best out of their chips, unless there is clear evidence to the contrary. I would start by fully understanding and analysing their circuit (and the similar Philips one). Only after that are you in a position to offer constructive criticism of it and suggest genuine improvements of it. That is my opinion.

I have given you a start by telling you in outline what each opamp does. You can continue by calculating corner frequencies; Q for the Sallen-Key; the effect of the added inductor in the Sallen-Key.

Alternatively, you can degrade the sound by cutting bits out or following the latest fashions in CD-player circuit abuse.

 

[matose]

I don't really understand the practical need for a LPF filter at a Q of 85-90kHz (usual values) when nobody can hear above 18-20kHz. I see just added stages and more components in the sound path 🙂

 

[rrrremus]

I don't really understand the practical need for a LPF filter at a Q of 85-90kHz (usual values) when nobody can hear above 18-20kHz. I see just added stages and more components in the sound path 🙂

In the old days some EMC experts used an audio band amplifier in order to locate EMC disturbance points in high frequency circuits. High frequency disturbance can get modulated into the audio band or even cause oscillations if using high bandwidth op-amps.
I have had such experience with a very high speed buffer from TI. Noise in the tenths of Mega Hz caused instability and overheating of the buffer.
Of course, you might not need the LPF, but please test before and after

 

[DF96]

I don't really understand the practical need for a LPF filter at a Q of 85-90kHz (usual values) when nobody can hear above 18-20kHz.

Yes, it is clear you don't understand. See my post 4 for an explanation.

I see just added stages and more components in the sound path

I guess that means you believe in simplicity for the sake of simplicity. Putting simplicity above sound quality is a common newbie error. The simplest possible CD player would just connect the laser pickup directly to a loudspeaker. It would not work very well.