Op Amp Swap in a CDP Question - OPA2134 for the stock NJM4580D

[seventy1]

I just finished a full recap on a Sony 508ESD CD player and had so much fun that I'd like to try an Op Amp upgrade that I read about in a very old post elsewhere on the interweb. The TI OPA2134 looks like an easy drop-in replacement for the original NJM4580D, but what has me puzzled is the advise I found on upgrading the power supply caps for the OPA2134s. The advice was:

..great results running OPA2134's with at least 470uf Nichicon & 0.47uf PP film directly from each chip's supply pins to ground. Going from 220uf to 470uf with generous film cap bypassing really brought the bass into 'full reality' and gave the whole picture much more depth of tone. (edited)

I see (3) 220uF caps in the power supply circuit: C378 and C379 in the OPA vicinity, C378 goes from +5v to gnd, C379 goes from -5v to gnd. Then C920 from -16v? to gnd near the power supply area of the board - I assume this one is not in play.

Would the recommendation be to replace the 220uF C378 and C379 caps with 470uF and PP bypass caps? Or to leave the 220uFs and add (2) off-board 470uF caps: one with the "+" lead wired directly to the V+ pins on the OPAs and the "-" lead to gnd; then the other with the "-" lead wired directly to the V- pins on the OPAs and the "+" lead to gnd? This probably sounds like a stupid question to someone who knows what they're doing, but it's not clear to me. Thanks for any help!

Sony CDP 508ESD Schematic

 

[dr*audio]

C907 and C906 in the power supply are already 470uF and are on the +5V and -5V supply lines. No need to add caps. The opamps in this player are not putting out a lot of current and will not benefit from adding the caps. Do use 8 pin sockets so you can try other opamps. I have had excellent results with LM4562 opamps.

 

[Blue Shadow]

I put the LM4562 in an Onkyo and an OPA2228 in a Sony CDP-C705. Used sockets but can't take the Sony out of the rig to make a swap as I have a spare 4562 to try, it was that much better than the 5532. No other change to the Sony.

 

[mbz]

..great results running OPA2134's with at least 470uf Nichicon & 0.47uf PP film directly from each chip's supply pins to ground. Going from 220uf to 470uf with generous film cap bypassing really brought the bass into 'full reality' and gave the whole picture much more depth of tone. (edited)

The above quote is almost certainly talking about C378,9.
Increasing these to 470uf would probably trigger a need to upgrade C906,7 to 1000uf?.
I've seen some vintage CDP's that used blackgates in these (C378,9) positions so I'd replace with something like Nichicon KZ.
Certain use an opamp socket for rolling but keep in mind the supply voltage/current for selecting possible replacements.

 

[dr*audio]

This player uses +/-5V supply for the opamp so anything should work with that, unless you pick one that requires more than +/-5V, which would be unusual.

 

[seventy1]

C907 and C906 in the power supply are already 470uF and are on the +5V and -5V supply lDo use 8 pin sockets so you can try other opamps. I have had excellent results with LM4562 opamps.

Thanks, great suggestion on the sockets. I was thinking I'd replace them and I'd be so thrilled that I'd never touch it again. Fat chance, right? I'll find these at Digikey or Mouser?

Ok, so it sounds like no one sees a benefit to wire large capacitors directly to the opamp power pins? The original post I saw on this was very emphatic about big caps directly to pins, even mentioned: When I have the room, I routinely go as high as 2200uf on 2134 supply pins. I didn't want to mention the poster's name since it appears he's been somewhat controversial, but very experienced in these Sony CDP mods (S. Sank).

The above quote is almost certainly talking about C378,9.
Increasing these to 470uf would probably trigger a need to upgrade C906,7 to 1000uf?

In full disclosure, in my recap, I did already replace the 220uF caps at C378, 9 with 470uF Nichicon FG, no PP bypass. C906, 7 got 470uF Nichicon HE. I've been playing it ~30 hrs with the original opamps, would I notice anything if they were incompatible?

Thanks all for the input, very helpful!

 

[dr*audio]

I would add 1uF film bypass caps to the 470uF. You can mount them under the board.

 

[seventy1]

add 1uF film bypass caps to the 470uF

I have some spare 3.9uF Erse PulseX PP caps, overkill?

 

[dr*audio]

I have some spare 3.9uF Erse PulseX PP caps, overkill?

That's fine. It's just to compensate for the higher impedance of the 470uF caps at higher frequencies.

 

[seventy1]

So, the original is in the middle and the two TI op amps are what I plan to try. But I'm questioning compatibility - the three are very different in slew rate, gain bandwidth product, input bias, output current and voltage. Slew rate, higher is better, so I guess no problem. GBP, anything >30kHz is ok for audio? Current input bias and output, no idea. Voltage, the symbols are a little confusing, but my guess is when they are used with V+ and V- inputs (like this application), then they're good down to 2.5 volts. Let me know if I'm missing anything.

Also, for the sockets, is the preferred style the round pins with the gold plated ID like below? I wanted to be sure the Op Amp pins were compatible with gold, I know that gold and tin don't like each other, but the TI pins don't look like they use tin. Thanks!

 

[dr*audio]

GBP of 30kHz is pretty crappy. That means that if the gain is unity the bandwidth will be 30KHz but if the gain is 2 the bandwidth is 15Khz.
In this application the input offsets and current don't matter much but you want distortion and noise to be as low as possible, higher slew rate, higher bandwidth.

 

[FileFixer]

U can use OPA1612 or OPA1642 instead of NJM4580 or M4580 from Mitsubishi. If you have schematics for Sony 508ESD look what capacitor you have on negative feedback on that NJM4580. I have 100pF and OPA1612 work grat without oscilating on my Pioneer DV-410V DVD Player.

 

[dr*audio]

The thing to do is check the data sheet for the opamp you use. If it is unity gain stable you don't need a compensation cap in the feedback. If it is not unity gain stable or not stable into a capacitive load, it may require a resistor in series with the output, or a compensation cap. the datasheet will tell you what value you need. It is usually in the Applications section.

 

[Retrovert]

So, the original is in the middle and the two TI op amps are what I plan to try. But I'm questioning compatibility - the three are very different in slew rate, gain bandwidth product, input bias, output current and voltage. Slew rate, higher is better, so I guess no problem. GBP, anything >30kHz is ok for audio...

Higher speed means the opamp may oscillate with higher-frequency input, whereas a slower unit would not. The oscillation, BTW, will be in the MHz, so unless a search is specifically conducted it will likely be missed. This is why TI will not provide tech support to opamp rollers. It is also why opamp rolling often "inexplicably" results in hot or overheated opamps (oscillation), howling in the output, or a "warm" sound; all of this is a sign of oscillation.

Tubes are happy to oscillate at the same frequencies, unlike the older transistors which were slow. The same issue consequently arises when slower transistors are replaced with faster ones. The bandwidth goes way up and oscillation, formerly impossible, now rears its ugly head.

Arguments abound about the risk of using more than 10 V/µSec and how unnecessary it is.

 

[dr*audio]

It's not hidden information, it's in the datasheets. If they don't say the opamp is unity gain stable, assume that it is not. They will specify (usually) the capacitive load that it can drive. The interconnects have capacitance so that represents a capacitive load. If the opamp can't drive more than 100pF, a 1K resistor should be added in series with the output, or oscillation will result.

 

[W9TR]

I think it’s fair to state that if you are going to roll op-amps you need to have a decent 100 MHz ‘scope to determine if there are any oscillations in the output of the modified circuit. The option is to follow in the footsteps of someone who has made the exact swap-out you are considering and has the test equipment to validate the swap.

A lot of folks won’t want to recognize this caveat as it is expensive and potentially inconvenient.

Tom

 

[Retrovert]

I think it’s fair to state that if you are going to roll op-amps you need to have a decent 100 MHz ‘scope to determine if there are any oscillations in the output of the modified circuit.

This is the key point. The oscillation can be range from low frequency (hundreds of kHz) to high frequency (MHz to tens of MHz), and unless one looks for it the issue never surfaces. The same problem occurs in tube amplifiers or transistors which have been upgraded.

Oscillation is very, very destructive for an amplifier and can burn it out. Giant wave of voltage running to the rails can easily blow output devices, and the high frequencies can blow tweeters. Snubbers or Zobels will be needed at various stages to ensure a faster device doesn't take everything to Valhalla.

The option is to follow in the footsteps of someone who has made the exact swap-out you are considering and has the test equipment to validate the swap.

Ahhh, but that pre-supposes that one might find such an upgrader and be able to trust the statements.

Many times the claims about improvements in sound are subjective and dubious upon examination. Sometimes the upgrader will self-disqualify.

I had an equalizer from the 1980s which used discontinued RCA opamps. (Since regifted to a new home.) I spoke with a rebuilder who had advocated using adapters to install better opamps for lower noise and claimed the sound was superior. Ok, I'll buy that. Why not? Given that the new opamps had twenty times the bandwidth I asked about frequency compensation to prevent oscillation. "What's that?" was the reply. Uh-ooooooh. Never a good sign. Ummm, ok, if you didn't look for it on a scope, how about the opamps got really, really hot? "Oh, that's why they got so hot! I just kept swapping until I found ones that didn't get hot enough to burn my fingers." Yeaaaaaaaaah. Okaaaaaay. That's me backing away from the upgrade, with slow and deliberate movements to avoid antagonizing the animals.

A lot of folks won’t want to recognize this caveat as it is expensive and potentially inconvenient.

True that. Also interferes with the ability to put one's imprimatur on equipment by randomly swapping parts and claiming an "improvement" which means the swapper was somehow greater skilled than the original designer. A lot of audio is just nonsense, where poor acoustic memory and a desire for bragging rights turns into a giant smoking ruin.

 

[seventy1]

It's not hidden information, it's in the datasheets. If they don't say the opamp is unity gain stable, assume that it is not. They will specify (usually) the capacitive load that it can drive. The interconnects have capacitance so that represents a capacitive load. If the opamp can't drive more than 100pF, a 1K resistor should be added in series with the output, or oscillation will result.

From the OPA2134 datasheet: The OPA134 series operational amplifiers are unity-gain stable and provide excellent dynamic behavior over a wide range of load conditions, including high load capacitance... These devices operate up to 36-V supply rails and offer ultralow distortion and noise, as well as 8-MHz bandwidth and high capacitive load drive.

From the LM4562 datasheet: The LM4562 is unity gain stable... driving complex loads with values as high as 100pF.

So both of these should be safe to try?

 

[seventy1]

I think it’s fair to state that if you are going to roll op-amps you need to have a decent 100 MHz ‘scope to determine if there are any oscillations in the output of the modified circuit.
Tom

So, I just have a PC based scope, says it's capable of sampling 1 Mbyte/channel, whatever that equals. Even if it was capable, I'd be like a monkey with a handgun.

 

[W9TR]

That PC scope is sure better than nothing. The spec you are interested in is the sampling rate, measured in mega samples/second. Divide by three (roughly) to get the useful bandwidth.

I sure don't want to discourage you from experimenting. Lots of people do it and don't blow stuff up.
The ones that do blow stuff up don't ever post their results.

It really does pay to check the results quantitatively. Maybe you have a friend with a scope?

Here's why. While unity gain stable op-amps should work, they don't always.
There are too many circuit variables to make an unqualified statement that you can replace one unity gain stable op-amp with another and not have the resulting circuit oscillate.

Take the two op amps above. The OPA 134 has 8 MHz gain bandwidth product (GBP) and the data sheet shows it is stable into over 10 nF loads. That's 10,000 pF.

The LM 4562 has a GBP of 45 MHz and is stable into 100 pF loads. It's really easy to pick up 100 pF. A few feet of audio cable and you are there.

So the OPA 134 is a much safer bet - 5 x lower GBP and 100x the capacitive drive capability.

Hope this helps,

Tom