Can I increase capacitance on these TONE BOARD caps?

[nashvillebill]

Noise when the amp is idle, they (to my ears) have a high noise floor, so more "white noise" in the background. The NE5532 (again to my ears) is much quieter at idle with a higher SNR.

Now this is all somewhat relative, so a single replacement of a 4558 with a NE5532 may not suddenly bring sonic bliss and eliminate all background hiss (at idle). Resistors have noise too. Also, the inherent design of the circuit comes into play....that may in fact be the biggest contributor!

A final note, again I am a mechanical engineer not a EE so I'm not a guru on op-amps, but a note of caution on swapping op-amps. It's tempting to go for some of the expensive new op-amps with gobs of bandwidth and specs out the wazoo. However, the circuit design may not be stable with these op-amps. The circuit design may allow oscillation at high frequencies (I'm talking hundreds of kHz) which wasn't a problem with the older op-amps since they didn't operate into that range. Hence no high-frequency oscillation. But a newer op-amp which DOES amplify up into the MHz range will go into oscillation. Big problem. So specs alone are not the end-all metric to look for here.

The NE5532 (sorry I keep gong back to it) generally plays nice in circuits. Fast enough to give great audio results but not too fast to provide oscillation. In general it can be subbed for most op-amps without worry of circuit issues, but the other op-amps out there now are not always as forgiving.

 

[BinaryMike]

It appears that the PCB was designed to accept dual op-amps with the standard DIP-8 footprint. Please confirm, and try to make sure there's room for a socket. Here's a datasheet on TI's version of the 5532: http://www.ti.com/lit/gpn/ne5532

I don't see any good reason not to bump the power supply rails up to +/-18V, unless you want to use op-amps with a lower limit. There are one or two high-end ones that specify 36V as absolute maximum total supply voltage, but most of them are 40~44V like the 5532. Are you up for some serious hacking?

 

[roger2]

......Are you up for some serious hacking?

Good question. I don't know. First, I would need to identify and verify a problem and, secondly, a reasonably certain solution.

I have zero experience with opamps. On the other hand, I have been engaged in a pretty thorough restore of this amp, far beyond just recapping. If these opamps are adding to the noise floor, and if I was convinced that noise floor could be noticeably improved, then yes I might want to do something about it.

There is some hiss in this amp. Not horrible, but a little more than average compared to several other amps/receivers I have of similar vintage, power, and general quality level. I will investigate the physical parameters. Based on pics I have seen, the pin spacing on the PCB looks too small to fit a standard DIP-8.

Would it be necessary to bump supply rails up to +/-18V? I can envision lots of changes required to make that happen. I have no scope to verify my work.

 

[roger2]

Physical dimensions of the pins on the NE5532 do not match the pins on my 4558 TO-99's

From the datasheet you linked (thanks) in inches:

Distance from pin 1 to pin 4 of the NE5532 on centers is .3"

Distance from pin 4 to pin 5 of the NE5532 is .43"

Corresponding measurements on my 4558's are .3125" and .1875" (approximate)

 

[EchoWars]

Too bad the TO-5-8 package is such a rarity nowadays.

 

[nashvillebill]

You can get standard DIP-8 sockets at Radio Shack cheap enough to examine, I even think they carry NE5532 op-amps. I think the pin 1 to pin 4 spacing wouldn't be a problem, but the distance the other way...a socket *might* have long enough leads to bend them inward underneath the socket to get them to fit.

Or...on the side closest to us as we look at your picture, can you drill a row of 4 more holes spaced out enough for the socket to mount, then do something with the traces underneath the board? Or...mount the socket on a small "flyout" board with short (keep them as short as possible) wire leads going back to the existing 8 holes.

 

[roger2]

Mouser lists several Nat. Semiconductor TO-99's starting from $8.33 and going up. I think I recognize some of the mfgr model numbers, maybe these also come in the standard DIP package.

The pins appear to be arranged in a circular fashion as opposed to two rows like mine, but that is probably not an issue. (I think one of the pics at that link is a mistake)

I am not concerned about the money if there is something to be gained. But would it be worth some effort to investigate these available TO-99's? I mean, would it quiet some background noise in my amp?

What are the key parameters to look at?

 

[EchoWars]

Mouser lists several Nat. Semiconductor TO-99's starting from $8.33 and going up. I think I recognize some of the mfgr model numbers, maybe these also come in the standard DIP package.

Most of those are ancient design or single-channel devices, and others are simply not suitable for your purposes. BUT...I forgot about the National Semi LM4562 and the LME49720. Expensive in the metal can package, but you can drop them right in as soon as you ID the leads and where they go.

The pins appear to be arranged in a circular fashion as opposed to two rows like mine, but that is probably not an issue. (I think one of the pics at that link is a mistake)

Come again? The original leads on the stock 4558 are in a circle too...they've just been neatly divided to fit into the PC board holes.

I am not concerned about the money if there is something to be gained. But would it be worth some effort to investigate these available TO-99's? I mean, would it quiet some background noise in my amp?

What are the key parameters to look at?

1st question: It might be. There are a large number of 4558 opamps out there...some not bad, and others that sound gritty and/or are noisy. 2nd question: You don't have a lot to choose from. The two I listed above are pretty much the only viable alternatives unless you want to look at SMD devices using some sort of adapter. Of those two, I really like the LME49720, and have used the DIP version in numerous pieces of gear. And, it's a modern device.

 

[BinaryMike]

The key parameters for noise are Vn and In, but Vn dominates in this case. The 5532 is actually very quiet at (typically) 5nV per root Hz in the midrange. As Doug Self likes to point out, this op-amp is a huge bargain among its competitors.

It's very likely that your PCB was designed to accommodate DIP-8 packages, and the TO-99 leads were simply formed to match that footprint. This was common practice in the industry for a while, because it gave the buyer an extra card to play when compatible ICs got scarce. Spacing between the two rows of holes will be 0.300" and pin pitch will be 0.100" if that's true. The 0.430" dimension on the TI drawing has nothing to do with PCB design. Those pins are always formed inward to 0.300" spacing prior to PCB installation.

 

[roger2]

@ nashvillebill
Thanks for bringing the 4558's to my attention. I have wondered about those since I did searches a long while back on my amp (KA-4006) and found some negative references.

I can say with confidence that drilling into the PCB is beyond what I am capable of or would want to attempt.
-----------------------------------------------------
@ BinaryMike
In a similar vein, your suggestion of increasing rail voltage makes me a bit queasy too...although I might possibly consider that option.
-----------------------------------------------------
@ everyone

Having found that TO-99 op amps are still available (linked in my last post) I would prefer to explore that option first. I would need guidance on how to choose from among those op amps available, which parameters to match to my 4558's and which parameters are OK to surpass vs the 4558's (Noting nashvillebill's previous warnings)

EDIT: was writing while EW and BinaryMike were posting...

 

[roger2]

...BUT...I forgot about the National Semi LM4562 and the LME49720. Expensive in the metal can package, but you can drop them right in as soon as you ID the leads and where they go....

EchoWars...I am guessing that you're the kind of guy who doesn't like repeating himself...and I understand the phrase "drop them right in"

But would you please verify that the LME49720 will run correctly on +/- 9v and that no other components on that board would need to be altered?

No worries about oscillation or adding ceramics to prevent it?

EDIT: I looked at a datasheet for the 4558 that says slew rate is .5v/uS. The LME49720 is 20v/uS ? Does that sound right, 40x faster?

 

[EchoWars]

The key parameters for noise are Vn and In, but Vn dominates in this case. The 5532 is actually very quiet at (typically) 5nV per root Hz in the midrange. As Doug Self likes to point out, this op-amp is a huge bargain among its competitors.

It is, but all the 5532's and 5534's are repo's of Signetics original design, and IMHO, none are as good. I'd go with a modern design rather than an aging copy of the Signetics chip.

It's very likely that your PCB was designed to accommodate DIP-8 packages, and the TO-99 leads were simply formed to match that footprint. This was common practice in the industry for a while, because it gave the buyer an extra card to play when compatible ICs got scarce. Spacing between the two rows of holes will be 0.300" and pin pitch will be 0.100" if that's true. The 0.430" dimension on the TI drawing has nothing to do with PCB design. Those pins are always formed inward to 0.300" spacing prior to PCB installation.

Some of the early Kenwood's had a trap...the pin layout on the PC board was mirrored for the opamps. I've run into this a couple of times where the original device contained TO-99's (or TO-5-8's). So to mount a DIP opamp, I had to bend the leads backwards, so that the opamp ID side was flat against the PC board. Wierd, but I've seen it a couple of times. It has been a while, and I can't remember what the gear was specifically, but back then I couldn't find another TO-99 that I liked. Now, I'd probably use the LME49720 in the metal package.

EchoWars...I am guessing that you're the kind of guy who doesn't like repeating himself...and I understand the phrase "drop them right in" :yes:

But would you please verify that the LME49720 will run correctly on +/- 9v and that no other components on that board would need to be altered?

It's only 9V at the current draw that the original device presents...Ri27 and Ri28 see to that. I'd replace these 680 ohm resistors with a jumper wire (or a cute little zero ohm resistor), and run the device directly from the +/-14V supplies.

No worries about oscillation or adding ceramics to prevent it?

Oscillation is always a possibility if the supply is not properly decoupled. A high-quality low-impedance cap should be use at Ci15 and Ci16, and bypassing these with small film caps (like a 0.22µf stacked film) would be good. Better to go from pin 4 and pin 8 on the PC board as close to the opamp as possible with these small film caps to ground. FYI, I have not had stability issues using the 49720, but my experience is no guarantee of what yours might be, especially considering the very lengthy feedback loop for the tone. My instinct is that you'll have no problems, but by all means the output of the opamp should be scoped. Lastly, if the resulting offset of the replacement opamp is low enough (5mV or less, tone on and off), you might consider replacing Ci5 and Ci6 with another wire or zero ohm resistor.

EDIT: I looked at a datasheet for the 4558 that says slew rate is .5v/uS. The LME49720 is 20v/uS ? Does that sound right, 40x faster?

Yep. 20V/µS is pretty pedestrian nowadays...the slew of many opamps is measured in hundreds of volts/µS. The stock device's 0.5V/µS is damned pokey as things go.

 

[roger2]

I'd replace these 680 ohm resistors with a jumper wire (or a cute little zero ohm resistor), and run the device directly from the +/-14V supplies....

The 14v zeners that supply this board are going to be replaced. Could easily go to a higher voltage as BinaryMike suggested...

...Oscillation is always a possibility if the supply is not properly decoupled. A high-quality low-impedance cap should be use at Ci15 and Ci16, and bypassing these with small film caps (like a 0.22µf stacked film) would be good....

Which leads back to the OP where I was asking about increasing the capacitance of these from 100 to 220. I also mentioned that I will use Panasonic FM. Is FM what you would suggest as the best cap for this circuit? (You're not gonna say Silmic II are you? ara

And yes I could do the bypass. ECQ-V?

...considering the very lengthy feedback loop for the tone...

Is it long because it runs through the switch?

...My instinct is that you'll have no problems, but by all means the output of the opamp should be scoped....

HERE is where you may have lost me. You mean that it should be checked on an oscilloscope? I don't have one and don't know how to use one

...Lastly, if the resulting offset of the replacement opamp is low enough (5mV or less, tone on and off), you might consider replacing Ci5 and Ci6 with another wire or zero ohm resistor....

Getting rid of Ci5 and Ci6 would be a bonus as I believe they are the weakest link in my signal path. I would require instruction on how/where to measure the offset of the opamp.


Recapping (no pun intended) everything, the scope may be the only deal breaker. Unless "scope" means something other than what I think it does...

 

[EchoWars]

The 14v zeners that supply this board are going to be replaced. Could easily go to a higher voltage as BinaryMike suggested...

The LME48720 is only rated for +/-17V, so I guess you could go to 15V zeners if you want.

Which leads back to the OP where I was asking about increasing the capacitance of these from 100 to 220. I also mentioned that I will use Panasonic FM. Is FM what you would suggest as the best cap for this circuit? (You're not gonna say Silmic II are you? ara

Lol! A 100µf or 220µf FM will be fine. I'd also go to a 25V cap, both here and for Ck3 and Ck4 on the power supply board, where I'd put in a 680µf FM cap.

And yes I could do the bypass. ECQ-V?

Sure.

Is it long because it runs through the switch?

Yep...clear off the board and back again.

HERE is where you may have lost me. You mean that it should be checked on an oscilloscope? I don't have one and don't know how to use one

I don't know what to tell you. Any AK'er near you with an oscilloscope?

Getting rid of Ci5 and Ci6 would be a bonus as I believe they are the weakest link in my signal path. I would require instruction on how/where to measure the offset of the opamp.

Just like measuring offset anywhere else...with a decent meter that reads accurately in the mV range.

 

[ken kantor]

I'll try one last time, then leave you to enjoy:

The designers of this amp were not total idiots. They understood how circuits work. They would >not< add to the parts count and cost of the product unnecessarily, and further run the opamp at +/-9V, for no reason! It is unheard of to run a 4558 via 680 Ohms unless you >want< to have the effects that this will create. If everyone here thinks they are more competent or clever than the designers, the comments in this thread would suggest otherwise.

If you want to add bypass caps, fine. If you want to put in a more modern opamp, fine. But, when you start changing values or topologies, be careful to understand how the circuit really works, and why it might have been designed that way.

-k

 

[EchoWars]

There will be zero difference in the way the circuit works whether it is running on +/-9V supplies, or +/-18V supplies. I suspect that, due to the simplicity of the zener regulated supply, and the middling PSRR of the stock opamp (as low as 70-something db), that the resistor was added in order to filter any residual ripple. Just a guess, but I suspect I'm right.

No, they weren't idiots. Matter of fact, Kenwood liked to 'borrow' the best ideas that they saw in other gear, while still pushing the cutting edge with their top end gear. But they, like everyone else, had to work under the constraints of the parts available at the time, and with the physical constraints of fitting what were fairly large caps into a space that was workable. Too bad they didn't have the high-temp, low-impedance caps that we do, which are much smaller than the originals...and they didn't have a National Semiconductor LME49720 with a PSRR of 120db either, which should make the use of those resistors superfluous.

 

[Oilmaster]

Oh well, when your budget permits, you could also opt for something more discrete... bought here

This thread is drifting (nicely) away from your simple cap question :scratch2:

Sorry for the massive pic earlier

 

[BinaryMike]

It is, but all the 5532's and 5534's are repo's of Signetics original design, and IMHO, none are as good. I'd go with a modern design rather than an aging copy of the Signetics chip.

Doug Self and friends tested an assortment of 5532 clones for the recent power amp project in Elektor magazine, and reported that TI's version is the cream of the crop. Their goals were not quite the same as our current context, however. I believe they paralleled 64 op-amps (32 chips) to drive an 8R load directly, so price per unit was a key feature.

 

[ken kantor]

Yeah, National sent me a handful of those when they came out a few years ago. They are excellent amps, possibly a response to TI’s INA series. I doubt either are going to get a ton of design wins at those prices, though! I was an early advocate of the 4558 when it appeared, and have been designing with it 3+ decades. Probably you have, too. I think it was the first IC opamp to have broad usefulness in audio. (I liked 318‘s in high performance designs before that, but they were hell to stabilize.) These days, I’m using mostly JRC4560‘s for routine audio apps in mass production.

I can see both sides of the Kenwood design, and they aren’t mutually exclusive. The filter shown will be down about 22 dB at 60Hz; 28 dB at 120 Hz. Moderately useful for ripple filtering. In the time domain, things are complicated by the non-linearity of the supply current. As a first approximation, RC is about 55ms, and 3*RC is about 165mS. These sound to me like just about right for mitigating turn-on and turn-off transients.

Personally, I have never seen an RC ripple filter on an individual opamp line stage. Have you? For example, 7815-based rail yields a >worst case< ripple and noise that is more than 60dB below a volt. Run this into a PSRR of 70dB and we are down into the thermal noise anyway. If the supply needs to be really clean, say for a phono stage, the standard would be to use a beta-multiplied cap, rather than large resistances in series with the supply. This assumes there are more opamps in the design, which I am not familiar with. If there is only one opamp, who knows what is being done with it.

Again, I don’t think the functions are mutually exclusive, but I wouldn’t go so far as to say changing the the cap is guaranteed to yield no change in operation. One man’s opinion. I’m kind of hoping the OP tries it, to see what actually happens!

-k

 

[BinaryMike]

Odd that you should mention the 318, Ken. I've used it in audio circuits as well, because I was very familiar with it from my work in video. I've also used video buffer chips like the LM6321 to upgrade older preamps (as main output and tape output buffers).