All 4558's are not created equal! - Mu-Tron Phasor II proves it

[Mark Hammer]

I'm looking at the schematic for the pedal/circuit in question, and I see one light source for 6 LDRs.  That, in itself is not a problem.  What IS a confound here is the variation in response one can get from LDRs when they or their shared light source is jostled around.  I learned all about this when repairing a buddy's Morley wah 15 years ago.  In that case, the light source was a bulb not an LED, but the same principle applied: jiggle the light source a bit so that it points 5 degrees to the left or right or up/down, and the response of the LDR is altered.

So, when I hear someone swapped out the op-amps in such a pedal, I have to ask what the likelihood is that either one or more of the LDRs budged during the process, or the light source was nudged.

Again, I am not claiming that all of this op-amp talk is a waste of time, or that light-source-to-LDR alignment is necessarily the source of the perceived change.  All I am saying is that when multiple LDRs and a shared discrete light source enter the mix, and all of that is stuffed into a small package, one has to factor in (or out) physical alignment of components as well as any electronic changes before drawing strong inferences about the electronic mods.

[R.G.]

Good point, Mark. Also, the output current drive of opamps is a big variable. It's possible that with one light source for six LDRs, if one opamp drives it, then one version of opamp might have more or less ability to drive the current. Or more or less stability with a heavy load.

I guess I ought to look at the schemo. But seeing different part numbers on the opamps in question pretty much iced it for me. Different brand opamps which only share numbers in the part type are different acting? Uh, yeah!

[Paul Marossy]

So much pure - um - mysticism gets spread around about magic parts. The single unifying thread seems to be that "you can't get them like that any more, and these few I have left are very expensive".

That sort of thing ought to be left to the old NOS Ge transistors which are not being manufactured and/or not available anymore. The expense part anyway. The only thing magical about those old Ge transistors is finding ones with the right gain factor to be useful in a specific circuit to get "the sound". Or maybe the old BBD chips that are very scarce now. They are hard to come by now, so off they go to the highest bidder.

I wonder how much of these percieved differences go along with personality types. A guy who sees the glass as half-full will think that it sounds much improved with this replacement part whereas a guy who sees the glass as half-empty will say that he doesn't hear any difference at all with that part, or maybe that it sounds worse.  

[MoltenVoltage]

tthe output current drive of opamps is a big variable. It's possible that with one light source for six LDRs, if one opamp drives it, then one version of opamp might have more or less ability to drive the current. Or more or less stability with a heavy load.

I wouldn't be surprised if that's what's happening.  This is more or less what I was getting at in my last post, that both types of op amps can get these sounds, but the circuit needs to be properly tuned for the differences in current or gain.

Molten Voltage doesn't build distortion or phasers or any "classic" pedal designs, so we have nothing to "gain" (excuse the pun) by hyping 4558 chips.  Like everyone else, I had read the hype and don't really believe it, but then the different chips sounded distinctly different in my old Phasor II with all other things being equal.  The point of this thread was to try to figure out why, not to build some sort of "mojo" vapor buzz around these chips.

I appreciate all the thoughtful responses, and will set up a sinewave test when I get a bit of time. 

R.G. - you suggested that I should "set up the phaser in question with LFO sweep disconnected, input a sine wave and diddle the sine wave until I got as deep a null on the output as possible. I'd then change opamps and re-null, noting whether it nulled at a different frequency and/or different ultimate depth."

Questions:

1) What do you mean by "diddle the sine wave"?

Change the frequency?  Change the level?

2) What do you mean by "null"?

Make the combined input and output wave cancel?  Make the output flat?

3) What do you mean by "ultimate depth"?

4) What do you want pictures of?

Will simply comparing the outputs at the jacks get it done, or is there some better procedure?  Do you view the input and output separately on the scope or do you add the waves together and look at the difference?

Also, how will disconnecting the LFO make the phasing visible on the scope?

[StephenGiles]

Interesting - I remember talking to the EH UK electronics guy about finding 4558s in so many EH products. His answer was that all the main manufacturers made them, so no problem with supply, and they were 100% interchangeable from any manufacturer without affecting the sound or performance of any particular pedal!

[R.G.]

Molten Voltage doesn't build distortion or phasers or any "classic" pedal designs, so we have nothing to "gain" (excuse the pun) by hyping 4558 chips.  Like everyone else, I had read the hype and don't really believe it, but then the different chips sounded distinctly different in my old Phasor II with all other things being equal.  The point of this thread was to try to figure out why, not to build some sort of "mojo" vapor buzz around these chips.

Actually, I didn't think you were trying to hype them. Your post was a model of clarity - "hey, I found this difference; what gives?" I expected, and got, albeit very gently the "well, some 4558s are magic" response from someone else.

I appreciate all the thoughtful responses, and will set up a sinewave test when I get a bit of time. 

Thanks for taking the time. The real facts are powerful dispellers of vaporware, vaporposts, and so on.

1) What do you mean by "diddle the sine wave"?

Mess with amplitude and frequency, primarily frequency. The amplitude thing is just to keep it small enough that distortion is not changing results.

2) What do you mean by "null"?

Phasers work by introducing a progressive phase shift over a frequency range while not affecting the amplitude much if at all. They're allpass filters, passing all freqeuncies but changing phase. This phase shifted signal is then added to the original signal. At frequencies where the two reinforce (that is, mutual phase is zero or N times 360), they add up to double the size. At frequencies where the mutual phase is N*180 degrees, they cancel, producing as nearly zero output amplitude as the cancellation can support. At frequencies in between, the added results have a variable amplitude. This makes for a "comb filter" with a few notches and broad, flat peaks. Feedback can make the peaks sharper and higher, but it can't make the notches much deeper.

The "null" is the notches. Put in a sine wave of about 100mV peak for a test signal, verify that you get this at the output. Now engage the effect. You'll see the output sine wave change size, getting as much as 200mV big and dropping to zero, perhaps reversing phase after nulls. This is the visible effect of a swept comb filter.

The sweeping of the filter messes with any possible measurement of the peaks and nulls because they're always moving around. Good for ears, bad for o'scopes. So you have to disengage the sweep and give the sweeper/LDR elements a fixed non-sweep so it'll hold still to measure peaks and valleys. Once you get this rigged up, you have a phaser that doesn't sweep, so it'll display the comb filter effect at the output as you sweep the input frequency, the output getting bigger and smaller as you change input frequency.

3) What do you mean by "ultimate depth"?

How close to true cancellation you get when you adjust the sine wave for the lowest signal at the output. Generally -20db is good enough for good phasing. If the phase line has a gain that's not exactly unity, when you add the dry signal back in, the two may not cancel perfectly at any frequency because one is bigger than the other. The closest you can get to truly zero at a null is the "ultimate depth" of the null. The example of this is the two mixing resistors on the univibe that can be adjusted to get best phasing.

4) What do you want pictures of?

Dry/bypassed signal at test frequency; Output signal when bypassed; output signal when engaged with input frequency adjusted for biggest output; output signal with input signal frequency adjusted for smallest output; notation of the input signal level (100mV is fine), frequency of biggest input, frequency where signal drops to lowest level. For extra points, how many places in the audio spectrum you get nulls. This last will probably be half as many as you have phase stages.

Will simply comparing the outputs at the jacks get it done, or is there some better procedure?  Do you view the input and output separately on the scope or do you add the waves together and look at the difference?

Input signal and output signal. The effect will do the adding internally, and this is diagnostic of whether the signals are adding correctly. If and when you see ugliness on the output, more work can be put into looking at the internal signal levels that go to the output adder that makes peaks and nulls.

Also, how will disconnecting the LFO make the phasing visible on the scope?

The nice thing is that disconnecting the LFO - holding it fixed and unmoving, actually - does is let it quit wobbling the peaks and nulls around so you can measure them. The peaks and nulls are still there if the LFO signal is a DC level; they just aren't as noticeable to the ear. But fixed LFO level lets you find and measure the peaks and nulls on a scope.

The point of all this is then to compare what happens with the audibly-good opamps, then changing nothing else, put in the audibly-bad opamps and see what happened to the peaks and nulls. If the "bad" ones are, in fact, not doing right (NDR as us engineers say...  ) then they will produce not as good peaks and nulls, since that's what your ear hears. Finding out what they *are* doing lets us track down why.

[analogmike]

Well, here is the internal architecture of the RC4558 and the NJM4558.  The RC version is the one with values listed.  The NJM diagram shows none.  I have no idea about the values of the internal components, nor whether, even if the values were identical, the fabrication process resulted in any functional differences, or whether currently available datasheets for Fairchild semiconductors accurately represent Raytheon chips from 30 years ago.

But, unless you can spot something that I didn't, I can't see any differences in their design.

Not all transistors sound the same or have the same specs. The op amp is not perfect, it's made up of a bunch of imperfect transistors.

I don't say that old JRC chips sound better than the new ones on my website. Maybe the diode that analogguru pointed out does make a difference. But certainly there are differences in various op amps.

Right now I have my best tech working on a new JRC chip under a microscope to find and remove that diode 

[Mark Hammer]

Right now I have my best tech working on a new JRC chip under a microscope to find and remove that diode 

I was always under the impression that one had to be a very petite Indonesian lady with tiny hands to do that sort of work.

[MoltenVoltage]

you have to disengage the sweep and give the sweeper/LDR elements a fixed non-sweep so it'll hold still to measure peaks and valleys

I was ready to give it a go (including disconnecting the LFO) until you suggested that I send voltage into the LDR array.  From everything I have read, those parts are unique to Mu-Tron and I don't want to fry my pedal that I've had for almost 30 years.  Especially now that I finally got it working again!

Isn't there a way to just test the gain, frequency response, and other important parameters of the two chips using a breadboard?  I'd be willing to take out one of the RC4558 chips and test it against the JRC4558 in a controlled experiment to find out why they behave so differently.

[aron]

I'm with the others thinking the RC vs JRC means in this case, different manufacturer.

As I said before, for me, I like the TI RC4558.

[R.G.]

I was ready to give it a go (including disconnecting the LFO) until you suggested that I send voltage into the LDR array.  From everything I have read, those parts are unique to Mu-Tron and I don't want to fry my pedal that I've had for almost 30 years.  Especially now that I finally got it working again!

Not the LDR array - the driver to the light that illuminates the LDRs.

[R.G.]

Not all transistors sound the same or have the same specs.

True.

The op amp is not perfect, it's made up of a bunch of imperfect transistors.

True. In fact, this is the whole point of opamps - to make the passive components outside the amplifier control what happens, not the imperfect stuff inside. You pay for this with the "excess" gain that's used to hide the imperfections. This process has limits, many of which are well known and studied. This studying process has been continuous from the 1960s through today, uncovering ever-smaller differences.

I don't say that old JRC chips sound better than the new ones on my website.

Haven't checked.  Just to be clear, what DO you think about old JRC chips versus new ones?

But certainly there are differences in various op amps.

There absolutely are. Otherwise, there would only be one type number of opamp manufactured. The trick is to find them and figure out what the differences are.

Right now I have my best tech working on a new JRC chip under a microscope to find and remove that diode 

I want that microscope. 

[newperson]

To- MoltenVoltage

Is there any chance of you posting a few gut shots of your pedal? 
Thanks,

[MoltenVoltage]

I was ready to give it a go (including disconnecting the LFO) until you suggested that I send voltage into the LDR array.  From everything I have read, those parts are unique to Mu-Tron and I don't want to fry my pedal that I've had for almost 30 years.  Especially now that I finally got it working again!

Not the LDR array - the driver to the light that illuminates the LDRs.

It's all sealed together in one unit in a weird little canister.  I don't know what it looks like inside, but from the schematic it appears there is one LED and a number of photocells or other types of sensors.

I will post gutshots soon.

[brett]

Hi
+1 for RG's theory about no miracles in electronics.
So... what really is happening here?

Someone hand painted little orange dots on each one - I don't know whether that was Mu-Tron or the manufacturer, but it was pretty sloppy, so I am guessing Mu-Tron

By far the most likely reason for the dots is that there has been some sort of matching.  Therefore, we might assume that *even among the old 4558s* there was enough variation in one or more characteristics to affect the operation of the pedal.  So, it is not at all surprising that a new, unmatched 4558 did not work as well as the matched 4558.  The fact that there was LESS phasing with the new 4558 also indicates a degree of mismatch.  However, there is no evidence in any of this that an old chip (unmatched) would have made the pedal work properly.  Although the subject of the thread is true, the evidence isn't for old vs new, it is for one vs another variation. (There MAY be old vs new differences, but this wasn't a valid test for that, because it also tested unmatched vs matched, etc).

So - what is the objective of the match?  Here's my guess - as mentioned above, the gain-bandwidth product isn't too high, and is variable.  Maybe GBW or phase shift was being matched?  That could be tested (or the importance of it maybe deduced from the scheamatic).
cheers

[Paul Marossy]

By far the most likely reason for the dots is that there has been some sort of matching.  Therefore, we might assume that *even among the old 4558s* there was enough variation in one or more characteristics to affect the operation of the pedal. 

This sounds quite plausible to me.

[newperson]

I will go first,


http://sites.google.com/site/mutronphase/home


The black circle to the left of center should be the light enclosure.

[MoltenVoltage]

You beat me by about 15 minutes!  I can tell yours must be an earlier model, as the layout is somewhat different, as are a lot of the components.  Mine is serial #18480.


I replaced the diodes at top right and bottom left a long time ago on an earlier repair attempt.  Same with the 10uF tantalum cap at top right.  The two big power caps are also replacements.


At top right you will see a SmallBear JRC4558 - I lost one of the original RC4558 chips so I figured the LFO one should be the one to replace.  Everything else it original as far as I can tell.  I bought is used in '83.


That can holds the LDR array


I assume the lower numbers are batch numbers - two different batches in this pedal



Top of the line in '79!




The white wire was added to repair a trace and the black one to make a better ground connection.

[newperson]

Thanks for the pictures.  I added a picture of the backside to the one I posted.  Looks like mine has two different chips, but most of them seem to be from 1976 if that is the date code.  Very different layout.

[Scruffie]

Regards the coloured dots, I have read of some people marking different chips by colour on arrival just to speed the process of assembly up so they don't have to read each one... but that's just speculation.