Implementing mods on a CE-3

[miketbass]

Alright, so I dug out my thrashed CE-3 today and was playing with it a little. It certainly doesn't sound as lush as a CE-2 or a CS-9, but it's not bad. It has a very "80s" sound to it but sounds duller than some of the other choruses I've tried. Anyways, I thought it might be interesting to try out some of the mods for a CE-2 on this guy to get greater tonal flexibility. I am certainly interested in the "chorus intensity cap mod", "effect level mod" and  the "wet bass mod" found on the  Tonepad Ce-2 layout. The chorus intensity cap was pretty easy to locate as it's a universal location on every chorus and flanger I've looked at. On the CE-3 factory schem it appears to be C-25. However, the appropriate cap for the wet bass cut has me scratching my head a bit.....granted I'm by no means a master of reading schematics. However I am wondering if the stereo out on the CE-3, through what looks like 2 mixing stages ( IC1 and IC2 near the outputs) means that I would need to switch out two caps. Or maybe not - I'm kind of off the cuff at this point. And of course the appropriate resistor for the effect level mod escapes me too . ANY help would be greatly appreciated, I will post the following schems with credit and thanks to the original creators/posters.....

Tonepad CE-2 Layout and schematic
http://tonepad.com/getFileInfo.asp?id=101

Boss CE-3 factory schematic
http://www.dirk-hendrik.com/BossCE-3.zip

Regards again to Tonepad, Joe Kramer, and Dirk Hendrik for the material to make this possible.

[Geoff Hunter]

Wet Bass Mod Cap
CE-2 = C14 = 0.033uF
CE-3 = C21 = 0.068uF

Note that the stock CE-3 allows more bass through the modulated section, which may explain why the CE-3 sounds more "duller".
I have not tried modding this as I use the CE-3 more in "Dry only" mode, and don't want my sound thinned out (of course it may not make much difference at all).

Effect Level Mod

CE-2 = R22 (+ 250KC Pot - as described in the Tonepad file)
CE-3 = R36 (for Dry+Wet), and R31 (for Dry-Wet).  In order to put level on both channels you would have to:

1. Lift the leg of C21 that meets the junction of R36/R35/R31
2. Put the pot between the leg of C21 and the pad.

[miketbass]

Thank you so much Geoff! I'm interested in what the CE-3 would sound like with that smaller wet-bass cap. Looks like it's time to place a small order from Mouser.....
Thank you again, I will post back with my results.
-Mike

[Mark Hammer]

Geoff's got most of it right.  The identified cap does feed both outputs, and reducing its value will take out some of the bass wobble that makes choruses so annoying sometimes.  You may want to reduce that by a great deal.  Perhaps even down to 1000pf or so, given what I suspect the existing rolloff is.  Also, note that C16 is a co-conspirator in the low end as well.  Dropping that puppy down to 4700pf would be a good idea too.

I would not suggest tampering with R31.  It is part of the signal-inverting stage built around IC2 that creates the sum-vs-difference split between the two pedal outputs.  When R31=R29, that stage is unity-gain.  Change either of those resistor values and the gain changes.  The resistor which IS functionally equivalent to R36 is R28.  You will note that it is a) the same value as R36 (one of the reasons why Geoff made the simple error of thinking R31 was the correct part), and b) situated immediately prior to the switching FET, just like R36.

Obviously the complexity of changing the value of two separate resistances was enough of a headache for Boss that they decided to forego it.  Sticking a dual-ganged pot in there is likely to be a royal PITA, so I would suggest maybe getting a 3-position DPDT instead.  How would you use it?  Simple.  Replace R36 and R28 with a much larger value, like 180k-220k.  This would be your "subtle" mode.  The "left" side of those resistors would go to the common (middle) on the switch.  The outside positions on the switch would then add other values in parallel with that base resistance.  So, if the base was 220k, then you could add in 180k to make 99k (medium level) and 62k in the other position to get 48k parallel resistance (close enough, eh?) and "normal" effect level.  Be sure to use 1/8W resistors so it will all fit in there.

I realize not everyone takes the stance I do, but I generally find that 3 choices satisfies my thirst for variation.  It may work for you too.

[Geoff Hunter]

Ah of course! Stupid me   

Thanks Mark for correcting my pre-coffee error

[Mark Hammer]

As I understand it, it was a pre-coffee error that resulted in the incursion into Iraq, the decision to decline signing the Beatles as another "guitar group", and the decision for Texas Instruments to launch a personal computer several years after Apple and Tandy.  As pre-coffee errors go, yours was negligible and you were easily suckered in by the resistor value.  Entrapment + no coffee = thoroughly excusable behaviour in my books.

[Joe Kramer]

Hey MikeT,

Plenty of people agree with you that the CE-3 sounds "thin" or "metallic," especially compared to the CE-2.  I've got both and to my ear the CE-3 does lack a certain "body" compared to the CE-2.  But what's interesting is, a reading of the actual circuit seems to suggest that there would be more low end in the CE-3, as Geoff and Mark point out. 

My hunch on a possible reason for the difference in tone has to do with the 2SK30 FET at the input of the CE-3 versus a bipolar in the CE-2.  Please take this WAGOS since it goes against conventional wisdom, but  FETs for guitar often sound thin and scratchy to me (maybe with the exception of the higher gain types like  NTE458/2SK117/J201).  Same goes for input impedance--I don't find higher always sounds better.  That said, the higher 1meg input impedance of the CE-3 compared to the 470K impedance of the CE-2 might also add to the impression of thinness.

I haven't tried this yet, but subbing something like a 2N5088/89 and a 470K (R4) into the CE-3 might make a helpful difference.  The swap would be easy, since the SK30 is D-G-S, and the 5088/89 is E-B-C and would drop right into the PCB without a lot of lead-bending.  Maybe the extra muscle of the bipolar and the lower input impedance would get the CE-3 sounding a little "thicker."

Regards,
Joe

PS: Don't forget the importance of that trimpot in the CE-3.  It sets the bias for several stages of the circuit simultaneously.  Run a line-level (loud) sine wave test tone into the CE-3 and adjust that trimpot for the least amount of distortion, which can be heard as an octave above the fundamental test tone (second harmonic distortion.)   

[miketbass]

Thank you Mark and Joe, I am eager to try out all of these suggestions. I am looking at rotary switches for the effect level at Smallbear, but I am a little worried about fitting it all in there. I figured I could probably shoehorn in a 16mm alpha pot right next to the input jack, but some of these rotary switches look a little big. Any suggestions? My idea is to mount a rotary on the side and then use two toggle switches for the other two mods mounted right underneath the depth and mode pots. As to transistor idea that Joe recommended, this is an interesting idea that I will try out ( I do have some 2n5088/89 and NTE458 kicking around). I don't know how much of a difference lowering the input impedance will make but it's worth a shot too. For a quick one that I did that worked good, I socketed the two 4558 opamp chips and started plugging in what I had lying around. I tried JRC4558, NE5532, TL072, and a BB OPA2134 for a nice variety of ICs. The NE5532 made things a bit darker and rounder, not the sound I was after (and waaaay more current draw). I thought the TL072 and BB chip sounded the best, I guess the FET input on these helped to add a little more high end shimmer. Well, I'm pretty settled on giving this all a whirl as soon as I find a proper rotary switch. Any help would be appreciated in this area. Oh, and for instant tone I put in a superbright green LED - I swear it sounds better now
-Mike

[Mark Hammer]

Personally, I find discussion about the tone on these things a little premature.  There is so much filtering, pre-emphasis and de-emphasis, going on in there, all with 5% resistors and Lord knows what tolerance in the caps, that until one finds out what the nature of the tone shaping (intended and unintended) is on that particular unit, everything else is a shot in the dark.  For instance, just what IS the nature of the lowpass filtering post-BBD?  Do any of us know?  Is there some sort of Q-factor in there that could be easily altered to tame unwanted peaks?

[stm]

I did a thorough comparison between CE-2 (Tonepad schem) and CE-3 (Boss schem), and found the following:


DIFFERENCES:

1) Input buffer: CE-2 with transistor and 430k equivalent impedance; CE-3 with JFET and 1M equivalent impedance.  Subtle effect, expect CE-2 slightly warmer.
2) BBD stage: CE-2 with MN3007/MN3101; CE-3 with MN3207/MN3102.  The former is reported to sound better, has higher dynamic range and less total noise.  Notice there is a diode in series with one of the supplies of the MN3102 IC in the CE-3, which is not present in the CE-2.


SIMILARITIES:

1) Pre and de-emphasis networks: identical values.
2) Pre and post BBD low-pass filtering: identical values.
3) LFO and interface to BBD oscillator IC: identical values.
4) Low-frequency corner of capacitor before output summing stage: CE-2 is 33n and CE-3 is 68n, however the former feeds a 47k resistance to a virtual ground (negative input of opamp) while the latter feeds two 47k in parallel each to virtual ground, thus having the same lower cutoff frequency, which is 100 Hz.
5) Output components: identical values.


SUMMARY:

Actual differences between Tonepad's CE-2 and Boss CE-3 that can objectively affect tone are:

1) Different input buffer stage
2) Different BBD IC's
3) Finally, let's not forget that Tonepad's CE-2 is true bypass and has no series JFETs in the signal path to do the switching!

[Mark Hammer]

4) Low-frequency corner of capacitor before output summing stage: CE-2 is 33n and CE-3 is 68n, however the former feeds a 47k resistance to a virtual ground (negative input of opamp) while the latter feeds two 47k in parallel each to virtual ground, thus having the same lower cutoff frequency, which is 100 Hz.

Thanks for that tidbit, and systematic review.  I had been calculating the lower cutoff frequency using the 1M fixed resistor to Vref, not knowing that the 47k resistor would have been the one to use.  Naturally, that changes my recommendation for what value to replace the .068uf cap with.  Clearly, a .001uf cap will be too small.  Start with .015uf and work from there.

Of course, the other thing that comes from Sebastian's review is that if there is nothing particularly different in the designed filter specs, buffers, etc., then maybe some of the perceived differences stem from unit-to-unit component tolerances, rather than the design itself.  I honestly can't see how the MN3007 vs 3207 difference would result in any substantive change in tone under normal signal-level conditions.

[joelap]

Wow... thanks STM for that breakdown of differences.  I had always thought the CE-3 was much more similar to the CE-2 than that.  Could the choice of BBD's in the CE-3 be the reason for the hiss the CE-3 inherently has?  I removed about two-thirds of the electrolytic caps for films which was supposed to make the pedal quieter.  It helped, but its still hissy, especially with a distortion pedal behind it.

[Joe Kramer]

Nice rundown/comparison STM, thanks.  The 3007 vs 3207 is sort of a wild card.  The 3007 specs better, but as Mark says, the practical difference is another matter.  I still lean toward that FET in the CE-3 being a big factor in the tone.  If this component were swapped, then the only substantive difference between the two circuits (notwithstanding a generous tolerance margin) would be the BBDs.  If I get some time today, I'll try the mod and get back here with a field report. 

Also, I repeat the importance of that bias trimpot: if it's off its optimal setting, it can affect the balance of dry/wet signal and change the overall tone.

Joe

[joelap]

Joe, I'll see if I have an NPN around to try the swap with.  The bummer is I'm out of desoldering braid and don't have a solder pump, so I might risk runing the board by trying to pull the components through the hot solder.  I've popped pads off of two boss pedals that way when I was new at this.  Not that I'm much better now, but you know...

I do love my CE-3, but I've never had it next to a CE-2.  I can get 2 or 3 CE-3's for the price of one CE-2 though (picked up the CE-3 I'm using now for 20 bucks shipped to my door cause the LED and switching were funky... fixed it in 10 minutes!)  There's something cool about those CE-2's though.

[stm]

To Mark Hammer:  since both MN3007 and 3207 are fed from aprox. 8V, apparently there is no actual difference in performance to be had, at least theoretically.  I checked both datasheets and all relevant parameters (THD, noise, dynamic range, etc) are basically identical at 8V supply.  The important difference appears if you feed the MN3007 with 15V, since it almost doubles input dynamic range, which can be better in order to tolerate modern hot humbucker pickups.

To Joe LaPorta:  if you feel handy, I guess you could build a CE-2 clone yourself and enjoy the true bypass benefit and extra mods described by tonepad. In this case it might be cheaper than buying a 2nd hand CE-2.

[Joe Kramer]

All right, here's a CE-3 modification round-up:

Q1 2SK30 replaced with MSPA18 (didn't have a 5089 at hand)
R4 1M replaced with 470K
R2 10K replaced with 1K

Before the mods, I recorded an arpeggio figure on guitar into two tracks of CoolEditPro, with a buffer/splitter feeding one track through a CE-2, the other track through a CE-3.  On playback, I set up volume envelopes and looped the guitar figure, so that I could listen back to alternating tracks with automated muting.  There was a considerable difference between the two choruses, with the CE-2 having more midrange body and a less tinny high end.  The CE-3 lacked low mids and sounded somewhat scratchy on top.

The first mods on the CE-3 were Q1 and R4.  I recorded a second time and compared the CE-3 to the CE-2 as well as to the first recording of the CE-3.  There was some improvement in the body and low-midrange of the sound, but the CE-2 still sounded thicker and overall better.

The second mod was sort of an afterthought, the changing of R2 in the CE-3 from 10K to 1K, which is the same as the CE-2.   After another recording and some freehand playing and A/B-ing, I found that this mod made a substantial difference.  Now the CE-3 has almost the same overall tone as the CE-2 in the low midrange, though the CE-2 is still the slightest bit darker on top.  For the most part though, now it's much harder to tell them apart with these three mods in place.  The only other difference in the input sections is that the input resistor (R2 on the CE-3, R1 on the CE-2) precedes the .047 cap on the CE-3, and comes after it in the CE-2.  They would be fairly easy to swap around, but the question is: would the order of a series cap/resistor makes much difference here?     

As far as noise, the two units are nearly identical.  The CE-3 is not noisier--at least mine isn't.  I think it's pretty safe to say that the differences between 3007/3207 at the nine-volt level are practically nill.

I'm happy with these mods--they're keepers.  In a stand-alone playing situation, without doing an actual A/B comparison, I doubt I could tell them apart.

Joe

PS: Hey joelap:  risk not a broken trace.  Get thee to Ye Olde Radio Shack, and buyeth a desoldering bulb with a teflon tip, approx.  $3.  Held me in good stead for many a year.

[Mark Hammer]

To Mark Hammer:  since both MN3007 and 3207 are fed from aprox. 8V, apparently there is no actual difference in performance to be had, at least theoretically.  I checked both datasheets and all relevant parameters (THD, noise, dynamic range, etc) are basically identical at 8V supply.  The important difference appears if you feed the MN3007 with 15V, since it almost doubles input dynamic range, which can be better in order to tolerate modern hot humbucker pickups.

All true.
Of course, even at 15v supply, the dynamic range isn't exactly fabulous.  Even with easily measurable differences in their performance and handling of transient peaks, the two chips should not present any audible differences in "tone" when fed manageable signal levels (which is what they get most of the time).

Though the MN3207 can be run at 8v, what you tend to see in designs is that they get run at 5v (from a regulator).  My sense is that the very basis or intent of this low voltage capability was so that the biasing set for a V+ of 5v would be valid over the lifespan of a 9v battery even when it sank to 7v (the point where the 78L05 regulator would choke and say "I don't think so").  The problem with the MN3007 was that  you couldn't use a 3-pin 9v regulator with a 9v battery, and it wouldn't work at the 5v supply range.  That meant you could either power it from 12v-15v externally (common in desktop and rackmount, but not in floormount pedals), and assure that the tweaked bias voltage was bang on, but if you tried to power it from a 9v battery the biasing would start to be off once the battery starts to drop below maybe 8.5v.  I doubt that the MN3207 draws that much less current, but it allows you to use a battery for a longer period of time.

It's funny how battery-power considerations shape the design of pedals. Once the design team answers the question "Do we want people to be able to run this off a battery?" with either a yes or a no, that sets in motion a whole stack of other things.  A few years ago, I corresponded with the programmer/developer working on the Line 6 Echo Park.  He mentioned that they had considered setting it up to be a "true" stereo pedal (i.e., two completely independent parallel delay lines, input to output), but the number of clock cycles required to provide decent sound from a single DSP while handling two independent channels made battery operation impossible.  Consequently the choice was made to pool the two inputs to mono, process, and "redistribute in stereo".  In truth, a 9v alkaline doesn't last all THAT long in an EP, but the difference between being able to last 2hrs and only being able to last 20 minutes means a great deal in terms of both demo-ing and sales potential.  Many of the most cherished Boss pedals are from the "hammertone" cast aluminum 1 series (PH-1, CE-1, BF-1, etc.).  So why didn't they just keep making them in smaller packages?  Simple - few of them would run off batteries, so the design changed.

[stm]

...The second mod was sort of an afterthought, the changing of R2 in the CE-3 from 10K to 1K, which is the same as the CE-2.   After another recording and some freehand playing and A/B-ing, I found that this mod made a substantial difference.  Now the CE-3 has almost the same overall tone as the CE-2 in the low midrange, though the CE-2 is still the slightest bit darker on top.  For the most part though, now it's much harder to tell them apart with these three mods in place.  The only other difference in the input sections is that the input resistor (R2 on the CE-3, R1 on the CE-2) precedes the .047 cap on the CE-3, and comes after it in the CE-2.  They would be fairly easy to swap around, but the question is: would the order of a series cap/resistor makes much difference here?

Funny how I missed the 1k v/s 10k series input resistor when I did the comparison.  It seemed so far from producing a change that I didn't even check their values.

The big question now is: how is it possible that changing that (relatively small) series resistance that goes into a (relatively high) input stage can make such an important difference in the low-midrange?  I see no logic in it.

To Mark H.: Good points about battery operation trade-offs in guitar pedals.  It's a funny synchronicity that during the past few days I've been thinking precisely in departing from the single 9V supply into +/-9V or +/-15V in favor of audio quality and performance.

One thing is having to deal with a single supply instead of dual supplies.  This usually increases the need for decoupling capacitors (sometimes electros) and requires a Vcc/2 voltage divider, again bypassed by an electro.  If you analize this type of circuit carefully you'll see there are many places where audio is forced to travel through capacitors, sometimes coloring the sound and/or affecting low frequency response.  In fact, the audio will travel through the Vcc/2 decoupling capacitor as well during its return to GND.

On the other hand, having a single supply of a relatively low voltage such as 9V is even worse.  In this case most opamps are severely limited in terms of clipping-free dinamic range.  You shouldn't expect to get closer to each supply rail than 1.5V in the best case, so in a 9V battery system your output dynamic range is only 6V (9-1.5-1.5).  There is very little room for clean amplification.  Moreover, when battery voltage falls down to 7.5V due to aging your range squashes down to 4.5V.

That is what you trade off when you decide to go from mains supply (typical in rack units) to battery power.

Just some thoughts to add to what Mark H. just mentioned.

[Dr.maligno]

dont froget R18, this divisor give to the ce3 pedal less gain that the ce2, changing this resistor to 1M, will jump to life a ce3.
cheers

[petejt]

Hi there. I've been reading these posts and getting a lot of good information. Thank you.

Instead of just swapping caps and resistors to improve the tone, I am considering to install a variable crossover into my CE-3, so the treble frequencies get filtered to the BBD circuit, and the bass frequencies are bypassed to the end.
This way I can either set a vibrato effect (full signal gets modulated, with no dry signal), or varying degrees of a light chorus effect, with the bass intact (not muddied up). 

And when running the CE-3 in Mode 1, I get a full-range output, and a bassy, dark sounding output without modulation. This is so I can run two amps with contrasting tones.

Does anyone know how I can implement this? I am hoping that it is as simple as a variable capacitor installed as the frequency splitter, after the input jack.