freestompboxes.org

IvIark wrote:We could do it with 3 rotary switches. If two are 4 pole 3 position we can switch between 3 different sets of caps for the 0.22µF and 3 for the 4.7µF, and then use a 2 position 6 pole to switch between 2 sets of tone caps. I've also got the switches unless people would prefer 3 sets for the tone as well, but we'd need to get hold of a 6 pole 3 position if anyone knows a supplier?

all of those caps are going to be replaced with .1uF- except the tone caps. i'll draw a schematic in the next couple of days so we can avoid all of this confusion.

here is the factory schematic. it's close enough for our purposes. all caps except the .01 in the TC and the 25uF bypass caps will be changed to .1uF. the .01 in the TC will be changed to .02 so we can use elytics. the bypass caps will have to be decided on, but will not be switched, and the 220p will not be switched either.

of course it would be best to switch out every coupling cap in the circuit. how to do this, we will have to figure out. also, we have to figure out if we are going to switch out the TC section as well. up until now, i have been assuming that we will.

IvIark wrote:What would everyone else like to see used for the values listed?

Bear in mind that we can sub in different values, so that we aren't killing ourselves trying to get different dialectrics, and then being able to sort and spec to get matching sets.

If we concentrate on just one or two sizes, I think that we'll be fine. In the case of the Overdriver, I think that we could sufficiently use just 100nF and 1uF, which are readily available in ceramics, electros, tantalums, and all sorts of film caps.

Then it comes down to which values we are willing to modify. For instance:

- The pair of 220nF caps could be reduced to 100nF

- The pair of 4.7uF caps could be reduced to 1uF

- Most 10uF/22uF/25uF/caps could be reduced to 1uF as well

Remember - the pedal doesn't have to sound awesome. Film caps in 4.7uF and larger? And enough to match to spec? How large can we get ceramics in? I know that monolithics will take you up to 220nF or so. But much beyond 1uF? And electros or tants in 10nF?

I'm not saying that we can't find them, but it could be harder than we need to make it.

I think that we should stick to the following dialectrics:

- ceramic (probably monolithic, available in lots of sizes, cheap enough to match to spec)

- electro (cheap, lots of sizes down to about 100nF or so)

- film of whatever type the pcb will accomodate, since some 1uF films can get large. Poly box caps or similar would be an obvious good choice, as they are compact and pretty easy to match to spec (as opposed to greenies and such - tolerances all over the place)

I guess that tants could be subbed for electros, if we can find enough of them to match at the sizes we need.

I really don't see the need to switch out every single cap in the circuit. IMO, concentrating on a few coming off the collectors on the trannies would probably be sufficient, and also maybe one or two feeding into the bases.

We really need to whittle this down, or we could drive ourselves nuts.

I agree that simplicity is the best strategy here. It would be fun to test a pedal with 2-3 rotary switches, after that it might get confusing and you'd drive yourself crazy trying out all possible combinations.

I'm not that experienced to suggest which capacitors in the circuit would be the most influential but from a practical point of view I'd like to see the input capacitor being switchable. It's one component that I almost always fine-tune when I want to mod a pedal and it would be nice to know if the dielectric type really matters there.

Surely the use of rotary switches is going to invalidate the test from the point of view of the golden-eared brigade, as that will introduce extra components and wiring (cloth-covered or not)?

Just saying...

NickS wrote:Surely the use of rotary switches is going to invalidate the test from the point of view of the golden-eared brigade, as that will introduce extra components and wiring (cloth-covered or not)?

Just saying...

What would be the alternative??

I don't see what's wrong with the rotary switch. All capacitors in a given set would be connected to the same switch so there are no extra components. You don't even need extra wiring, you can just solder the caps to the switch lugs.

MoonWatcher wrote: We really need to whittle this down, or we could drive ourselves nuts.

yes, i think this tactic is probably right. although, as far as dialectrics go, if we stuck to 100n, even simpler, we could use some more exotic caps which generally measure better in testing. i am thinking specifically of polystyrene. i think the difference between a cap like this, and something like a ceramic or elytic would theoretically be easier to hear. i have some russian 100n p/s like the large green box caps pictured on this page that i think would work well.

http://babelfish.yahoo.com/translate_ur ... =Translate

i really don't think that the difference between .1 and 1uF would make much of a difference when dealing with guitar frequencies, but hey, what do i know?

RnFR wrote:i really don't think that the difference between .1 and 1uF would make much of a difference when dealing with guitar frequencies, but hey, what do i know?

Depends on place of use.Calculations of cutoff frequency of high pass RC network (interstage decoupling) 0.1 uF/10 K = 160 Hz, 1 uF/10 K = 16 Hz

KMG wrote:

RnFR wrote:i really don't think that the difference between .1 and 1uF would make much of a difference when dealing with guitar frequencies, but hey, what do i know?

Depends on place of use.Calculations of cutoff frequency of high pass RC network (interstage decoupling) 0.1 uF/10 K = 160 Hz, 1 uF/10 K = 16 Hz

yeah, i'm aware of that, but i'm just much too tired and lazy at the moment to actually figure it out for this circuit.
thanks for clarifying though!

KMG wrote:

RnFR wrote:Calculations of cutoff frequency of high pass RC network (interstage decoupling) 0.1 uF/10 K = 160 Hz, 1 uF/10 K = 16 Hz

I could be getting ahead of myself, but for the purposes of this test, does one suspect to hear a largely audible difference in the range of 16Hz to 160Hz anyways? Unless you've got 8 12's pumping away through ported cabs, you could probably safely bump the HP ledge up to around 200Hz or so, IMO.

The most "overt frequencies" IMO are between around 400Hz to 2500Hz or so. And I'd venture a guess that when the average guitarist hears a difference between any two pedals that are being demoed that it's probably the interaction between a midrange frequency and the upper order harmonics that accompany it.

RnFR wrote:if we stuck to 100n, even simpler, we could use some more exotic caps which generally measure better in testing. i am thinking specifically of polystyrene.

Yes - one of the cap types should be easier to tweak to the other two, IMO - something to take up the slack, if you will.

Which brought me to another thought. If one of the proposed dialectrics is all over the map when spec'ed, we could conceivably use other types that are fairly commonly found at 82nF, 120nF, maybe even 150nF, all in an effort to get them within a reasonable matched capacitance.

The only reason that I bring it up is that someone may really think that "only vintage ceramics sound good." So if anyone has enough older disks, it might be a little tough to spec the other two sets around them. Electros are certainly cheap enough that we could procure 50 or more until we find matches. And polystyrene could "take up the slack" with the other two types.

Of course, we could always use monolithics for the ceramics, but I've got to wonder if enough folks might be boo hooing them as a somewhat invalid choice of ceramic dialectric. For the purposes of this test (just being able to perceive differences), I think that it's irrelevant, but if we could kill just one more bird with an existing stone, it might not hurt (except for the bird ).

umutgokcen wrote:I don't see what's wrong with the rotary switch. All capacitors in a given set would be connected to the same switch so there are no extra components. You don't even need extra wiring, you can just solder the caps to the switch lugs.

Yeah - this is exactly how it's done in many handbuilt amps. Unless you want another turretboard with an assload of caps on it, you mount them all right on the rotary, and then you can have just a singular pair of wires leading back to the pcb to where the singular hardwired cap was. When it's done right, you aren't contributing any more to a potential rat's nest of wires than you would with a design that has a pile of pots. 4 caps = 8 wires from the pcb to the rotary. Have nice twisted pairs for each cap, color code them, and it will be neat as a pin.

...And with pedals and toggles, it's pretty common to see this practice with caps, clipping diodes, etc. And if you're smart and design around it, you can even have a more compact and smartly laid out main pcb.

A multi pole rotary can really be such a cool thing in a pedal, IMO. Really great for hearing a comprehensive set of changes on the fly. The hard thing for this test will be finding one that has at least 3T's for the different dialectrics. But 4P3T's do exist, and aren't that uncommon.

i don't think a 160Hz cutoff is unacceptable for guitar purposes. a guitar supposedly goes down to 80Hz, so even that lowest note may still be included, although attenuated a bit. there are many a pedal out there that have cutoffs higher than that i'm sure.

yeah, i don't think fidgeting with the value a bit is that big of a deal. if it makes it a bit easier, i say we do it. damn the naysayers!

if we want to use those big green p/styrenes i have, i believe they are very high tolerance anyway, so it shouldn't be that hard to get matching sets(hopefully!)

also, when matching, are we going to be exact to the letter, or is there going to be a bit of leeway? say an average could be used over all of the caps in a set? as well as a minimum difference that we could set up. the average value would make sure that the values aren't skewed too far +/- each way.

regarding ceramics- it's true that many of them measure very different, i remember one specific type was declared to be avoided at all costs by FSB member gus(i keep bringing up his name lately!) because of one of the studies i posted in the other thread. i believe it is the small size/large value ceramics. i could go back over the study to figure out which model number, as i think it would be a good candidate for the test. not to mention that they can be had for cheap.

i think we should try to go from the extreme "best" to the extreme "worst" types in order to make it as easy as possible to hear the perceived difference, and we should definitely hash this all out as much as possible before anything gets built.

regarding the rotary sw- i think it's a good idea. and remember, we can always have more than one rotary. you would just have to set all of them between each test.

I had another idea that could really help for the purposes of this test. Sorry for all of the posts, but I only tend to stop in here once about every 12 hrs or so. And one huge reply is something that many folks won't want to read (just a guess on that, though).

This idea proposes the use of a terminal block with at least 8 of them (maybe a pair of 4's?). The wires from the pcb could be routed to the terminal strip, with the secondary set of wires to the rotary switch.

Why would this be beneficial? Since the caps are going to be mounted right on the rotary, we could conceivably wire up different "cap modules" on different rotaries. I.E. different dialectric sets, and then replace the rotary with a relative amount of ease.

RnFR wrote:also, when matching, are we going to be exact to the letter, or is there going to be a bit of leeway? say an average could be used over all of the caps in a set? as well as a minimum difference that we could set up. the average value would make sure that the values aren't skewed too far +/- each way.

Exact dead on would be the ultimate, as there is a lot less room for claims of perceived differences due to even minute capacitance variations. That said, I personally think that it's impractical and unnecessary. That said, I think that a range needs to be established. I would prefer 3%, but could be persuaded up to 5%. And I think that either one of those could be done with a minimum amount of effort.

As far as an "average global variance," I think that things could get a little sticky if we do that. Since we have the advantage of sorting the different types for each component in the circuit that will be replaced, I think that we should do that instead. I.E. slot 1 might have all 3 types matched at 93nF, slot 2 could be 102nF, and so on.

RnFR wrote:regarding ceramics- it's true that many of them measure very different, i remember one specific type was declared to be avoided at all costs by FSB member gus(i keep bringing up his name lately!) because of one of the studies i posted in the other thread. i believe it is the small size/large value ceramics. i could go back over the study to figure out which model number, as i think it would be a good candidate for the test. not to mention that they can be had for cheap.

By all means, I think that this is our ceramic dialectric of choice! This would serve two purposes. It would either lend credence to the cited study or make references to the study somewhat irrelevant for our specific purposes. And it also might serve as a sort of fine tuning tool, if they are common caps that will find their way into the average build. Kinda strikes right at the heart of what we are trying to do with this.

MoonWatcher wrote:

RnFR wrote:also, when matching, are we going to be exact to the letter, or is there going to be a bit of leeway? say an average could be used over all of the caps in a set? as well as a minimum difference that we could set up. the average value would make sure that the values aren't skewed too far +/- each way.

Exact dead on would be the ultimate, as there is a lot less room for claims of perceived differences due to even minute capacitance variations. That said, I personally think that it's impractical and unnecessary. That said, I think that a range needs to be established. I would prefer 3%, but could be persuaded up to 5%. And I think that either one of those could be done with a minimum amount of effort.

As far as an "average global variance," I think that things could get a little sticky if we do that. Since we have the advantage of sorting the different types for each component in the circuit that will be replaced, I think that we should do that instead. I.E. slot 1 might have all 3 types matched at 93nF, slot 2 could be 102nF, and so on.

RnFR wrote:regarding ceramics- it's true that many of them measure very different, i remember one specific type was declared to be avoided at all costs by FSB member gus(i keep bringing up his name lately!) because of one of the studies i posted in the other thread. i believe it is the small size/large value ceramics. i could go back over the study to figure out which model number, as i think it would be a good candidate for the test. not to mention that they can be had for cheap.

By all means, I think that this is our ceramic dialectric of choice! This would serve two purposes. It would either lend credence to the cited study or make references to the study somewhat irrelevant for our specific purposes. And it also might serve as a sort of fine tuning tool, if they are common caps that will find their way into the average build. Kinda strikes right at the heart of what we are trying to do with this.

I've just gone through a bag of twenty 100nF capacitors, and it's easy to find several the same! In the bag, there were five that were 100nF +/- <<1nF! There are no excuses to get all the types of capacitor to have exactly the same measured values - they just have to be selected.

mictester wrote:I've just gone through a bag of twenty 100nF capacitors, and it's easy to find several the same!

Of this bag of twenty, how many different diaelectric types were there? Was it a mixed bag?

I've found that some types of dialectric tend to always be on the high side of the rated value - mylar films come to mind almost immediately. Yet others always seem to come in low - certain monolithic types that I had did just that.

I just measured about a dozen ceramic 220nF caps - the lowest was 278nF! Many were 290nF or higher.

Some 220nF metalized polyester films measured:

-222nF
-216nF
-216nF
-213nF
-227nF
-219nF
-213nF
-212nF
-215nF
-220nF
-213nF
-214nF
-217nF

...so out of 13 caps rated at 5% tolerance, I got one that was dead on. Two were over spec, the rest clearly under spec, by as much as 7nF, except for a single one that was off by 1nF. At 5%, they could go as low as 209nF.

I don't mean to go against what you're saying - just not sure how practical it would be. Not everyone can fire up a bag of 20 caps and get as close as you did.

Maybe if a few more folks can test three different dialectrics from a modest amount and get at least two dead on matches, it might help determine the amount of wiggle room (if any - maybe I was the odd man out with my measurements).

if i've got some time on the couch after getting off work tonight, i'll go through some of the caps that i've got.

I'd say that within 1% of the value is good enough, come on let's get on with the build!

I was trying out some caps in a big muff tone stack today and I could barely tell the difference between 0.01 and 0.012. That's 20%. Even if we assume that the 0.01uf was actually 5% higher, that is 0.0105, and that 0.012 happened to be 5% lower (0.0114), that difference is still 9%. I doubt that anyone can hear 1%, or in other words the difference between 0.01uf and 0.0101uf.