Some more questions on the Tubescreamer/RubyAmp circuits.

[slashandburn]

Somehow I've managed to find the time and motivation to start making things again. Last Autumn I discovered things like the Ruby and the Smokey Amps and was so blown away with that little 386 chip I digged out all my old college electronics paperwork.  Life got in the way before I'd really got a chance to play properly with all the possibilities and different ideas. This last week or so I've breadboarded up a Ruby amp again and stripped everything back to see where I really want to go with it.  I won't pretend to really know what I'm doing, to be honest. Last time round I think I got a little carried away dropping in basic elements from different stompboxes to pimp it out a bit and got a little lost in all the numbers. 

Anyway, back to the point.  This is what I have. Excuse the crude sketch. Basically a Ruby Amp with half a Tubescreamer tacked on in front of it. And a Big Muff Tone control thrown in for good measure. I'm itching to finalise something so I can show it off properly (and inevitably move onto something else) but I'm trying to take the sensible approach this time and trying to plug some more of the gaps in my knowledge first. 

https://drive.google.com/file/d/0Byueee9ZKwqJTDhUU3lEWEFjb2RIOEpzN2pQZzg4enQtNWI4/view?usp=sharing


Questions then. Ponderings, rather. If any of you would be so kind.

First up, that big question mark (assuming my link worked) in the feedback loop around the opamp. I've over-thought this and think I'm probably missing something. Is this resistor really necessary? Besides playing a somewhat subtle role in the Low Pass filter in the feedback loop, does this guy really just serve to ensure that there's always some resistance across these pins? I can just ditch this part, and get something close to unity gain when the pot is as '0'?  Full of doubts.  Feel like I'm missing something fundamental.

Next up then. The section between the Opamp and the poweramp. This is where I had the most fun, swapping in and out different tone circuits and constantly changing my mind over whether I wanted a volume pot and/or another clipping stage in there. My real question here, with adding passive circuits like this and signal, how much is too much and is there a simple way to compensate for this kind of signal loss?

And last I think, the 8 ohm load to from the attenuator to ground.  Most circuits I've seen mention this as optional. I'm not having any issues without it, but I guess it's protecting something.  The circuit, from the speaker is my best guess. But again, how necessary is this part? What is the risk without it? Should I alter it's value to match the speaker it's driving? i.e if I'm driving a 4 ohm speaker should this peice of the puzzle also be 4 ohm?

Thats about it for now, hope someone can shed some light!

[GibsonGM]

Hi Iain,

1) ? resistor is just there to always have an Rf (feedback resistor) present, so you always have some gain there when the wiper shorts the pot's resistance, yes.   It COULD be doing a little more, I dunno, I don't build much in this style so just go with the flow....play away - take it out, see what you get (a buffer of sorts? ha ha)    Won't hurt anything, we're non-inverting...

2) You can simulate a tone stack, etc that you would like to put between TS and 386, and see what the INSERTION LOSS is.    Then, you can see if the stage driving it (TS here) can in fact...drive it ok.     If you're going to lose too much, you can add a RECOVERY STAGE....the last stage of the BMP works nicely, in fact, or you can use another opamp section (great if you are using a dual for the TS 1/2).....what flavor you use is up to you!   FET, BJT, opamp...as long as it gives you enough gain to get to where you want for feeding the 386.   You can get even more tonal variation by what 'type' of recovery stage you choose - more experimenting!    One easy way to know what's up is to put the tone stack in, and just listen, as that will not hurt anything.    If it sounds like poo - you know you need more, ha ha.   You'd notice a lack of highs, noise - just not good sounding.      If you used say, the BMP recovery stage, and found it had too much drive, you can reduce that (altho - it goes with the BMP tone control, he he!)....

Dunno what's up with the 8R to ground...to prevent shorting the 386 output to ground?  Why a magic "8 ohms" and not 100, tho?  Or 10?    *shrug*    Won't let you turn it all the way down, at any rate, but close to it...hope this helps! 

[slashandburn]

Cheers Mike! Helpful as ever (I'm sure I remember you from when I last frequented these parts!).

So, "question mark" resistor can get dumped. I'lll have a play around, I did at one point have a spst to short The feedbackloop in the opamp (a unity gain buffer/preamp gain toggle switch of sorts) so I'll try comparing the output  with the switch shorting pins 2 and 6 to that of the switch open and the pot at zero.

I'll also get reading on recovery stages. Having quickly glanced at the BMP circuit,  is that basically just another jet impedance buffer tacked on the end?

I'll forget the 8ohm load for now too unless anyone else can shed some light? No problems without it, no point trying to fix something that seems to work fine as is.

Thanks again!

[GibsonGM]

Yeah, I do recall you too from a few months back Iain, glad you're digging in!    Good idea with the switch, I think that will give you the info you want. 

I got to thinking that perhaps the ? resistor is an artifact from the fact that there is another section to the TS, and that may set min. gain, which appears to be about 11 (with 51k/4.7k to Vr), so that you always are providing *some* drive to the tone control/2nd opamp section.   If not there, you might turn down to silence when using the gain pot, is all.     OK to cross input to output in this case, seeing what a plain buffer would sound like.   Don't think I'd allow Vr into that loop after the cap, tho - that is just acting as a ground for that resistor....

The last BJT transistor stage in the BMP, from its .1u input cap to vol pot, is just a gain stage, to take the now-lowered signal from the tone stack and "re-boost" it (hence, 'recovery stage') for proper output level.    It's  a  good snippet to keep around for these purposes, like a generic black box of boost we can re-use without having to reinvent every time!   Without doing the math, it probably has *ok* in and output impedances, generic....does the job....good enough for rock n roll!   

Re. that 8 ohm load, I think it was to make sure you don't short the output directly to ground, you you might want to include it in case you do that.  I'm not sure how a 386 reacts if you send its output straight to ground.   Sorry I wasn't clear on that!    Now I want to try it to see if it kills the chip, ha ha!   Could heat it up, I s'pose....

   

[slashandburn]

Ah! I think I understand you better now that I'm sitting comfortably at home (not reading on my phone at work).

That 8ohm resistor on the output pot, is for when the pot is turned to zero? That would short the output of the 386 to ground, and that could possibly be undesirable depending on the nature of the chip feeding the signal?

The recovery stage sounds straightforward enough. I like these generic circuits, make me feel smarter than I probably am. Are.  Is?  I can be quite pedantic so I'll look properly into the numbers  tonight. 

And the ? resistor. I think I have good grasp on that now, but perhaps not the V ref issue.  I asked about that here before (wiring that cap to Vref rather than ground) I should skim back over that thread before thinking too much about this, but are you saying to avoid doing this if I were to  leave out ? resistor, or only if I were to short input and output?

Again, greatly appreciate you taking the time!

[GibsonGM]

Ah! I think I understand you better now that I'm sitting comfortably at home (not reading on my phone at work).

That 8ohm resistor on the output pot, is for when the pot is turned to zero? That would short the output of the 386 to ground, and that could possibly be undesirable depending on the nature of the chip feeding the signal?

The recovery stage sounds straightforward enough. I like these generic circuits, make me feel smarter than I probably am. Are.  Is?  I can be quite pedantic so I'll look properly into the numbers  tonight. 

And the ? resistor. I think I have good grasp on that now, but perhaps not the V ref issue.  I asked about that here before (wiring that cap to Vref rather than ground) I should skim back over that thread before thinking too much about this, but are you saying to avoid doing this if I were to  leave out ? resistor, or only if I were to short input and output?

Again, greatly appreciate you taking the time!

Still haven't 'tried' a 386 with output sent to ground; I will assume it's not a good thing, so don't do it, ha ha!   Yeah, the vol. pot will do that with no add'l resistance there.

You could build that BMP recovery circuit in a simulator like LT Spice or Tina and being messing with it, getting a sense of what the parts do, and not have to analyze it too much


The 100n cap connected to Vref**  is blocking DC - but to AC (your signal), that is the same as going to ground - since we 'raised' the signal by 4.5V at the + input, and is the purpose for that cap to be there - so if you just cross output to input, buffer-style, don't go placing a wire such that Vref will flow thru that loop (stay on the diode end of that cap, or disconnect Vref, or lift one end of the cap (same thing)).    Probably undesirable to put 4.5V DC in that loop.   Don't think it'll ruin the chip, but would possibly make a loud ugly noise or silence the circuit....think of this as "if the Vref is still there, don't remove that cap".    If you REMOVE Vref tho, you can only make a buffer as the opamp has no path to ground (Vr) any longer for the gain resistors. 

Hope that makes sense!   Just - everything going thru that opamp is referenced to 4.5V as being its "center point"...its bias point....about which your signal swings, both positive and negative (that is how we're able to amplify BOTH sides of the signal with a positive-only power supply).    Vref is "pushing everything up" by about 4.5V.   You can see that on the input, where the  1M resistor is feeding Vr to the input....47n input cap blocks it from 'traveling'....


** sorry, just looked at your document again.... I use the TS808 schematic for quick reference, and just noticed you have that 100n cap going to ground - any reason why?  My belief is it should go to Vref, logic for this already written above!!    In essence, putting it to circuit gnd is like putting it to -4.5V - again, hope that makes sense!   Vref is the center of where the audio signal is.

[slashandburn]

.............just noticed you have that 100n cap going to ground - any reason why? 

It was like that when I got here.  Ah, seriouly though, I don't really know. I'd asked a few questions months ago as to why sometimes (in the same or similar circuits) I've seen it going to ground, others it goes to vref. I'm guess there is still a hole in my theory there. I'm not questioning your logic in any way, just not entirely following you. If either vref or ground works here, Should this cap goto Vref for the sake of consistency? I want the "audio" signal to think of 4.5v as it's center point, so it makes sense to point the feedback loop there as it's ground reference?

I think I'm there, so if I want this bit of the circuit to be switchable between a simple buffer and a tubescreamer gain stage, then send it to ground as a precautionary measure. If it's only going to be a gain stage (as it is designed to be, given it's lifted from the tubescreamer) then send it to Vref.   Would that be right?

[GibsonGM]

Yeah, Iain - I can't find a rule of thumb about whether or not to run the feedback loop to Vr or ground, either.  For a grin, I simulated both ways in LT Spice, and don't see any real difference.  The input must be tied to Vr, though, as you have done with the 1M resistor.  Perhaps there is some other reason it's often tied to Vr, relating to impedance or something.  I really don't know!    I would have suspected it to be necessary due to (as you note) the desire to have the audio signal reference be at 4.5V, but that doesn't seem to be the case.   I run these to Vref, for consistency, yes, FYI. 


To make this a simple buffer, you would open "that line to Vref" - open the right side of the 2.2K resistor.  To cut Vref out of this part of the circuit completely, it's not needed for a buffer.    As we agreed, that leg is just going to ground...it's actually doing some filtering.
Then, you need to get the output tied to the "-" input....you need to bypass the gain pot and the "?" resistor if you choose to leave it there.   So, probably a DPDT switch.    But you're still left with whatever comes after this stage, your BMP tone control and maybe gain recovery stage...

[ashcat_lt]

Yeah, Iain - I can't find a rule of thumb about whether or not to run the feedback loop to Vr or ground, either.  For a grin, I simulated both ways in LT Spice, and don't see any real difference.  The input must be tied to Vr, though, as you have done with the 1M resistor.  Perhaps there is some other reason it's often tied to Vr, relating to impedance or something.  I really don't know!    I would have suspected it to be necessary due to (as you note) the desire to have the audio signal reference be at 4.5V, but that doesn't seem to be the case.   I run these to Vref, for consistency, yes, FYI. 

It doesn't really matter because the cap blocks the DC component anyway.  Tie to +9V and it's the same thing.  The only real issue might be if the cap from Vref to ground isn't big enough, and/or the Rs from the feedback loops to Vref isn't big enough, you can get some cross-coupling between the opamp stages.  I don't know how to tell you what "not big enough" means in real numbers...

[GibsonGM]

+1   Yup, to AC, pos and neg appear to be dead shorted due to the presence of filter caps and so on.  It doesn't care.   Both +   and  -    will work for filter paths and so on, just like with tubes, and also works at Vref....it's just a return.   

Glad to hear there is no special "rule".....you know, the thing you totally forgot about, maybe relating to the PS internal impedance or something, lol....the great thing about this hobby is that there is always another layer underneath the "introductory" or "beginner's" layer.  So there's always more to learn, and to use later in a different way!

>> So there we have it, Iaian....no matter if you tie that 100n to Vref or Gnd.   

[slashandburn]

Great stuff! I think I have all I need for now. Thanks a lot!

....the great thing about this hobby is that there is always another layer underneath the "introductory" or "beginner's" layer.  So there's always more to learn, and to use later in a different way!

Oh no. Please don't. You're right though. I think that's part of the appeal for me. Unlike the anticlimax that comes with learning to play something and realising it's easy and not witchcraft after all, it's more akin to hillwalking. Reaching the summit, enjoying the view for a second before realising you haven't even started on the hill you intended on climbing. And you're not sure whether that big bastard mountain that only just came into view is the one you want to be on, or merely concealing the one you want to be on top of. Thankfully at least this can be an indoor hobby and my shocking fitness levels shouldn't be too much of a handicap.

The only real issue might be if the cap from Vref to ground isn't big enough, and/or the Rs from the feedback loops to Vref isn't big enough, you can get some cross-coupling between the opamp stages.  I don't know how to tell you what "not big enough" means in real numbers...

That does somewhat answer another question I nearly asked. I've wondered about the value of that cap and how to decide on what value to use, since it seems to vary a little and I'm short on 47uf caps.

I'm off to get a big jug of coffee on the go.  Not sure tonight will present much of an opportunity to play about, but enough caffeine might "free up" an extra hour or two.

Thanks again!

[karbomusic]

That does somewhat answer another question I nearly asked. I've wondered about the value of that cap and how to decide on what value to use, since it seems to vary a little and I'm short on 47uf caps.

Does it somehow need to be big enough to encompass the bandwidth in the gain loop based on the resistance used? Newb question, apologies.  

[slashandburn]

Pass. But I've had no trouble subbing in anything between 22uf and 100uf.    (just incase: I'm refering to the Electrolytic Cap from Vref to Ground, at the voltage divider. Wandered quite far from my original questions!)

[karbomusic]

Pass. But I've had no trouble subbing in anything between 22uf and 100uf.    (just incase: I'm refering to the Electrolytic Cap from Vref to Ground, at the voltage divider. Wandered quite far from my original questions!)

Ah, oops. I think that one just needs to be large enough for smoothing out VREF but not so large it takes all day to charge/discharge. If too large it's going to take awhile for the bias to stabilize AKA one side of the wave is going to clip until the cap is charged when initially powering on the circuit. For 9V I typically use 100-220 and for a 4.5V Vref I typically use 47uF.

Speaking of charging, still guilty as charged on that being a newbie assumption.

[GibsonGM]

The one in the power supply divider, at Vref itself?  10u or above, I've never had a single issue.     Whatever is around...10, 47, 100....there is probably a point that it would be ridiculous....3300u?    If this was for an LFO in a trem or something that 'changed a lot', I'd be all about making that cap biggish, but not for this type of application.   47ish is the safe bet.  Bigger won't hurt as long as it's not insane.   

The circuit isn't drawing diddly beans from Vref; it just ASSURES us of a nice clean reference voltage at that point with proper decoupling.   It's good practice.  We don't want what's happening with the 9V supply to couple to our Vref, which could manifest itself as noise or even oscillation in *certain* circuits.     

The theory here is that if you suddenly had a spike (a big need for current) on your power rail, you'd have a 'short reprieve' because that cap is there as a reservoir and acts to smooth things out.   LFOs driving LEDs and such - they'll cause that (the oft-complained about ticking being one side effect) far more than a gain stage.

[PRR]

> Is this resistor really necessary?

Muntz it.



Muntzing -- "Muntz ...carried a pair of wire clippers around and when he felt that one of his builders was overengineering a circuit, he would begin snipping components out. When the TV stopped functioning, he would reinsert the last part and leave the TV as it was."

So cut it, and see if the picture fades out or you get green smoke.

By head-games, the predicted effect is that no-signal gain goes to infinity. The tiny input hiss is amplified to a monster output hiss, limited by the diodes. When you pluck and let-fade, the fuzz does not get smaller, it just morphs from tone to hiss.
_________________

The 8 Ohms under the pot makes little sense. It pretends it is going to keep a load on the '386. But What when the 500r pot is set half-way? We have 258r one way and 258r the other way, for a 129 Ohm load. If the '386 was gonna blow-up for a too-light load, this might do it. In fact the '386 (like nearly all chip power amps) is thrilled to run no-load, just like the mule when you take the plow off him.
_________________

> how much is too much and is there a simple way to compensate for this kind of signal loss?

This is Basic Audio.

When the listeners complain the sound is weak, you've tossed too much sucky-stuff in the path. The "compensation" is another Amplifier. Or find some un-tapped amplification in the stages you already bought. In fact I don't think you could possibly need more guitar gain than is "possible" with a TL071 and a '386. The '071 can push a gain of 500. The '386 can do 200. 500*200 is a very big number -- 100,000? OTOH soft guitar is 20mV or 0.020V, the 9V powered '386 can't output more than 3V, 3V/0.020V is a gain of 150 needed. You could throw in 666:1 of loss and still slam the poor '386.

It may not be quite that simple. Any guitar input with gain of 500 will distort all to heck because loud guitar is several tents of a Volt.

But play around, use the wire-snippers, learn.

[GibsonGM]

^  I love it, Paul!  LOL   MUNTZ it.      If I'm inserting a tone control/stack, I will pretty much always just put it in there, see what the *sound* is like, and then decide if I need to goose it.    Might be ok with no changes at all, or minimal changes to the feedback loop....

Thanks for the answer re. 386 with NO load - how about a load shorted to ground, though?  I haven't been able to find any info on 'output short circuit protection', or if that's even an issue.    I might just DO it, see what happens, if I can verify that I have more than 1 in stock  

[PRR]

At 6V, an LM386 can face a dead short and survive quite a while. At higher voltages, temperatures, and times, it *may* be possible to blow one up. Basically it is too weak to quickly melt itself at design conditions. Push it hard and long, you could be in for a big $0.69 replacement.

[GibsonGM]

Thanks , Paul...another ?  of the universe now understood      You know up where we live (Dirigo, baby)...more like $3 at the 'local' (20 mile drive) Radio Shack.

I really need to put 10 of those into my next parts order!!

[slashandburn]

Love this place! Thanks again and excuse the slow response. I made the jump last night and threw this all together on stripboard carefully, checking and double checking each connection and, low and behold!!! WOULD YOU BEIIEVE IT!!!!!!???

It don't work. I fully expected this, though. My soldering and testing/fault finding skills need to be refined, but I'm more confident this time that I will be able to get this guy singing. I'm off to get more 9Volt batteries and will report back with photo's of the board and voltages if I still have no luck.

Off on holiday next week so I'll likely need to put this away and forget about the "practical" side of things until we get back. Could be a decent oppotuninty to finally get my head around LTSpice and simulating circuits along with more general reading and overthinking things, away from the the distraction of my chaotic workbench.

You never know, maybe a fresh 9v battery is all this thing needs to spring to life.  I was at least smart enough to leave my breadboarded circuit intact this time around so I can refer to my "working" circuit for an idea of what voltages I should be getting.   Hopefully just a bad solder joint or a short somewhere. Every chance I've messed planning the veroboard though, or the offboard wiring.  Oh god yeah, the offboard stuff. Why do I always think that bit will be childsplay?

Now there's a point.  Wire length. And guage, perhaps.   I try to keep the offboard stuff trimmed as short as possible. I'm often torn in between using stranded and solid core and inevitably end up just grabbing whatever i can reach without moving from my chair.   I've just noticed  there though that the breadboard is much neater and I've stuck to the same guage and type of wire.  I should probably make an effort to do the same when I'm soldering to stripboard too. I can't really think of a good excuse for not having done that in the first place.

Mike, if I wan't so scatterbrained I might be able to save you a few bucks by being able to say "yeah, that would explain all those fried 386's from my last experiments.".  And Paul, absolutely fantastic suggestion.  My inital reaction was "but no! I've just hours/days carefully planning all this! You're crazy!"   But yeah.  It's no different really from lifting something off the breadboard to see what difference it not being there makes, except it takes slightly longer and a couple of drops of solder to put it back in there.

Cheers again!