Boss VB-2 LFO analysis?

[Joe Kramer]

Hi Friends,

I'm borrowing the LFO section of the VB-2 for use in a phaser, foot-controllable modulation pedal, etc.  It puts out a very nice sinewave and with the addition of a darlington xstr, drives an LED/LDR combo well.   With gratitude to "godiksennet" here's the schematic:

http://www.godiksennet.com/images/sch/BOSS_VB2.JPG

Two questions:

First, I'd like a better understanding of how the circuit works.  I can see something like a twin-T network in there, and it appears that the RATE control is changing the center frequency and hence the speed.   Also D4 appears to be a "temp comp" for Q8's grounded emitter, but beyond this I don't understand what Q8 is doing.  (BTW, the VCA business after the LFO does not concern me in this application.)

Second, I'm having a problem with the RATE control.  When I change speeds, especially from fast to slow, sometimes the modulation becomes very shallow, and occasionally stops altogether.  Another nudge of the RATE control snaps it back to full sweep again.  I've made sure to match the 1Ms and .047s in the twin-T.  I'm using a TL061 op amp (3.5 slew rate, which should be plenty fast).  Also using the correct 2SC945 xstr.  Since I'd like to use this circuit in a foot-controller, reliable speed control is important.  Any ideas on how to somehow "stiffen" the oscillation potential so it doesn't die out with speed changes?

Many thanks!

Joe

[DiyFreaque]

Hi Joe,

Looks like a Twin Tee to me, too.  I imagine someone here can chime in about that LFO - I sure don't recall seeing one like that.

My approach would be to try a different op amp rather than the TL061.  I notice the original is a TL022, which is a low power op amp, but I believe it's not a JFET like the TL061. 

Reason I say try a different op amp is part mojo/part past experience.   One time, bench testing a circuit for Thomas Henry, he pointed out that I should use a bipolar op amp instead of the FET op amp he had on schematic while I was working on the breadboard.  This was (IIRC) in an expo converter.  I followed his intruction and used a 1458 instead of a TL072, and it worked.  For the hell of it, I tried the TL072, and quickly figured out he wasn't woofing.  On perfboard it worked fine, but not on the breadboard.  I suppose that was the one instance I've run into where the stray capacitance definitely wasn't my friend.

The Mojo part is I see if I can work it with the original mojo parts, then go for the alternates (if desired).  Sometimes those designers will take advantage of a particular quirk of a particular part, and that's what makes the thing tick.

Then again, I could just be full of beans  .  But, I'd try it anyway (if you haven't already).  Something more bipolarish and less JFetish.  Mouser has the TL022 for about twice what the TL061 is.

Scott "JFetish" Stites

[Geoff Hunter]

Hey

Sorry can't help with the technical side of your question, but on a similar note to your problem with the rate:

When i take my VB-2 out of its cabinet  and plug it in, normally when I switch it on, I don't get any wobble.  However if it touch the rate and depth knobs (can't remember which one), then the  LFO springs to life and works perfectly fine after that!

AFAIK my VB-2 hasn't been modded, so it might seem that the LFO is not.....perfect to begin with.  Mind you its 20 years old now.

Geoff

[George Giblet]

It's this common type of tunable sine oscillator:

http://www.national.com/ms/LB/LB-16.pdf

LB-16, figure 2 is the easiest one to translate.

There few small twists which might throw you out:

- LB-16 fig 2, R1 is replaced with R45, R47, R48
- LB-16 fig 2 first inverting opamp, R5, R6 etc is replaced with the inverting transistor stage.
- No zeners, I think the inverting transistor stage is allowed to clip instead.

The whole thing is based around the well know bandpass circuit (second opamp in LB-16 fig 2).

Article should explain the rest of the details.


Edit:  Oh the basis for the band-pass circuit is a *bridge-T* network in a feedback loop.  Don't let R45, R47, R48 make you think it's twin-T.

[DiyFreaque]

When i take my VB-2 out of its cabinet   and plug it in, normally when I switch it on, I don't get any wobble.  However if it touch the rate and depth knobs (can't remember which one), then the  LFO springs to life and works perfectly fine after that!

Well, so much for the op amp mao-jao. 

It's this common type of tunable sine oscillator:

http://www.national.com/ms/LB/LB-16.pdf


How cool is that?  Thanks, George!

Cheers,
Scott

[Joe Kramer]

Scott: My approach would be to try a different op amp rather than the TL061.  I notice the original is a TL022, which is a low power op amp, but I believe it's not a JFET like the TL061.

Reason I say try a different op amp is part mojo/part past experience.   One time, bench testing a circuit for Thomas Henry, he pointed out that I should use a bipolar op amp instead of the FET op amp he had on schematic while I was working on the breadboard.  This was (IIRC) in an expo converter.  I followed his intruction and used a 1458 instead of a TL072, and it worked.  For the hell of it, I tried the TL072, and quickly figured out he wasn't woofing.  On perfboard it worked fine, but not on the breadboard.  I suppose that was the one instance I've run into where the stray capacitance definitely wasn't my friend.

Thanks!  Wise words, which I have taken to heart.  The closest thing I have on hand to a 022 is a NJM2904, similar specs but intended for single-supply.  Just to be sure, I breadboarded an alternate version using a dual op amp to include the buffered voltage divider.  Tried 062 & 1458 with pretty much same results.     By now I'm pretty sure (judging by Geoff Hunter's testimony), that at least slew rate has little to do with the matter (022 = SL .5)

Scott "JFetish" Stites

Geoff: When i take my VB-2 out of its cabinet  Grin and plug it in, normally when I switch it on, I don't get any wobble.  However if it touch the rate and depth knobs (can't remember which one), then the  LFO springs to life and works perfectly fine after that!

It helps me a great deal to know that the circuit in the stock pedal behaves the same way (at least it's not something I'm doing wrong).  Thanks!  Now take your VB-2 out of the cabinet for a while for a great way to start out the New Year! 

George: It's this common type of tunable sine oscillator:

http://www.national.com/ms/LB/LB-16.pdf

LB-16, figure 2 is the easiest one to translate.

There few small twists which might throw you out:

- LB-16 fig 2, R1 is replaced with R45, R47, R48
- LB-16 fig 2 first inverting opamp, R5, R6 etc is replaced with the inverting transistor stage.
- No zeners, I think the inverting transistor stage is allowed to clip instead.

The whole thing is based around the well know bandpass circuit (second opamp in LB-16 fig 2).

Article should explain the rest of the details.


Edit:  Oh the basis for the band-pass circuit is a *bridge-T* network in a feedback loop.  Don't let R45, R47, R48 make you think it's twin-T.

Thanks George, this is great!  I had GOOGLED for sine wave generators and didn't find this.  Your comparison of the two circuits makes it very clear.  If I'm reading this right, the xstr (Q8) is putting out a rough sinewave which is then filtered by the bridge-T network (functionally equal to a twin-T but much easier to tune, IIUC now).  It strikes me that the frequency of the filter and the frequency of the sinewave would have to be moving in parallel for equal filtering at all pitches.  Is this the case?  ( Digression: If it is, I could see somehow adapting this to a pitch-tracking filter of some kind.)

My only question remains how to punch-up the oscillation potential so that is doesn't fall asleep now and then. (This might be of benefit to Geoff and other VB-2 owners who are just waiting to tear into their vintage pedals and upgrade them.  ) Anyway, looking at the 2SC945 specs, it's an AF xstr, not slow but not too fast.  Maybe a higher freq xstr, more prone to oscillation?

Thanks again to all you guys!

Moe Joe

[Joe Kramer]

Hey Friends,

Lab Report: Lowering R51, the resistor at Q8's collector, has helped with the oscillator lag problem.  Using a 22K instead of 33K, now when I change the RATE from fast to slow, there is a little loss of sweep depth for a fraction of a second, but then it catches up a returns to normal pretty quickly.   I suppose this change in the bias helps keep the xstr "falling forward" a little more easily, instead of arriving at a balanced stasis point.  A 22K also increases the sinewave amplitude somewhat, but I find this is easily compensated for by backing off the DEPTH control.

As far as I can tell this cures the problem.     Thanks again for all the helps guys!

Sined,
Joe

[DiyFreaque]

Cool beans!  Really nice to know you can pull it off with a different op amp (I did notice in the TL022 datasheet that it was not recommended for new designs).  And with little or no Moe Joe.  Unless those are, like, cork resistors or something.


Charrio,
Scott "Bipolarish" Stites

[George Giblet]

Dropping the value of the 1M resistor R53 should fix the problem without impacting on the too much.  Thelowering R51 trick is slightly more instrusive but achieves a similar task - I'm surprised the output went up so much (maybe it supposed to be high like that?).

[Geoff Hunter]

Geoff and other VB-2 owners who are just waiting to tear into their vintage pedals and upgrade them.  icon_wink

Just had a nightmare thought about performing surgery on my VB-2......... 

Perhaps behringer can make a VB-2 clone for £15!

Geoff

[Joe Kramer]

Just had a nightmare thought about performing surgery on my VB-2......... 

Perhaps behringer can make a VB-2 clone for £15!

Sorry--I didn't mean to give you nightmares!  I should be more careful about what I say.     Easy now, buddy, it was only a dream. . . .

Don't know what's on Behringer's to-do list ( though their shrewdness is not to be underestimated), but if  you're still looking for a VB-2 alternative, this DIY option is coming soon:

http://www.buildyourownclone.com/vibrato.html   

BTW: I intend to try out George's suggestion about lowering R53, just to cover all the bases.  Will report back when this is done. 

Joe

[Mark Hammer]

Not sure how the BYOC folks intend to get around the fact that the BA662 is a proprietary part made for Roland.

[puretube]

http://www.1sourcecomponents.com/partinfo/ba662.htm

[Geoff Hunter]

LOL i've put it back in its case for now.......

Looked on the web for BA662 info and found this: http://www.technologytransplant.com/, but there doesn't seem to be any more info on that web site.  Just a teasing statement........... 

Can't the BA662 be designed out if making a clone?  I think its just used to allow the LFO to be ramped up and down from Vcc/2 so that the transition from on to off is smooth? Is there another way to achieve the same result with an IC that can be got more easier?

[Joe Kramer]

Looked on the web for BA662 info and found this: http://www.technologytransplant.com/, but there doesn't seem to be any more info on that web site.  Just a teasing statement........... 

Yes, I've come across that at some point before.  Apparently all talk, unfortunately. . . .

Can't the BA662 be designed out if making a clone?  I think its just used to allow the LFO to be ramped up and down from Vcc/2 so that the transition from on to off is smooth? Is there another way to achieve the same result with an IC that can be got more easier?

Seems like a 3080 would work with some component changes, though it wouldn't be "original," FWIW.

BTW, I may be reading wrong, but it appears on the VB-2 schem that IC4 (the VCA) has TWO pins marked as "1," one of them going to ground, one of them to the emitter of Q9.   Is this correct?  A typo?

Joe

[byoc]

http://www.buildyourownclone.com/ba6110.html

This explains how to substitute the BA662A with the more common and much less expensive BA6110.  The VCA is not something you can amputate from the circuit.  The LFO doesn't function properly without it.

The BA634 (IC6) can be done away with completely if you're gonna build it true-bypass.  It's just a SIP5 flip-flop.  The rotary switch can be replaced with a SPDT where the pole is connected to the same place as the pole of the second half of the rotary switch.  The "unlatch" throw would still connect to the same place.  The "latching" throw connects to ground.  The fisrt half of the rotary switch can be done away with completely including the "bypass" mode buffer.  Q11/LED can of course go too.

There are errors in the "roland" schematic.  The duplicate pin1 to ground of the BA662 should be pin5.  Pin 9 is shown going to +9v.  It really goes to +6v off the emitter of Q7, but will still work just the same with +9v.

[puretube]

welcome to the forum, byoc! 

[Geoff Hunter]

Many thanks from someone who just ordered and received a couple of  BA6110's in the post last week 

[no one ever]

http://www.buildyourownclone.com/ba6110.html

This explains how to substitute the BA662A with the more common and much less expensive BA6110.  The VCA is not something you can amputate from the circuit.  The LFO doesn't function properly without it.

The BA634 (IC6) can be done away with completely if you're gonna build it true-bypass.  It's just a SIP5 flip-flop.  The rotary switch can be replaced with a SPDT where the pole is connected to the same place as the pole of the second half of the rotary switch.  The "unlatch" throw would still connect to the same place.  The "latching" throw connects to ground.  The fisrt half of the rotary switch can be done away with completely including the "bypass" mode buffer.  Q11/LED can of course go too.

There are errors in the "roland" schematic.  The duplicate pin1 to ground of the BA662 should be pin5.  Pin 9 is shown going to +9v.  It really goes to +6v off the emitter of Q7, but will still work just the same with +9v.

hey! nice of you to join us 


will you be offering your vb-2 boards separately? bravo on this accomplishment,btw, this is a landmark project.

[byoc]

Many thanks from someone who just ordered and received a couple of  BA6110's in the post last week 

I hope you didn't get your BA6110 from Newark/MCM InOne.  They've been knowingly selling defective BA6110's for almost a year now.  If you've got batch code 420 178 you should send them back.