Audio probe, BBD, Zombie--need advice.

[gorohon]

I've made R.G.'s audio probe and was using it to poke around on my dead zombie.  Everything was O.K. until I probed pins 7&8 pf the BBD then through the filter section until the signal got mixed.  I thought that the BBD was fried, so I put it in my working Ultra Flanger.  No problem there.  My guess is that there is a voltage/bias problem here.  I'm wondering, what else could cause the signal to be lost through the BBD when it is not broken?

[Mark Hammer]

Bias is a good guess.  The Zombie uses a single bias voltage to serve as both bias voltage for the BBD and Vref.  As has been discussed here on many an occasion, that's an elegant compromise, and it certainly pares down the parts count, but it doesn't always work well.

There are a few ways you can tackle it.  If you look at the schematic for the CE-2, you will see that Boss does what Hollis does, and that is to insert the bias voltage earlier in the audio path, and make sure there are no DC-blocking caps between that point and the input to the BBD.  Boss inserts the bias voltage or the BBD at the input to the filter just ahead of the BBD, so that the signal rides on the bias voltage through the lowpass filter and arrives at the BBD input on top of that voltage.

At the same time, Boss uses a separate V+/2 bias voltage for the LFO, where Hollis uses just the one bias for every part of the circuit.

If you look at the Small Clone schematic (try the Heladito at Tonepad), you'll see that a standard V+/2 bias voltage is used throughout the circuit except for the BBD, but here too the bias is inserted just at the input to the lowpass filter before the BBD.

Looking over a variety of schematics, the smart thing to do is probably to set the Vref resistors at 10k/10k for the existing board (instead of a non-midpoint Vref using 10k+15k), and supply a separate bias voltage to the BBD at its input (pin 3) on some kind of little daughter-board.  Use a 10k trimpot, with a 2.2uf-10uf cap from wiper to ground, and a 22k-100k resistor from wiper to BBD input pin.  The noncap end of the trimpot goes to V+, and the other to ground.  The wiper feeds the BBD a bias voltage between 9v and 0v.

Since there already is a bias voltage fed to the input of IC1a, you probably want to "reset" things by sticking in a DC-blocking cap between the output of IC1a (pin 1) and the input of the BBD (pin 3).  This will let the existing Vref come along for the ride with the dry audio signal to the mixing point, but cut it out for the BBD, which will have its own bias voltage.

That should let you dial in whatever a suitable bias voltage is for that BBD chip, without producing problems elsewhere in the circuit.

[gorohon]

Thanks, Mark.
I tried doing this external trimpot setup before, but I did not use a DC blocking cap.  I think I'll give that another try (the right way) and see what happens.   BTW, I'm going to build the Tonepad Small Clone.  I'll keep the zombie for experimenting.  Plus, it would be nice to do some A/B comparison between the two.  I'm sure that the Small Clone variation will be more refined, quieter.  That said, I wonder what the minimal filtering lets through on the zombie besides noise--anything special? Probably not, but I'm curious.

[gorohon]

Mark,  It's working now.  I used a 1mF electrolytic cap to block the DC voltage.  I've got the bias dialed in for the least possible distortion. It is the quitest that its ever been.  What I'm wondering now is whether I should put another decoupling cap after the BBD pins 7&8?  There's still some ticking. I have not used any shielded wire, so maybe It would be wise to shield the input to stompswitch to board then the wiring to the LFO speed pot.  I'd like to know what you, or anyone else, has to think about this.

[Mark Hammer]

Whenever you have HF clocks and audio living under the same roof, shielding is recommended - a bit like the good-fences-make-good-neighbours maxim.

Whenever you have bias or reference voltages, you need to decouple them from the overall supply with a suitable cap, like 1-10uf between the reference point and gnd.  Think of it like a teeny-tiny uninterruptible power supply.  That cap smoothes over any momentary draw on the same power source from another part of the circuit, by providing an itsy-bitsy reservoir of current stored up in the cap.

This is kind of crucial when there are LFOs involved.  Although we think of the LFO as producing a triangle waveform - something that ascends and descends gradually - in reality it is a squarewave oscillator. The rising and falling edges are smoothed out by an integrator stage.  What that means in practical terms is that when the square wave shoots up from the negative to positive half-cycle (or vice versa), there is a big sudden draw on the current.  That "ticking" you hear is the electronic equivalent of trying to take a shower when other people in the house keep flushing the toilet, and turning the dishwasher and washing machine on and off.  If the shower had its own reservoir of saved-up hot and cold water, you wouldn't notice the variation in water temperature, despite what everyone else in the house is doing. (And yes, this IS autobiographical! :lol:  ).  Sticking a decoupling cap on both the Vref subcircuit and the bias circuit for the BBD, and sometimes individual chips (e.g., the usual 100nf cap you see between V+ and gnd) helps to give everyone their own water supply that will not interfere with others.

make sense?

[gorohon]

Thanks, I need to get my terminology straightened out between decoupling and blocking.  I used the 1mf cap to block the dc voltage from IC1A to the BBD. The daughter board has a 10mf electro for decoupling.  So, I was mixed up on my terminology.  I'm wondering if I need to block any voltage that might come out of the BBD through the signal chain from entering IC1B.  That said, would it be wise to go ahead and make independent vrefs for the LFO, even though the main is set at  V+/2 and the BBD now has its own bias? Or, can a decoupling cap just simply be run from the vref point on the IC to ground?  So, instead of independent vrefs, just decouple at the pin as well as decoupling at the main point where the V+/2 is set?  I guess that I'm wondering if decoupling vref's always have the two resistors with the decoupling cap from the center of the two resistors going to ground, like I see with your detick mod & the zombie's main V+/2?  Could it be that simple as putting an extra cap + on the vref and - to ground at the individual points to have their own reservoir?

In plainest of words, can I give the vref points a reservoir without having
to make each one have indepenent V+/2 sources?

BTW, I finally made a "Y" cord to run into my amp's effects loop, and the zombie did not tick!  It was still a hair swooshy, but no ticks  Used at the front of the amp, it still has tiny ticks :?  The problem with running the zombie and other effects in the effects loop is that a huge popping sound occurs when engaging or disengaging an effect.  I think that Crate didn't do some tying down at these points, but I'll have to go somewhere else to ask that one (OT).

[Mark Hammer]

The short answer is yes, although other folks like Tona dn RG will have much more to say about whether this would deliver the lowest possible immunity from supply-line noise.

Take a look at the schematic for the Boss BF-2 and you'll see that it has separate Vrefs for the audio path and the LFO, and a third bias for the BBD.  The CE-2, on the other hand, has a single trimpot-adjusted bias for the BBD+audio path, and a separate V+/2 Vref for the LFO.

"Blocking" caps simply prevent DC from being treated as if it were audio signal.  The input cap on any pedal is there because you have no idea if there is any DC mixed in with whatever you're feeding it with.  That cap rolls off bass as well, but in with that bass is the lowest possible frequency - DC.

A decoupling cap, on the other hand is like the hypothetical showerhead reservoir I described earlier.  Its job is to smooth out fluctuations in power supply demands so that the audio path is not impacted this or that part of the circuit requiring an instantaneous shot of current from the same battery/source.  It will always go between ground and the power-supply connection.

Bias/Vref circuits are best thought of like mini-power supplies.  They divide down your battery power, the same way a wallwart divides down the wall supply.  In that respect, the junction between the two dividing resistors, or the wiper of any trimpot fulfilling the same role, IS your "power supply" connection.  It just happens to be a lower voltage than the main supply, and is being fed to one of the inputs to the chips rather than to the supply pins.  It benefits from a cap to ground just like the chip's supply pins do.

[puretube]

some observations from my side, here:
the fact that you hear no ticking in the FX-loop, doesn`t mean that the Zombie doesn`t tick anymore: it just doesn`t get amplified that much as it would be hooked to the amp`s input...

Imho, the "ticks" are not neccessarily Vref-related in first instance, but more "B+", i.e. "rail-to-rail" clicks, that do their destructive work firstly here, and secondly get transferred through the resistive divider to the Vref.

So to me, the most important thing is to de-couple the local "B+" of the LFO from the rest of the circuit by R&C (electrolytic).
Of course this should go hand in hand with a thoroughly designed routing of the PCB-traces (similarly to "star-grounding", use "star-supplying").

Important for an LFO-Vref is a low impedance of the R-divider, where I prefer to also shunt the upper R with a big cap (B+ to Vref).

The other source of ticking can be wires or traces from the Schmitt-trigger-output-part of the circuit, which induce the spikes into the audio-path. This concerns the leads of the speed/rate/frequency-pot and or possible tri-/sqr-wave switches, as well as LEDs when hooked to the sqr output. Keep those wires/traces as short as possible and away from the signal route.

Another thing: the higher the clock of the BBDs, the more current it sinks.
This will shift its local B+, if that one is provided by a seriesresistor from the main power supply.
Also, the BBDs tend to shift the out-put DC level on which the audio "rides" , with rising clock freq.
(More with the older SAD.. and TDA.. chips than with the MN..s).
This calls for a DC blocking cap before the next stage (and not to forget to bias that stage anew).

[gorohon]

Thanks to you both.
I realized that the effects loop didn't cure my zombie, just that it worked well there, minus the popping in and out.  BTW, the same tick reduction occured by putting my EPFM compressor before the zombie, when going into the front of the amp.  Again, this isn't a pedal fix, just a setup that worked well; however, in the dirty channel, going through the front isn't quiet at all, no matter what you use.

So, am I to understand that there are limitations to eliminate ticking altogether, due to the layout of the zombie PCB?  

Puretube,  to block the DC voltage after the BBD, would you recommend blocking after the filter stages before IC1B, or directly before the filter stages?  All of the other stuff you said was on a higher level than I'm able to understand at this point.  I'll have to study it all some more.


This is the next plan of action:  shield input wires, wires to the LFO speed pot,  maybe look into shielding a cap in the LFO; decouple outgoing DC voltage from the BBD.  I'll see what improvements there are and let you'all know.

[puretube]

I`ll try if I can find the schem and/or the layout...

[gorohon]

Now I've gone and done it.  I've made my Zombie 100% tick free

Here are all my mods, thanks to all that have given me advice/ideas on the Zombie especially Mark Hammer and Puretube:

--V+/2 using two 10k resistors for R12&R13.

--Set independent bias for BBD (NTE1641/MN3007) using a trimpot on a daughter-board.  See Mark Hammer's description for this.  On my board I used a 10k trimpot, 100k resistor, and a 10mf electro cap.  I cut the trace after U1a pin #1, R4, and R5 that leads to pin#3 of the BBD.  There is a long single stretch of this trace where you can easily cut a small gap, drill two small holes and install a dc blocking cap.  I used a 1mf cap to do this, negative side towards the BBD.  Make sure that you figure out where you have enough space on the top of the board for your cap, that's where you'll want to cut the trace.  Also, make sure that the bias point for pin#3 of the BBD is between the cap and that pin.  I adjusted it using my ear to tell me when it was the least distorted, then zeroed in on the sweet spot.

--I shielded all three wires to the LFO speed pot, shielded the entire input from jack to stompswitch to board.  Finally, I shielded C8 (0.01mf) using a foil and electrical tape sandwich surrounding the cap.  I managed to solder  a ground wire to the foil (very hard).  If you do this, make sure that the foil and cap are insulated from each other as well as the surrounding parts.
I didn't have to shield any other wires, in my case.  I don't know if shielding the cap was necessary.

The results:  Absolutly no freakin' ticking, even when put into the front of my amp in high gain mode.  There is only the mild swoosh/hiss BBD sampling noise that is left to take care of now.

[puretube]

fine! - congrats!

[gorohon]

Hey,
I've made some illustrations using MS Paint showing what I did (sans the shielding stuff).  I wouldn't mind posting it somewhere so that anyone interested could look at it.  The problem is that I used R.G.'s images, so I'm reluctant to do anything unless it's alright by R.G.

[d@vide]

can you send me your image??

[gorohon]

Sure.  Check your messages, d@vide.