Gretsch Controfuzz build report ...sound samples?

[LucifersTrip]

I finally got around to building the Gretsch Controfuzz and it really is a fun one.

If you're not familiar, it's a mix-in fuzz similar to a Fuzz Rite. The big difference is with a Fuzz Rite the clean single diminishes as the fuzz is mixed in, so when you're at 100% fuzz, the clean is gone. On the Controfuzz, the clean remains the same as the fuzz gets mixes in.

I decided to have some fun with the vero I won in the Turkeyshoot:
http://www.luciferstrip.com/fuzz/turkey-prize.jpg

...and Renegadrian made a tight little layout:




It's a very cool fuzz, but I'm surprised you can't put the fuzz higher in the "mix". I did read a bunch of old threads with similar complaints and some were actually annoyed by the clean signal being there at all. That's what makes it unique!  

The clean signal is great, but at full fuzz, mine sounded like a Fuzz Rite (just the mix, not the fuzz) when the fuzz is at about 40-50%. Is there something wrong with the layout? Did I make a mistake somewhere?

Unfortunately, once again (like the Dr Q voltages I asked about a few days ago), I couldn't find any working voltages for this thing...I couldn't even find a sound demo from an original!

With a 9.6V supply, I have ~ 4.8V on all pins except 4 & 8, which are 0 & 9.6V, respectively.

Anyway, since I couldn't find any errors on quick inspection and the circuit functions as I would expect otherwise (good fuzz, good decay, no hum, noise, etc), I looked for a solution.

Unlike others, I don't want the clean gone and I don't want the fuzz high in the mix all the time. I just want to be able to mix it in from low to very high at my discretion...to have the original Controfuzz fuzz + more.

Fortunately, the solution was so insanely simple, I'm almost surprised it wasn't designed that way.

Mark Hammer was pretty much the only one that did any type of proper build reports with some great info, but I didn't see anyone post this exactly:



For mine, at 0% (full ccw pt1) it seems identical to the stock version at 0%. It's basically a boosted clean signal with only the most miniscule fuzz barely audible in the background. Is that what an original sounds like with the fuzz pot at 0% (full ccw)?

At about 10% cw on mine, the fuzz mix-in is about where the stock version is at max (full fuzz). The final 90% is a slow fuzz mix-in and by the full 100K, the clean is barely audible over the super high gain fuzz. I wanted a slower transition, so I used a log pot there.

Edit:
just realized I forgot to try subbing a 100K pot for the 10K stock pt1...It works the same as the 100K var resistor to ground. So, is the 10K wrong in the schematic or did they not want the fuzz ever take over?

There are many ways to increase/decrease the clean & fuzz, so experiment away. The only thing I would do to improve this is to put a control on the clean signal, so I can turn that up and down at will, also. I did that for a Fuzz Rite and it's amazing.

[rutabaga bob]

I'm pretty sure there are sound clips through the runoffgroove or home-wrecker pages.

[LucifersTrip]

I'm pretty sure there are sound clips through the runoffgroove or home-wrecker pages.

nope...just clones, but thanx for checkin'

[Mark Hammer]

The original used 748 op-amps with compensating caps.  The comp cap in tthe clipping stage likely restricts top end, yielding a different quality tone. You may want to put  small cap in parallel with R3.

[LucifersTrip]

The original used 748 op-amps with compensating caps.  The comp cap in tthe clipping stage likely restricts top end, yielding a different quality tone. You may want to put  small cap in parallel with R3.

Thanx for bringing that up.

The info I read said that the original 748 needed 25pf's for "stability".  Did I foolishly read that to mean that with a common modern TL072 (or LM458) it wasn't necessary? Is what they meant by "stability", taming the high end? If so, why did Beavis leave that out of their clone?

I'm an opamp novice, but on the 748, the caps are across both + and out on both stages.


edit (I should've posted the original first):

[Mark Hammer]

Well, given the gain of the clipping stage (>2000), stability is an issue, but primarily in the upper end.  The comp cap keeps the gain in the upper end under control.  See the datasheet for what the gain bandwidth product is.

[LucifersTrip]

Like I wrote, I'm an op amp novice, so I don't know how to interpret or compare "gain bandwidth products", nor pull that info from a datasheet yet, unless it says specifically "GBWP".

I'll read some at wikipedia and some articles:
http://en.wikipedia.org/wiki/Gain%E2%80%93bandwidth_product
http://masteringelectronicsdesign.com/an-op-amp-gain-bandwidth-product/

But in the meantime, does anyone have a specific answer to either of my above questions:

Why did Beavis leave the stabilizing cap out of their clone?

Is placing a stabilizing cap across R3 (220K from "-" to out) as Mark Hammer suggested above the same as placing stabilizing caps are across "+" and out in both "stages" of the original?

thanx

[WaveshapeIllusions]

The newer opamps don't need usually need compensation caps, since they are more stable.

You could get a similar lowpass effect by adding a cap in the opamp feedback loop. I'd suggest the same value as the compensation cap as a start.

[LucifersTrip]

The newer opamps don't need usually need compensation caps, since they are more stable.

what is the definition of a stable op amp, relating to a circuit like this?

why use any compensation cap if not needed, unless the cap is doing something else in addition to stabilizing?

[WaveshapeIllusions]

The newer opamps don't need usually need compensation caps, since they are more stable.

what is the definition of a stable op amp, relating to a circuit like this?

why use any compensation cap if not needed, unless the cap is doing something else in addition to stabilizing?

A stable opamp would be one which doesn't oscillate at higher gain. My knowledge of opamp internals is a bit limited, but I would assume it has something to do with the gain crossover. That is, when gain gets cut low enough to avoid oscillation.

At high frequencies, the delay through the circuit nears 180°, which would turn negative feedback into positive. Most circuits use a cap to cut the HF gain to keep it from oscillating. I would assume they couldn't fit the cap on the die, so it was made external. Newer circuits probably have the caps on the chip.

Since the cap serves as a lowpass filter, it may also smooth out the tone a bit depending on where the cutoff is. At 25 pF, it would be pretty high. It might mellow some of the upper harmonics, so it'd be less harsh.

[Mark Hammer]

Re: gain/bandwidth product, and open-loop gain....

We set the gain of op-amps by providing varying amounts of negative feedback from the output back to the input.  The more feedback we apply, the lower the gain.  The "open-loop gain" is how much gain the chip can provide when all negative feedback is removed (i.e., an "open" feedback loop).

That gain depends on the frequency of the signal being amplified.  The higher the frequency, the less gain can be provided.  If you set the gain low enough, then the chip can provide that amount of voltage gain across the entire audible spectrum.  But as you start to boost the gain (by providing decreasing amounts of negative feedback), that gain starts to be applied across a more and more restricted part of the spectrum.

This is why we talk about "gain-bandwidth product".  It designates how much flat bandwdith you can expect when you apply this much gain, that much gain, and so on.

If I look at the datasheet for the LM748, I see that with a 5pf comp cap, the most I can aim fror in terms of flat bandwidth is 40db (gain of 100x) at 100khz.  If I increase that cap to 30pf, I see that the most flat bandwidth I can hope for at that same gain is 10khz.  I can get 60db gain (1000x) at 1khz and 80db gain (10,000x) at 100hz, but not a whole heckuva lot at 10khz.

Now, consider that, with a 220k feedback resistor and 100R ground-leg resistor, the aimed-for gain of that stage is 2201x.  The gain-bandwidth plots in the datasheet tell us that we're dreaming in technicolor if we think we're gonna get that sort of gain for stuff in the 4khz and up range, using a 25pf compensating cap.

Now, I have no idea how the chip will behave when asking for that sort of gain in the upper mids and treble, but for sure it isn't going to behave normally.  If I look at the datasheet for the NJM4558, I see that I can expect a gain-bandwidth product of about 5okhz at 40db gain, and at a gain of 2000 (= 66db), I can expect a usable bandwidth of around 3khz or so, before I start losing top end.  For a 5532, I can actually get that 66db gain at up to 10khz.

All of which leads me to say that I have no bloody idea whatsoever about the extent to which a feedback cap in an internally compensated op-amp mimics the compensation cap in the externally compensated LM/ua748.  But suffice to say that not using ANY feedback cap in parallel with that 220k resistor - when using an internally compensated op-amp like a TL072 or JRC4558 - is unlikely to nail the tone of the original.