How does one guesstimate what sort of series resistance will yield useful sag?

[Mark Hammer]

I was tinkering with an old AMZ Mosface last night, and thought I would throw a couple more options into it.  One was varying the feedback resistance in the Q2-emitter/Q1-base path as ZVex does in the Woolly Mammoth  (I used a 3-way toggle to add either 360k+330pf, or 180k in parallel with that, or no additional resistance to the stock 100k.  All positions yield pleasingly different and useful tones.  Recommended.) 

The other was a 3-postion "sag" control.  I started out with 47k and 10k, I series with V+ and these were clearly major overkill.  47k killed all sound, and 10k got me a sputtery mess (louder, for some inexplicable reason, but still sputter, and not especially usable). 

Obviously, I'm going to have to reduce the values when I get home.  But what should I aim for?    I was pondering just using a pot and noting what series resistance yielded useful differences, but laziness demands (well, actually, laziness says "If you're up, and going into the kitchen anyway, could you bring me back...?") that I have a hunch of what two additional resistance ranges I might introduce, on top of the stock unrestricted +9v.

Should I stay well within the 1k range?
Does the series resistance of relevance depend on the current draw of the circuit?
Is there a rule of thumb for how much more series resistance one could aim for? (e.g., make the most sag 3x the resistance of the moderate sag)

[knutolai]

most starve controls I've seen in commercial products are either a 10k or 5k pot in series with the supply. No idea about any rule of thumb though

[midwayfair]

On the 'face on my board, it's a 10K for the starve -- and I have to rebias below halfway to get any sound at all (I actually run it on a slightly low voltage, biased properly, because it works like a tone control turned down a hair, and I use an active pedal in front of my fuzz, but that's another show).

But that's a BJT, not the MOSFace. I found the MOSFace to be waaaaay more sensitive to being biased properly than a BJT version -- a little variation on the trim pot and Q2 would just sputter out completely. I would guess that it's not going to be overly happy being starved no matter how you do it.

[GGBB]

Does the series resistance of relevance depend on the current draw of the circuit?

Taking a stab ...

If voltage sag is merely about lowering the voltage into the circuit via series resistance, and a dying 9V battery is the model, then we want to drop the voltage by about 2V (arbitrary) for that circuit. Since adding a sag resistor essentially forms a voltage divider, and the voltage divider rule tells us that the ratio of the resistors is equal to the ratio of the voltage drops, if you want a 2V sag on a 9V circuit, your sag resistor should be 2/(9-2) * circuit resistance.  Ohm's law tells us that the overall power supply resistance of a circuit will be equal to the supply voltage over the current draw (R=V/I), so from there we can calculate the resistance required for whatever sag voltage we desire.

In other words yes.

[R.G.]

The ideal resistance... (wait for it ) ... varies.    

"Sag" isn't one thing. It's the combination of how the circuit responds to lower voltage from its power supply and higher impedance, and those two actually do different things, depending on the details of what the circuit does inside.

If it was as simple as faking a failing battery, the answer would be easy - measure a battery as you methodically drain it and then do that. But once failing batteries are modelled and pots substituted, people are off to the races on sag, and this may be beyond what real batteries do, and may have other sonic effects that sagging batteries don't.

Yep, it's the humans that have made this hard.    

To complicate matters, the AC part of the sag equation - the internal impedance of the battery to AC as it ages - can be made irrelevant by local power supply bypassing, filters from +V to ground in the circuit itself. If they're big enough, the "sag" will be like a low impedance power supply voltage that may vary over time as it's loaded, or even not that if the filter is big enough. All those 1000uF caps to eliminate RF and power supply hum and such are working on the results of sag too. And most people don't mention or notice them as doing something other than eliminating hum.

[R.G.]

Simulpost

[PRR]

> AMZ Mosface .... what should I aim for?

Link to schematic??

http://www.muzique.com/schem/fuzzface4.gif

Ah, thanks!!

This circuit "normally" runs about 0.6mA. It can't possibly suck more than 9V/(1K+5.6K+1K), or 1.18mA (plus a wee bit for Q1) because that would imply Q2 is totally slammed. It probably does not run a LOT less, like 0.1mA, because there would not be enough voltage across R2 to get any good output. Dart-throw half of that, call it 0.6mA.(*)

Now if a thing takes 9V at 0.6mA it is somewhat like a 15K resistor.

What is "sag"? As R.G. sez, it..... depends.

Some circuits crap-out with the least sag. Others work more/less well down to quite small voltages. This can be predicted from study, but thinking is hard.

If we want to "sag" a 15K resisty-thing down to half, we use a 15K resistor.

A 7.5K resistor will sag it down to 2/3rd; a 30K sags to 1/3rd.

So I'd gather parts from 5K to 20K and have a trial session.

(*) Actually we know from casual inspection (if we casually analyze circuits) that the idle current *IS* just-about 0.6mA. Q2 bias is set by Q1 Collector, and Q1 Base watches Q2's Source. The only way it is happy is with ~~0.6V across 1K R5.

[composition4]

Simulpost

Watch yourself R.G, Randall Smith could be reaching for his lawyers' speed-dial right this very moment...

[greaser_au]

bizarre how what was once a limitation of the available technology is now a desirable element.... 

david

[Mark Hammer]

So, last night I dickered around and chopped the resistances I used for the 3-way toggle by a factor of 10.  One of the toggle settings now imposes 4k7 in series with V+, and the other settings either bridge the 4k7 with straight wire (stock) or a 1k resistor (combined parallel resistance of 825R).

The 4k7 is still a little sputter, and might be a little better at 3k9.  Interestingly, the sputteriness is at maximum when the gain is up high.  Turning down the gain improves its behaviour noticeably.  The 825R setting is pleasing, well-behaved, and different-sounding than stock.

I wasn't aiming for any particular sound, just some variety of usable tones.  Between the 3 different feedback resistances, the two added current-limited settings (and I hesitate to call them sag, because I can't really hear anything "sagging" or sounding more compressed), and a third toggle to select input cap values, there is a fair mount of tonal variety, though, for a two knob fuzz.

Based on Paul's and RG's posts, though, clearly the starting point is to know how much current the circuit normally draws.