'Tone Shaping' Between Stages

[exclamationmark]

I realize that the attenuation of certain frequencies between stages has an immense effect on tone. I plan to experiment a little with LED/diode bias to see what sort of sound it gives (also in part due to my irrational hatred of capacitors), but I notice that this sort of removes the option of tweaking the sound through the shelving filter formed by a partially bypassed stage.

When I look at most "high-gain" designs, they use coupling capacitors of .02uf, which provides a -3db rolloff at frequencies of around 7 to 20hz depending on grid leak/stopper values, most of the time not even being able to attenuate ANY of the signal of a low E string (or at some absurd value like 0.01db).

Is this simply because of the shelving filter attenuating/not boosting the low frequencies, so essentially all the coupling capacitors do is... couple? So in a diode biased design would it be wise to reduce the value of the coupling capacitors to at least get some low frequency roll-off? I'm thinking if I actually employ the coupling capacitor as a tone shaping tool I can make a reasonable approximation of a shelf filter. Obviously they will sound different, and I don't mind that (I'm not looking to develop any 'classic' sounds here), as long as it sounds good.

Oh and I attached a rough schematic of an example gain stage. If you notice, the capacitor between the grid and the cathode is there to make a low pass filter, in place of the miller capacitance filters usually seen. This is because the miller capacitance of the valves I'm using is only 15-20pf so it's not actually possible to roll-off audible frequencies without the additional cap. As you can see from the schematic, the grid leak/stopper are configured in a voltage divider arrangement... My question is, how exactly do you calculate the roll off frequency in this arrangement? Is the total resistance value simply the parallel of the grid stop/leak resistors ?

[Steve Conner]

I've used coupling capacitors down to 4700pF in high-gain preamps. There were also amps (Selmer Treble'n'Bass come to mind) that used a tiny coupling cap in the "Treble" or "Normal" channel, instead of Fender's classic approach of the bright cap on the volume control. Can't remember if it was 4700 or 470pF, but it was pretty small.

In other words, coupling capacitors are commonly used to shape the frequency response, so I don't know where you got the idea that they are all 0.02uF. They're not.

You can make a coupling capacitor into a shelf filter. Just use two coupling caps in series: the normal sized one and the undersized one. Put a 4.7M resistor or suchlike across the undersized one.

[exclamationmark]

In other words, coupling capacitors are commonly used to shape the frequency response, so I don't know where you got the idea that they are all 0.02uF. They're not.

I was specifically looking at the SLO100/5150 schematics and noted that they use the same coupling caps. Although since they're practically the same amplifier, this is rather a moot point. I know that in other designs they use the coupling caps to change frequency response, but apparently not so much in these particular amps, and I'm assuming the partially bypassed design is why.

[Wilder Amplification]

I was specifically looking at the SLO100/5150 schematics and noted that they use the same coupling caps. Although since they're practically the same amplifier, this is rather a moot point. I know that in other designs they use the coupling caps to change frequency response, but apparently not so much in these particular amps, and I'm assuming the partially bypassed design is why.

Most amps use the partial bypass technique for the majority of tone shaping so it's OK to use those values. But if you can't partially bypass a stage due to LED/diode bias, you do it in the interstage coupling.

The difference is that when partially bypassing a stage, the shelf isn't a continuous shelf like it would be with shaping via inter stage coupling. The bypass cap shelves between bypassed and un-bypassed gain whereas shelving via coupling caps provides a continuous shelf.

I've used coupling caps as low as 1nF (0.001uF or 1000pF) in conjunction with partial bypassing. Yields excellent results when you're limited on gain stages (i.e. customers who don't want the extra hole punched to add an extra preamp valve).

For shelving highs you can use local feedback on a gain stage as well.

Diode biasing works very well for stages that need to be full band pass, which can be implemented later on in the gain chain.

[kg]

you might want to take a look at a schemo of my preamp

http://ken-gilbert.com/images/pdf/newpreamp.pdf

as it has quite a few cathode LEDs for bias

i haven't made any substantive changes in the past 9 years, so it's still pretty accurate.

obviously this wasn't the first iteration of a preamp. at some point during the design i basically threw out everything i had learned about guitar preamp design and just started experimenting. the output goes straight into the BAGA (a very large all tube power amp).

[tubeswell]

Thanks for sharing that Ken. (Looks like a gain monster). Curious about what it sounds like? (Got any soundbytes?)

[kg]

here's one, a few (too many) years old:

http://ken-gilbert.com/audio/920last.mp3

it's not a song or anything.. just a jam. we never played it before or since!

past the halfway part you get a little cabinet resonance.

[exclamationmark]

Thread resurrection!

First off, thanks for the help everyone, I actually got my pre-amp to sound decent with 'only' 4 gain stages (I've got another 2 tubes left sucking up the 48v heater voltage, 4 tubes total in series). One neat thing I've found is because the gain of the 12au7's is so low I haven't had to dump any excess signal.

Pondering the idea of shelf filter I went and researched a bit and came up with a design that I *think* would give me a slight mid boost. If I've calculated things right it should give me a 6db shelf at around 1khz relative to 500hz. Then the 150pf cap across the grid/cathode in combination with the grid stopper should start rolling off highs at 4khz~. In theory....

As I understand it, the shelf filter works as a voltage divider in combination with the grid leak, with the small cap acting as a bypass. For any frequencies above 1khz, the 1.5nf capacitor should have practically no impedance, effectively short circuiting (ignoring) the 220k resistor in parallel with it, so it shouldn't interfere with my predicted 4khz high roll off circuit, should it? Basically, I'm a little bit confused with how all these things interact with each other.

I'll probably build it and find out first hand, but since I'm ordering some more components for other projects I thought I'd ask first to avoid getting the wrong oddball values for the smaller bypass caps in case I'm out by several orders of magnitude .

[guitarmike2107]

Well to help you along you can either download and use a spice sim program, or take a look at the calculators on ampbooks.com

There is a bright cap calculator which is essentially what you have there.

I could look up the correct formula’s if needed but the web stuff makes it straight forward.

BTW I have used coupling caps as low as 500pf, the -3db roll off point caused by the coupling cap is dependent on the impedances around it.
Those calculators show that too…..

Mike

[exclamationmark]

I had a bash at using LTSpice, and the frequency response looked mostly like I expected, except the peak was an octave higher than I expected... I'd actually been avoiding using sim programs because they looked to be complicated, but it actually took me all of 5 minutes to work out how to use. Thanks!

[guitarmike2107]

Cool, I am no spice pro, probably worse than yourself, but I did find these tutorials helpful.

gaussmarkov: diy fx » An LTSpice Tutorial
gaussmarkov: diy fx » LTSpice AC Analysis with the BMP Tone Stack
gaussmarkov: diy fx » LTSpice Analysis and the DOD Overdrive 250