tube interstage impedance bridging for frequency response shaping

[tapehead ted]

I understand that the output impedance of v1, say, and the input impedance of v2 form a potential divider. However, I keep reading about the effects of impedance bridging ratios on frequency response, but I've been unable to find an explanation of exactly what goes on as the ratio shifts from 1 to 1 to the sometimes recommended minimum of 1 to 5 to the more commonly expressed 1 to 10. I read that it affects the frequency response, but how?

I'm trying to design an instrument amp that has a very minimal phase shift in the guitar frequency band. I'd like to be able to approximate a gentle high cut and low cut by some other means than RC filtering, mostly because of my allergy to phase shifts. It seems like this impedance bridging ratio could help.

One reason I believe this is because I've used a very nice tube DI made by Manley that switches impedances to do just this.

Anybody?

 

[JMFahey]

my allergy to phase shifts

You'd better take some pills for that, or get used to it, because frequency response altering involves plase shifts.

By the way, how do you *know* you are "allergic" to them?
How do you even notice them?

 

[doozerdave]

I agree with JMFahey. You're not going to avoid phase shifting. Impedance is all RLC, so even if you're not using discrete C's or L's as filters the characteristic impedance of the plates/cathodes/grids is all because of C and L anyway, so you're using phase shifters.

 

[funk1980]

Impedance matching is not the way to tone shaping. Some capacitance or flux is always needed. By e.g. connecting a 10k ohm output impedance to a 10k ohm input, you get half the signal level, but with the same frequency response. The input- and miller capacitances of a tube are to be taken into acount, but extreme resistance values (several mega ohm) need to be used to let them have any 'tone-shaping' effect (large risistors = hiss, so try to avoid them).

It sounds like you're going to try to DC couple V1 to V2. A common cathode stage DC coupled to a cathode follower, when properly designed, can give very nice effects indeed, and gets rid of at least one cap.

Low cut can be achieved but choosing the right cathode bypass cap in e.g. V1, so the cap is not actually in the signal chain. By reducing the value, so only higher frequencies are bypassed, the lower frequencies will get attenuated by local feedback at the cathode.

 

[tapehead ted]

I developed this allergy doing purist minimalist recording of acoustic instruments. Using the same ultra-bandwidth equipment to record electric instruments, and performing with some of it, has let me hear how lifelike instruments can sound through a loudspeaker, and even distorted instruments can have a kind of vivid healthy sound that makes all the difference, I find. EQ was never part of shaping the brightness or darkness of the sound. It always sounded far more natural to make other choices, most notably just the gain structure of a path. The best sounding chains usually had sky high and gully low bandwidth.

Plenty of twiddling experience behind these proclivities, pristine and wildly overdriven... and just what feels right when playing- it all seems to go together. I do have a sense of when and how to tickle or hammer various kinds of gain stages, speakers, et al.

I believe gain structure will be, once again, the key to producing a sound that doesn't have me looking for a high cut.

I am definitely addressing all the miller capacitances and so on, and using a switchable coupling capacitor for a low cut (with a 3 hz option as well), but I'm sure there are other ways to darken the sound appropriately.

I note when turning a step attenuator ( I have one in my guitar amp), the sound is much darker and more solid feeling with any amount of attenuation at all. This is not nearly enough to swing the miller capacitance. It's a whole different sound than using the stepped RC filters in the EQ section. Those filters are actually rather bright and clean compared to with the EQ bypassed.

Is that not a relative impedance shift creating that very different spectrum of sound?

I'd like to be able to darken the sound without introducing a low pole. Not every means of darkening the sound comes with the side effects I'm trying to keep to a minimum.

That said, can we talk about the subjective effects of varying impedance ratios? I'm all ears.

Ted

 

[Struth]

Hi Guys

There is an implication that the "stepped attenuator" in your amp is one between the PA and the speaker. If so, then its effect on tone is well-known and expected. The interaction of the amp with the speaker is corrupted by the interposed resistance, isolating the reactive load from the amp and essentially killing the magic of driving a speaker with a tube amp. TUT4 explains how this happens in detail and also ways to patch it up to sound better.

In the rest of the amp, interstage attenuators are necessary if one is to avoid grid rectification. The alternative of using very high grid-stop values adds quite a bit of noise. Note that grid-stops and screen-stops should be present everywhere in MI amps, and should be of a reasonable value to do what they are supposed to. The engineering ideal of a zero value for them does not apply to MI as MI amps are typically used in an "abusive fashion" taken from the usual hifi perspective.

To darken sound simply requires a high frequency rolloff. MI speakers are essentially woofers with sometimes extended midrange response. This is the biggest EQ influence of the system. Anode bypass caps will reduce treble response, and as others have stated quite well, phase shift is present anywhere there is capacitance.

Impedance matching is a hifi concept that really does not apply to tone creation - although it can be used to that effect. Placing the "wrong" load on an OT tap provides treble rolloff or a bit of peaking. Our amps are designed to be used this way as it is perfectly safe. Impedance matching between stages won't be easy to take advantage of given the extremely high input impedance of a tube grid.

The other thing that is not quite clear from your posts is whether you are talking about the design of the MI amp itself or of part of the recording path?

Have fun
Kevin O'Connor

 

[tapehead ted]

There is an implication that the "stepped attenuator" in your amp is one between the PA and the speaker.

This is not the case in this amp- the stepped attenuator replaces what was a volume pot, with tremendous sonic improvement, I find.

"Impedance matching is a hifi concept that really does not apply to tone creation - although it can be used to that effect. Placing the "wrong" load on an OT tap provides treble rolloff or a bit of peaking. Our amps are designed to be used this way as it is perfectly safe. Impedance matching between stages won't be easy to take advantage of given the extremely high input impedance of a tube grid."

When the current limiting kicks in could be a good time for a high rolloff...

 

[tapehead ted]

And this is all envisioned to be taking place within a MI amp, not as part of a recording chain.

Thanks for your reply.

I'm seeing a possibility of sizing a grid leak resistor to create various impedance "matching" effects. Possibly a variable grid leak resistor (opto type, so as not to put an increasing resistance in series as well), or a switch to choose a couple values- a variation on what Valve Amplifiers calls a type "C" attenuator, with no series resistor other than the grid stopper.

The idea once again is to have some minimal phase shift alternative to a high cut. Any scenario where the tone darkens as the amp distorts seems likely to be that much more pleasant to listen to.

This is all being designed to work with a certain given 12" guitar speaker, and the cabinet design is addressing the frequency response as well.

My intuitive understanding is that the tone will be darker approaching a 1 to 1 interstage impedance ratio and brighter with a more typical impedance bridging ratio- again keeping the high pole way out there basically out of the picture. Is this the way that curve goes?

Thanks again.
Ted