tube preamp input impedances

[ec]

I'm modding a Roberts/Akai reel to reel for mic pre/DI duty. I'm generally following RodC's Boris schematic, but I want both of the hi-z inputs (i.e., both the EF86 and 12AD7 channels) to look like instrument DIs. In the Boris, the input impedances are given as 100K and 500K for the EF86 and 12AD7 channels respectively. My plan was to make both of these 1M. Here's a sketch over the original Boris schematic:



Changes circled. Compared to the Boris: In the EF86 stage, this introduces a new 1M resistor, Rinput. In the 12AD7 stage, it replaces the 500K resistor R10 with 1M. I assume this won't matter to the tube, but will it change the impedance seen at the mic input? (Assume the same transformer.)

I think there's something I don't get here. Cinemag gives the impedance ratio for the CMMI-7C as 200 (or 50) : 9.8K. I'm reading that as saying a 9.8K impedance across the secondary will be reflected as 200 (or 50) ohms to the microphone. Then I downloaded the data sheet for the CMMI-7C and it says it's "designed for 97.6K load on the secondary." That's an order of magnitude difference---do they mean something else by this?

The turns ratio is 1:7. Suppose we want the mic to see 200 ohms off the 200 ohm tap. Flipping it around to 7:1, the square of the turns ratio is 49; 49 * 200 = 9.8K. But the Boris has it connected to a 100K volume pot. (Much closer to the 97.6K figure?) Wouldn't that reflect a ~2K impedance back to the mic? Wouldn't the 1M input impedance I sketched reflect a ~20K impedance back to the mic?

Like I said? I'm getting the impression that I don't understand this correctly. Thanks for any assistance you all can provide.

[floid]

someone else will hopefully chime in, but as i understand it this follows the 1:10 bridging rule of thumb. calculate two 200 ohm resistances in parallel, then 200 parallel with 2000, to get an idea of the principle.

[ec]

Yeah, I was wondering if that's what it was, given the 10x rule of thumb and 10x discrepancy---i.e., that I didn't understand what they meant by "200 ohm input." Let me try this again.

In the Boris, the EF86 channel is connected like this:

trafo secondary -> 100K -> grid

Suppose the mic input is wired to the "200 ohm" tap on the primary. Calling this a "200 ohm input" means that it's designed to take a mic with a 200 ohm output impedance, NOT that the impedance seen at the jack is 200 ohms, correct? (That is: the actual impedance seen by the mic at that input will be ~2K, as my math suggests, which is 10x the intended mic impedance.)

Assuming the above is correct, and that my 1M resistor would reflect a 20K impedance back to the mic, should I be looking at using a different input transformer? Say, a 1:20: by the above math a 1M impedance on the secondary would be reflected back as a 2.5K impedance to the mic.

[Edit: Is showing a 200 ohm mic a 20K impedance even a problem? Or is that high enough to be worrying about noise/levels/etc.?]

Thanks again.

[The Scum]

I think the thing you're missing is that you're adding a second resistor in parallel - on the EF86, the 1M in your schem is in parallel with the 100K input attenuator. When they're in parallel, the smaller resistance dominates.

100K || 1M ~= 100k (read this aloud as "100K in parallel with 1M is approximately equal to 100K")

I think what you're trying to do is accomplished if you swap out the 100K pot for a 1M one...but then the load on the transformer is incorrect...but a slug resistor and normalled jack could solve that.

This also assumes that the EF86 grid is significantly higher Z than these resistors - otherwise, it's another parallel load.

[ec]

Gah, you're right about the parallel loads. Thanks! Yeah, I'm assuming the tube grid impedance will always be very high.

After further thought, I'm wondering what would happen if I just used 500K pots on both channels. (Edit: this isn't a pot on the second channel, it's R10.) I got to wondering what's "too high" for a general purpose mic pre, and noticed that my RND Portico preamps are 10K on their mic inputs. (Other notes: the Gyraf G9 looks like it presents a similarly high impedance through a similarly ratio-d transformer; Gordon mic inputs are 2M.) So impedances of >10K wouldn't seem to be a problem in and of themselves.

If I'm thinking about this right, 500K on the secondary of that Cinemag would reflect 10K2 on the high tap, and ~2K5 on the low tap. Which sounds like it could be pretty useful. Meanwhile, 500K should be fine for the DI. (I could keep the 1M input on the other channel, too.)

Thoughts? Again, much appreciated. This is the last issue I'm sorting out before ordering parts. I'll be sure to document/share the process.

[emrr]

The problem with higher pot values is you lose increasing amounts of high frequency when the pot is turned down any significant amount. You want to use the smallest value that's reasonable for necessary loading. As well, many tubes have a minimum grid resistance which is easily looked up, and should be considered. Might or might not operate fine when higher in value.

[ec]

The problem with higher pot values is you lose increasing amounts of high frequency when the pot is turned down any significant amount.

Yeah, I was thinking about this. I play electric guitar, so it's a familiar phenomenon. As long as I'm asking questions:

(warning, brain dump. i gave myself a homework assignment, hopefully helpful to somebody else...)

In a guitar, the pot (a resistor R) together with the cable (a capacitance between signal and shield C) forms a low pass filter, and the cutoff is inversely proportional to (R * C), so if you raise either of them (turn the volume down, use longer/crappier cable) the high frequencies roll off, right?

I was thinking in this case, that at least on the mic input, the capacitance of the wiring to the transformer could be kept low enough to allow a higher pot value. Then I remembered that in other circuits (suppose there's no knob here) there's often two resistors, one from signal to ground that sets the input impedance, and the grid stopper, in series with the tube. The grid stopper (R) plus the tube's Miller capacitance (C) is used as a low pass filter to block radio frequencies.

It looks like in the Boris EF86 channel, the pot is doing both. (Tell me if I'm getting this right?) Normally wired, the pot will always show 100K to ground. (the input impedance) Depending on where the knob is, it will show between 0 and 100K to the tube. (the grid stopper) The issue of the volume knob attenuating high frequencies is a consequence of the pot introducing resistance into the signal path and raising R, correct?

Next question is, when turning down drops the treble, where's the capacitance coming from? The wiring/the tube itself? I assume it's both, but don't know if they're additive, or what. At any rate: I'm sure it's dependent on other variables, but these folks got 10pf from their EF86. After playing with the online calculator for a bit, it looks like the pot values we're talking about are already pretty close to the line: e.g., at 500k and 20pfd, the cutoff is ~16k, well within the audible range. So regardless of the wiring, it seems reasonable to think this alone could be a cause for concern.

If I'm getting this right, it seems like any normally wired volume pot at this point---between the input and the tube---will place an upper bound on both the input impedance (because it shows a constant resistance to ground---the parallel loads issue) and, at least when not dimed, the frequency response (because it will introduce resistance into the signal path, i.e., raise R). It seems like this isn't a problem in the B channel of the Boris, because the knob goes between the two stages of the triode. You can set the impedance of the channel via R10, independently of the value of the pot, so the two values aren't joined at the hip. Does that sound right?

I thought about a few approaches:

1. In a guitar, people sometimes use a capacitor to bypass treble frequencies around the pot. This seems like it would address the volume-knob-frequency-response issue at the cost of letting RF signals through, and maybe sounding equally weird.

2. Just let the damn thing be a mic channel. The only reason this is an issue is that I wanted to raise the impedance for the hi-z input. If I can do that on the B channel inputs, I can just use those as DIs. Also, 100k should be reasonable for hi-z Radio Shack-style mics, from what I gather---this box is supposed to be an analog Swiss Army knife anyway, so cool. I can just let it be what it is. If I really have the burning need to run guitars direct into these channels, I can use an external DI. (This approach would seem to hold even if I've failed to grasp the basics of electromagnetism...)

3. Use this as an opportunity to voice the channel by trying to control the amount of HF rolloff in the volume knob. Any thoughts on something like this?



This seems like it should provide a 300k input impedance, (a lower impedance guitar input, like the #2 inputs on a Fender?) and a grid stopper/R value of between 150K and 300k, depending on the position of the knob. The particular values are just examples, but if you could determine the C component, you could maybe "tune" the volume knob in some sonically useful way. (I'm a guitarist, so maybe I think there's something intuitive about turning up enhancing the high frequency response.)