Nuke-'Em E2 Octal Tube Mic Pre

Presented here is an unusual Microphone Preamplifier design that I have been working on. It is based on a Newcomb E2 Mic Pre, but has been modified to suit my needs. It has tons of gain but isn't very clean. The distortion it has is nice and works great on some sources but I would like to have adjustable feedback so that I can tweak the distortion even further. With the three tube stages and the two gain controls I can really cook the input and then trim back the output for some gnarly sounds. Anybody have any suggestions on how to put in an adjustable feedback loop to tame the distortion? Here is the schematic: http://tablebeast.com/schematics/nukemE201.jpg



Jesse

Off the cuff ramblings that will reveal what I do and do not know:

I personally like distortion less when feedback is used. To me, it would seem to be a 'hard/soft' kind of adjustment. Or 'smooth/harsh'. whatever you want to call it.

It will have to be localized loops from plate to grid (cap and resistor) of the last two stages to avoid interaction with the volume controls. If you do a loop around the last two stages the final volume control will interact and partially negate the feedback when used. Lower volume AND headroom at the same time, I believe.

If you put this type of feedback on the first stage, then you have to use an isolation resistor between the input and the feedback return / grid point. Otherwise, you are trying to drive your microphone like a tiny speaker with the output of the first tube. Your impedance reflection to the mic goes all wacky, along with other unmentionables I can't mention because I don't understand them all. I usually see something like a 47K isolation resistor on old pieces with 1st stage grid return. The other thing you see is feedback hitting the bottom of the secondary winding, and thus traveling through the input transformer secondary. The bottom of the secondary is stood off of ground with a resistor in these cases; something like 100K seen. I haven't thought about these two methods much, because you rarely see them. Usually the only time you find feedback returning to stage 1 it's two stage loop feedback hitting the 1st stage cathode, since it's negative for two stages and doesn't affect the grid so much.

I guess you could get experimental and set a direct resistor feed from plate to grid that also sets the correct bias, and do away with the cathode resistors altogether. Wouldn't that work if carefully selected? Would of course dictate the feedback amount with no ability to change.

The input cap should probably be bigger; at least a 0.22.

Is that grid leak bias on the first stage, with no cathode resistor? Aren't there a million arguments against this, and for a standard cathode bias arrangement? Am I seeing that incorrectly altogether?

6SF5; the 12AX7 of the octal family. Higher gain and noise probably for a first stage tube in something that's already high gain. Maybe only necessary for that certain distortion signature; maybe too much gain and noise for the kind of levels you are putting into this thing in the first place. I've heard a lot of noisy 6SF5's; shot noise I think they call it.

That 27K resistor on V2 is an interesting bit of isolation; is it necessary?

Bigger cathode bypass caps will improve low end and phase response of low end. Like 100 to 470 mfd. I tend to always replace 25's with 100's since I have a big bag of those.

From something PRR said about the Collins 6Q, I believe the 0.25 filter cap on V1 plate feed gives some bass boost.

Check the phase relationship of input xfrmr versus DI input. May be reversed for DI. If the transformer favours a wiring that is opposite polarity from the DI, there's not much you can do other than use a phase reverse on the output side of the amp to correct. The assumption there is that you could reverse the input secondary winding to match, but the sound may not be as good due to the capacitive effects of the winding in this orientation.

10K output is more of a power stage load for that 6J5, which would give higher undistorted output into something more like 15K-20K. That's nitpicky and you may be set on the output trans.

A suggestion;
Put a switch that switches the electro in & out on V2 cathode. This will reduce the gain.
You need to charge it up, so use 100k resistor wired from pin 8 to the cap.

The contact bias input stage presents a very high impeadance load to the I/P
Tx which has fair bit of output due to the step up ratio.
You might find V1 is overloaded by high level microphone input signals.
If you dont wont to rearrange the input stage, then consider using a switchable input pad.

Sorr

Cool, great suggestions. Now to respond to your inquiries to get this whole ball rolling.

First off the output transformer is written out as a 10K to 600/150, but now I remember this was just a guess. I drew this months ago when I knew much less about this stuff than I do now. It is probably is more like 15k to 20k on the primary. I'll check and report back.

I DID change the input capacitor on this since I drew it out, it is now a 0.5uf 600V

The 27k resistor is like that because originally V2 was a 6SN7 with the other half of the bottle devoted to a phono input and the two inputs met here. I will swap out the 27k and the 100k next to it for a single 120K on my next go round.

The input transformer is the Newcomb TR-91. Anyone know if it flips phase? I was going to put a phase reverse on the output anyway so it doesn't really matter.

No cathode resistor may seem weird but I've had several Newcomb and Bogen circuits with 6SF5's in them and they all used the same huge (10M to 18M) grid resistor with no cathode resistor. I was thinking lowering gain here would be the easiest way to reduce distortion. You nailed it though, sorr, it does distort with high level signals. In a cool way on some sources like bass guitar and kick drum but sucks on snare and tom. Maybe I could have a DPDT switch that puts the cathode bypass cap in and out of V2 while lowering the gain of V1 by adjusting its bias?

I'm really loving the sound of the 6SF5 and find them to be a real bargain in the market as they have incredible sound for little money. I got a lot of 5x Military JAN NOS Tung Sol 6SF5 GT's that I have been using in the prototype and they are quiet and focused compared to the metal jacket RCAs that came with it which are dirty and gnarly. I have a three channel Newcomb H4 mixer that has three inputs to one output and it has three different tubes in there for varying colors. One of the cooler things about this setup is the 9-pin plug-in transformer that it uses is fairly standardized and I have been able to swap out various similar (and not so similar transformers to get very different sounds and colors. I want this to be my go-to 'color' pre and I want to have the ability to get a wide range of sounds from it. Your suggesions are very helpful, any more before I open the prototype back up and lay down some mods? I will then redraw the schematic and repost. Thanks so far!

Jesse

> isn't very clean

For small inputs, it can be plenty clean. Fiddle the several gain pots so nothing overloads.

The problem is the input overloads easy and can't be adjusted. Too lazy to look-up 6F5 gridleak data tonight, but overload is probably around 0.5V peak. 300:50K input ratio is 1:15 voltage ratio. 500mV/15= 30mV input overload level. Output level from mikes used around sources which can be made "gnarly" without harm tends to be 100mV to 1,000mV.... no wonder you have uncontrollable distortion.

Put a 100K pot between T1 and V1 (before the 0.1uFd cap!). Use a Audio Taper pot. Set it to "5". By now you probably know the sound of V1 overload very well; trim that pot for "nice" V1 overload, not grossly smacked, not down in the noise.

Noise level takes a hit, but you aint doing harpsicord in a large concert hall. V1 noise is probably the least of your worries.

With condenser mike on bass instruments, T1 may be overloading in bass before the new gain pot. You could try a 1K Linear pot before T1. Yeah, that unbalances the source.... keep the leads from pot to T1 short and don't tell anyone, they'll never know.

The coupling network into V3 grid has a 30Hz -3dB point, -1dB at 60Hz, enuff to shave deep bass. Is that what you want? (You can always bump it in ProFools.....)

OK, so what's happening isn't that the overall circuit is overloading its really this first stage that is the culprit. I was really trying to avoid having a third gain control. PRR, are you suggesting that I use the pot just to determine what would be a good value for a fixed resistor to put in there ahead of the input cap? As for the network ahead of V3, I plan to move the 27k resistor out of the audio path and replace the 100K B+ resistor feeding the plate of V2 with a 120K. Am I correct in thiking that this would keep bias the same but increase the gain slightly? I was also going to possibly change the value of the .022 coupling cap ahead of V3 to a .05 or .1uf. If I up it to .1 I will also change V2's coupling cap to .1 as well, maybe even .5. Thanks so much for the help so far. As for my original question, does anyone have an idea of how to optimally implement a single, switchable (as in having an on/off switch on the pot), feedback loop?

Jesse

I sort of intimated that the distortion was likely all in that first stage due to the tube choice. If you want it lesser there, you'd do best to change to a lower gain tube, which will almost certainly be a less noisy tube.

PRR's method is not to determine a fixed resistor after the input trans; it's a variable attenuator with a fixed 100K load to the transformer. Set on half (sweeper to grid) it will be -6db, so not much change and at the expense of high end when you go further. Remember a fixed resistor (or pot) reflects Z back to the mic; so if you go as low as a 47K resistor the mic sees a 188 ohm load. Which will work, but with worse noise and a high frequency roll-off for most dynamics. Nitpicky, as PRR said in this circuit and for this purpose.

The 27K makes sense knowing it's mixer stage isolation. You could just strap it out rather than change physical resistors. It won't change much. Or just move the cap to the plate rather than the resistor junction. Try both and see what you like. It's currently a voltage divider throwing away a tiny amount of gain for the sake of isolation from another now non-existant stage.

V2's input cap is fine; doubling is the most you should do. At some point you get so much low end passing that it starts coupling to the PSU and motorboating.

The input transformer polarity; you'll have to do some polarity tests. No one knows these things. There are standards, but that doesn't mean they were followed.

Feedback will require experimentation with a 500K or 1M (2M if you have one) audio pot (as rheostat) and some variably sized caps, as mentioned before from individual plates back to grids of V2 and V3. 1mfd being the biggest necessary I'd think. Plate, cap, resistor, grid. Take a lot of measurements or make the cap larger 1mfd and leave the pot in. This will still require you figure out the maximum allowable or desirable feedback and put a fixed resistor inline with the pot so you never go to 0 resistance (100% feedback at frequencies passing the cap) and shut everything down.

Single switchable doesn't jive with the fact that you have two volume controls; maybe headed for three. If yo want more control, you get more controls. Do you want feedback at V2 or V3? Or both? Only way to know is try it. How much; what do your ears tell you? Do you like several positions at each tube? That's two variable feedback pots to complement the two volume controls. Or a more complicated dual switch to change local feedback on both stages simultaneously. It's like we're building a double GERMANIUM PRE in here. :wink:

Notice I'm doing a lot of seat of the pants guessing based on experience. You can do the math, but I can get out the parts and come up with a solution much faster than I can figure out what math and where. Buy a set of decade resistors and capacitors for this purpose. You'll use them a lot. Ebay, $5-$10 each when they show up. You'll probably have to find an older cap type to get the voltage rating needed.

We are also all in the dark without some way to send test tones and take db measurements to measure real world results. And check polarity through ANY unit. Saying 'this pre sounds better than that one' is pointless if one has a polarity reversal affecting the result and you don't know it.

Of course we should all learn the math, but using it enough to remember in the field is another matter. Experimenting and measuring will give you real world experience that will give you a sense of how these things scale.

Well, if you really want total control over the distortion, you MUST have control of every signal level arriving at each gainstage, so I guess the only good solution would be another pot at the input. If you really do not want to use three pots in a mic pre, just get that output pot out. Then you just have to adjust the input level (the pot after the transformer) and the output level (the pot btween 1st and 2nd stage)...

By the way, what´s the output transformer you are using?

> I was really trying to avoid having a third gain control.

Understood. But the input overload level of 6F5 is well below the output level of many sources in a contemporary recording studio. So you MUST be able to pad-down, if you don't want gross distortion. OTOH, some sources are softer, you must run them unpadded for decent S/N ratio. So you MUST have a front-panel input pad.

And as you say, this should be a useful "gnarly-box". But you need a way to tastefully trim the amount and type of gnarl. Since a lot of that will be in the first stage, and depends critically on the source level and desired gnarl level for that specific track and take, you need an adjustable input pad.

I'd say you also want an output pad, so V3 can be flogged into power-amp flavor overload, without smoking your recorder. Yeah, four knobs. Many hours of fun finding different colors, and re-finding similar colors for different source levels.

Ya know: a simple dead-clean mod is to switch the input tranny onto the grid of V2, leav V1 out of the fight. Maybe lose the cathode cap on V2. Even then, an AKG414 near pop percussion will push V2 pretty hard, couple-percent THD.

> If you really do not want to use three pots in a mic pre, just get that output pot out.

Then only V1 gnarls, V2 V3 run clean. You want control of two anti-phase stages so you can adjust for asymmetric or semi-symmetric clipping, even or odd order.

> implement a single, switchable (as in having an on/off switch on the pot), feedback loop?

No. If the feedback does any good, overall gain drops. Now the gain control must be implemented in the feedback, and the forward path must be opened-up to maximum gain for maximum clean. Silk purse from sow's ear? Not gonna happen; anyway the sow would rather have her ear than a purse, she can't hear the music (food hitting the trough) with a silk purse.

What I'd say is: record clean with a clean preamp. Many on the market and DIY world. Then use this gnarly-box in mix-down. Change the input to 10K:10K so it'll be "on the edge" with recorder-level outputs (if you were awake in tracking, the track levels are much more consitent than mike levels). Give it lots of knobs, and tune the sound.

The design of distortion effects is Advanced Thinking and Listening, and really a very different problem than the design of transparent audio boxes.