Transformerless tube mic preamp schem

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Gulistan

Active member
Joined
Oct 22, 2004
Messages
44
Location
France
I found this schematic by arranging my office. (I had found it 2 years ago on Internet and printed. It is not any more on my PC and I thus have scan it).
What do you think of this schematic ?
I remember the guy who made this schem claimed it work verry well...
http://alterneo.free.fr/schematics/preamp.gif
[Edited: Ethan]
 
Hey, wasn't somebody asking about all-balanced topologies just a few days ago?

Peace,
Al.
 
The gain is too low for certain mics, distortion will be fairly high, output impedance is fairly high, and the source impedance of a mic is too low to obtain best noise performance running directly into the grid of a triode. I wouldn't expect professional-quality results from this circuit.
 
Well let's see,

1- I wouldn't use an LM-334 as a current sourse... too noisy. But do use a current source here.
2- I would direct couple the diff-amp to the cathode followers.
3- I wouldn't use 12AX7 for the cathode followers (use 12AT7 or 12AU7).
4- 2.2k phantom power resistors are too small. Use 6.8k instead.
5- With 1M grid resistors you do not need 10 mfd coupling caps.
6- Use 63v rated coupling caps on the input not 400+ volt caps.

Using the bootstrapped input configuration of the cathode followers reduces the load on the diff-amp plates and that is good, but I would want to look at (or rather, listen to) that configuration and compare it to other non-bootstapped cathode follower input topologies.

The total gain for this circuit will be appox +40 dB and will not be time/age stable. The gain is also static. The HF response probably won't be all that great either. I'd think about a wrapping a few dB of feedback around the whole circuit to improve the HF response and reduce the effects of time/aging.

Other than that, it looks alright to me.
 
> 2- I would direct couple the diff-amp to the cathode followers.

> Using the bootstrapped input configuration of the cathode followers reduces the load on the diff-amp plates and that is good, but I would want to look at (or rather, listen to) that configuration and compare it to other non-bootstapped cathode follower input topologies.

Looks good to me, except with that whopping B+ there is some danger of heater-cathode breakdown. And overload will be different, probably nicer with the R-C coupling.

> 3- I wouldn't use 12AX7 for the cathode followers (use 12AT7 or 12AU7).

For small load and cathode current, 12AX7 gives higher Gm and lower output Z than AU or AT. Also 12AX7 is in-stock everywhere there are Fender guitar amps. But for even 10K loading, I think the 12AX7 will run out of poop to drive the load.

> 5- With 1M grid resistors you do not need 10 mfd coupling caps.
> 6- Use 63v rated coupling caps on the input not 400+ volt caps.


Looks like someone had an excess of 10uFd 415V caps.

> The total gain for this circuit will be appox +40 dB

Only get that high if you bootstrapped the 270K resistors.

> will not be time/age stable.

Will vary a dB or two. Good enough for rock and roll, where the musicians vary +/-9dB depending on mood, drugs, and line-voltage.

> The gain is also static.

Low, but if you need less you gotta get a pad.

> The HF response probably won't be all that great either.

There is no Miller-effect and the 12AX7's plate resistance will be much less than that 270K plate load resistor. Depending a LOT on layout (and how physically large those coupling caps are) it could be over 50KHz or only 10KHz. Still good enough for rock and roll.

> The gain is too low for certain mics, distortion will be fairly high, output impedance is fairly high

Zout is around 1K-2K. Gain is essentially 80 or 37dB. I doubt distortion is an issue (if load impedance is high): If output voltage is much above 1V to 10V your next stage will overload, so you'd need an input pad. Working with 395V B+ and push-pull, THD will be far below 1%. Good enough for rock and roll.

> noise performance running directly into the grid of a triode.

Input noise will be several microvolts. Compared to 0.2uV for a dynamic mike in a silent room, this sucks. But my AKG 414s output 1uV of noise, equivalent to a pretty quiet 14dB SPL. With those hot mikes and 3uV of tube noise we are at 24dB SPL, which isn't that bad, lower than many rooms, and insignificant in the face of 104dB SPL guitar amps with 40dB SPL hum/hiss. Good enough for rock and roll.

If I remember the origin of this plan: it WAS a rock-band tool, using available parts in former-USSR lands, and was said to be very successful in that use. That may be a happy accident of band, mikes, and plan, but isn't an uncommon need.
 
If the tubes short or saturate, and there running at Mesa Boogie voltages, you will have 400 into your LA2a tranny due to catastrophic 63 volt cap failure. For live, they wanted bullet proof as possible.

Nice thing is it wires up in ten minutes on an old Fender chassis and costs nothing.
 
CJ, the caps would still explode in that case, regardless of whether they're 63V or 415V caps, because they'd be reverse-biased by quite a few volts! (I assume we're talking about polarized caps since I can't remember the last time I saw a 10uF, 415V unpolarized cap!).
 
There is no Miller-effect

Why do you say that?

For there to be no miller effect one would have to use a cascode topology for each side of the diff-amp, or connect one side of the diff-amp as a common-plate (cathode follower) amplifier. In this design there are plate resistors on both sides on the diff-amp.

Maybe I am missing something here...
 
I'll bet he was referring to the coupling between the input stage and the cathode followers. The drive to the cathode followers is from a high source Z, but the CF plates are at AC ground, so no Miller effect there.

There's Miller effect in the input stage, but the source impedance of the typical mic is low enough that the HF rolloff might be fairly inconsequential.

I still don't think it's a great design in absolute terms; but as alluded to earlier, it seems a reasonably OK lashup of available parts to do the job at hand.
 
Will vary a dB or two. Good enough for rock and roll, where the musicians vary +/-9dB depending on mood, drugs, and line-voltage.

...Still good enough for rock and roll.

...Good enough for rock and roll.

So it can be used!!!

cheers!
Fabio
 
> Why do you say that?

What Dave said.

The first stage plates see only a cathode follower, no Miller there. Two plate-grid capacitances, just a few pFd plus wiring strays. Ballparking node impedance as 30pFd and plate resistance lower than 150KΩ (too tired to look it up) I get 50KHz. If these are gigundo film-caps with another 50pFd stray capacitance, I get 20KHz. The only thing above 10KHz in a rock-band is singer-spit and snare/cymbals; we usually have too much of that and anyway it is all EQ-trimmed to taste in the mix.

Yeah, for "classical" or "accurate" I would sharpen a pencil and also bench-test. But I think the non-minimal noise level discourages "critical" use.

The funky simplicity does suggest use on rock with hot mikes. How bad can it be? Bad is sometimes good.

Yes, I should have said the inputs are Millered about 80:1, but with a 200Ω or even a 3KΩ mike the few-hundred(?) pFd is a non-issue.
 
Not exactly the same as the circuit discussed here
- but it reminded me of a circuit (somewhat more opamp-like)
that was discussed back at the previous forum:
http://home.hetnet.nl/~chickennerdpig/FILES/Tek122/Tek_Type122_circuit_ampsection_rt.jpg
(file more readable than the pic below)

Tek_Type122_circuit_ampsection_rt_scl.jpg


FWIW, these are from:
http://home.hetnet.nl/~chickennerdpig/FILES/Tek122/Tek_Type122_125_frontpic.jpg
Tek_Type122_125_frontpic_scl.jpg


I thought Marik(?) had more than a few of these boxes as well.

Peter
 
[quote author="FredForssell"]1- I wouldn't use an LM-334 as a current sourse... too noisy. But do use a current source here.
[/quote]
There are no noise specifications on the large IC series stabilizers,
but I thing, they are not very bad.
I did virtual battery with cascaded Toshiba LM7812 and capacitor
multiplyer (BC550C in common base with 220uF to the ground (and big
R s in base bias divider). I have a look to the output of 7812 and output of
BC550C follower via oscilloscope. No change. It means, that output noise
of (TOSHIBA) 7812 is as good as from low-noise cap multiplyer ? what s this ?

xvlk
 
PRR, I'm curious about something. How are you coming up with a gain of 80 for the input stage? A common-cathode 12AX7 with those component values and voltages can do about 60 running into a high-Z load. And since this circuit has a current source tail, I would expect high cathode degeneration and less voltage gain than that.

Since the input and output are differential, the single-stage gain (minus the small loss in the cathode follower) is the same as the overall insertion gain of the preamp.
 
edit... removed some funky "I must have been brain-dead at the time" posting error.

My point was that the LM-334 current source has too much voltage noise for low noise applications, in my opinion.
 
> since this circuit has a current source tail, I would expect high cathode degeneration and less voltage gain than that.

When driven and loaded differentially, the gain comes out the same as single-sided with fully bypassed cathode. There is zero signal voltage at the cathodes. No degeneration. The diff-amp has the same gain as a single triode.

Of course if you take only one output, gain is half.

> How are you coming up with a gain of 80 for the input stage? A common-cathode 12AX7 with those component values and voltages can do about 60 running into a high-Z load.

You may be more right than me. 80 sure is optimistic. But it can be higher than the book-value RC-amp chart, because the next-stage input impedance is darn-near infinite (around 50MegΩ) and because it has a very-high B+ (395V).

On second look, they are under-feeding the 12AX7. Only 1mA in the tail, so 0.5mA in each side. That forces the plates to sit around 260V, something like 258V across the tube. A 12AX7 will pass around 4mA at that voltage; to hold it down to 0.5mA takes like -2.5V on the grid. Pretty heavy bias, both grid and plate. Most triodes will give better voltage gain with less bias.

Eyeballing the dubious published curves, I figure cathode resistance around 1K, plate resistance around 100K. So (270K||100K||50M)/1K= 73K/1K= about 73.

Oh: using this 73K node impedance with the 2.2pFd of capacitance in each tube (5pFd), I get a -3dB frequency of 400KHz. That's not going to happen in real life, but careful design with a physically teeny coupling cap (or no cap) could get around 100KHz. That's higher than my dog hears. (Yes, with casual layout and overkill caps, it could be 10KHz.)

The bias seems to optimize output voltage versus linearity. Which is insane: this thing will make about 160V RMS across the two plates with less than 5%THD. You are unlikely to overdrive it with a mike; OTOH you are unlikely to find a board input to swallow 160V. And the output stage gives up much sooner, especially if loaded with less than 100K.

So no: this isn't a universal preamp. But it may be a happy little thing in selected applications. With dynamic mikes, any distortion you hear is in the line-input following this circuit. Many stage acts won't overload it even (not much) with big hot condenser mikes. Of course the Children's Choir I'm doing this weekend, with dynamic-class mikes, will need much more gain than a mere 37dB. I expect to use 52dB gain to get them to -10dBv nominal, 2V peaks.
 
And since this circuit has a current source tail, I would expect high cathode degeneration and less voltage gain than that.

For differential inputs the impedance of this current source doesn't exist. It does for common mode inputs and you will get better common mode rejection figures as the impedance of the current source increases. CMRR performance can be improved in such a circuit by connecting a JFET from the output of the current source to the cathode connections of the diff-amp (kind of a cascode configuration) thereby increases the source impedance of the current source as seen at the diff-amp cathodes. But then there is (in my mind) the noise issue.
 
Back
Top