Tube mic noise (once again)

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

pasarski

Well-known member
Joined
Jan 29, 2010
Messages
292
EDIT: I changed the title cause this diverged to a general discussion

I know this has been discussed before, but I haven't found answer to my specific question/questions.

I've read that a value for a capsule bias resistor in a condenser mic should be 100M-1G. Or even 3 gigs. But still in some classic designs it's much lower. 60M in U67 and even 8M in a ELA M 250/251 schematic I found. That's quite of a range.

I built a circuit following the G7 circuit but using 5840 tube, different transformer and different cathode resistor value. Now I'm testing it with a chinese K67 type capsule. First I tried 50M  for the capsule bias resistor which I had lying around. Sounded good, brittle though, but that's what I expected for that capsule. Then I swapped it to 1G, same as in G7. Much more low end, which is nice, but also more hiss. Now I'm considering swapping the 50M back, cause I can easily EQ it for more bass. And because I'm going to get a Thiersch M7 type capsule, which haven't got as much high's as K67 type, I'm probably gonna need to boost highs for my pop mixes. But can't do that with that 'hissy' 1G capsule bias resistor. And I'm going to use the mic for close miking vocals anyway, so I probably get enough low's with the proximity effect.

Am I over simplifying things? Is it just that different type of capsules need different values? Or did I mess up my circuit with flux or something? Tried not to, and cleaned everything with isopropyl alcohol. Is hiss a normal side effect of large capsule bias resistors?

I should of course test different values when I get the Thiersch in, but still, I'd like to learn more about effects of a capsule bias resistor.
 
You capsule is a capcitor. Your grid resistor drops the input impedance of the valve and so forms a high-pass filter whos cut-off lowers as you decrease the value of the resistor.

Capsules vary in capacitance and bass response, so suit different values. Sometimes higher values can increase noise noticeably as you pointed out.
 
Ok, thanks. That means my question has been answered before :-[ I just always thought grid resistor refers to the grid leak resistor or a resistor between grid leak resistor and grid (which I haven't seen in tube mics).

Speaking of the grid leak resistor, I tried the mic without it first, and it worked. I didn't notice much of a difference after adding a 1G. Somewhere I've gotten the impression that if you have a grid coupling cap you need a grid leak resistor too?
 
If you look around this site there are are posts about things like this.

Now I don't know your testing methodology

One 5840?  I have over 100 and sometimes I get a noisy one
How was it heated?  Look up grid current and how a tube is often used in a tube microphone
How long did you  run the tube 5 mins or days before testing the  noise?
What are you plate cathode and B+ reading with the different resistors(or missing resistor like a Royer 2001 mod)?

All kind of things matter with a tube microphone.

Look up tube specs and note the recommended MAX grid to ground resistor value.  Tube microphone are often > than the value listed

When cleaning be careful if you use polystyrene caps

Did you touch anything with your hands in the High Z area

If you look around the web there was a ela m251 schematic  that had a 8 meg grid to round IIRC

 
Gus said:
If you look around this site there are are posts about things like this.

Now I don't know your testing methodology

One 5840?  I have over 100 and sometimes I get a noisy one
How was it heated?  Look up grid current and how a tube is often used in a tube microphone
How long did you  run the tube 5 mins or days before testing the  noise?
What are you plate cathode and B+ reading with the different resistors(or missing resistor like a Royer 2001 mod)?

All kind of things matter with a tube microphone.

Look up tube specs and note the recommended MAX grid to ground resistor value.  Tube microphone are often > than the value listed

When cleaning be careful if you use polystyrene caps

Did you touch anything with your hands in the High Z area

If you look around the web there was a ela m251 schematic  that had a 8 meg grid to round IIRC

I have to admit I haven't been very scientific in my measurements :-[ I changed the tube at one point cause I pulled out a lead by accident. Haven't burned in the new one long enough yet. I have couple of more and if the noise don't go away I'll try an other one. I'm only measuring noise by my ears, and a peak/RMS meter. EDIT: the first tube seemed quiet right away, it's possible that I have a noisy one in now, not sure at what point the noise came. I work on this at nights, very tired, cause I have a two months old boy, that's why I hassle too much;)

I underheat the tube a little, 6V (6.3V in specs). I'm using a chinese PSU and I increased the last RC filter cap value, didn't make any difference.

Have measured the plate (forgot it, about 70V I think) and B+ (115V) voltages without a grid leak resistor and with 50M capsule bias, but not yet with 1G capsule bias and 1G grid to ground. Value of capsule bias shouldn't affect these cause the capsule doesn't draw current,  but the the grid to ground will, am I right?

I tried not to touch high Z parts, but it's possible I touched something.

Didn't know that about polystyrene and isopropyl. I actually see a small hard drop on it's surface (1n grid coupling cap). Maybe I corroded it.

In ELA M schematic I have the capsule bias is 8M and grid to ground 30M.

I'll burn in the tube longer, try different values for capsule bias and change the polystyrene. I try to do one thing at time to find the cause of the noise (it's not very bad noise, but maybe bad when [stacking vocals and had to edit his old post cause I started to think about stacking and S/N ratio..] boosting the hi's).

I'm still wondering if I need the grid to ground resistor at all cause it seemed to work good with out it?

 
I removed the 1G grid to ground resistor and now the noise is better again. I think it was not so much the level of the noise than the nature of it. With 1G grid to ground there was more high pitched "electrical" sounding hiss in my circuit. Without it the noise is more gentle kind of hum. And no mains/ground buzz, I'm happy. I will not mess with the capsule polarization resistor until I change the capsule.

I wonder if the "no need for a grid leak resistor" is due to the lowish (115V) B+? I also connected the tube leads like Royer does in his MXL 2001 mod; I read Gus talking something about leakage paths via pins, don't really understand it, but my mic seems to work.
 
The resistor value is a compromise with the grid-gate leakage current.

The resistor must make hiss. Higher R is higher hiss voltage. BUT the capsule capacitance bypasses the hiss. The same capsule C bypasses better if the resistor is higher.

The tradeoff is non-intuitive, does give various hiss spectrums, and does depend on the device leakage.

Common tubes have enough leakage that 100Meg may be as high as is wise.

Even early JFETS (when cool) had low-enough leakage to favor 500+Meg gate resistors.

There is also, of course, interaction with capsule capacitance and bass limit. A 30pFd capsule against a 100Meg resistor is-3dB at 50Hz, a noticable bass-shave.

Leaving-off the resistor: device bias becomes "unpredictable". Real tubes have different balances of contact-potential and grid leakage, and open-grid operation can let the tube wander-off to an unworkable (non-)operating point. This happens in self-built guitar amps when we forget to put in the grid resistor. At lower current with very good tubes, there is a tendency for most tubes of a type to "find" a specific zero-grid-current bias, which "may" be stable (at least on the lab-bench), but may not be right for your other parameters.

And yes, microscopic filth is the enemy of low-hiss hi-Z work. I use 151 Everclear spirits for general cleaning, but I dunno if that is perfectly-pure enough for Gig impedances.
 
PRR said:
Leaving-off the resistor: device bias becomes "unpredictable". Real tubes have different balances of contact-potential and grid leakage, and open-grid operation can let the tube wander-off to an unworkable (non-)operating point. This happens in self-built guitar amps when we forget to put in the grid resistor. At lower current with very good tubes, there is a tendency for most tubes of a type to "find" a specific zero-grid-current bias, which "may" be stable (at least on the lab-bench), but may not be right for your other parameters.

And yes, microscopic filth is the enemy of low-hiss hi-Z work. I use 151 Everclear spirits for general cleaning, but I dunno if that is perfectly-pure enough for Gig impedances.

I've been searching the web my butt off but this is the first time I read what's the effect of leaving the grid leak resistor off in clear words, all they talk about is the right value. Thanks. Now I understand what burdij ment by "I think you need to stabilize the bias in the Royer circuit...". He recommended 200M-1G resistor, and said "I find the lower end of the range to sound better." I see not all are fans of "oversized" resistors in microphones?

I'll keep the mic on and keep testing it. I think when I last tested it after couple of days burning in, it maybe sounded like the bias has changed, but I'm not sure, was just listening the noise mostly. I have a pot at the moment as a cathode resistor so I'll wait if the tube settles in and try to find a working bias point.

I use isopropyl alcohol from a pharmacy for cleaning. And I have the high Z components in the air.
 
one of the first links using google
http://www.w8ji.com/fusing_and_floating_grids.htm
 
On the firs link using what index word? Not "grid leakage", "grid leak resistor", "grid to ground resistor",  "omitting the grid leak resistor" etc.  anyway.

And I don't see how that article explains in clear words the effect of leaving the grid leak resistor out? To the beginner with no extensive knowledge in electronics anyway.

EDIT: "grid leakage", third link, what a dumb ass I am, sorry (but I still don't grasp the connection between that article and grid with no grid to ground resistor in a cathode biased circuit).
 
> don't grasp the connection between that article and grid with no grid to ground resistor

It starts with fused grids. The myth is that "in a fault" the fuse will blow, the open/floating grid will de-bias the tune to cut-off, the tube won't melt. That's probably wrong anyway. However the tubes "we" use have grid wires smaller than any fuse-wire: fusing is futile. That page is about POWER tubes, big transmitter tubes. I like to say "tubes is tubes" but that's a different world.

It goes on to cite textbooks. Well, the best textbooks steer you to middle of the road conservative design. The young engineer has enough problems without getting into fringe design. Use <1Meg grid resistor and at least you KNOW what the grid wants to do.

But HIGH-impedance circuits like the condenser capsule "must" veer into "fringe design".

There are tube-maker plans for 10Meg grid resistors; these work reliably with voltage-amplifier tubes worked with enough series resistance that worst-case dissipation will not damage the tube. (Example: 150V supply, 50K plate resistor, worst-case would be 75V across tube 75V across resistor, 1.5mA, 0.113W dissipation, our tubes don't come smaller than 1W rating, safe.)

Anything higher is very "fringe". You are on your own.

> I see not all are fans of "oversized" resistors in microphones?

It Depends.

On the active device (thermionic or silicon, general-purpose or low-leakage), on the capsule self-noise spectrum versus grid noise spectrum, and probably the phase of moon.

There is very little reliable documentation for LARGE grid resistance. I think experimenters found that it sometimes worked, but didn't write it up because they were not sure of the "why?" or because it was useful commercially.

Radiotron 4th has a plot of typical grid I/V curve, you can infer the open-grid point(s!) by seeing where it crosses zero current.

There is an article from AES, reprinted in the AES "Microphones" book, which plots mike noise spectra for several variations. Heavy reading.

> maybe sounded like the bias has changed

"By ear" doesn't really work. Measure bias.
 
Thanks again. Cathode biased circuit without a grid leak resistor must be in that "fringe design" category?  I'm a beginner so I think I don't go that far to the fringe!

I tried the mic with 200M grid to ground. It was the noisiest combination this far. But it was a cheap resistor from a cheap mic. I put the 1 gig back. I think the noise is OK now, maybe I've been just a little too demanding, it's a tube mic after all. I think the noise level is about the same as it was in the body donor's original circuit, maybe a little bit higher (but the sound is already more balanced and detailed even with the stock capsule I think).

I also realized that it's probably not a great idea that I have the capsule's grounded diaphragm connected to the same turret (I did the circuit on a "Fender style" turret board) than the heater ground lead. I'll wire the heater ground directly to my "star ground", maybe it'll make the mic even quieter.
 
Finally got to multimeter. With 100K plate resistor and about 750 ohm cathode resistor I have:

B+ 129V, after RC filter on top of the plate resistor 119V
on plate 40V
over cathode resitor 0.58V

That seems about right?
 
I have built a tube mic similar to the one you are mentioning, mine makes no sounds and if i turn my volume all the way up i get a little hum.
Any ideas?
 
could be a number of things...

capsule connected wrong, tube connected wrong, bad solder joints, broken caps...


styroflex caps don't like heat... got any pics?
 
nice discussion. use gloves when working on this stuff and get rid of ALL flux, and don't destroy polystyrenes with heat or chemical. If any mechanical connections to the capsule are slightly loose you can also get noise.
 
pasarski said:
Thanks again. Cathode biased circuit without a grid leak resistor must be in that "fringe design" category?  I'm a beginner so I think I don't go that far to the fringe!

I tried the mic with 200M grid to ground. It was the noisiest combination this far. But it was a cheap resistor from a cheap mic. I put the 1 gig back. I think the noise is OK now, maybe I've been just a little too demanding, it's a tube mic after all. I think the noise level is about the same as it was in the body donor's original circuit, maybe a little bit higher (but the sound is already more balanced and detailed even with the stock capsule I think).

I also realized that it's probably not a great idea that I have the capsule's grounded diaphragm connected to the same turret (I did the circuit on a "Fender style" turret board) than the heater ground lead. I'll wire the heater ground directly to my "star ground", maybe it'll make the mic even quieter.

as far as I understand, the higher the grid to ground resistor the better the snr. Brauner uses 3 Gig to ground and the mics are very quiet and have an excellent bass response. with 8 m grid to ground in an Elam 250 you will have almost no bass, even 68m will do a roll off, I put 250m inside and I like that
 
Optimum grid resistance is a compromise between tube grid leak current and low-frequency cutoff point. Going very-high tends to exclude quite a lot of potential tubes - . i.e. the higher the grid resistance, the fewer of a batch of tubes will behave good in the circuit (and the shorter a quality operational life it'll probably have).

For wire between capsule and tube grid, I prefer a thin variant of the silicone flexible cables that you often use for test leads. Most important parameters are probably a)Easily cleanable and b)Low amounts of self-resonance in the dimensions used between capsule and electronics.

Jakob E.
 

Latest posts

Back
Top