THAT1512 based pre

[aortizjr]

That one does look excellent. The problem for Carl and I is finding it in the US

I have been browsing around the digikey and mouser catalogs/websites for a while now. The only one that I have been able to find is that big one mentioned earlier that is $8 each.

Now there are a ton of EMF/RFI filters, ferrites, etc etc. There are also some nice common mode inductors, but getting the right specs has been difficult. Plus digikey has a minimum order in the 100's for some of them. Not good for some quicky DIY stuff.

I think for this particular circuit, if I can't find a good common mode inductor for a decent price, I might just go with ferrite beads. I figure with the beads, the cap shunts, and the THAT1512 built in Common Mode rejection it will probably be good enough.

 

[cuelist]

[quote author="aortizjr"]That one does look excellent. The problem for Carl and I is finding it in the US [/quote]

You could try contacting;

VAC MAGNETICS CORPORATION
2935 Dolphin Drive
Suite 102
Elizabethtown, KY 42701
Tel: (+1) 270-769-1333
Fax: (+1) 270-765-3118
E-Mail: [email protected]

 

[Carl_Huff]

Thanks for the address. If I'm gonna go the hassle, I'd much rather go with common mode than just a ferrite bead.

 

[bcarso]

Or just get a little ferrite ring and some fine enamel wire and wind your own. Twist two conductors together and thread the pair through the core several times.

 

[aortizjr]

[quote author="bcarso"]Or just get a little ferrite ring and some fine enamel wire and wind your own. Twist two conductors together and thread the pair through the core several times.[/quote]

I guess my biggest concern would be getting the inductance right.

Perhaps I just don't understand, but would it be possible to screw it up causing audio degradation?

 

[bcarso]

[quote author="aortizjr"][quote author="bcarso"]Or just get a little ferrite ring and some fine enamel wire and wind your own. Twist two conductors together and thread the pair through the core several times.[/quote]

I guess my biggest concern would be getting the inductance right.

Perhaps I just don't understand, but would it be possible to screw it up causing audio degradation?[/quote]

The precise inductance is not critical. What does matter is the equality, which is pretty well insured by the bifilar winding. The equality insures that the choke "isn't there" for differential signals, i.e., the signals you generally are wanting to amplify.

Normal-mode chokes, single ones in each line, followed by little C's to common, are more critical as to their effect on the audio. But usually the common-mode ones that we're mostly talking about are getting rid of the major antenna effect coupling garbage.

Too high of inductance will have a too-low self-resonant frequency, thus not choke out higher frequencies. Sometimes even two common-mode chokes in series will be needed depending on the interference frequencies.

 

[aortizjr]

OK...

So I updated the schematic with the addition of the common mode choke coil and some ferrite beads.

I decided to add both since from looking at spec sheets, common mode will handle a certain frequency range, then the ferrites will cover in the upper MHZ region and above. Plus when searching around I found Power supply filters and such that had both. I figure it couldn't hurt...

But I did some looking around and E-mailed places. I found some more info about the inductors and chokes.

I think that some of that info would be better off in another thread. So I will start one up here in a minute with some of the info I found.

Now with this circuit... I am going to try and add a buffer and the line balancer today. The buffer is good for unbalanced output, even though not necessary for the line balancer.

 

[mhelin]

The common mode choke (in updated schematics) should not be connected like a transformer, it should be connected like two (separate) coils like the f1 and f2 (but using different symbol).

 

[aortizjr]

[quote author="mhelin"]The common mode choke (in updated schematics) should not be connected like a transformer, it should be connected like two (separate) coils like the f1 and f2 (but using different symbol).[/quote]

Thanks! I have this schem in another post too.... I think I got it right this time

 

[aortizjr]

Here is the other post regarding the chokes and ferrites.

http://www.groupdiy.com/index.php?topic=10471

 

[mcs]

I'm working on a similar project also. I just have one simple question - why are the input caps in most circuits 22µF or 47µF?

I would think a smaller cap would work. The "load" is 10k as far as I can see, so wouldn't 2µ2 be enough?

(2*pi*10k*2µ2)^-1 = 7Hz, or am I missing something?

Best regards,

Mikkel C. Simonsen

 

[moamps]

[quote author="mcs"]....I just have one simple question - why are the input caps in most circuits 22µF or 47µF?....[/quote]

From 1512 datasheet:

"Note that the values of R1 and R2 in these figures
should be kept small to minimize pickup of unwanted
noise and interference. A value of 1 k is
often used, since some microphones require a differential
input impedance of this magnitude."

Regards,
Milan

 

[mcs]

If you read a bit further, they recommend 10k - look at page 6 :grin:

Best regards,

Mikkel C. Simonsen

 

[PRR]

> why are the input caps in most circuits 22µF or 47µF?

I believe you will find that it is most BJT transformerless mike preamps that call for giant input caps.

BJTs make noise current. This is shorted-out by the source impedance. 200Ω will short-out a lot of noise current. If you put a 1uFd cap in there, the impedance shorting-out the noise current is ~10K at the bottom of the audio band, doesn't get down near 200Ω until halfway up the audio band. No hiss, but a lot of random rumble.

Combined with 1/f noise, and wide-band systems, it can be a LOT of subsonic noise. I have seen a BJT phono preamp with "big enough" input cap throw so much subsonic noise that speaker-amp DC protection was tripped before reaching realistic reproduction level.

In a mike-amp, room rumble will be large too, so it may not be necessary to keep input Z down to 200Ω to below 20Hz. 2*22uFd= 11uFd seems small to me: ~1K at 20Hz and approaching 200Ω at 100Hz, which is actually better than most rooms so I suppose it is OK.

If you only use high-output condensers, the internal amplifier noise will usually overwhelm "bad" mike-input noise. If you only record Fender amps and hi-energy drums, noise may be a non-issue. Recording pipe-organ in a very quiet church, I could "see" the mike-amp 20Hz noise in the recording in the softer passages.

Noise current in FETs and tubes is so very low that it is a non-issue at mike-line impedance. Of course we see few FET or tube transformerless mike preamps, because the noise voltage of any practical tube or all but the hottest FETs is much higher than thermal noise at mike-line impedance.

 

[mcs]

[quote author="PRR"]I believe you will find that it is most BJT transformerless mike preamps that call for giant input caps. [/quote]
I was referring to SSM (and clones) preamp circuits.

But thanks for your reply. I wanted to use film caps instead of electrolytics, but the 22µF ones are a bit big. So electrolytics it is...

Best regards,

Mikkel C. Simonsen

 

[PRR]

> I was referring to SSM (and clones) preamp circuits.

BJT = Bipolar Junction Transistor, in case that is not clear.

Yes, the Green, SSL, SSM, THAT, TI, and nearly all transformerless mike preamps are BJT input. They have all had the "same" transistors for the last 25 years. Oh, you used to have to select a vendor and weed-out hissers, now several discrete parts are reliably low-hiss; SSM pioneered good large-area integrated input BJTs, THAT and TI and others have carried-on.

They all face the same issue: to get very low noise voltage, they must run at high emitter current. This leads to large base current and large base current noise. High Beta helps but only (if all is optimized) as a square-root effect.

> I wanted to use film caps instead of electrolytics

Sadly, that isn't easy. You could try 2uFd: it may not cause enough LF noise to be a problem in your work. You might try that plus reducing input-pair current: noise voltage will rise, but noise-current induced noise should drop.

 

[PRR]

Off-topic: MCS, I want to thank you for preserving the "Improving Williamson" article. I read that when it was nearly new, and thought it was an excellent analysis of an excellent design. I'm not sure I now agree with every opinion and detail, but it is still very worth the time to read.

In an idle moment, I collected the three page-scan images into one file, ImprovingWilliamson.PDF, about 290KB. This may be a little easier to re-size and print. Feel free to grab it.

 

[mr coffee]

Hi 12afael,

you wrote

why the C9 should be a low ESR cap?
I see on a couple of designs that put a few caps on parallel instead a big one. the ESR is in serie with the gain pot so a couple of ohms don`t affect too much, or i`m wrong?

And no one seems to have an answer so far. I don't know either, unfortunately.

My guess is it has to do with other aspects of the construction of the low ESR capacitor, rather than the series resistance per se.

Anyone actually know?

 

[bcarso]

ESR is due to a complex of physical details in the capacitor and is hardly an ideal resistance per se. So you want it to be truly negligible compared to your gain-setting one(s).

Parenthetically, another reason sometimes for coupling caps to be a lot larger than the low frequency signal response of interest is one of common-mode rejection---the caps are not usually tight tolerance, and an imbalance translates to poor common-mode performance unless the highpass poles are way down there.

 

[mcs]

Here's my PCB layout - put two of those in a 1U rack, and you will have a nice, big box of air :green:

I didn't include the 6800µF cap in series with the gain resistors. Is it actually nescessary? I haven't used on in my previous SSM pres, and haven't had any problems...

Best regards,

Mikkel C. Simonsen