Active ribbon-mic

[bcarso]

BTW, my motivation in this is partly to get these parts into awareness and I hope, long-term availability. Toshiba has already been discontinuing them right and left, and long ago stopped supporting them in parts of the world not involved in manufacturing, e.g., as the US consumer electronics industry dwindled. Other vendors are probably even more difficult to deal with.

It's true that ICs have gotten so much better that they will serve in a variety of apps. And you can learn a good deal by using them. But for me there is nothing as instructive and fun as designing and building with discretes. Where we discrete users have a disadvantage is in higher lead inductances, and in lacking the intrinsic matching of nearby on-chip parts. But SMD helps a lot on the first issue, and duals and arrays, such as there are, on the other.

 

[tmbg]

update:

I have a layout done, decided not to do the SMD thing...

It is a layout of the 2N4401 version, the layout was done assuming 10uF 10V caps, and the footprint is 5.5mm diameter, 2.5mm lead spacing.

The caps I have in my BOM are Xicon XRL series 5x11mm parts, 2mm lead spacing. That series goes all the way up to 100uF in that package at 10V, and some of the slightly bigger and/or higher voltage parts are 6.3x11, 2.5mm, and those may fit as well.

The board is 0.9x0.9", so it's under a square inch, so it should be $2.50 apiece plus shipping to have fabbed at Sparkfun.com.

The layout seems done, and passes DRC on my layout tool, but I haven't yet discovered whether I have my gerber output in order so that sparkfun can process it. As soon as it passes DRC, I'll order a couple.

If anyone is interested in this, let me know what file formats you can handle and I'll try to get something put up.

 

[Gus]

Xicon? I would try other brands

 

[bcarso]

Ditto Gus. The Japanese -cons are pretty good: Nichicon, Rubycon, United Chemicon. Other far east vendors are spotty. Get the 105C rated parts too---even if you never approach that temperature, the dielectric is better quality, typically.

 

[tmbg]

the nichicons are in the same footprint, and same or smaller height.

In fact, Nichicon 105F 100uF 10V are 5x11mm and 15 cents apiece in tens.

 

[Rossi]

Getting good electrolytics seems even harder than getting the right transistors. A while ago I spotted BC components 'lytics in the Conrad catalog (a chain somewhat like RS). So I went to there, ordered quite a number of those. When I came home I discovered they put the same old Hermei stuff in my bag. Same brand that they always sold...

 

[bcarso]

I've heard good things about BC film caps too.

 

[Rossi]

I've used BC polyprop (1000 pf) for capsule to FET connection in a couple of microphones and was very happy with the results. I usually use Wima caps, but the polyprop ones are physically too big for most microphones.

 

[Gus]

I am not a fan of United Chemicon or IC or Hitano electros. Elnas are good. US made stuff tends to be good. The samwha and samyoung stuff I have tested seems to be OK. I tend to like rudycon at over 100V of the cons.

Funny thing some stuff sounds fine, measures good at the 85C

For the guitar amp thread here, measure the ESR of a cap at elevated temp I think this is a VERY big part of the sound of tube guitars amps as they "heat up" I have place electros in a temp controlled oven and then tested them at different temps.

 

[bcarso]

Anyone tried the Sanyo Oscon caps? They tend to be fairly low voltage so limited in application, but if low loss were the determining factor (hotly debated issue) they ought to sound good.

I remember hearing about how Sanyo did a demo for some people at Harman---they had changed out the caps in a ghetto blaster and A/B'ed the before and after versions. Although the sales people were enthusiastic in their conviction of a huge improvement, by all accounts the results were not impressive. A ghetto blaster---what were they thinkin'?

Audio Note swears by their oil-paper-copper foil caps, supposedly because of (or at least associated with) an absence of audible piezoelectrical effects. I used some for a friend's design as output coupling C's and I quipped that they were perhaps the first non-electrolytic caps that needed a d.c. servo.

 

[Gus]

Does the oil used have a dipole movement(if this is the right term) in an electrostatic field? I don't "get" the oil cap stuff.

FWIW at may job I have removed and replaced thousands of caps over the last 21 years. I would often test them both the removed and replacements.

Beside brand, mass, dia to L and seals are a big part of how they measure and last and even sound IMO.

 

[bcarso]

I suspect the leakage is due to ionic impurites in the oil, maybe arising from how the copper foil is handled---it wouldn't take much sulphate and copper ions to give some leakage.

(dipole moment is the term you want I think).

That's interesting about the effects of form factor.

 

[Rossi]

I'm still waiting for the more fancy parts to arrive. But since I had some time over the Easter hollidays, I fooled around some more with what I had.

I tried paralleling several BD139s. I didn't do any fancy matching, just Hfe. The BD139-10 seems to be low tolerance. I got 10 or 12 of them and they're all between 93 and 96. Paralleling two reduced noise a bit. Adding a third one resulted in no appreciable difference. Next, I tried reducing the 150 ohms resistors (R1, R2) to 100 ohms. Quite surprisingly that lowered the noise floor considerably, especially when used with the Siemens V272. I'm not sure what to make of that.

Another thing I've noticed is that current draw is only about 1.6 mA. Can this be? Wasn't the circuit supposed to draw 5 mA? I measured between where R5, R7 and C6 meet and ground.

BTW. I got some 2N5550 transistors which the Conrad catalogue lists as equivalents of 2SC2547E which in turn look quite usable on paper. But noise was worse than with BD139-10. But as I said, Conrad often sell junk that's supposed to be equivalent without even asking if that's okay with you.

 

[PRR]

> I tried reducing the 150 ohms resistors (R1, R2) to 100 ohms. Quite surprisingly that lowered the noise floor considerably

That lowers gain by 3dB. It also drops output impedance, and if the post-amp has very high noise current that would be another few dB noise.

Since you are changing gain, both signal and noise drop about the same amount, no net improvement.

> current draw is only about 1.6 mA. Can this be?

Proper 48V 2x6K8 Phantom power? Or something less?

I don't have my thinking cap on, but should be like 30V at the collectors and 2V-2.5V at the top of the 470 emitter resistor. 2.5V-3V at the top of C6.

At 1.6mA total, I don't see why paralleling BD139 should reduce noise much. We certainly are not flooding electrons in that 500mA part. Maybe Brad has a clue to this observation.

In general theory, for same current in each device, noise voltage goes down as square-root of number of devices. So 2 devices is 0.7 times the noise voltage, 3 devices is 0.578 times the noise voltage or 0.82 of the 2-device case. If you are hearing that 3dB drop, I'd think you would hear another 2dB drop, unless it is masked by post-amp noise. However you are not keeping the same current in each device, you are splitting the same current to more devices. Until parasitics dominate the noise figure, paralleling "should" not reduce noise. I dunno. Maybe the BD139 is so big that at 0.8mA-0.4mA only a little bit in the middle is really working, with high parasitic resistance.

 

[Rossi]

Thanks, PRR. It's proper 48V 6K8 phantom power with all the preamps I used..

Yes I figued reducing the resistor values would lower gain. But since I had only one circuit, I had no side by side comparison. I simply used as much gain on the preamp as needed to get the same loudness when I was standing in front of the mic with my headphones. So I'm pretty sure it was not just 3db less gain but a real reduction in overall noise level, at least when it comes to the V272. I *think* it was lower noise with the other preamps, too. My other preamps were a Mindprint TRIO, a Trident 4T and an ancient B*ringer Ultragain 2000 and a B*ringer Eurodesk MX8000 (an oldish grey one, not the blue A-Version). I now wish I had made recordings to be absolutely sure.

The two paralleled transistors were definetly lower noise than one. Not by very much, maybe not even 3 dB. A third one was just a little less noise, certainly no 2 dB improvement. I'm really quite sensitive to low level noise, I should add. At typical singing/speaking levels, the overall noise level with the doubled transistor 100 ohms circuit was comparable to that of a 20 dB-A self noise condenser mic on the same preamp at the same overall output level.

I'll check the voltages on the weekend. I don't have the circuit here, right now.

 

[clintrubber]

Just some bookkeeping: tmbg has threads about a PCB for the BJT-circuit from earlier up this thread at:

start:
http://www.groupdiy.com/index.php?topic=6755

continued:
http://www.groupdiy.com/index.php?topic=6775

 

[Rossi]

Hi folks, I've been away for a while because of the Frankurt Musikmesse & stuff. So I didn't have much time to work on electronics.

PRR was right (as always): I made a silly mistake when I built his circuit. Nonetheless, the performance was nearly identical to the correct circuit. :?

However, the effect of paralleled transistors seems diminished. But I still seem to hear a slight improvement in noise. Yet even less of a difference than before.

I'll do some more tweaking as soon as I have time. I'm still quite busy right now compiling Musikmesse news for next month's edition of Keyboards magazine.

BTW: I was introduced to Hans Thomann at the Frankfurt Music Fair. Turns out he's quite interested in microphone techonlogy, not just business-wise.

 

[Rossi]

Just a brief update, as I still don't have much time for electronics projects these days.

I tried those nice 2SC3329 transistors that Brad was so nice to send me. In combination with my ancient Siemens V272 the bipolar circuit worked perfectly. Noise performance was excellent. Strangely enough, noise was not better in combination with my B*ringer Ultr@gain 2000 which without the booster preamp was considerably lower noise than the Siemens. In fact the BD139 transistors seemed to be a little lower noise here, or at least the noise contained less high frequency content. I'll do some more testing as soon as I have time. Still, I'm intrigued by how low noise this circuit can be in combination with the right preamp. These ultra low Rbb transistors are cool stuff!

 

[bcarso]

Glad to hear it! I still haven't deployed them yet but I see an opportunity in Mr. Groner's mic preamp, which I have done some optimization on (resulting in a quite different topology compared to the Linsley-Hood).

The ultimate use of those bad boys would be in direct ribbon mic preamps and moving coil cartridge pre-preamps. For the former we need to get a bunch in parallel and run some serious current.

 

[Rossi]

Yes, the more I think about it, the more I think a direct transformerless ribbon preamp would be a great experiment. It seems doable with these parts. I'm curious to see if you can pull it off and how it performs in real life.

Another application for these fine transistors would be a nice (almost) no compromise preamp that would perfom exceptionally well with normal transformer ribbon mics. I always liked the preamp PRR developed in this thread: http://www.groupdiy.com/index.php?topic=2074
require a second stage, though.