Too much GAIN.

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

Yea, you know I might just have to throw in the towel on this one. making a buffer would be just as much trouble as making a melcor or somthing.

at least I have a box with +/- 15V and phantom...

I'm going to try it with no output transformer like PRR suggested. It might not sound half bad.
 
Hmmm I just noticed that my sound cards balanced input impedance is 24K.

using the 20:600/1:5 output transformer would be about right. 5K X 5=25K...

I still don't understand the low current thing.
 
> add a high turns ratio output transformer like 6:1 or 8:1 for lower gain and lower output impedance. So the low current output of the opamps will see a high impedance on the output transformer secondary.

Well, yes, but what is the max output voltage now? 11v peak from this amp, step-down 6:1, is 2V peak or just about +6dBm, and we want like +18 to +28dBm.

The 5K rated load and the 600 ohm pro-audio load suggest a 2.9:1 transformer. That gets us 4V peak or +12dBm. I've got stuff that level, but the whole system is designed around it.

We really want at least 14V peak and over 24mA peak, maybe twice that much current if feeding iron. That's just beyond this module's ability no matter how you transform it.

> Hows about something like this?

That is the most direct way to get pro line level out. The drawback of I.45 is that its power efficiency is around 5%: big heat for reasonable output. A practical compromise for 600 ohm load is Re=300 ohms, which gives 15V/300= 50mA standing current. Heat in transistor and in Re is about 0.75W each, suggesting a 2W resistor and a 2"x2" heatsink. 2N3055 would be suitable (and an appropriate vintage). For many reasons (including blow-up protection), add 50 or 100 ohms in front of the 600 ohm transformer.

+/-15V supplies are a bit low for high line levels; that's why all that gear that runs at +/-24V or with step-up transformers. You could go to a 150:600 transformer. It adds 6dB of gain you have to lose somewhere else. Your Re must be lowered to 75 ohms, maybe 100 ohms, idle current 150mA-200mA, 3W dissipation in transistor and in resistor, over 12W dissipation in a 2-channel box. Will be warm. Will need more than a 24VA power tranny to feed this.

The "problem" with this is that you already have a lovely package of vintage parts, and need to tack-on this booster. However it is just two/three parts, and nothing that could not have existed back in The Day.

The text promises better performance with I.46 below. You have not posted that but I can guess. And shudder. I fear it is the second coming of crossover distortion. If not, then it is a major project in itself and will clutter your neat package.

I said it elsewhere and I'll say it again: you could hide a 5534 under the Phil Brick and get modern perfomance with vintage smell.

Go back. What gain do you need? Lo-output mikes on female choir direct to disk, I sometimes need +57dB. But hot mikes 6 feet away from large percussion, I've run less than +6dB and still clipped. If you work with hot mikes in a small room with loud drums, you hardly need gain: AKG 414 will deliver around a volt directly. Use a 150:600 or 2K:10K transformer for a little gain and a place to put phantom, and run right into a Line Input.

And what parts do you have? The 1:2.8 input iron is actually pretty good for an op-amp of this class. I'd go as high as 1:5 for ab-fab noise performance, but 1:10 would be too much, and the difference with 1:2.8 isn't any problem unless both your room and your performers can be ultra-quiet. The 20:600 output iron is just wrong for this use. The 20 ohm side can't be fed more than about 3V to 5V peak before you get too-high output level on the 600 ohm side, and probably significant distortion. And the cascade of 1:2.8 input iron and 1:5.5 output iron gives gain of 1:15 or 24dB, -without- any voltage gain in the amplifier (but it must have enough current gain to buffer 10K into 20 ohms). Many many hot-close mike situations need less than 24dB gain. While they are very pretty things, a 600:600 is more what you need here.
 
Those pictures are mighty darn wide; can you link them instead of displaying them in the message?

> my sound cards balanced input impedance is 24K. using the 20:600/1:5 output transformer would be about right. 5K X 5=25K...

If the voltage ratio is 1:5, the impedance ratio is 5*5 or 25. So if this were an ideal 1:5 transformer, 25K on the output reflects as 25K/(5*5)= 1K on the primary. That's a bit low for this Philbrick.

And the impedance scales, but not for huge changes of impedance. Mid-band, you might get 1K on the "20" winding. But at the "rated lower frequency limit" the transformer will be 20 ohms inductive reactance and the driver will have to be able to drive about 20 ohms. For various reasons the real lower-limit may be below 20CPS and way below your lowest note, but expecting a "20 ohm" winding to act like 1K ohms is overly optimistic, unless you accept bad sag and distortion below about 500Hz-1KHz. (However the high frequency limit may go way beyond the audio range.)

I'm assuming audio transformers. I gather you found "geo" iron. I'm not fully sure what that is. I suspect it has extended "bass response" to amplify earth-shake for seismometers. If it will go down to 1Hz as 20 ohms, it -may- work as 500 ohms across the audio range. But I'm not at all sure what the low frequency limit on a seismometer is. Obviously it "can" go to "DC", when the earth actually moves. But unless you are right over a shifted fault, the quake averages out to zero, and before really good DC amps it wasn't practical to get DC response. (Traditional seismometers can't register DC anyway.)

The Philbrick will be quite happy driving the sound card input directly. However here we have a different problem. Most consumer sound cards will NOT accept Pro Line Level. Properly designed cards will take the 2V Max of a hi-fi CD player, or about -10dB "semi-Pro level", with no tolerance for anything more. And I have a couple cards that don't really take 2V Max. I have to attenuate even ordinary hi-fi levels before I go into them.

Hate to say it, but I don't see a practical use for these output transformers.

> I still don't understand the low current thing.

As hinted in Philbrick I.45: "general-purpose Operational Amplifier (designed primarily for high voltage gain and economy of power consumption...". It only draws about 6mA of power, and can obly divert 2mA of that to the load. In the original application, it only drove medium-Z impedances to solve equations, and you might have dozens of op-amps in a computer, so power economy was critical. As also hinted in I.45, Philbrick was happy to sell you other models with higher output current (but not any more).

Note that these sand-state op-amps replaced 2-twin-triode hollow-state op-amps. These were typically good for 100V and 1mA-2mA peak. The transistor op-amps had significantly lower output, but also lower input error, so you could scale your whole network down, still get all the accuracy of the tube op-amp, at about 1/20th of the heat and size. So even when tranny op-amps cost more than two 12AX7 in a brown base, the system cost was less (or you could build a bigger system at the same price).

The Philbricks were obsoleted by more complex modules (and later chips) that could hold low power consumption yet deliver more output current, allowing one model to be used both in large masses to compute many-factor equations and to drive external loads like chart recorders. The 709 and 101/741 were modeled after the Burr-Brown modules: over 10V and 20mA peak into 2K loads. Loads lower than 2K were generally not wise on chip op-amps because output heating ran right over into the input stage and unbalanced it, causing DC error (and bass distortion).
 
holy PRR! THAT was one for the archives, I just learned like a million different things from those posts. Thanks! I need to print those out and stick 'em up on the bench to reread all that until I fully understand it.

dave
 
I'm tellin' you....

PRR, I feel honored to have all that explained to me. Thanks so much for spending the time on this one. I've learned alot.

You were right about the bass distortion. The kick drum sounded very funny through it. I could see the waveform compared to the 1272, and it looked "drunk".

I picked up these opamps for pretty cheap and wanted to feel like I got some kind of deal. Oh well, you can't always shove a cube into a round hole.

On the bright side, I have two Langevin cards that you (PRR) actually commented on a long time ago. They were used for makeup gain on a summing buss of a broadcast board. They might make good canidates for a mike pre designed for close miking of loud sources.

l.JPG


schematic courtesy of rafafredd

Hey, I already have the box! I'll just use the 20:600 as input transformers.

It looks to be set with gain. a pot on the input will do.

Thanks again :thumb: :green: :thumb:
 
> use the 20:600 as input transformers.

Even assuming that the "20 ohm" winding is more like 100+ohms when unloaded, that is a fairly low load for most microphones. (2K is more what some mikes want.)

The AM4700 is set for gain of 1:40 or 32dB. Using the 1:2.8 (9dB) input tranny and no (or 1:1) output iron, that gives gain of 41dB, still too much for many close-mike applications. Big condensors make about one Volt near drums. You don't need any gain to go into most Line Inputs, and if overload level is around 10V then you don't want preamp gain over 1:10 or 20dB. 40dB is what I use with 414s hung 20 feet out from a string-trio, and on orchestra I run more like 20-26dB. You probably do not want anything like 40dB.

And you can't reduce the AM4700 gain by much because it screws-up the DC bias.

The Philbricks are probably unity gain stable. With 1:2.8 input and 1:1 output, that's 9dB (up to 49dB) gain, which isn't bad. (When you need less than 9dB, signals are hot enough to use the pad.) So you want 1:1 output iron and a buffer between the Phil Brick and the output iron. The I.45 note buffer will do the job with minimal added parts.

> It looks to be set with gain. a pot on the input will do.

That is equivalent to a pad. You are throwing away signal to noise ratio. With condensors on drums, you can afford that, but solo voice or guitar may be hissy.

> I picked up these opamps for pretty cheap and wanted to feel like I got some kind of deal.

As amplifiers, they are similar to 741 (cost under $0.12 each) except with less drive ability. The Phil Bricks were fabulous for their time, but that time was 40 years ago. And they certainly were not made for Audio, but for the specialized field of Analog Computing.

However they do look GREAT. And they may not have some of the small flaws of $0.12 741 chips.

Please post Philbrick note I.46, which will be a complementary boosted output. Since I can't think of any elegant way to use your prettily-mounted parts, let's see what they suggest. It has the "advantage" of being "period correct", and I like that concept. (If you don't have I.46, I can fake it, but I'd rather see how they did it.)
 
Here ya go professor...

b.JPG



But if they are similar to a 741 I might as well use the box for something else. I mean I could get the same sound by hooking up a transformer to my MXR pedal. Hey might be cool.

The input transformer straight into the sound card Idea might be neat. Could be the purest way to get loud drums to my hard drive. I have a "pro" sound card (lynx two) the specs for the card are here:
http://www.lynxstudio.com/lynxtwospecs.html

The geo transformers have many taps on the primary and could be used as 3 different gain settings starting with 1:2.8 up to about 1:8 winding ratio. I think there was a thread about this somewhere.

Specs on the geos courtesy of CJ...

tri2.JPG


tri1.JPG


Thanks again PRR !!!
 
Good stuff PRR! :thumb:

Langevin AM4700B Specs:

AM_4700_1.jpg


Input sensesivity is 40mV for +4dBm out. 4dBm = 0,3544V, so gain would be 0,3544/0,04 = 8,86 which is 18dB.

Is this correct? :?
 
rafafredd, do you want me to translate that SuperBuffer article for you?

I also have a schematic for a simple symmetrical Borbely JFET buffer that sounds really nice too. Let me know if you, or anyone else, are interested in that..
 
> Input sensivity is 40mV for +4dBm out. 4dBm = 0,3544V, so gain would be 0,3544/0,04 = 8,86 which is 18dB. Is this correct?

+4dBm in 600 ohms is 1.23V. Gain is 1.23/0.04= 1:30 or 30dB.

I'm not willing to give up on your Philbricks yet. They may not sound quite like a 741, especially when abused, and they are already mounted.

I can't fit the 20 ohm - 600 ohm transformers in anywhere, given +/-15V supplies and pro-audio standards. I hate to say it, but dump them and find some simple 1:1 600:600 iron.

Philbrick Note I.46 is as bad as I'd feared. Great simple DC amp, but will be pretty nasty on audio. And the wonderful text misses a few points. (It really would be better to ask your Philbrick Rep for one of those wonderful "Philbrick Booster Amplifier" modules...)

I hear AP groaning "it's Class-B!" already, but here is something in a similar pattern that won't sound too bad (been used in old but respected gear), won't sound like a 741, is simple, and pretty blow-up proof.
Philbrick-II.gif


The transistors buffer a 600 ohm load up to over 10K so the op-amp is happy. The buffering fails for currents less than 0.3mA; below that the op-amp drives the load directly through the 2K2 resistor. No it's not perfect: a $2 NE5532 is more likely to be more transparent. But the old stuff is sometimes surprisingly interesting, once you tame it. And sure looks impressive.
 
[quote author="sismofyt"]rafafredd, do you want me to translate that SuperBuffer article for you?

I also have a schematic for a simple symmetrical Borbely JFET buffer that sounds really nice too. Let me know if you, or anyone else, are interested in that..[/quote]

off coarse I want it...

If it´s not a pain also, I would love to get this borbely buffer schem.
 
Mr. PRR,

I would love to hear your comments about this one that I´m using after my langevin boards. What do you think about this simple booster?

booster_only.gif
 
Sweet!

Alright I'll get rid of the 20:600 transformers. I think I have some 600:600 laying around somewhere.

What transistors should I use? and I'm assuming +/- 15V right?
 
Rafa,

I have tried one of Borbely's followers (figure 15c). http://www.borbelyaudio.com/ae699bor.pdf
One of the best uses for it I have found has been matching a high turns/high impedance transformer to a low impedance amplifier stage, such as mating a 115KE to a JH990. It has many other possibilities too like adding a high impedance instrument input to an amp.

Tamas
 
> What transistors should I use?

I'd use 2N3055 and 2N2955 in metal TO-3 cases. They are a little over-kill for the job, but if you shop around they are cheap, will be blowup-proof, and have that vintage look.
 
Ahh yes the vintage look...

Api front end, Neve back...but not really...

this ought to be interesting...

Thanks PRR I will let you know how it all turned out... :thumb:
 
[quote author="tk@halmi"]Rafa,

I have tried one of Borbely's followers (figure 15c).
Tamas[/quote]

Thanks Tamas. Those looks great. One more reason that I should be able to get some 2sk170 transistors. :evil:
 
Wow!!! It worked!!

Hey this is a one of a kind frankenstien pre....

I have yet to really test it out but every things sounds fine in my head phones with my little computer mixer..

I used some parts that are a little different.

I used NTE180 and 181's. They seem to work fine.

http://www.nteinc.com/specs/100to199/NTE180.html

I used 43 ohms resistors insted of 47 ohms EXCEPT for the one connecting to the output transformer.

and all I had were 330uf bipolar insted of 100uf nonpolar.

I used UTC A-21 for output transformers.

Well, big thanks to PRR for hookin me up. :green: :green: :green:
 
Back
Top