Maker Pro
Maker Pro

Copying Op Amps to Make Mic Amps

C

cledus

Jan 1, 1970
0
cledus said:
Why is distortion not specified on the data sheet? Is the distortion
good for this part? Or has a tradeoff been made?

-c


There are graphs for distortion on Page 9. I overlooked it on the first
pass. My bad.
 
J

John Larkin

Jan 1, 1970
0
"cledus"



** THD is rarely specified in op-amp data.

There are too many variables affecting THD to make one figure meaningful or
even one graph.

These include frequency, stage gain, output level, output load condition and
topology ( inverting or non inverting).

See this site for a more info on the topic.

http://www.dself.dsl.pipex.com/ampins/webbop/opamp.htm


....... Phil

With a typical voltage gain of 30 million and a typical gbw of 75 MHz,
the LT1028 should be pretty good. The datasheet does have thd curves,
hitting numbers like 1 ppm at 1 khz, g=-20, 20 v p-p out. Not bad!

John
 
J

John Larkin

Jan 1, 1970
0
"John Larkin"
"Phil Allison"


** Shame how that figure is not typical.

For the non inverting mode, with 600 ohms load, gain of 1000 ( as used in a
mic pre-amp ), frequency 20 kHz - the THD figure rises by 1000 times to
a mediocre 0.1%.

Sure. 75 MHz / 1000 is only 75 KHz of available gbw. Two stages of
gain, 32x each, and the distortion would be back in the single-digits
of ppm.

I use these opamps in nmr gradient amps, to amplify the current shunt
signals. To reduce thermal effects in the shunts, it's best to keep
their resistance and hence voltage drop as low as possible, and we
need ppm wideband noise, so a good opamp really pays off. My first
stage out of the shunt is an LT1028 running gains like 15. The
feedback resistor values are so low (to keep the Johnson noise down)
that the 1028 would have to source a lot of current into them and
transiently self-heat, so I put another, cheaper follower opamp inside
that loop, so it's the one that gets hot. At that point, one
appreciates the fact that the 1028 has a compensation adjust pin. Nice
amp.

Maxim made a MAX1028 for a while, but LTC made them drop it, somehow.
Analog Devices now has a similar part.

John
 
E

Eeyore

Jan 1, 1970
0
John said:
Sure. 75 MHz / 1000 is only 75 KHz of available gbw.

Which is where using op-amps exclusively falls down. I can't think of many
situations where I use them for > 30dB gain. Rarely above 20dB in fact.

Graham
 
F

Fred Bartoli

Jan 1, 1970
0
Eeyore a écrit :
Which is where using op-amps exclusively falls down. I can't think of many
situations where I use them for > 30dB gain. Rarely above 20dB in fact.

And what's wrong with compound?

I have a fA input bias amplifier with a 100 gain that have 100MHz GBW
product.
Pretty easy in fact.
 
E

Eeyore

Jan 1, 1970
0
Fred said:
Eeyore a écrit :

And what's wrong with compound?

What's wrong with discretes ? I design with discretes quite a lot in fact. Another
of my specialities is high power audio amplification. ICs can't do that either.

I have a fA input bias amplifier with a 100 gain that have 100MHz GBW
product.
Pretty easy in fact.

Irrelevant to audio. I'm sure it's fine for what you need it for though.

Graham
 
J

John Larkin

Jan 1, 1970
0
Which is where using op-amps exclusively falls down. I can't think of many
situations where I use them for > 30dB gain. Rarely above 20dB in fact.

Graham

You're not going to get a clean gain of 1000 from a single discrete
stage, either.

John
 
E

Eeyore

Jan 1, 1970
0
John said:
You're not going to get a clean gain of 1000 from a single discrete
stage, either.

I'm not trying to am I ? The 'worst' example I posted was ~ 40dB from a single stage
(the remainder came from the op-amp) using the compound pnp-npn arrangement.

Graham
 
A

Andy Peters

Jan 1, 1970
0
I think this is a point that Graham has been trying to make. For a musician
or sound engineer, there is a certain sound to micpres, etc., and maybe
that's why his design was successful. In technical terms, I suppose it's
probably some lack of quality (I'm not saying this about Graham's design
BTW) in the design that imparts a difference in tone to whatever's being
recorded. This explains why you see Neve channel strips going for a couple
grand on eBay, even though Rupert himself admits that today's designs are
far better. This may also help explain why the high end boutique micpres can
fetch a similar price, even though there couldn't possibly be enough spent
on actual components or R+D or whatever to justify the cost. Maybe Phil's
right and the engineer types are deluded, but they seem convinced that there
is a difference, and maybe there is - this mysterious tonal quality that
certain designs impart. Whether it is better for the sound or not is a
matter of opinion and personal taste. I'm a musician myself, and what I
record is largely classical guitar, which requires a very low noise micpre,
which the SSM2017 is able to deliver. The noise contributed by the micpre
itself, with whatever resistors or other noisy passives I have in there is
miniscule, and for all intents inaudible, so this one works great for me.
I've not had the fortune to A/B it with a Focusrite to see if the latter has
a better 'sound', although a local studio owner has invited me to bring my
micpre and go head-to-head with his Focusrite. Maybe I'll take him up on it
some day, but I figure that my playing itself needs more work than my
recorded sound anyway. ....

I think that the mic preamp has reached its state of the art, and
further improvements are probably found in reducing the cost while
keeping the performance. It's not all that difficult to build a mic
preamp that measures flat from single-digit hertz to 200 kHz over a
reasonable gain range.

Anyways, I think the most obvious reason for differences in the sound
of mic preamps is simply the input loading. Change the input
impedance and how it reacts with a dynamic mic changes.

-a
 
E

Eeyore

Jan 1, 1970
0
Andy said:
Anyways, I think the most obvious reason for differences in the sound
of mic preamps is simply the input loading. Change the input
impedance and how it reacts with a dynamic mic changes.

That's one of the ways. Any 'boutique' mic pre I may ever make will have selectable input Z.

The other main things influencing their sound is transformer colouration and (for old designs) the small but
measurable amount of THD from the Class A stages.

Graham
 
J

John Larkin

Jan 1, 1970
0
I'm not trying to am I ? The 'worst' example I posted was ~ 40dB from a single stage
(the remainder came from the op-amp) using the compound pnp-npn arrangement.

Four transistors is not a single stage!

Pbbbbtttt!

John
 
E

Eeyore

Jan 1, 1970
0
John said:
(the remainder came from the op-amp) using the compound pnp-npn >arrangement.

Four transistors is not a single stage!

It is in my book. If they'd been ordinary darlingtons would you still have quibbled ?

Pbbbbtttt!

Ppffftttt !!!

Graham
 
P

Phil Allison

Jan 1, 1970
0
"Eeysore"
That's one of the ways. Any 'boutique' mic pre I may ever make will have
selectable input Z.

The other main things influencing their sound is transformer colouration
...


** The * PROBLEM * is that source impedance alters the HF response of the
transformer.

A given design may be "flat as a tack" with a 600 ohms mic as the source
but peak by 6 dB at some high audio frequency when changed to 200 ohms.

Then, if a modern condenser mic with say 20 ohms impedance is used, the HF
peak goes right off the wall.

Getting rid of that damn tranny was the best thing to ever happen to mic
pres.



........ Phil
 
Top