LM394 Instrumentation amp as a Mike amp

I was looking at the LM394 data sheet, and I found this straightforward instrumentation amp. It looks like a diff amp with 2 current sources at the emitter feeding a unity gain opamp in differential mode.

What I can see is that the 394 diff amp is not in the feedback path of the opamp (unlike a double balanced pre), so I dont know how low THD could be attained with this arrangement. Also the noise spec for the LM118 is far from stellar, so I would be using another opamp.

Can anyone explain why are the 500 ohm resistors there at the base of the input BJTs?  stability?

Also, where does the 200K come from in the gain formula?

Thanks!

If you want low THD and low noise, the microphone preamp circuit has pretty much been "solved". It's called CFIA as described here:

http://www.thatcorp.com/datashts/AES129_Designing_Mic_Preamps.pdf

The CFIA puts op amps around the input transistors among other things. For a transparent pre I'm not aware of anything better. And you can just buy the entire chip (THAT 1510 / 1512) for $7 USD on mouser.

user 37518 said:

I was looking at the LM394 data sheet, and I found this straightforward instrumentation amp. It looks like a diff amp with 2 current sources at the emitter feeding a unity gain opamp in differential mode.

What I can see is that the 394 diff amp is not in the feedback path of the opamp (unlike a double balanced pre), so I dont know how low THD could be attained with this arrangement. Also the noise spec for the LM118 is far from stellar, so I would be using another opamp.

Can anyone explain why are the 500 ohm resistors there at the base of the input BJTs?  stability?

Also, where does the 200K come from in the gain formula?

Thanks!

Click to expand...

That is from a very old national semi application note from back when LM394 were low noise.  Note the LM394 is designed as a very accurate log/anti-log device... The full app note is good reading.

Back in the late 70s/early 80s there were discrete bipolar devices available with ein more than 6dB lower than LM394.

Some of those low noise discrete devices are already obsolete as everyone just buys the off the shelf IC solutions from That corp, TI, etc. for professional mic preamps.

JR

user 37518 said:

Can anyone explain why are the 500 ohm resistors there at the base of the input BJTs?  stability?

Click to expand...

I'm guessing they make an R-C low-pass filter with the  transistor's built-in capacitance between base and emitter, making the input less susceptible to RF in the input.

Also, where does the 200K come from in the gain formula?

Click to expand...

It appears to be the 100k from the + op-amp input to ground, plus the 100k in the feedback (from output to the - input).

And the other thing:

What I can see is that the 394 diff amp is not in the feedback path of the opamp (unlike a double balanced pre), so I dont know how low THD could be attained with this arrangement.

Click to expand...

These were advertised as "super transistors" so perhaps had more linear (unchanging) current gain with changing collector current than "ordinary" transistors, thus would give lower THD.

These were discontinued, I forget if it was before or after TI bought National, regardless there's a new manufacturer for this:

http://www.alfarzpp.lv/eng/sc/transistors.php

benb said:

I'm guessing they make an R-C low-pass filter with the  transistor's built-in capacitance between base and emitter, making the input less susceptible to RF in the input.

Click to expand...

Makes sense, I knew it had to be some sort of base stopper resistors

It appears to be the 100k from the + op-amp input to ground, plus the 100k in the feedback (from output to the - input).

Click to expand...

Thats what I thought, since those are the only 100k resistors in the circuit, but I hoped someone knew for sure before I had to do a KVL/KCL analysis to be sure

And the other thing:These were advertised as "super transistors" so perhaps had more linear (unchanging) current gain with changing collector current than "ordinary" transistors, thus would give lower THD.

These were discontinued, I forget if it was before or after TI bought National, regardless there's a new manufacturer for this:

http://www.alfarzpp.lv/eng/sc/transistors.php

Click to expand...

Still, I would be happier if it werent operating outside the opamp NFB loop. The LM394 was used in the Jensen 990C, according to John Hardy, they replaced it with the SSM2212 which appears to be the same thing http://www.johnhardyco.com/pdf/990.pdf

squarewave said:

If you want low THD and low noise, the microphone preamp circuit has pretty much been "solved". It's called CFIA as described here:

http://www.thatcorp.com/datashts/AES129_Designing_Mic_Preamps.pdf

The CFIA puts op amps around the input transistors among other things. For a transparent pre I'm not aware of anything better. And you can just buy the entire chip (THAT 1510 / 1512) for $7 USD on mouser.

Click to expand...

I've used the 1510 before, the INA217 aswell, both are really good, BTW thanks for the link, it was a great read,. I was hoping on trying a homebrew version which is why that IA attracted me, call me a romantic if you may. Which makes me wonder, is this the end of the line for solid state pre's? is this as good as it gets? a jelly bean IC with 3 resistors?

JohnRoberts said:

Some of those low noise discrete devices are already obsolete as everyone just buys the off the shelf IC solutions from That corp, TI, etc. for professional mic preamps.

JR

Click to expand...

Really?, are most commercial solid state pre's INA 217 or THAT 1510 ? I always thought that many manufacturers still made their own version of the Graeme Cohen pre or something completely different and esoteric.

user 37518 said:

Really?, are most commercial solid state pre's INA 217 or THAT 1510 ? I always thought that many manufacturers still made their own version of the Graeme Cohen pre or something completely different and esoteric.

Click to expand...

Back last century when I was managing a mixer design group for a major manufacturer I was still using a variant Cohen (I hate calling it that, because I used it for decades before ever hearing his name attached to it) discrete  + op amps mic pre  that was already in the Peavey system. My version before Peavey used one more op amp than what Peavey used, but their version worked satisfactorily (I checked) so I went with the flow.

Managing an engineering group is about re-using old circuits that still work, until you can't, but when the low noise devices your preamp is based on go away, that is a major hint to cut bait and move on... The devices didn't go obsolete because people were still using truckloads of them.  :

In the early days the IC preamps were not very cost competitive but over time the production volume increased dropping the cost. Not to mention one pop, takes the place of multiple component touches, so cheaper to assemble, smaller taking up less PCB real estate, etc, etc...

KISS.

Of course I have been out of those trenches since last century so I may be full of sierra.... 8)

If I was selling some exotic mojo preamp, using a canned IC solution would be death for the marketing department. 


JR

> the noise spec for the LM118 is far from stellar

The transistors run at gain up to 480. There's no opamp with hiss 480X the hiss of the '394. The opamp is chosen for grunt and slew.

> Can anyone explain why are the 500 ohm resistors there at the base of the input BJTs?

So it will suck for low hiss on 200r sources. (Actually to protect the BE junctions; but for low hiss from 200r you need to scale everything lower, and then you really do want to look at alternate topologies.)

> where does the 200K come from in the gain formula?

There's two precision 100k resistors. The output appears across *both* in series.

For the specific case of 200r mikes and "good THD"-- the recipe is now standard. This app-note for higher source Z may save 19 cent opamps, but the Usual Plan works better.

PRR said:

> the noise spec for the LM118 is far from stellar

The transistors run at gain up to 480. There's no opamp with hiss 480X the hiss of the '394. The opamp is chosen for grunt and slew.

Click to expand...

I forgot about the transistor gain, how are you getting the 480 gain figure?  following the gain formula I get 133

I have seen an average gain of hfe 700 or more in LM394's. Still have a small stash of them.

> following the gain formula I get 133

You may be right; I may have slipped a bead. Still enuff to overwhelm 2nd stage faults.

> I have seen an average gain of hfe 700 or more in LM394's.

I think we are figuring voltage gain, not current gain. The devices run at about 0.3mA so hIE is near 90r-100r. They have 12k loads. There's two of everything which is where my beads got off-track, but Gv=~133 is a good number.

The "700" matters in-that the 500r base protection resistors reflect-in as 500/700 or another 0.7r in series with hIE. But this is only 1% error.