Microphone boosters?

[abbey road d enfer]

Doing a search of JFET cascode circuits I found almost the exact circuit

The only objective reason for using the cascode arrangement is that it solved the problem Triton (FetHead) had with FET's blowing because they are rated at 30V. See https://groupdiy.com/index.php?topic=74016.msg939589#msg939589
An additional problem was that, due to the rather high drain-to-gate voltage, there was significant noise current, that resulted in significant noise when used with high-ish impedance sources.
Because the output impedance of the cascode is even greater than the former two-FET circuit, it leaves the actual gain utterly dependant on the input Z of the mic pre, which can be a problem if this input Z is not linear. In addition it provides less-than-adequate loading of the mic pre's input.

I believe the solution is to buffer the FET's outputs and manage to protect the FETs from over voltage; that's what I and several others (Bumblebee, SE Electronics) have done, with good results.
I understand Triton could not do it within the limits of the very small enclosure they use and manufacturing process.

 

[user 37518]

Because BJT's have more THD, so require NFB.

Operated with almost no NFB, FET's offer acceptable THD performance, as well as adequate noise level.

I always thought of BJTs as having lower THD because of their higher gm, but I was thinking when used with NFB, without it the square law of the FET is less linear than the exponential law of the BJT so it makes sense that the FETs have lower THD when used without NFB.

 

[JohnRoberts]

I always thought of BJTs as having lower THD because of their higher gm,

perhaps in the context of NF and loop gain margin (more open loop gain translates to lower distortion). 

For TMI look at OTAs (operational transconductance amps). The input stage is a simple bipolar LTP operated open loop. To keep distortion down to usable levels, audio signals are padded down to maybe 20 mV. This creates a classic trade off between signal to noise and distortion.

but I was thinking when used with NFB, without it the square law of the FET is less linear than the exponential law of the BJT so it makes sense that the FETs have lower THD when used without NFB.

I am generally not very tolerant of distortion and use copious NF when practical, but have one glaring example when I didn't. That is my last phono preamp (to end all phono preamps) the P100. I operated the input gain stage open loop but increased the current density of the input device to a few mA, so the current change caused by the mV of input audio was small relative to the total current. The goal was to make a preamp that was literally impossible to slew rate limit.  You could inject a 2V HF square wave and it would be rolled off nicely by the RIAA playback EQ. 

This was arguably way over-engineered back decades ago but people are still trying to plow that well plowed field today.  8)

JR

 

[user 37518]

perhaps in the context of NF and loop gain margin (more open loop gain translates to lower distortion). 

For TMI look at OTAs (operational transconductance amps). The input stage is a simple bipolar LTP operated open loop. To keep distortion down to usable levels, audio signals are padded down to maybe 20 mV. This creates a classic trade off between signal to noise and distortion.
I am generally not very tolerant of distortion and use copious NF when practical, but have one glaring example when I didn't. That is my last phono preamp (to end all phono preamps) the P100. I operated the input gain stage open loop but increased the current density of the input device to a few mA, so the current change caused by the mV of input audio was small relative to the total current. The goal was to make a preamp that was literally impossible to slew rate limit.  You could inject a 2V HF square wave and it would be rolled off nicely by the RIAA playback EQ. 

This was arguably way over-engineered back decades ago but people are still trying to plow that well plowed field today.  8)

JR

OTAs are great for many applications such as VCAs , VCFs, and in general musical instruments, but the THD is way to high for pro audio, some of them have input linearizing diodes to reduce THD. There are not many models that I know of, the CA3080 is long gone, the most popular is the LM13700, but a more modern is the OPA860

 

[JohnRoberts]

OTAs are great for many applications such as VCAs , VCFs, and in general musical instruments, but the THD is way to high for pro audio, some of them have input linearizing diodes to reduce THD. There are not many models that I know of, the CA3080 is long gone, the most popular is the LM13700, but a more modern is the OPA860

The RCA 3280 (?) Had Gilbert linearizing diodes integrated in. I think I used that one an ill fated CX record NR decoder (expander) kit, last century. I used a proforma circuit provided by CBS, but didn't copy their mistake (time constant was wrong).

I used lots of OTA in companding noise reductions wrapped around BBD delay lines and they were serviceable. BBDs are not very low distortion either.

Back in the 80s I used OTAs in a commercial Noise Gate/Limiter... When not gating or limiting the audio path was a unity gain op amp so pristine. When gating or limiting the OTA distortion was a lesser evil.

OTAs are commonly used in power amplifier clip limiters, again in topologies there they are not in the audio path below clipping.

JR

 

[user 37518]

OTAs are commonly used in power amplifier clip limiters, again in topologies there they are not in the audio path below clipping.

JR

Interesting, do you have any circuit example ?

 

[L´Andratté]

Merlin Blencowe´s Engineer´Thumb compressor:
http://www.valvewizard.co.uk/engineersthumb2.html
not a power amp, but a guitar pedal, same principle nonetheless,
turning OTA NF gain up turns audio signal down, so distortion/noise
get´s insignificant...

 

[JohnRoberts]

Interesting, do you have any circuit example ?

#1. I don't have schematic handy but they are published. Peavey's widely used DDT (amplifier clip limiting) has a resistor feeding into the high impedance + input of a non-inverting gain stage. The output of an OTA also connects to that node. When the circuitry detects onset of clipping it turns on the OTA opposing the resistor and reduces the input voltage. 

#2 Back in the early 80s I did a quad gate limiter (LOFT/Phoenix audio lab). In that circuit I put the OTA in the negative feedback path of a simple inverting op amp. Again the OTA was only affecting the audio path when limiting or gating, so the un-gated audio path was as clean as a unity gain op amp (very clean).

I do not recommend OTAs for serious dynamics work, but even then the sonics will typically be dominated by the side chain manipulations that effectively multiply the audio signal even when using a perfect gain element. 

The modern generation of THAT corp VCAs are quite good.

JR

 

[ThreepE0]

Every user claims a noise improvement, and no manufacturer discourages these proclamations. Some make them themselves. It’s impossible. Use a line amp after a decent preamp if you need more. Better yet, digital gain in post.

But. Now the kidz also believe such devices are essential for “driving the preamp”. How do you argue with that? They all want 5%+ distortion on everything.

But that’s exactly what all of these devices are designed to do: provide gain as cleanly as possible in order to boost signal away from the noise floors of preamps that are maxed out and have high noise floors in those scenarios. So yes, these devices can and typically do help drastically with noise.

 

[emrr]

But that’s exactly what all of these devices are designed to do: provide gain as cleanly as possible in order to boost signal away from the noise floors of preamps that are maxed out and have high noise floors in those scenarios. So yes, these devices can and typically do help drastically with noise.

You
Are
Wrong

Defies all natural law. It’s extremely tiresome.

Add noise in front of an existing noise floor = more noise.

It is extremely disingenuous to suggest there’s some other magic ‘clean gain’ that no maker of preamps ever sought to incorporate; it’s insulting to all makers of preamps.

It will never be more than a band-aid on a head wound.

Snake oil. If you believe the ad copy you are misguided. If you are writing the ad copy you are a liar.

If I ever walk into a tracking session with a bunch of this crap in line, it will all be unplugged and go in the trash, and a proper lesson in gain staging and recording levels will be applied.

 

[Bo Deadly]

But that’s exactly what all of these devices are designed to do: provide gain as cleanly as possible in order to boost signal away from the noise floors of preamps that are maxed out and have high noise floors in those scenarios. So yes, these devices can and typically do help drastically with noise.

The noise floor of a small signal audio chain will depend entirely on the design of the amplifier and performance of the amplifying semiconductor junctions. So a mic line booster would help if it had superior noise performance to the downstream preamp. However, unless the preamp is complete junk, in practice, the opposite will be true. Because of the limitations of phantom power, mic line booster designs are always inferior to even a vaguely good mic pre. Even cheap mixers from Behringer and Mackie have CFIA preamps which approach the limit of noise and distortion performance in a high gain amp. Mic line boosters might be useful for some scenario like a ribbon mic that needs a little more gain than the preamp can provide. But I say "might" because most folks are probably going into an audio interface with a noise floor far below what the mic pre is putting out and therefore you can add any amount of gain digitally with zero impact on noise performance.

 

[JohnRoberts]

Dead thread talking...

JR

 

[soapfoot]

I don’t think the utility of these devices is “clean gain”—I think it’s high load impedance. Right?

Something like an RCA 44BX will obviously perform quite differently depending on its load, and its source impedance varies pretty dramatically with frequency

At 1k it’ll be nominally a few hundred ohms, but at 100 Hz it approaches 2k. This is obviously pretty challenging for a typical preamp with a few kOhms input Z

if the ribbon is over-damped due to a low preamp input Z, mic output will be very low (and more gain will be required… which means more noise from the preamp, right?)

If you’re using a U87 or 414, OR if you’re already using a preamp with an astronomical input Z (like the one from AEA, which is intended to maximize the performance of their ribbons), such a device would be less-than-useless.

But for many ribbons (and, plausibly, a few moving coils) the extremely high input Z of the cloudlifter or FetHead does indeed seem to make a difference on performance.

We have two vintage Neve desks at our place (an 8088 and an 8058) which both use the 31102 input module. For a given level-to-tape, using a Coles 4038 or RCA 44 directly into those input modules does result in a higher overall noise floor than when adding a Cloudlifter to the chain.

The mics sound subjectively better with the FET device added to the chain, as well--transients are a bit brighter and snappier and more open-sounding, etc.

But we’re under no impression that any of this has anything to do with the “clean gain.” We attribute it to the mics being allowed to operate in a bridged impedance condition throughout their entire bandwidth.

YMMV and I’m obviously open to being wrong about WHY we’re hearing what we’re hearing, but there is zero dissent at our place about using them with our older ribbons.

As a bonus, using them allows us to safely hot-patch tie lines on our TT bays with the Neve's global phantom enabled... but that's just a nice side-benefit.

 

[abbey road d enfer]

I don’t think the utility of these devices is “clean gain”—I think it’s high load impedance. Right?

Something like an RCA 44BX will obviously perform quite differently depending on its load, and its source impedance varies pretty dramatically with frequency

At 1k it’ll be nominally a few hundred ohms, but at 100 Hz it approaches 2k. This is obviously pretty challenging for a typical preamp with a few kOhms input Z

if the ribbon is over-damped due to a low preamp input Z, mic output will be very low (and more gain will be required… which means more noise from the preamp, right?)

If you’re using a U87 or 414, OR if you’re already using a preamp with an astronomical input Z (like the one from AEA, which is intended to maximize the performance of their ribbons), such a device would be less-than-useless.

But for many ribbons (and, plausibly, a few moving coils) the extremely high input Z of the cloudlifter or FetHead does indeed seem to make a difference on performance.

We have two vintage Neve desks at our place (an 8088 and an 8058) which both use the 31102 input module. For a given level-to-tape, using a Coles 4038 or RCA 44 directly into those input modules does result in a higher overall noise floor than when adding a Cloudlifter to the chain.

The mics sound subjectively better with the FET device added to the chain, as well--transients are a bit brighter and snappier and more open-sounding, etc.

But we’re under no impression that any of this has anything to do with the “clean gain.” We attribute it to the mics being allowed to operate in a bridged impedance condition throughout their entire bandwidth.

YMMV and I’m obviously open to being wrong about WHY we’re hearing what we’re hearing, but there is zero dissent at our place about using them with our older ribbons.

As a bonus, using them allows us to safely hot-patch tie lines on our TT bays with the Neve's global phantom enabled... but that's just a nice side-benefit.

That's correct.
These vintage ribbon mics were designed to operate in conjunction with unloaded secondary input xfmrs. It is well specified in RCA datasheets.
This resulted in very high impedance (>10k) at higher frequencies.
In contrast, modern preamps have a much more constant impedance, so, in comparison, they tend to choke high frequencies from the microphone. They do the same at the resonant frequency of teh ribbon, which makes them somewhat lacking in the "chest" register.

 

[soapfoot]

That's correct.
These vintage ribbon mics were designed to operate in conjunction with unloaded secondary input xfmrs. It is well specified in RCA datasheets.
This resulted in very high impedance (>10k) at higher frequencies.
In contrast, modern preamps have a much more constant impedance, so, in comparison, they tend to choke high frequencies from the microphone. They do the same at the resonant frequency of teh ribbon, which makes them somewhat lacking in the "chest" register.

Thanks!

So if a ribbon mic is over-damped (at some or all frequencies) and has lower-than-optimal output as a result, it will require more overall gain from the preamp(s), right?

So the gain provided by the FET device is a bit of a red herring. The real benefit is a fully bridged impedance condition that necessitates less overall gain, yes?

I feel like I've noticed this with our Cloudlifters and FetHeads in conjunction with the Neves and older ribbons.

Adding the 22dB (IIRC) of the FET device to whatever gain the 31102 is providing almost always seems to add up to a lower total than would be required from the 31102 alone to achieve the same level-to-tape.

Add to this the fact that the 31102 isn't particularly quiet, and you have a not-really-all-that-subtle difference in noise floor.

 

[emrr]

Oh dear, now I've offended soapfoot and all other Neve owners. You are correct on the loading aspects with the Neve, and with many other modern preamps in regards to ribbon behavior.

It points out that there are STILL multiple theories of mic and preamp design to contend with.

There's the approach that assumes condensers. There's the approach that assumes ribbons. Amongst the ribbons, there's RCA with very high bridging, and there's Western Electric and Altec with matching Z. These things are all jumbled in practice and used interchangeably without much thought.

The Neve's were built in an era assuming lots of German condensers that didn't need big free step-up gain, and would be fine with a predictable more linear input impedance that works with said mics. Put the Hardy 990 types in that camp too.

The RCA/etc were built assuming a need for light loading and max 'free' step up gain from a high ratio input transformer, both to overcome low source levels and tube noise. The extremely light loading moves the collision points on the impedance graphs far enough out of range to ignore, mostly. The performance differences between old cheap versus old expensive transformers can highlight this.

I don't own anything like a Neve. I own lots of RCA/etc. Downside of the RCA/etc is....all the 'free' gain, usually needs a pad with loud sources or high output mics. Heck, I have to pad ribbons in some cases. Upside is....all the ribbons sound fine and noise free, ambient sound is always louder than preamp noise.

 

[moamps]

This thread intrigued me to go to the Cloudmicrophones website and find the Cloudlifter CL1 specifications, especially the input impedance but without success. Maybe today is just a bad day for me.
But I found this:
"The Cloudlifter allows you to cut through the mix on stage with increased gain before feedback."
This is something new for me.

 

[emrr]

This thread intrigued me to go to the Cloudmicrophones website and find the Cloudlifter CL1 specifications, especially the input impedance but without success. Maybe today is just a bad day for me.
But I found this:
"The Cloudlifter allows you to cut through the mix on stage with increased gain before feedback."
This is something new for me.

Oh God. WTF are they talking about?!?! Word salad ad copy.

 

[emrr]

A device like the Mogaine has a more believable chance of increasing gain without noise, given it's transformer step up gain, but with a 3K3Ω input Z, does not totally address the loading issues.

 

[soapfoot]

Oh dear, now I've offended soapfoot and all other Neve owners. You are correct on the loading aspects with the Neve, and with many other modern preamps in regards to ribbon behavior.

It points out that there are STILL multiple theories of mic and preamp design to contend with.

There's the approach that assumes condensers. There's the approach that assumes ribbons. Amongst the ribbons, there's RCA with very high bridging, and there's Western Electric and Altec with matching Z. These things are all jumbled in practice and used interchangeably without much thought.

The Neve's were built in an era assuming lots of German condensers that didn't need big free step-up gain, and would be fine with a predictable more linear input impedance that works with said mics. Put the Hardy 990 types in that camp too.

The RCA/etc were built assuming a need for light loading and max 'free' step up gain from a high ratio input transformer, both to overcome low source levels and tube noise. The extremely light loading moves the collision points on the impedance graphs far enough out of range to ignore, mostly. The performance differences between old cheap versus old expensive transformers can highlight this.

I don't own anything like a Neve. I own lots of RCA/etc. Downside of the RCA/etc is....all the 'free' gain, usually needs a pad with loud sources or high output mics. Heck, I have to pad ribbons in some cases. Upside is....all the ribbons sound fine and noise free, ambient sound is always louder than preamp noise.

hahaha no offense taken, of course!

But yes, a fair amount of “Neve-inspired” preamps are on the market today