Very simple Audio compressor for FM transmitter

[ronsimpson]

The other thing I don't like about this circuit. The gate voltage must charge up from 0 to 1.6V before anything happens. (long time) Then form 1.6 to 1.7V it goes through its entire range. (short time).
So I added a circuit that brings up the gate voltage to -6bd reduction (with not signal). Then the peak limiter just needs to charge from 1.6 to 1.65V. (better response)

 

[MrAl]

Mr Al, 50% audio attenuation is only -6dB which is a small drop in level. Maybe you mean 50% of the range of resistance change? A fast Vactrol has a range of 50dB so you mean limit it to 25dB?

Gordon, the Vactrols (never heard that word before) are very good.

Hi again AG,

Yes my mistake, a spec of 50 percent audio attenuation can only be the from the test result of a particular circuit. Changing the circuit and we'd get 1ms to 25 percent attenuation for example

 

[MrAl]

The other thing I don't like about this circuit. The gate voltage must charge up from 0 to 1.6V before anything happens. (long time) Then form 1.6 to 1.7V it goes through its entire range. (short time).
So I added a circuit that brings up the gate voltage to -6bd reduction (with not signal). Then the peak limiter just needs to charge from 1.6 to 1.65V. (better response)

Hi Ron,

Yes, once you have established the right operating conditions you can usually customize the circuit to fit. If there is any really critical aspect though then a per-device adjustment may be needed.

 

[Gasboss775]

This application note from Siliconex has some good ideas about how to get the best results from fet based voltage controlled attenuators.

Note: There is a mistake, under figure 6 the formula should be the same as the one in the right hand column marked (3)

 

[ronsimpson]

Gordon, Thanks for the PDF. The same thing (more or less) is true for the MOSFETs. I have used the circuit to make the distortion better. (linearity) The idea of pre-charging the gate is what I was talking about in post #121.

 

[Gasboss775]

Something like this would ( see attachment ) allow control of the MOSFET from 0 volts upwards.

 

[audioguru]

A Fet has its resistance changed by the signal voltage that causes the severe distortion. A simple distortion reducer is to feed half of the signal at the drain to the gate but since a coupling capacitor must be used then it takes a long time to charge that causes a very slow attack time.

 

[Gasboss775]

I am thinking that provided the source and load resistances are fixed ( perhaps by using buffers ) then it should be OK to DC couple the distortion reduction network. If R2 & R3 are large in comparison to R1 and the source and load resistances then the control voltage only needs to be doubled as the point at the fet's drain will be around gnd potential and its resistance would be insignificant.

See attached.

 

[audioguru]

DC coupling and using the distortion reducing resistors results in a loud POP when a fast attack time is used. Because the DC control voltage to the gate feeds to the output.

 

[Gasboss775]

DC coupling and using the distortion reducing resistors results in a loud POP when a fast attack time is used. Because the DC control voltage to the gate feeds to the output.

I am wondering if an equal but opposite signal could be added to the output signal to cancel out control feed through. It must be possible as there are monolithic vca's that manage good control signal breakthrough. However, I'm not an expert on how these types of chips are designed, I'm just making ( hopefully! ) intelligent guesses as to how this type of thing is achieved.

 

[Tony Stewart]

Using Audacity and the right audio drivers with record on "What you hear" or "Stereo mixer" as an input, one can use your PC and Audacity as a compressor many variables and neglible distortion on CD quality input.

 

[ronsimpson]

one can use your PC

So the this thread started out using two transistors and one IC to make a limiter. Now we are talking about using a PC and its millions of transistors. LOL Your are right there are DSP limiters that do a very good job (in software). I do have a number of DSP audio effect boxes.

there are monolithic vca's that manage good control signal

Most certainly!

DC coupling and using the distortion reducing resistors results in a loud POP when a fast attack time is used.

Probably so. I used two 10meg resistors (no cap) and run the MOSFET at about 1k to 100 ohms so the "pop" would be very small. Doing the math. We have 20meg to the "gain" voltage and 500 ohm to ground. So not much of the gain voltage will get into the audio.

Something like this would ( see attachment ) allow control of the MOSFET from 0 volts upwards.

Gordon, Sorry what you show injects some part of the negative supply into the audio. At low gain most of the negative supply is in the audio and at high gain a small amount of supply is in the audio. When the AudioGuru talked about a pop when the gain changes. This is it in a big way. The source of the MOSFET must set on audio ground. You need to add the voltage to the gate.

Here is the way I got the gate DC worked out. I do not like it but.....
MOSFET M1 and M2 need to be a matched pair.
M2 is to make a voltage that just gets the MOSFET turned on. (actually twice that voltage) The gate is at turn just on voltage, and via the two 4.7meg resistors "Vref" is at twice that voltage.
The top of R15 is at ground DC. This is where audio is injected at mV levels. R15, R12 make a 2:1 divider so this MOSFET is also at just turned on.
So really what I did was take R13 (470K) and remove it from ground and bump it up to a voltage where the MOSFET is just starting to work.

R12, R15 were 10meg in the limiter/compressors I built 1975-1979. I can't remember the transistors but I know it took many volts on the gate to get it turned on.
-----edit-----
R7 should be 0 to 1k. I am using that to adjust the attack time. 100k is wrong.

 

[Gasboss775]

Just to clarify, the 2n2907 is being used as both rectifier and voltage to current converter, is that correct?

 

[ronsimpson]

Just to clarify, the 2n2907 is being used as both rectifier and voltage to current converter, is that correct?

Yes. More or less. If the output of the op-amp is too large the B-E junction works as a diode (rectifier). The base current causes collector current, that charges up a capacitor.

 

[Tony Stewart]

So the this thread started out using two transistors and one IC to make a limiter. Now we are talking about using a PC and its millions of transistors. LOL Your are right there are DSP limiters that do a very good job (in software).

the point of a PC is it is readily available, free and allows you to define the characteristics you want before you do brick and morter in hardware.

No design is any good without specs.

Here are some go-to reference settings:

• Threshold: -26db
• Ratio: 3:1
• Attack: 1ms
• Release: 10ms
• Output: +6dB

 

[Tony Stewart]

YOu can make a Really Simple Compressor with a CA3280A (RSC)

But if you need a the world's best pro compressor, you can go retro with vacuum tubes.
http://www.attackmagazine.com/reviews/the-best/top-20-best-hardware-compressors-ever-made/20/
http://www.triodeel.com/urei.gif

The infamous units built in Solid State
http://www.gyraf.dk/gy_pd/1176/1176sch.gif

 

[MrAl]

Hi Tony,

It is interesting that you brought this device up as i had almost forgotten about the OTA's from the past.
I've never used the CA3280 but i have used the probable predecessor the CA3080 for a solar panel max power tracker back in the 1980's. It was used to calculate the power of the array output due to it's multiplication type properties (modulator). I also used a "four quadrant multiplier" chip but cant remember the part number now, and that would also work for an audio compressor i think.

Unfortunately, the CA3080 was discontinued, and the CA3280 was supposedly marked to be discontinued also but im not sure of the status of that now (that was several years back) and i do see them being sold on the web here and there for prices anywhere from about 2 dollars each to 10 dollars each (USD).

Another possible chip is the LM13700 (about $2 USD), which i believe is still in production. Doesnt have quite the performance of the CA device, but it can probably work in an audio application ok. Also, a host of applications appear on the data sheet which includes a limiter circuit. This chip also has the linearizing diodes, i dont think the CA devices had that feature.

I think i have one CA3080E chip left from circa 1980.

 

[Gasboss775]

Intersil have started making the ca3080 again. I think that they realized there was a huge demand for Chinese copies of the chip, predominately from hobbyist and professional designers/builders of electronic music related products. e.g. guitar fx, analog synthesizers ( vca, vcf, ring modulators, etc

**broken link removed**

 

[Tony Stewart]

OTA now is more commonly referred to as Over-the-Air, and few remember how 6 decades of gm control can be used for VCA,VCF,VCR in the audio range with high SNR

I am an analog guy at heart and Interesting Shapes of Input Error Caused by Output Heat Flow, when cross-plotted vs. Vout on an X-Y scope. and how this all related to MHO's the unit of Transconductance but has nothing to do with this question but more to do with the ""What's All this Output Impedance Stuff, Anyhow?") and common centroid balancing.

https://www.ti.com/lit/pdf/snoa471

**broken link removed**

Harris Corporation Limited, formerly named Wing Fung Company, was established by Mr.Paul Yeung in 1977

**broken link removed**

 

[MrAl]

Tony:
Yeah, in this context it stands for "operational transconductance amplifier".

Gordon:
Where did you get that news from, about the remake of the part? I see their data sheet still is marked "obsolete", and none of the main suppliers seem to stock any although they stock the LM13700. Do you have a link to the article where Intersil states this or something like that?
You got my hopes up that this chip might make a come back

Intersil have started making the ca3080 again. I think that they realized there was a huge demand for Chinese copies of the chip, predominately from hobbyist and professional designers/builders of electronic music related products. e.g. guitar fx, analog synthesizers ( vca, vcf, ring modulators, etc

**broken link removed**

If you follow that link you get the "obsolete" marked data sheet.