That 4301 compressor

Thread Starter

Arjune

Joined Jan 6, 2018
169
I have attached a file of the THAT 4301 compressor. Will the circuit work as the data sheet schematic. I'm confused about the resistors should I ignore the values following the M or K. For instance R14 is 1K43. Is this 1K.
Screenshot_20210411-004330_Drive.jpg

I want to build this compressor for a microphone amp mixer. I'm waiting for my three THAT4301's to come from China and they are about $18 for one.
 

jwet

Joined Feb 29, 2020
9
Component values like this put the K or M designation where the decimal point would normally go- its a European style that makes some sense. 1K0 is 1.00 Kohm or 1000 ohms, 1M65 would be 1.65 Megohms. Its sort of based on significant digits.
This compressor circuit is likely pretty good but is a very old design. The combination of an RMS converter and a Voltage controlled amp with 3 buffer amps is pretty unique. THAT specializes in this kind of analog processing stuff, most is done in the digital domain these days but a compressor/expander is a good case for an analog solution. The company THAT is still around and still makes similar parts- the 4305 has the RMS converter and the VCA without the amps, you could use a few nice quality audio op-amps for the rest- you might have to do a bit of experimenting but it would be reproducible. The 4305 is listed at $2.49 at 1000 pcs and is still being made. Maybe they'll sample you a few for your project. Tell them you'll publish your results on line.
 

Ian0

Joined Aug 7, 2020
3,281
Resistors specified as 1K43, 383k etc. are E96 series. I wish people wouldn't use values outside the E24 series except were absolutely necessary. The circuit works perfectly well if these are rounded to the nearest E24 (or even E12) values,

so use 1.5k for 1k43, 390k for 383k, 5.1k for 4k99, 560k for 590k and works absolutely fine. I've built one.
 

Ian0

Joined Aug 7, 2020
3,281
THAT is telling customers to use the THAT4305 and three op-amps. Profusion has the THAT4305 for £4.19
https://www.profusionplc.com/parts/that4305q16-u
The signal only goes through one of the op-amps, the other two are on the sidechain, so don't need to move very quickly.
You could use an NE5534 and an LM358, and they would only cost 79p and 38p respectively for one-offs.
Profusions's product is more-than-likely real!
 

Papabravo

Joined Feb 24, 2006
16,836
My pet hate - designers using E96 values when they are not needed!
Ever looked at a circuit diagram and wondered "why is this 221 ohms? - it must be doing something that requires real accuracy" only to find it's driving an LED.
That is manifestly not the reason.
At the dawn of the electronics era resistors were commonly 20% or 10%. Over time as the process got better and the manufacturing economies of scale improved 1% resistors actually became less expensive than their cousins with looser tolerances. This process was accelerated by the arrival of SMT processes in the early 1990's. I don't know if a similar transition is taking place to 0.1% resistors. I have been out of the game for a decade and a half. At some point the looser tolerance parts become pure unobtainium. That's the real reason. You did notice that the E6, and E12 series are not even on the list any more.
 
Last edited:

Ian0

Joined Aug 7, 2020
3,281
That is manifestly not the reason.
At the dawn of the electronics era resistors were commonly 20% or 10%. Over time as the process got better and the manufacturing economies of scale improved 1% resistors actually became less expensive than their cousins with looser tolerances. This process was accelerated by the arrival of SMT processes in the early 1990's. I don't know if a similar transition is taking place to 0.1% resistors. I have been out of the game for a decade and a half. At some point the looser tolerance parts become pure unobtainium. That's the real reason. You did notice that the E6, and E12 series are not even on the list any more.
You have a point!
(But I still have to pay quite a premium for a 0.1% resistor over a 1%, but 1% and 5% are about the same price)
 
With "real blue prints", the decimal point tended to no duplicate well and dissapear into the paper. 1K0 is preferable to 1K;

Another thing, you have designations of 5N and 5N5 pure with thins like gasses. 5N is 5 nines or 0.99999%; 5N5 is 0.999995%

Ratios and temperature coefficients are much easier to do in IC form. The resistors are on the same substrate so the temperature coef. is better. laser trimming occurs after encapsulation. The resistors are trimmed through the encapsulation.

5% might be normal manufacturing. 1% standard trimming Lower tolerances might require cooling and multiple passes to get the value right. That's just a guess.

Some capacitors used to be in uuF or micro-micro-Farad. Non-polarized capacitors did have a marking as to where the outer foil is on a capacitor. Now, it's typically not marked. It is especially important in tube designs. The foil needs to go to the low Z point.
 

MisterBill2

Joined Jan 23, 2018
9,308
I do notice that the circuit diagram gives no pin numbers for the IC, and at $18 each you certainly need to get the connections right the first time. So I hope that you can get some additional information. And it is not at all clear why such precise values are needed in a single channel audio device. Normally circuits would be designed to work with 5% tolerance parts if they woulf be mass produced.
 
Top