THAT Corp DN115 Attack & Release Circuit

[Bertu]

Greetings,

I have been struggling to understand how the attached attack & release circuit from design note 115 by THAT Corp. is actually supposed to to work. The design seems to assume that the comparator at the input swings low when the compression kicks in, and the constant current sources that are mirrored then discharge the timing cap. When the compression releases and the input signal goes back up to 0, the upper current mirror charges the cap back up.

This works as advertised in LTSpice. However, in the prototype, the comparator output is stuck at V+, and the output of the circuit hangs around -10V. I fail to understand how a comparator could ever have an output voltage of 0, or basically anything other than (near) V+ or V-. Is there something special about how the 393 is used here that I am missing?

Thanks for any insights or suggestions on how to think about this.

- Bertu

 

[abbey road d enfer]

I fail to understand how a comparator could ever have an output voltage of 0, or basically anything other than (near) V+ or V-.

The comparator output is an open collector. If the output transistor is not conducting, the output can be anything, depending on the current injected by R35 and Q5.
Having tested it myself, I can testify to the correct operation of this circuit.

 

[zamproject]

The 393 is open collector (already stated by abbey)

maybe not your issue but If you are using only one unit of a 393, datasheet state:
-All pins of any unused comparators should be tied to the negative supply-

As this comparative circuit have a feedback, maybe for some reason as soon as it hold from unexpected behaviour you can't unlock the comparator ?
Maybe as a start try to use a fixed voltage at Vref (-in) and only swing +in to see what happen to the comparator out ?

 

[elskardio]

Have you tried testing the prototype with a low offset precision opamp instead of the LM393?

I built this circuit years ago and used half of a NE5532 with great success.

Roger Foote was using an OP07 in his Pico compressor.

 

[JohnRoberts]

If you substitute a low impedance output op amp you may want to add a diode in series to mimic the open collector output of the comparator (cathode pointed toward op amp output). Or you could just copy the application note circuit verbatim.

JR

 

[Bertu]

Thanks everyone for the tips and feedback, great to know that someone else has built the circuit successfully.

Regarding the open-collector output of the comparator: if the output transistor is not conducting and R35 and Q5 inject current, where does that current flow? Q3 based on my understanding is a current mirror of a very low current CCS composed of Q1/Q7, so it wouldn't be able to sink much current?

I am not questioning the circuit, I am trying to understand it so I can get an idea what went wrong in my prototype, which is (intended to be) a faithful representation of the original design note.

Thanks!

 

[JohnRoberts]

Thanks everyone for the tips and feedback, great to know that someone else has built the circuit successfully.

Regarding the open-collector output of the comparator: if the output transistor is not conducting and R35 and Q5 inject current, where does that current flow?

With the comparator output high impedance there is no current drawn through R35. The current fro q5 will charge up C18

Q3 based on my understanding is a current mirror of a very low current CCS composed of Q1/Q7, so it wouldn't be able to sink much current?

Q3 is a current sink (basically to discharge C18).

I am not questioning the circuit, I am trying to understand it so I can get an idea what went wrong in my prototype, which is (intended to be) a faithful representation of the original design note.

Thanks!

By looking at the voltages at the different nodes you should be able to figure out what is wrong.

JR

 

[Bertu]

Mystery solved. Originally I couldn't find the J113 Jfet that the design note specified, so I used what I thought was a close enough alternative, but it wasn't close enough. Finally found the correct Jfet (not sure how I missed it the first time on mouser.com) and everything seems to work as intended now.

Thanks for the help and encouragement to keep looking at LTSpice.

- Bertu