Noise Reduction System using 1176 Style Circuit

Hi All,

I have been playing with the Urei 1176 preamp and gain control circuits in LTSpice and I have found that with a few minor tweaks I can make a noise reduction circuit like the DBX or Dolby noise reduction systems or LM1894 chip.

The following is the standard 1176 preamp but I added a smallish capacitor to make the low pass filter (and inductor to make filter 2nd order and lower noise further):

[click on these to make them larger]



The gain control circuit has been changed so that the bias is 0 (meaning the JFET is biased on instead of nearly off) and the diodes are reversed so that the rectified control voltage goes down and not up (meaning the JFET turns off instead of on in reaction to the presence of high frequencies).



And of course the sidechain has to come from the input (connection not shown) and get boosted instead of coming from the preamp out. And of course the sidechain is high passed in a complementary way to the preamp low pass filter so that only high frequencies turn "off" the low pass filter.

This plot of a white noise burst shows that it's working:



Gray: The input signal (input to R2)
Green: The output signal (actually net of inductor and capacitor connected to JFET)
Red: The JFET gate
Blue: Unmixed control voltage (output of gain control circuit)

So you can see that the high frequencies are filtered until the gain control circuit detects high frequency signal and then it drives the fet off so that high frequencies are passed.

This pic shows a closeup of the noise reduction in action:



This pic shows the AC response of the preamp LC filter and of the gain control LC filter which are of course inverses of each other.



Green: The low passed input when no high frequency signal is present (meaning JFET is on)
Red: The high passed sidechain input means only high frequencies will trigger JFET

The question is, for anyone with knowledge of the aforementioned noise reduction systems, do you think this will work?

I'm thinking about making some PCBs for the 1176 preamp and GR circuits (IO and power handled elsewhere) but I'm wondering if I should add the few components necessary to have an option for this "noise reduction" twist on the 1176 Yeah, it's a little kooky but it might be useful for something like a spring reverb or stand-alone noise reduction unit like the Rocktron HUSH.

DBX, Dolby and DNL are VERY different entities. You are probably better off by getting a thorough understanding of the details in these systems first.

Jakob E.

ioplex said:

Hi All,

I have been playing with the Urei 1176 preamp and gain control circuits in LTSpice and I have found that with a few minor tweaks I can make a noise reduction circuit like the DBX or Dolby noise reduction systems or LM1894 chip.

Click to expand...

Ok those are three different NR systems DBX is full range 2:1 and 1:2  encode/decode  NR, Dolby uses 2:1 ratio encode/decode but just below a threshold.

DNR (1894) is single ended and uses a sliding LPF just on playback.

You forgot the downward expander, similar to a noise gate but more gradual. and the combination sliding low-pass with downward expander (Dynafex). 

The question is, for anyone with knowledge of the aforementioned noise reduction systems, do you think this will work?

Click to expand...

It will do something...Will it work, depends on how you define "work". The simple sliding LPF can sound unnatural when you hear the nature of the noise floor changing with level. There is also a psycho-acoustic influence when HF hiss in the noise floor makes the signal appear more wide-band than it is.

Probably the best approach for single ended NR (IMO) is to combine downward expansion with a sliding LPF. That way you can use less of each for a more natural sound while getting excellent combined NR. This was made popular by the old Dynafex.  I made an AMR version in the early '90s that used two dbx VCA per channel, one for the LPF and the second VCA for the downward expander. Hard part for these is opening back up fast and cleanly to handle transients.

I'm thinking about making some PCBs for the 1176 preamp and GR circuits (IO and power handled elsewhere) but I'm wondering if I should add the few components necessary to have an option for this "noise reduction" twist on the 1176 Yeah, it's a little kooky but it might be useful for something like a spring reverb or stand-alone noise reduction unit like the Rocktron HUSH.

Click to expand...

I suspect an 1894 would be cheaper and not impact the path sound quality of a spring. Do springs have that much HF content to accommodate? (rhetorical). A fixed LPF and noise gate or downward expander might be more effective for a typical spring return.

JR

Thanks for the answers. So if this sliding LPF circuit is really not that interesting, it sounds like I should skip this for now and just do a standard 1176 circuit. If I break out the sidechain, that should by itself provide for some experimentation.

At some point I would like to make a nice spring reverb which I think would dictate a limiter in front of the tank driver as well as something after (or preferably as part of) the recovery amp. Based on your comments it sounds like a downward expander would be the first thing to try.

I'm not sure about the difference between noise gate and a downward expander but the nonlinear behavior of the 1176 fet / GR circuitry might be interesting wired as a noise gate. It would be the inverse of the 1176 as a compressor so that would make it a downward expander? It would be simple enough to make. It's the same as the circuit above but without the LC filtering. Although there's no threshold control in my circuit and I'm not sure if restoring the bias circuity would remedy that.

ioplex said:

Thanks for the answers. So if this sliding LPF circuit is really not that interesting, it sounds like I should skip this for now and just do a standard 1176 circuit. If I break out the sidechain, that should by itself provide for some experimentation.

At some point I would like to make a nice spring reverb which I think would dictate a limiter in front of the tank driver as well as something after (or preferably as part of) the recovery amp. Based on your comments it sounds like a downward expander would be the first thing to try.

I'm not sure about the difference between noise gate and a downward expander but the nonlinear behavior of the 1176 fet / GR circuitry might be interesting wired as a noise gate. It would be the inverse of the 1176 as a compressor so that would make it a downward expander?

Click to expand...

A noise gate is generally on above threshold, off below. A downward expander drops the gain X dB for every Y dB the signal drops, so the output drops X+Y dB.

A simple above threshold feedback style limiter can not be simply converted to an expander since the gain element is very non-linear but works because it's inside a feed back path.  As I mentioned already I used dbx VCA for mine.

For a spring reverb downward expanding the output will actually shorten the reverb time.

JR

It would be simple enough to make. It's the same as the circuit above but without the LC filtering. Although there's no threshold control in my circuit and I'm not sure if restoring the bias circuity would remedy that.

Click to expand...

Well just for kicks I simulated a noise gate twist on the 1176. The preamp is identical to the 1176:



and, as before, the gain control circuit polarity is reversed:



and I fixed the threshold circuitry.

But as JR explained it is not possible to just invert the 1176 and make a downward expander because there is no feedback. This circuit uses feedforward. If it used feedback the attenuated signal would hinder triggering the rectifier. Meaning it would have positive feedback and not the desired negative feedback. The circuit as shown does make a noise gate. But it's not particularly special or useful. To make a downward expander I'm guessing the gain control circuit would have to be quite different entirely.