Next SINUS project??

[radium98]

does it seems to to work as a compressor and limiter , i mean amplify weak voice and decrease high voice .to make them at constant level as possible near 0db or 75khz.I made the simple limiter and the mono version of pira but none works , i did not see any difference .i will post pictures sooner.

[reverend]

Just to be clear you can get as much roll-off as you want from an active filter, you'll just need more stages, same as with a passive filter. But if you only have the one opamp to spare then the best you'll get is 12db/octave. Two opamps gives you 24db/octave, three would be 36 etc. Not really much point going beyond 24db normally though

With a single op-amp it's possible to get 18 dB/octave roll-off. This site https://www.beis.de/Elektronik/Filter/A ... ilter.html is really helpful for designing op-amp based filters and has a separate page for 3-pole (18dB/oct) filter design. Bear in mind that for anything more than about a 5 pole filter, you will need to start using very high tolerance components (ideally 1%) which is easy enough for resistors but for the capacitors can be difficult and expensive to get hold of.

[sinus trouble]

Thank you all for the comments!

I have made some good progress on the limiter circuit! The 2N3819 is very hard to get these days, So i chose the BF256B as a replacement!

The latest schematic below!

[sinus trouble]

Schematic_Sinus Audio Limiter_2023-01-31.png

[sinus trouble]

I also have a sneak preview of the PCB!

[sinus trouble]

Limiter Preview.PNG

[sinus trouble]

Lim Prev 2.PNG

[Albert H]

Sinus - that's fine. However your design lack three things - 15 kHz lowpass filter, an overshoot clipper and options for stereo connection.

I'll be happy to provide a really excellent 15kHz lowpass filter circuit, and I can advise on how to interconnect two of your limiters for stereo operation. The clipper isn't too complicated, either, and guarantees that you'll never go over modulated (at the expense of distortion if you persist in sending very over-level audio!). Adding a clipper (and clipping indication) is certainly something you should consider. Your clipping indication will need some kind of pulse stretcher for the indication, so that very brief peaks are indicated.

[sinus trouble]

Cheers Albert!

Limiters are something that i have used often, but never really built before so i am still learning?

I used the redundant Op Amp of the TL074 as a "Sellen Key" active filter to roll off @ 15Khz! But again this is quite modest!

How the circuit will perform in regards to overshoot, Attack and release remains to be seen? The LED limiting indicator looks to be a neat feature? Hopefully i can get some prototypes ready for testing soon!

[Albert H]

That's "Sallen & Key" - the two chaps who derived the filter!

I use either cascaded S&K filters or M-derived gyrator filters. The problems with S&K filters is that they're relatively low "order", reducing the steepness of the rejection curve, and if you try to design a higher order S&K filter, they're prone to ringing.... Simple op-amp filters are always compromises (like all electronics!) and it's a set of trade-offs between phase accuracy, ripple, steepness and distortion artifacts....

That's why I (and most commercial designers) prefer gyrator-based filters. The gyrator networks emulate inductors (very effectively) and it's fairly easy to design for the shape you want with minimal unwanted phase shifts, ripple, distortion and noise. The circuit I prefer uses 1% resistors and (ideally) 1% capacitors, but you can get away with 5% capacitors with little adverse effect. I like the NE5532 dual op-amps - they're very low noise and distortion, and have plenty of output current available, so they don't run out of grunt.....

All the capacitors are the same value throughout (giving economy of scale) and the slightly weird resistance values are made up of series or parallel pairs of resistors. My in-house stereo coder has one of these filters in each channel, but built with a couple of extra stages, giving a very steep cutoff, allowing the audio to go up to 16.3 kHz. However, after extensive listening tests, the difference was so marginal that it really wasn't worth the expense of the extra stages!

The circuit that I'll put up here is around -55dB by 19khz, which I've found to be plenty steep enough. I'll tidy up the circuit diagram, and I'll put it up here later.

Incidentally, I appreciate that the Limiter circuit that you're playing with uses a single supply rail, but you can save components and get more headroom if you redesign for dual rail....

Circuit later today.

[radium98]

nice , one question do that gave agc to amplify weaken level and keep then to near 0db to not overmodulate over 75k

[Albert H]

The airchain that I favour uses a slow-acting AGC first, to overcome major disparities in programme source levels, followed by a pre-emphasis stage, then a two or three-bad limiter, with an overshoot clipper afterwards for safety, then the stereo coder.

[sinus trouble]

nice , one question do that gave agc to amplify weaken level and keep then to near 0db to not overmodulate over 75k

Hello Radium

No limiter is perfect, but if set up correctly should give good results!

Yes the limiter increases the gain of low level signals! If you look at the schematic IC1.3 is a low pass filter with a cutoff around 15khz

However, this IC is configured for "Unity Gain" which means that it merely keeps the gain "Original" there is no gain if that makes sense?

IC1.4 on the schematic has a very large gain! Especially when limiting is disconnected!

Q3 then controls the gain of this IC to keep everything as level as possible!

[sinus trouble]

The thing i find interesting about this circuit is that circuitry around IC1.1 and IC1.2 seems to sample the positive and negative peaks?

In turn i guess this would minimise offset and give a smoother limiting response?

[sinus trouble]

Sinus - that's fine. However your design lack three things - 15 kHz lowpass filter, an overshoot clipper and options for stereo connection.

I'll be happy to provide a really excellent 15kHz lowpass filter circuit, and I can advise on how to interconnect two of your limiters for stereo operation. The clipper isn't too complicated, either, and guarantees that you'll never go over modulated (at the expense of distortion if you persist in sending very over-level audio!). Adding a clipper (and clipping indication) is certainly something you should consider. Your clipping indication will need some kind of pulse stretcher for the indication, so that very brief peaks are indicated.

Cheers Albert!

Looking back on one of your previous posts, A stereo modification would not be too difficult to implement!

Your "Clipper" suggestion would be worth looking into, as overshoot can be a big problem!

Overshoot seems to be a tricky problem to address? Whilst the active components have no issue dealing with this! Passive components need time to react?

[jvok]

You need to sample both peaks because some audio signals can peak much higher in one direction than the other. Especially voice.

You can reduce overshoot by having a faster attack time, as long as you keep the release time long to stop it pumping too much. But you can't go much faster than 3ms or it will sound distorted. My limiter design uses a fixed 5ms attack and adjustable release time from 300-1000ms. Even then you'll still get overshoots so will still need a peak clipper. But having a fast attack time means the clipper won't be clipping for very long so you won't hear it. Also I found you need to set the limiter around 3-6db below the clipper to keep it inaudible.

[Albert H]

Jvok: You're exactly right. I found that the limit point had to be 4.5dB below the clipper for inaudibility. I also discovered that a green bidirectional LED made by Kingbright (essentially two LEDs wired back-to-back inside one package) had exactly the right conduction threshold for precisely going into clipping at PPM 6! The clipper I developed used the LEDs in the negative feedback path of an op-amp, and the did nothing until the threshold was reached, at which time they'd saw the peaks off very accurately (both ways, of course). It was a lot simpler than my previous designs that relied on the bias point of a pair of very accurately matched bipolar transistors!

The minor downside of using a clipper like that is that you won't see the LEDs flash, partly because you're only using them right at the very edge of conduction, and because the overshoot peaks are too fast. To overcome this issue, I came up with a simple pulse stretcher circuit (based on a quad comparator chip) that gave a nice, bright and highly visible blink from a bright Red "Clip" LED. I'll put all the details up here later today.

[radium98]

i would like at the tests see many or different
input level and how does it behave at the output .

[sinus trouble]

Thank you Jvok and Albert for the clarification!

I managed to find a list of specs for the original Veronica limiter? Ofcourse my version will be slightly different!

It seems to state an attack time of 1mS? I dunno how accurate that figure is?

Veronica Spec.PNG

[sinus trouble]

i would like at the tests see many or different
input level and how does it behave at the output .

I hope to carry out those tests once i have one completed!

In theory, it should perform ok? Or i could have messed it up and will be a disaster?

I will be ordering some PCBs soon! So one way or another? We will find out!