Simple diode negative peak limiter for line level audio?

[K1JJ]

I wonder if anyone has a simple circuit to build a one diode negative peak limiter? It should be self contained, function with about 1 volt of audio and have a threshold adjustment.

I'd like to raise my audio level since the 4X1 modulators have the headroom. However, standard audio processing will not limit the negative peaks, thus the need for a neg peak limiter.  

In the past I used one in a class E rig and thought it performed well.

Also, is there a "soft" clipping vs: a "hard" clipping circuit?  


Thanks.

T

[ke7trp]

YES!  You are on to something here.. I have dreamed of this for the last year or so.  I would like to control my Neg peaks at the Audio rack and not in the transmitter.

Clark

[W1AEX]

That would interest me as well. Peavey has an interesting circuit they call "DDT compression" which prevents clipping in their audio power amplifiers. As I recall, it utilizes an op-amp that samples the output waveform and at a point slightly before the amp begins clipping, it imposes just enough limiting to keep the amp from bottoming out. I have no clue about the details, but then again, anything with electrons running through it confuses me.

Might make us sound less tubby Tom.

[WD5JKO]

I wonder if anyone has a simple circuit to build a one diode negative peak limiter? It should be self contained, function with about 1 volt of audio and have a threshold adjustment.
I'd like to raise my audio level since the 4X1 modulators have the headroom. However, standard audio processing will not limit the negative peaks, thus the need for a neg peak limiter. 
In the past I used one in a class E rig and thought it performed well.
Also, is there a "soft" clipping vs: a "hard" clipping circuit? 

Tom,

 All this can be done, but first consider that asymmetrical modulation done at a low level and then amplified to a high level will be subject to level shift in the r-c coupled stages. This will tend to undo the asymmetry created earlier. Also, if successful with passing the asymmetry through the active stages, then the modulation transformer will look at the asymmetry as AC voice with a DC component on it, so transformer saturation is a concern.

  So yes I think you can achieve what your looking for, but you might have to look at the design of your modulator first. Maybe try it out with a pulse generator that has a variable DC offset, and then see how that pulse shape and asymmetry is preserved on the final RF modulated envelope.

   I tend to believe that with plate modulation, doing the asymmetry thing is done best at a high level, and AFTER the modulation transformer. Just my opinion.


I did what I describe with a super modulated Viking I (200%) with "soft clipped" negative cycle where the progressive (1-2-4-8) attenuation kicked in at about 80-85% negative, and prior to that there was no asymmetry induced.
see reply 8 here, but in my case I had a solid state modulator with Rs=0:
http://amfone.net/Amforum/index.php?topic=22341.0

Jim
WD5JKO

[WA1GFZ]

Try a nice slow diode like a LED or two in series. Want it softer add a series resistor

[K1JJ]

Thanks for the info, Jim.

So you're saying that when using a low-level neg peak limiter, I may have problems cuz the asymmetry is not preserved thru the later chain and I'm using mod iron in the chain vs: using say a class E rig that will not be affected by the DC component, etc?


I've tried the classic high level 3-diode circuit in the past and it worked, but found it created splatter that was not there before. I've tried it on three different rigs with the same poor side-crud results. I think Don/KYV came to the same conclusion.  Though, others seem to have had better results than me.   I noticed your experiments with the 3-diode circuit from back in 1987. What are your conclusions today?  Have you run some modern spec analyzer tests that shows the more complex circuit you described is a better bet?  I notice you had a splatter filter drawn in.


Here's the schematic of the latest  QIX class H modulator he posted recently. The neg peak limiter is in the middle of the page. It uses a few transistors and is more complex than just a single diode and resistor. I wonder if the added complexity is needed, Frank, and what is the advantage?  I'd imagine it would be to soften up the attack time.

http://amfone.net/Amforum/index.php?topic=23014.0




Maybe I'll try the simple LED/ variable pot technique you suggested, Frank, as a first trial to see what happens.

It seems that if boxes like the Volumax, which use a low level diode neg peak limiter in commercial service work OK, then maybe low level limiting might work in my 4X1 mod-transformer-laden rig.

More ideas about this?

T

[WD5JKO]

So you're saying that when using a low-level neg peak limiter, I may have problems cuz the asymmetry is not preserved thru the later chain and I'm using mod iron in the chain vs: using say a class E rig that will not be affected by the DC component, etc?

   Yes!   A speech clipper by itself due to phase shift after the clipper will diminish the clipping action. Add to that asymmetry, and the R-C coupled stages and transformer non-linearities will lessen the offset of the +/- peaks, if not eliminate them. You either need to DC couple, or have some sort of DC restoration. Taking the R-C time constants stage by stage and decreasing them (lower the -3db cutoff frequency down to 10hz or lower) will help for non periodic speech (pauses between syllables). If you do any of this, remember that transformers don't like DC.

I've tried the classic high level 3-diode circuit in the past and it worked, but found it created splatter that was not there before. I've tried it on three different rigs with the same poor side-crud results. I think Don/KYV came to the same conclusion. 

  I agree, I don't like the three diode circuit either. The circuit I linked to is different. Sorry for the crude schematic, but hey that was drawn in 1987 to Hoisy W4CJL on his provided letterhead.

Maybe I'll try the simple LED/ variable pot technique you suggested, Frank, as a first trial to see what happens.

Ok, just remember that you said line level audio at 1 volt. I presume RMS? The threshold of an LED varies on color, but is around 1.2 volts. The cool thing about LED's is the threshold is abrupt with a near perfect kneee. I'm not talking about the bright ones, but just the red/green/yellows that we've known for years and years. You might need to boost the signal to 10v peak, do your diode/resistor stuff, and then go back down to 1v. Could be done with a dual op-amp.


Give consideration to using a function generator or pulse generator with DC offset capability. See what your modulator can do. if the asymmetry goes away in the modulator, there is little point playing with low level clipping/asymmetry circuits.

One more idea to accomplish what you want. Take the 4-1000 RF tube screen voltage, and since it is modulated, add a clipper diode there with variable threshold, and series resistance. This will make a "soft-clipper". I tried this with a 6146 RF tube, and limiting G2 downward swing to +30v with some series resistance did the trick for me. The action was smooth without an abrupt change in the RF output envelope pattern as 100% negative modulation was approached. The idea could be scaled up to a big RF PA tetrode.

Jim
WD5JKO

[K1JJ]

Jim,

Good info, tnx.

I will try your 4X1 screen diode idea after the low level LED idea.  Using the 4X1, (screen voltage = 450v at 140ma with a series 15H self-modulating choke) could you draw out the circuit you described with suggested values and perhaps scan and post it?  Or describe it in detail like a schematic? I'll bet others here might like to try something similar too.


T

[W2XR]

Jim,

Good info, tnx.

I will try your 4X1 screen diode idea after the low level LED idea.  Using the 4X1, (screen voltage = 450v at 140ma) could you draw out the circuit you described with suggested values and perhaps scan and post it?  I'll bet others here might like to try something similar too.

Another question:  Since the famous Volumax is a low level box using a neg peak limiter, did it really work with the broadcash transmitter industry considering the AC coupling problem thru the audio chain and mod iron limitations as you described above?   I used to use one and cud see the low level neg clipping action, but don't remember how well it worked on my plate modulated rigs to limit neg peaks at the high level...

T

Hi Tom,

Actually the CBS Audimax and Volumax (referred to within the industry as the "Max Brothers") were designed in the days of plate modulated transmitters, and their effectiveness with these rigs was supposedly quite significant. They were considered a major step forward in terms of AM radio audio processing, compared with the wideband AGC limiters available a the time, such as the RCA BA-6A, the Collins 26U-1, and others of that ilk. The Audimax and Volumax really were the first processors designed to the unique requirements imposed by AM broadcasting, and CBS sold a boatload of them.

Their hold on the industry was largely complete, until the advent of the Dorrough DAP-310 in the early '70s, and later, the Orban Optimod.

I had the pair, but I did not like the audio quality thru them. As always, your mileage may vary. Sold them off a few years ago.

73,

Bruce

[K1JJ]

Hi Bruce,
Now that I think of it, it was the Dorrough DAP 310 I had that used the diode neg peak limiter. I don't remember if the Volumax had one or not.  But the same question is posed: Did the BC rigs maintain the asymmetrical waveform all the way to the high level modulators?

Right now I'm trying out the LED idea Frank suggested. I got it to limit the negative audio peaks at the 4V level really nicely. I'm warming up the 4X1 rig now to see if it works in the high level modulators as well. Will let ya know what happens..

T

[W2XR]

Hi Bruce,
Now that I think of it, it was the Dorrough DAP 310 I had that used the diode neg peak limiter. I don't remember if the Volumax had one or not.  But the same question is posed: Did the BC rigs maintain the asymmetrical waveform all the way to the high level modulators?

Right now I'm trying out the LED idea Frank suggested. I got it to limit the negative audio peaks at the 4V level really nicely. I'm warming up the 4X1 rig now to see if it works in the high level modulators as well. Will let ya know what happens..

T

Hi Tom,

Yes, they did.

When I was a young BC engineer during my college days in the early '70s, we had the CBS Volumax and Audimax pair in the station audio chain. We could routinely modulate in excess of 100% in the positive direction after the FCC amended the rules limiting positive-going modulation to 100%.

I hope that this answers your question. If not, perhaps I don't understand it.

73,

Bruce

[K1JJ]

Update:

I just tried using a low level LED as a neg peak limiter for the 4X1 rig.  It required a good 5+ volts of audio to operate as Jim suggested.

I was able to effectively limit the negative peaks of the class C final amplifier as shown by adjusting the pot on the LED and watching the 100% line of the neg peaks get fatter on the scope as I watched the RF envelope.  So it appears the low end coupling is good, though not DC coupled. However, it added a tiny amount of audio "fuzz" as expected.

Anyway, I want to try your screen diode idea too, Jim, and see if that is an improvement.  Though, I wonder if the phase shift caused by the self-modulation screen choke will create a problem...

T

[w3jn]

Ifn ya want a project to mess around with, Eric Babour (used to work for Svetlana tubes) ginned up a soft limiter using a 6BN6 gated discriminator tube.  It was part of a tube audio synthesizer.  It shows up on google.

[steve_qix]

Hi Tom,

The reason the single diode works in the class H modulator for the class E rigs is that the system is DC coupled, and the diode (in the new class H modulator) is working with very low power levels (it's ahead of the modulator MOSFETs).

I use a low level negative peak limiter in the PWM modulators as well, but again - DC coupled.

I have used the 3 diode high level neg peak limiter for YEARS in broadcast and other transformer-coupled transmitters with no addition of garbarge products.  Once the system starts clipping, you will obviously generate some audio harmonics which can make the signal wider, but this will occur with ALL waveform modification systems 

Regards,

Steve

[WA1GFZ]

Tom,
Most LEDs have a breakdown voltage of about 1.7V so I am surprised it took 5 volts unless there is an internal resistor or multiple leds in series. You might try a cap across the led to soften the effect of the clipping. Also a series resistor ahead of the clipper may help. I use a similar NPL as Steve uses in my PDM modulator.
There are a number of ways to build a low level clipper. Biased diode is pretty good but might be too fast causing distortion.
As Steve said clipping action causes wavefrom modification which is distortion.

OH, BTW, your rig sounds great
 

[K1JJ]

Tom,
As Steve said clipping action causes wavefrom modification which is distortion.
OH, BTW, your rig sounds great

Yep, nothing comes free... We'll have to see how little crud we can get away with with the limiting.

I think the 4X1's audio is close to what I've worked for.  Check out this recording made by Al / W1VTP today using his Flex and sync detector. I was limiting the highs to about 6kc with the new digital low pass filter.  The reverb is configured to work on the extreme highs only - here's the results below.

Al/VTP said: "Bandpass was set for 7 KC.  Your sibilants are real punchy.  A lot like WBZ (1030 AM). Detection was SAM"



T

[Steve - WB3HUZ]

Sounds very FB - crisp, well balanced and loud, but not obnoxious or in-your-face loud.

[WA1GFZ]

I agree. The highs were nice and sharp and lows sounded like you were sitting in the same room.

[Opcom]

A vacuum diode has a nice soft curve. Perhaps an instrument diode? or any diode with a very close spacing.
A small signal low V_Gth FET might work.

Your signal sounded extremely clear BTW, very admirable!

[K1JJ]

Thanks very much for the comments guys...

After a week of bothering everyone for audio reports, I'm finally happy. I just knew there were a set of optimum settings that would make my average voice sound decent.  With the 6-band processor and 31 band EQ, it gets overwhelming and frustrating at times...

To continue the thread...others are wanting to do a low level limiter for their plate modulated rigs too.  I did more reading about neg peak limiting and saw an older thread that discussed this.  As already discussed above, to neg peak limit a transformer based rig (including mod transformer) it must be done with a HIGH LEVEL 3-diode circuit (or equivalent) due to phase shift and DC offset problems. I may install the 3-diode high level circuit and use it as a safety net for occassional overmodulation and to protect the mod xfmr.  To use it for heavier modulation may work OK, but at a slight cost of some minor crud. I'm also thinking about doing it in the screen circuit as a simpler, lower voltage alternative.

I'll post the final results, whatever they are.

T

[WD5JKO]

I will try your 4X1 screen diode idea after the low level LED idea.  Using the 4X1, (screen voltage = 450v at 140ma with a series 15H self-modulating choke) could you draw out the circuit you described with suggested values and perhaps scan and post it?  Or describe it in detail like a schematic? I'll bet others here might like to try something similar too.

Tom,   I am attaching what I did with my Gonset G50. A Heising modulated 6146 is a far cry from a 4-1000 modulated by a pair, but the screen diode clamp may work in a similar way..

    On a big rig, I might try the diode (or two in series), a series resistor like 5K 10w, and a bench variable DC supply that goes zero to +150v. The idea is to see the effect of clamping the downward G2 swing to some minimum voltage. I'm not sure how that 15H choke will react to the clamp....

Jim
WD5JKO

[K1JJ]

OK Jim,

Let's see if I can adapt this to the 4X1 rig....

So, the cathode of the diode(s) goes to the screen pin (after or before the screen choke?)  and the anode of the diode connects to a 5K 10W resistor - the other end of the 5K resistor goes to the plus of the variable 0-150V supply. The supply negative goes to ground.    Is this correct?

Do I need any bypass caps or resistors anywhere else?  Should it be a flat function when I freq sweep it looking for neg peak operation - and if not, do I put a compensation cap across the diode, etc?

T

[WD5JKO]

OK Jim,
Let's see if I can adapt this to the 4X1 rig....
So, the cathode of the diode(s) goes to the screen pin (after or before the screen choke?)  and the anode of the diode connects to a 5K 10W resistor - the other end of the 5K resistor goes to the plus of the variable 0-150V supply. The supply negative goes to ground.    Is this correct?
Do I need any bypass caps or resistors anywhere else?  Should it be a flat function when I freq sweep it looking for neg peak operation - and if not, do I put a compensation cap across the diode, etc?

   Tom, yea I think you got the idea. The diode cathode directly to the G2 pin. The 5K resistor is just a precaution against too much G2 current if you hook it up with the plate voltage off. That said, some series resistance with the diode might be desirable.

   On the 6146 in the G50, I did not have a series R with the diode, and the G2 voltage minimum was around 30v, and the G2 limiting was abrupt. Quite surprising to me though was the RF envelope at the same point just had a slope change suggesting "soft limiting".

75% mod before diode conducts:
http://pages.prodigy.net/jcandela/G-50/75_mod.jpg

85% mod, diode threshold being met:
http://pages.prodigy.net/jcandela/G-50/85_mod.jpg

Normal speech showing diode action. Without the diode, I was brightening at the baseline frequently:
http://pages.prodigy.net/jcandela/G-50/YEA_AM.JPG

The R's and C's on the 6146 G2 where an attempt to compensate for audio phase shift on the modulated G2 voltage. As it is, a triangle wave into the modulator results in an exact shape on the RF envelope anywhere from 50 Hz to 5 Khz. A more typical circuit really messes up the upper frequencies due to phase shift. I believe in your case with a self modulated screen grid that this approach is non applicable (NA).


The old Gonset G50 in stock form had one of the worst high level modulators I've ever encountered. My change using a P-P output transformer the way I did (modified Class A Heising) really cleaned the thing up nicely.

Jim
WD5JKO

[WBear2GCR]

Not sure if it will help but one might apply some "soft recovery diodes" to good advantage here...

Thinking out loud, perhaps for the DC "bias" component, if there were two feeds of bias, one from a stiff supply and one from a rather wimpy soft supply (maybe via a high Z resistor) that might cause the initial threshold through the wimpy supply to sag (failing to have enough current) down to the stiff voltage supply causing a kind of "soft knee" clipping? Each voltage would have to be fed through a diode to keep them from interacting...

                      _-_-bear

[ke7trp]

Tom.  I would suggest you remove that CHoke and modulate the Screen directly with the Dropping resistor.  Your not going to believe the difference in sound quality and punch when you do. I did back to back tests on 4-400s this way. We where shocked to actualy hear the difference on first key up. 

C