The 3-Diode Negative Peak Limiter Circuit - Experiences ?

[K8DI]

I'm not familiar with your 4x1 rig's circuit type, does it use a capacitively coupled modulation transformer/modulation reactor setup, or a straight modulation transformer with DC on it? If the former, do you have this on the transformer side or the reactor side of the capacitor? If the latter, does it even work for transformer/reactor setups?

Ed

[K1JJ]

Hi Ed,

That's a good question.  Let's look into it more...

I am using a mod reactor that is coupled to the mod xfmr via a 2 uFd capacitor.  (so-called "modified Heising")    The NCL is connected to the mod coupling cap output, so there is B+ and audio superimposed at that point the NCL is connected.  IE, it is connected to the modulated HV line on its way to the final.

Are you thinking that the NCL should be connected to the mod xfmr output before the coupling cap or something else?

T
 

[K1JJ]

Ed,

I checked with a friend of mine who said:   "I agree the network will still provide a load on the negative cycle. It will just be AC coupled to the modulation transformer secondary."
I drew it out and agree.  Maybe you have a different take on it?

Another subject for the group - bear with me, this may sound confusing:  
I ran some tone tests on the NCL (negative cycle loading) and can go down to -99% negative with no additional IMD deterioration. Once -100% is hit, the IMD starts to climb rapidly, as expected.   In comparison, the 3-diode circuit needs to start working at about -90% to keep from going below -100%. When the keep-alive voltage is set at -99%, it does nothing to hold the cycle above -100%.  When set at -90%, the IMD starts to rise at -91% and gets worse as it approaches -100%.   So with the simpler NCL, we are gaining the distance from -91% to -99% by approaching the -100% point carefully with no high-level limiting.

I continue to think that a low level limiter is the best way to handle negative peaks, while expecting an occasional neg peak to flash the LED - but safely contained by the  NCL power resistor.  

IE, the NCL  is totally transparent until hitting -100% negative, and then the normal IMD goes up due to neg overmodulation.  I like its performance on my 4X1 rig cuz it adds no IMD unless below -99% neg but still protects the mod iron.   The NCL adds no high-level negative peak limiting, (or additional unwanted IMD)  just protection. Hope this is clear enuff.


T

[K8DI]

Regarding where it’s connected, and the topology, I’m thinking that in a transformer reactor circuit, it would be problematic to put it on the transformer side of the cap. The whole point of the thing is to keep a load on the iron when the modulated B+ is driven negative, and the tube cuts off/stops conduction/unloads the circuit.  On the cap input side there’s no B+, it’s driven below zero every half cycle...so what this would do is add the extra load R on the negative cycles. Your modulator tubes would be unequally loaded, one would get hotter, and depending on the excess capability in the modulator, you’d end up sagging the negative half cycles a bit...like an asymmetrical limiter might do. At the same time, the tube load would still go away at -100%, it would not prevent overmod (assuming the modulator is robust), BUT the mod iron would never be unloaded.

On the other hand, on the reactor side of the cap...as the audio frequency drops, it’s possible that an undersized cap may add enough reactance to the path that the modulator could be massively clipped but the limiter not even get hit, not that the circumstance of a big dc transient ever happens when you drop a cable plug or anything ;-)  but whether that actually happens is not the kind of destructive testing I enjoy doing with vintage mod iron and tubes...

Overall as an iron saver I think it’s a fine thing, a safety to protect the vintage parts that just can’t be had anymore.  As a limiter to increase average modulation, in my opinion, all the diode circuits suck in one way or another, be it high audio distortion, high rf imd/splatter, or just don’t work to prevent overmod.

Ed

[K1JJ]

Hi Ed,

Thanks for the second opinion. I think we are thinking the same here. Your precaution about a low value coupling cap (1-2 uFd is normal) could be a potential loophole for a DC transient to get thru. Good point. Hopefully the low level limiter (I use a DSP audio version here) will help to reduce these kinds of accidents.

My good friend added some more info to explain the circuit operation:

"Tom,
My circuit doesn't overload the negative cycle back to the modulator tubes unless you drive the signal below zero. The load resistor is referenced to ground, so how often are you driving negative. As you observed the 30 watt resistor barely gets warm on a modulator that easily makes 1 KW of audio.

This is unlike the 3 diode circuit when you load the negative cycle whenever the output goes below the plate voltage. That circuit does make the negative side of the modulator work harder to drive the resistor in parallel with the final and I bet it gets hot if the value is close to the operating impedance. The 3 diode increases the transformer losses and does slightly compress the negative cycle by making lots of heat. That actually works but a very inefficient way to compress the negative peaks and heat up parts.
 
The extra cap in series with the modulator adds to the value of the load resistor but compare the reactance of the coupling cap to the operating impedance of the circuit. the additional reactive load is peanuts.

The NCL resistor on the output of the Heising circuit will do a better job of loading the whole circuit against ringing. In a heising modulator the whole circuit is susceptible to ringing when there is a system transient. The more reactive components you have in a circuit the more chance you have for resonant frequencies being generated during a system transient.

That is the reason I went with a big single transformer rather than a smaller one heising connected. The core needs to be so much bigger to handle the dc offset. In a heising circuit you have the choke handle the DC while the transformer handles the AC. Since the choke doesn't have a primary winding you have greater design flexibility. "
 
---------

** Think I will tighten up the low level DSP limiter a bit. Currently I'm able to generate some VERY low frequency -100% negative peak LED flashes by bouncing the mic boom which tells me there needs to be some more limiting.  This is the kind of thing Ed warned about.   I usually try to use as little audio processing as possible, but peak limiting is important to parts safety.  I like that low level peak limiting works rarely, at the peaks only - and is mostly transparent.  DSP and sophisiticated peak limiters like the Optimod, CRL-300A types are what I am talking about rather than high level semi-clipping.


T

[K1JJ]

This is an actual on-air SDR video Rob/ W1AEX made of my 4-1000A plate modulated rig when I was running the 3-diode high-level limiter a week ago.  The first part shows me hitting the audio hard, leaning into the 3-diode circuit. It shows my signal spreading out about +- 20 Khz.

The last 1/3 of the video starting at about :34 shows the audio backed off so that the limiter is not being used. The negative peaks are now NOT touching -100% or being limited at all.  I am back to a normal +-  5 to 5.5 KHz.   The difference in IMD is tremendous.

I realize that the 3-diode circuit is not intended to be used to increase positive peaks, but many of us (me included) have tried to do this in the past and simply generated more splatter as a result.   High level limiting is extremely hard to do and would require a wheel barrel full of large parts to do right.  

Use a low level limiter and keep the audio away from -100% negative and we are golden.

* Important point: This demonstration is when using my low level   5.5 KHz DSP audio filter. This shows how a low level filter is useless when the IMD is being generated in the high-level area.



** Look to the left of my 3873 carrier for best view of the sideband.  Each division is 1 KHz.  There is another AM station on 3885 above. Notice the splatter he was receiving in the first 34 seconds of the video.

https://www.youtube.com/watch?v=2wSvYVbkKRE

:3 is worst bandwidth when talking directly into mic            :36 is best bandwidth with no limiting

Since this video I have added the new NCL and made some more changes that should help improve the IMD and bandwidth further.


T