A bit off topic, processing for AM

BillPaul · Post by **BillPaul** » Thu May 03, 2012 11:37 am

I know this isn't an ESSB topic, but I figure it couldn't hurt to ask this here.

I want to experiment a bit with asymmetrical audio processing for AM. That is, I want to be able to limit/compress the positive and negative peaks in my audio independently. When modulating AM, once you achieve 100% modulation, any further increase will result in distortion of the negative peaks, even if you have more headroom available for the positive peaks.

There are both audio-level and RF-level approaches to addressing this. My understanding though is that the RF-level negative peak compression mods only apply to high-level modulated transmitters (i.e. the 'big iron'). I'm not sure how they translate to low-level modulated rigs, like the Kenwood TS-850 in particular. (Note that I'm talking about the analog modulation scheme. I have the DSP-100 unit, but I don't use it for AM.)

When it comes to audio-level negative peak compression, commercial broadcasters tend to use something like this:

http://www.scmsinc.com/shop-item/studio ... processor/

This is nice, but somewhat expensive. And this is considered cheap: similar units from other vendors (e.g. Inovonics) can cost almost twice as much. Also, this unit does somewhat more than just asymmetric clipping, to the point where it duplicates some of the functionality I already have.

I'm just wondering if anyone is aware of a cheaper way to implement just the asymmetric processing feature.

-Bill, N1GPT

Post by **Voodoo Guru** » Fri May 04, 2012 3:15 am

Howdy Bill Paul,
You are hitting on a pretty deep subject here and without a doubt I know you have delved into this subject with intense research, so we'll refrain from stating anything obvious and just mention our own experience.

Running pure analog audio equipment here for our rack preserves the natural asymmetry in our voice. Of course the phase needs to be checked after each piece of equipment to be sure that it is correct all the way through to the transmitter. It has been noticed that the digital units seem to destroy this natural asymmetry and force it balanced to some degree. Some digital audio units tend to do this worse then others.

Note when we said digital units above, We are talking about digital rack gear for audio here, not the Digital AM Broadcasting processors which by design produce asymmetry via negative peak limiting, phase rotation, etc in the digital domain.

That being said, since all of our audio gear is analog, and it has been checked for the proper phase, It has a pretty good asymmetry. We also use analog split band compression for maximum voice density which gets kind of tricky in the analog mode because of the phasing that happens with the analog audio crossovers in the Multi-Band splitter and summer (Drawmer Three-Sum).

The cheapest solution for negative peak limiting for us was the Innovonics 222. This unit is totally analog (yea!!!) and will give you negative peak limiting at 100% and pass up to 130% positive peaks. Please note that there is no phase selective inversion (phase rotation) in this unit so for the most part the phase needs to be correct going in the input. This unit also has a selectable NRSC pre emphasis filter for the high frequencies along with a few more options. You can usually pick this unit up used from 250 to 400 bucks.

We think this is about the best economical approach for low level ANALOG modulation. You hit upon the crux of the matter Bill Paul when you stated that you were talking about the analog modulation scheme in the transmitter. Most of the low level AM transmitters out there are using DSP for the AM Modulation. In our experience it seems the DSP algorithms are forcing the asymmetry to be balanced, this includes the DSP in the TS-950sdx.

This is the reason that we here at the Voodoo-LABS are designing a totally analog 455kc AM Transmitter that will be retrofitted to the TS-950SDX. After prototyping many, MANY different AM Modulation schemes which include, balanced modulator IC's (yuk spit!!), Trans-conductance Amplifier IC's (better but still yuk), Differential Amplifier Modulation (better yet), Dual Gate MosFet Modulator (even better yet), a pair of Jfet's in Push-Pull - Series Drain modulated by a class AB amplifier (BEST so far).

What we have noticed is this, we were very hung up with trying to get the cleanest signal output as possible in both THD and IMD. It seemed the cleaner we got the signal, the more sterile the sonics of the recovered audio was. The Balanced Modulator IC's had the cleanest signals but were the worst sounding as far as the sonics were concerned. It seemed that when the transmitted signals THD was around 3% and the IMD was around -30dB, that this provided the richest and more pleasing recovered audio. It had more dimension and life to it.

Since we made this discovery, our next evolution is to build a low level tube transmitter prototype. This will be a Push-Pull pair of 6C4 Triodes in deep Class C being Plate Modulated by Class AB Audio. We already had a Modulation Transformer custom made for us by Edcore Transformers. The Mod tranny was made on a Split core and has a flat frequency response of 20cps to 18kc with 100ma of dc on the secondary. We will also be using a 40H Modulation Reactor rated at 100ma. This analog signal will be filtered and padded to the proper input level and substitute the DSP AM Modulated signal going in the rear of the TS-950sdx.

We are accumulating all the parts at the moment and should start building the prototype later this year.
It should be interesting to say the least!!!

Take Care Y'all,

BillPaul · Post by **BillPaul** » Fri May 04, 2012 10:55 am

Using my Kenwood SM-230 station monitor, I have in fact observed natural speech asymmetry with the TS-850 using analog modulation which is absent when using the DSP-100 or the digital modulation in the TS-950SDX. I'm using a Behringer DEQ2496 for equalization/compression which is fully digital, but it still seems to be preserving the asymmetry. The TS-850 modified for AM also has much better frequency response in analog mode, up to 10Khz compared to 6Khz in DSP mode.

I like to run the audio to the TS-850 through the ACC2 port, the same as I do with the TS-950SDX. (This lets me use the same isolation/matching transformer and cabling setup with both rigs.) I immediately noticed a phase inversion the first time I used it this way. Curiously, this same inversion is not present when I turn on the DSP-100 unit and it doesn't seem to be present when using a regular MC-60A desk mic via the front panel mic plug either. Luckily, the MIC2200 mic preamp has a phase reverse feature to compensate for this. (When I realized that the DEQ2496 could also provide phantom power for my mic, I thought maybe I had wasted my money on the MIC2200, since that was the main reason I got it. Turned out to be worth it though.)

I looked at the Inovonics 222. Although the BW Broadcasting DSPXmini-AM is digital and more expensive, it has a couple of advantages in my view. The asymmetrical clipping feature has greater range (150% positive/100% negative peak ratio), it has factory and user presets, and it has XLR jacks for its analog inputs and outputs where the Inovonics unit looks like it just has a bare terminal strip on the back. Okay, so I prefer plug & play -- sue me.

I haven't made up my mind yet, though I happen to have the money in my piggy bank for the BW Broadcasting unit, so... we'll see.

Also, I'm glad to hear you're making progress on the outboard AM modulator. I was wondering how that was going. I guess I shouldn't be too surprised you decided to go the tube route. Looking forward to hearing the results.

-Bill, N1GPT