A3E Modulation

Since the confirmation that the radio must be A3E, I've had to rethink my design, largely from scratch. I investigated PWM-based PSU modulation for the final amplifier stage, Class-E approaches and similar modern "plate modulation" techniques. While very attractive, I've decided to do something different. I know several other members are attempting these approaches, and I dislike working with "black box" chips where possible. While the PWM chips now available are very neat, and fairly easy to replace with discrete circuitry (i.e an Op-Amp triangle generator and a comparator, plus some interlock logic to prevent rail-to-rail shorts if you stuff it up), I'm looking at the low-level modulation technique.

Your basic single-balanced modulator is an easy way to implement amplitude modulation with only three transistors. So I've taken this approach. I have a fairly classical differential pair, with a third device as a controlled current source in their linked emitters. Ignoring biasing, the circuit looks like this:

AM Modulator Circuit Idea

You can think of this circuit as half a Gilbert cell. Unlike a Gilbert cell, the carrier does not cancel, so the plus and minus products products and the original carrier are available at the output. This is IMO simpler than using a normal diode double balanced modulator and then doing carrier re-insertion, which has some complications in phasing and amplitude matching.

Doing the AM at a few dBm is also much easier on the AF side of things, it means you don't need a very powerful AF amplifier and a large/heavy modulation transformer. Of course, it also means the entire transmit chain must be linear, and the inefficiencies associated with not using Class-C the whole way will be apparent. This general plan of attack also stems from my plans to make the resulting radio multi-mode. Building an A3E emitter seems, well, archaic. Peter explained it was to make the project accessible - I can't argue with that I guess, but IMO unmodulated carrier is just wasted power unless you are a real audiophile that dislikes product detection. On 160 metres I wouldn't argue at all, but on 80 metres DSB (or better LSB) seems more natural.

Breadboard AM Modulator

My trial circuit on a solderless breadboard works fairly well.

AM Modulation Waveforms

The modulation linearity is not fantastic once about 50% depth is exceeded. I figure I can fiddle with the biasing and injection levels to try and improve this, and perhaps add a bit of feedback to flatten the response. I can always consider feedback around the entire amplifier chain, to neutralize non-linearity anywhere in the transmit chain. That said, the audio quality is quite acceptable for voice as-is, and as an initial modulator this circuit should be fine.

Reasonable Linearity at Moderate Depth
Poor Linearity at Deep Modulation

The square-law demodulator in most AM receivers can't handle such deep modulation with better linearity anyway, but of course it is best to produce a clean signal, long before any receiver has a go at mangling it.

Leave a comment on this article.


title type size
AM Modulator Circuit Idea Source application/postscript 11.206 kbytes

Parent article: "2007 80m Homebrew Challenge".