Monday, 13 October 2014

Simpler Single-Signal Direct Conversion Receivers



Greetings!
I love the sonic impact of a well-designed direct conversion receiver. My best, a version of Rick Campbell’s R1 receiver with a Level 17 diode ring mixer still thrills me to bits. How do I describe the sound of a well-designed DC receiver?  Well, subjectively of course:  pure, raw, sibilant and dynamic come to mind. For contrast, after listening to 1 of my DC receivers, I’ll fire up 1 of those quotidian home brew superheterodyne receivers [NE612 — Cohn crystal filter — LM386 etc.] To my ears at least, the latter sounds cadaverous.

The story didn’t end with the R1, for Rick crunched the math, made careful experiments and raised the bar with his follow-on; the R2 which contained an image-reject mixer.  His work ushered high performance, single-signal DC receivers onto the modern day radio workbench with wide-reaching impact.

For image-reject receivers, instead of 1 baseband signal from a DC receiver, we get 2 baseband signals: Q or quadrature-phase that lies 90° out of phase with the I or in-phase signal (phase 0).

Modern DSP experimenters process the I and Q basebands with software, while the more primitive radio experimenter applies analog signal processing — usually some audio phase-shift networks and a combiner to reject 1 sideband before some more AF filtering/amplifying to drive a speaker or phones. Your brain also provides some neurolog signal processing.

The math and design of image suppression receivers gets difficult for some, but for those of you who like math — feast away since web articles abound!  I recommend you read Chapter 9 of EMRFD (written by Rick) — a chapter that offers telling insight for both the electrical/computer engineer and weekend QRP homebuilder alike.


Receivers with Image Rejection of 20-30 dB

Image rejection gets trashed when phase error or gain mismatch occurs in the I and Q channels and reducing these errors takes work. Above 20-30 dB rejection, design and often bench practices become more critical. For example, the audio phase shift networks are typically all-pass circuits with 1% or less tolerance parts built around 5532 or quieter op-amps. 

Further, to extract maximum sideband rejection, tweaks like a phase trimmer and amplitude balance control become vital.

I plan to experiment with image–reject DC receivers with 20 or so dB sideband rejection. This allows us to apply less stringent 2nd order transistor, op-amp, or pot core-based LC phase shift networks, however, receiver design remains a challenge.

Although an audio spectrum analyzer will help, the brain + ears = the main tools we'll use to assess sideband suppression.
Another gimme = no worrying about crystal filters, or an image with a zero IF!

I plan 2 receiver experiments: 1. Funster   2. TMP [Towards Minimal Parts]. I'll just briefly introduce each and then, with any luck, we'll build them on this blog. 

 1. Funster


Above — The Funster Block Diagram. I don't show the DC circuitry and built-in and secondary LO circuits. The AF stage designs include an active tone control. I hope it all works! My former QRPHB readers resounded this clearly in their emails: "we want to see your failures and successes".


Above — The Funster front end. All passive components. Mixers = MCL diode rings.


Above — The pot cores I ordered from Amidon Inc. Both are 77 material which should yield some Q and make nearly any coil I require. See http://www.amidoncorp.com/pot-core-sets/


Above — A pot core exploded for viewing. Although Rick, KK7B used a variety of different cores for the projects in EMRFD Chapter 9, he just wound his coils with whatever was at hand. You may certainly substitute any core in a mix that provides the target inductance and will accommodate enough turns of your preferred wire size. I chose the PC2213-77 for making my pot core AF Wilkinson combiners. I keep 2 different inductance meters on my bench and wouldn't wind a pot core without an L meter.

I asked Rick whether he twisted the bifilar windings on the bobbin — he didn't, nor did I.

2. TMP [Towards Minimal Parts]


Above — A piece of the TMP receiver concept schematic. The complete receiver includes a VFO, a home brew I/Q diode detector and a feedback AF preamp with LM386 AF final set in a lower gain mode via some emitter degeneration. It features amplitude and phase tweaks.

I'll establish the value for (Q2) RC and RE on the bench. RC poses a challenge because the all-pass values C and R came right out of the B & W Model 350 Type 2Q4 SSB audio phase shift network. Think Hallicrafters HT-37, or Johnson Pacemaker glowing in your shack!  High plate impedance and all.
In EMRFD Figure 9.45, Rick decade scaled these R C values to better support discrete component FET + BJT phase shift networks, although they may work fine as shown with care. I have most of these R and C parts in my collection and thus will give it a go. 


The B&W 2Q4 2nd order network and scaled versions in EMRFD may give nearly 40 dB opposite sideband suppression from 300 to 3000 Hz when done with care and vigor. 

I'll go on the bench and get working on the Funster receiver. Thanks for reading and I hope you enjoy a great Fall + Winter of bench experiments. I for 1, feel stoked!




2 comments:

  1. I don't have EMRFD, so I may be repeating information that is contained therein, but: at the Yahoo Group "r2pro" there is a folder by KK7B titled, appropriately, "KK7B". In that folder there are four files titled "IQcrystalSet{1-4}.pdf" that describe an all-discrete, popcorn-style phasing receiver. You need to be a member of the group to access the files, but the URL is: https://groups.yahoo.com/neo/groups/r2pro/files/KK7B/ .

    ReplyDelete
  2. Thanks qrp-gaijin, I have the file and recommend it and the Yahoo group you mentioned to others. It serves as a basis for #2 [TMP] experiments along with EMRFD.

    ReplyDelete