ETA: Analog Receiver

ETA Analog Receiver

Overview: Information about the ETA analog receiver (ARX). The goal is to replace the previous ugly, expensive connectorized design, and use the experience to think about how to do LWA receivers.

Maintainer: Steve Ellingson (Virginia Tech)
This page is http://www.ece.vt.edu/swe/eta/ARX/
Updates:
23 Oct 2006. Added pictures of rack-mounted ARX and measured response.
21 Sep 2006. Added pictures of ARX in enclosure.
22 Jun 2006. Started page

Design Documents

ARX Version 3 design files in Eagle.
ARX Version 2 design files in Eagle and Gerber (.bin). Note: Board needs to be drilled in two places to fix a layout error; see below.
ARX Version 1 design files: This was a 2 layer design which I made just to make sure I had the custom part definitions correct.

(Click image for high-res.) The first 16 version 3 ARX modules, mounted in rack. Front view. October 23, 2006.

(Click image for high-res.) The first 16 version 3 ARX modules, mounted in rack. Rear view. October 23, 2006.

(Click image for high-res.) Frequency response of "production" version 3 ARX, complete in enclosure. NOTE: This illustration shows gain minus 50 dB (i.e., add 50 dB to get actual response.) October 23, 2006.

(Click image for high-res.) ARX version 3, complete in enclosure. The enclosure is a Vector model, which is inexpensive and fits nicely in groups of 8 in Vector's CMA13 card rack. October 9, 2006.

(Click image for high-res.) ARX version 3, with back panel and connectors attached. The N-connector is RF in, SMA is RF out, and the DB9 conveys power and 3-bit attenuator control. October 9, 2006.

(Click image for high-res.) ARX version 2: It's a 4-layer board, which is 4 inches by 4 inches (any numerologists care to comment?), designed to fit in a Vector card cage-mountable enclosure. The big blue thing is a coaxial relay which physically separates the receiver from the cable when the power is turned off -- the idea is to reduce the likelihood of lightning- or static-related damage when the system is not being used. Actually, it's the $100K+ in digital electronics behind the receiver I'm really worried about. The rectangles appearing in the layout are fittings for FerriShield Model PS100EMC24 shielding enclosures (sometimes referred to as "cans"). FWIW, I noticed absolutely no difference in performance with cans on vs. cans off, so at this moment I'm thinking to leave them off. That might change after trying this out in a box in field conditions. Also visible are the two holes I had to drill to fix a bone-headed layout error. I left off the DB9 connector but you can see where it goes. Also I have MMCX connectors on the input and output just because I currently have a ton of these.

Measured Specs:
Max Gain: 52.3 dB
1 dB Bandwidth: 30.7 - 43.7 MHz
3 dB Bandwidth: 27.3 - 47.3 MHz
40 dB Bandwidth: 17.7 - 65.3 MHz
Attenuation in 4 dB steps, 3 bits, to 24 dB total
Power: 330 mA @ 12VDC (accepts 12-15 VDC)

I didn't measure noise figure or IP3. The GNI analysis, combined with the active balun and feedline, give a cascade noise temperature of 256 K and IIP3 of -41 dBm (OIP3 = +38 dBm). Appears to be unconditionally stable, even without the cans; however, I did not try it in a box. The bandwidth is determined entirely by the 2 custom BPFs. The bandwidth of everything else exceeds LWA preliminary specs by a large margin. Might be possible to develop an LWA version just by changing the component values, keeping the same BPF topology.
Cost: As shown: $260 each, quantity 2. The PCBs will be the big quantity price break; in quantity 2 I get the PCBs for $162 each; could be had for much less if I ordered larger quantities. The cans cost $22 each, so the total "as shown" cost goes from $260 to $304 if cans are used. The coax relay is $28; a bit pricey but I really like the idea of being able to electronically "stow" the system. The attenuator is $47; that can definately be done more cheaply for LWA.

(Click image for high-res.) The schematic.

(Click image for high-res.) Top layer of PCB.

(Click image for high-res.) Layer 2 of PCB (ground plane).

(Click image for high-res.) Layer 3 of PCB (power plane; a combination of 5V, 9V, and 12V).

(Click image for high-res.) Layer 4 of PCB (more traces). Mostly used to dive under the can boundaries. No components on the back side, but a few through-hole components.

	(Click image for high-res.) The first 16 version 3 ARX modules, mounted in rack. Front view. October 23, 2006.
	(Click image for high-res.) The first 16 version 3 ARX modules, mounted in rack. Rear view. October 23, 2006.
	(Click image for high-res.) Frequency response of "production" version 3 ARX, complete in enclosure. NOTE: This illustration shows gain minus 50 dB (i.e., add 50 dB to get actual response.) October 23, 2006.
	(Click image for high-res.) ARX version 3, complete in enclosure. The enclosure is a Vector model, which is inexpensive and fits nicely in groups of 8 in Vector's CMA13 card rack. October 9, 2006.
	(Click image for high-res.) ARX version 3, with back panel and connectors attached. The N-connector is RF in, SMA is RF out, and the DB9 conveys power and 3-bit attenuator control. October 9, 2006.
	(Click image for high-res.) ARX version 2: It's a 4-layer board, which is 4 inches by 4 inches (any numerologists care to comment?), designed to fit in a Vector card cage-mountable enclosure. The big blue thing is a coaxial relay which physically separates the receiver from the cable when the power is turned off -- the idea is to reduce the likelihood of lightning- or static-related damage when the system is not being used. Actually, it's the $100K+ in digital electronics behind the receiver I'm really worried about. The rectangles appearing in the layout are fittings for FerriShield Model PS100EMC24 shielding enclosures (sometimes referred to as "cans"). FWIW, I noticed absolutely no difference in performance with cans on vs. cans off, so at this moment I'm thinking to leave them off. That might change after trying this out in a box in field conditions. Also visible are the two holes I had to drill to fix a bone-headed layout error. I left off the DB9 connector but you can see where it goes. Also I have MMCX connectors on the input and output just because I currently have a ton of these.
Measured Specs: Max Gain: 52.3 dB 1 dB Bandwidth: 30.7 - 43.7 MHz 3 dB Bandwidth: 27.3 - 47.3 MHz 40 dB Bandwidth: 17.7 - 65.3 MHz Attenuation in 4 dB steps, 3 bits, to 24 dB total Power: 330 mA @ 12VDC (accepts 12-15 VDC)	I didn't measure noise figure or IP3. The GNI analysis, combined with the active balun and feedline, give a cascade noise temperature of 256 K and IIP3 of -41 dBm (OIP3 = +38 dBm). Appears to be unconditionally stable, even without the cans; however, I did not try it in a box. The bandwidth is determined entirely by the 2 custom BPFs. The bandwidth of everything else exceeds LWA preliminary specs by a large margin. Might be possible to develop an LWA version just by changing the component values, keeping the same BPF topology. Cost: As shown: $260 each, quantity 2. The PCBs will be the big quantity price break; in quantity 2 I get the PCBs for $162 each; could be had for much less if I ordered larger quantities. The cans cost $22 each, so the total "as shown" cost goes from $260 to $304 if cans are used. The coax relay is $28; a bit pricey but I really like the idea of being able to electronically "stow" the system. The attenuator is $47; that can definately be done more cheaply for LWA.
	(Click image for high-res.) The schematic.
	(Click image for high-res.) Top layer of PCB.
	(Click image for high-res.) Layer 2 of PCB (ground plane).
	(Click image for high-res.) Layer 3 of PCB (power plane; a combination of 5V, 9V, and 12V).
	(Click image for high-res.) Layer 4 of PCB (more traces). Mostly used to dive under the can boundaries. No components on the back side, but a few through-hole components.