[TenTec] MC1496 front end mixer in the Ten-Tec Argosy (first version; first revision)

[Kevin Purcell]

ARRL reviewed of the first version Ten-Tec Argosy serial number 525  
00026 in October 1982.

<http://www2.arrl.org/members-only/prodrev/pdf/pr8210.pdf>

This is the original version of the Argosy that uses a 1496 (after a  
RF amp)

Measurements made with the the 500Hz 8 pole 2:1 shape factor crystal  
filter setting the IF bandwidth (but with a 4 pole 2.4kHz crystal  
filter after the mixer though that shouldn't make a difference for  
these measurements)

80m
MDS -133dBm
=> NF = 14dB
Blocking DR 99dB
Two tone SFDR 64dB
=> IIP3 = -37dB
=> R = -51

20m
MDS -133dBm
=> NF = 14dB
Blocking DR 98dB
Two tone SFDR 64dB
=> IIP3 = -37dB
=> R = -51

As you can see not very good numbers at all.

R is the W7ZOI Receiver factor, IIP3 - NF. This is a really poor number.

Ten-Tec Argosy later changed the RX mixer to a discrete diode ring  
mixer (and later a discrete ring with diodes integrate on a chip so  
they are all matched). See this link for more details

<http://lists.contesting.com/archives//html/TenTec/2007-04/ 
msg00326.html>

So why did they think the original design would be OK. I think they  
read the 1496 app note.

> Another advantage of the MC1496 product detector is its
> high sensitivity.

> For a 20 dB (S + N)/N ratio, demodulated audio output
> signal, a 9 MHz SSB input signal power of –101 dBm is
> required.

That's an MDS of -121dBm. Assuming a bandwidth of 3000Hz then gives a  
NF of 18dB for the 1496. Which is not certainly great but not bad  
either below 30MHz. For example, this would only be above the  
atmospheric noise figure in a "ITU-R P.372-8 very quiet rural" on 10m  
(assuming a loss of 3dB in bandpass filters). But there's not much  
margin there.

<http://www.vk1od.net/fsm/FSAmbientNoise.htm>

Moto then go on to say

> As a result, when operated with an SSB receiver
> with a 50 ohm input impedance, a 0.5 microvolt RF input
> signal would require only 12 dB overall power gain from
> antenna input terminals to the MC1496 product detector.

Unfortunately Ten-tec took them at their word. Because they put the  
RF amp in front of the 1496 mixer. It pushes the noise figure down  
(but not by much). And it reduces the dynamic range by a lot. But few  
people buying the product understood dynamic range in those days.

They also stuck to the standard 1496 design with 1mA per side of the  
transistor tree (Gilbert Cell) mixer. They could have pushed this up  
to 10mA or more (though exceeding 10mA would be beyond the maximum)  
as G4COL later did and reported in Sprat 95. He found you can push  
the IIP3 of a 1496 to +20dBm at 13mA. A big improvement (the IIP3  
depending on the cube of the current in a Gilbert Cell).

This reveals a potential modification for the 1496 Argosy's: change  
the resistor that sets the 1496 current to bump the current. The  
caveat is to make sure the power dissipation does not exceed the chip  
maximum. It will probably change the gain and input impedance of the  
mixer too. The other potential change might be to reduce the 1496  
mixer gain to improve IIP3 at the risk of worsening the noise factor  
by changing another resistor (the pin 2 to 3 "gain adjust" resistor).

Of course it would be best to make measurements of the IIP3, MDS, NF  
whilst tweaking these resistors.

Get the 1496 datasheet and application note (yes, it's still being  
made but not by Moto).

<http://www.onsemi.com/pub/Collateral/MC1496-D.PDF>
<http://www.onsemi.com/pub/Collateral/AN531-D.PDF>
--
Kevin Purcell
kevinpurcell at pobox.com