Corsair Hiss Discussion

from the Ten-Tec reflector, May 2005; comments by N4LQ and N9DG (italics), postscript by N1EU

The Triton through the Omni series were all quiet receivers. Narrow xtal
filters were added to the Triton to make it an Omni. On the Omni C, a small
transistor amplifier kicks in when using the 500hz cw filter. This adds just
a touch extra noise. Otherwise, the Triton and Omni are neck and neck for
noise. Being single conversion rigs, they had few mixer stages thus less
opportunity for noise.

The 580 Delta is almost in this same camp though a dual
conversion design. I found its main limitation (other than
poor AGC) is that the audio chain is rather noisy. Ten Tec
radios of this vintage I believe all used the ubiquitous
LM380/383 series ICs for their audio amplifiers. Theses
aren't very good audio amps if you look their specs sometime.
I'd bet that the RX audio of all TT radios of this vintage
could be improved upon with an audio PA update.

Then came the Corsair. Passband tuning was added. To achieve this feat meant
another mixer and another IF stage. The extra IF at 6.3mhz eliminated the
17m spur but added noise and lots of it in the form of hiss or white noise.
In the original Corsair you just had to live with the noise.
Then came the Corsair II. Still had the extra noise but they added a tone
control. Turning this control to minimum treble cut down on the
hiss…..some. If you turn down the "RF Gain" on a Corsair, virtually all the hiss
disappears leading me to point blame at the IF chain.

Like the models that precede it I've also concluded that it
is the audio chain that accounts for 80% of this hiss in a
Corsair series radio. The Corsair has a number of op amps
that make up the active audio filtering, I suspect that the
cumulative noise and distortion of those parts is not
insignificant. They too use a somewhat noisy LM38x series
audio output IC. My experiments last fall using the Corsair
II as a 9MHz IF feeding a 9MHz detector (I/Q baseband audio)
feeding DSP IF processing chain (PC sound card with SDRadio
software) seemed to confirm this. I did not however try
tapping the 9MHz after the PBT tuning but I suspect that
there would be minimal differences than when I tapped the
9MHz signal immediately after the first 8-pole 9Mhz filter.

There are two sources of hiss, one is indeed the
RF/IF stages, and the other is clearly the audio chain.

As an experiment turn the RF gain all the way down (with
50-ohm load or no antenna) and then the audio all the way up
there's plenty of audio hiss. Can be further demonstrated by
lifting connector 45 (RX IF in) to the IF/AF board and crank
the audio full up again, about the same level of noise -
though the RF gain setting does have some effect. Also if you
lift connector 76 (BFO in) you get a similar result, this
time the RF gain has no effect.

This set me to wondering if I could actually characterize the
amount of noise contributed by the PBT board in the Corsair
II a bit better. After studying the schematics some I saw
that it is quite easy to do. At least for a quick test. All
that I needed to do to bypass the PBT board was to take plug
75 from the PBT board and plug it into the position 73 of the
XTAL filter board. That simple change bypassed the PBT board
completely. As it turns out the PBT board adds a good share
of the overall RX noise, - more than I had originally

Needless to say this jumper cable reconfiguration removes 8
poles of IF filtering at the 6.3 MHz IF and PBT goes bye bye
too. And there is only the 2.4 kHz bandwidth IF filter at
9MHz left, essentially makes the Corsair II into an Omni
A/D/C. Didn't study it any further to see what the
ramifications are to the TX side. So DO NOT necessarily
consider this to be a mod that you want to make.

Then came the Paragon…Same PBT, same hiss but now they added a tunable
audio filter which helps hide the hiss better than a tone control.
Then came the Omni V with the wonderful PBT but still hissy.
They came the Omni VI. Still hissy but now we have DSP to help hide the
hiss. Works pretty slick as far as hiss hiding goes.

After some playing with a recently acquired Omni VI (price
was too right, couldn't pass it up) I found a new variation
of audio chain limitations. From what I can tell the A to D
is only 14 bits, I've never been convinced that 16 bits is
enough for high fidelity audio, so needless to say I have
reservations about 14. The Omni VI does however use a more
modern audio PA IC; it is noticeably cleaner than the
LM38x's. It is a TDA8611A, if you look up its specs it is
considerably better than the LM38x. The Pegasus (Jupiter and
Orion too I presume) use the TDA7056B. I always found the
Pegasus to have excellent audio. The Peg's (Jupiter too) main
shortcomings are the high-ish 1st LO phase noise and high
frequency first IF, - therefore suffers from the typical wide
roofing filter dynamic range limitation. Can't say I've
noticed digital noise artifacts for the way I use the Pegasus

So this is why one can listen to the Triton or older Omni series and come
away smiling. This is why people have a hard time giving them up.
I played with the Flex-Radio SDR for a few weeks. I got the same buzz. It
was as if my ears were directly connected to the ether! Crazy thing was
useless as a transmitter and flaky as a june bug though. It's SINGLE
CONVERSION! Man you talk about quiet! Some day I hope others follow this
lead and perfect it.

Bottom line here is; too many mixers spoil the radio.

N1EU comments - I thought the above discussion was well worth reproducing here. From my experience:
1. I don't find the LM380/383 audio amp as bad as alleged - its offenses are two orders of magnitude less than the op amps preceding it in the audio chain
2. I've done the experiment of bypassing the 6.3MHz i.f. entirely and I found that the 6.3MHz i.f. was not contributing THAT much noise
3. I know two of us have replaced the Corsair II op-amps with a big reduction in hiss. See corsair-opamp
4. I've written extensively over on about how the Omni 6 uses dsp to filter out hiss. I've implemented my own 3KHz LPF for ssb rx and 1KHz LPF for cw rx with excellent results
5. You can engage the Corsair Notch Filter at "8" and decrease 3-6KHz hiss by up to 10dB without affecting the cw or ssb passband audio