To: laskorg@gpfn.sk.ca
From: Larry Filby, K1LPS
Subject: Drake 2880 MMDS Downconverter
Cc: t@tir.com, n8vna@gte.net, rein0zn@ix.netcom.com, k0vty@juno.com,
ezimmerm@erols.com, pogletre@mail.coin.missouri.edu,
kd6odu@amsat.org
by K1LPS, Larry Filby
This is a sort of summary of postings on my work with the the Drake 2880
downconverter. An extension of the article in the February 1998 QST, which
was aimed primarily at a quick and easy way to get reception of Mode S (2400
Mhz) signals from satellites, including the impending Phase 3D.
My first objective was to see if the unit would get down to 2304, and achieve
performance that weak signal operators are used to. (Ie; 1 to 1.5 dB or
less.) Those using the downconverter for satellite work may not need this
level of performance, although you can probably achieve 3-4 dB with the
simple change of crystal and the use of a good preamp. (More on preamps later.)
To put the receive performance in perspective, let's throw out a few
numbers, and you decide the level of "tinkering" you want to get into.
Remember, we're talking 2304 here, and not the 2400 Mhz of the QST
conversion, although it's all relative,
and applies to 2400 as well.
Considering the specs of an "average" MMDS external LNA (Just to keep
things simple.), which are rated at around 1.5 dB N.F. (Most are better.)
and 15 dB of gain:
If you simply change the crystal... and do the minimum post amp coil-cap
removal mods... the basic N.F. of the 2880 will be around 15 dB with a 2M
I.F. at 2304.
The LNA will bring this down to 3.77 dB N.F., overall.
With the mods I made to the front end, an easy 7 dB N.F. can be achieved for
the basic unit. Adding the same preamp to this system gives an overall N.F.
of 1.87.
I've got one 2880 that makes 6.0 dB... and the LNA is probably running close
to 1.0
dB... and my overall measurement runs right around 1.4 dB. So... you decide
how far you want to go with your system. My approach does require the
appropriate test equipment and experience, for evaluation.
My interest in the converter was for N.A. weak signal work down at 2304, and
got extended into a way to turn the converter into a transverter. First of
all... there was the question of: Would the RF stages hit 2304 at all...
and would the VCO/PLL work at the frequency required for 2M I.F.? (2160
Mhz.) Short answer: Yes to both.
The VCO/PLL has been operated as low as 1967 Mhz, although the chip cap that
sets the frequency of the VCO must be increased to 8-10 pf for operation
below 2000 Mhz. We know the system works at 2278 Mhz... and probably will
work higher for
other L.O.'s, but the VCO tuning chip cap would probably have to be made
smaller.
In addition to the QST article, some basic information on conversion is
available at the following web site:
(Original value unknown, but obviously less than 8-10 pf.)
The site has excellent pictures and info on other I.F.'s, but still oriented toward 2400.
Noting the N.F. specs given in Table 1, page 44 of the QST article, and that apparently no mods were made to the first RF stage, the question of performance at 2304 was still unanswered.
I made the mods of removing the red coils and associated chip caps, and I also removed the choke that supplies power to the unit via the "F" connector, and feed voltage directly to the + side of the polarity/transient protection diode. A small hole was drilled just to the right of the "F" connector to pass the supply lead. (Be VERY careful drilling this hole, not to go in too far and destroy components on the board.) (Been there... done that!)
The two red coils and chip caps removed, form a broadband tuned network, designed for post amp frequencies of 222 to 408, the original I.F. in MMDS service. Removing them opens up the post amp response so that you can use I.F.'s as low as 144 Mhz. If you were to use a 222 I.F., such modification may not be necessary... but you have to do the numbers to ensure that the VCO/PLL, etc. will work for your application. Opening up this post amp response has it's down side, but we'll not discuss that at this point. One thing for certain... if you don't do the coil-cap removals, the unit will not function well at 144 Mhz I.F.
The unit was tested with the new 8.4375 crystal and lock at 2160 was obtained without problems.
Next, N.F. performance was measured at 2304/144 Mhz I.F. The N.F. measured very high. (Around 15 dB.) To make a long story short... it was found that three things improved the N.F. at 2304: (1) Extending the length of the first element of the interdigital filter about 2.5 mm. (2) Adding a 2.5 mm stub to the output terminal of the RF amp device, in the only space where it will fit. (3) Scratching out the tiny little shorting strap across the end of the hairpin loop that is on the input side of the RF amp device. Item (1) has the most dramatic effect on the N.F. performance. It's relatively easy to do. The others give .5 to 1 dB improvement each... and considering that even an 8-10 dB basic N.F. will be covered pretty well by a good preamp... you may want to dispense with the other two. (This stuff is SMALL... and may intimidate some of you.) These mods were done with the aid of a noise figure meter, and may be difficult to see the improvement by any other method. However, I have done four of them to date and all were consistent. And... such nitpicking may only appeal to the ".1 dB N.F. weak signal cult" - hi.
Note that replacing the inner cover usually causes the N.F. to increase by
about 1 dB.
I have run the units without the inner cover, but that can have it's plusses
and minusses too.
Continuing with the receive portion... the next step is to add a preamp.
There are numerous publushed H.B. designs, kits and built from Down East
Microwave, etc.
A very simple, effective and inexpensive solution is to simply use external
LNA units available to the MMDS industry. These are made to mate right to
the downconverters, and achieve gains of around 14-15 dB... and noise
figures between 1 and 1.5 dB. Considering that we've been able to buy these
at $25 to $34... it's hard to imagine why you'd want to do anything else.
We're talking "plug 'n' play"... no building, testing or packaging. Two
that have been used are the California Amplifier
#30947, or the Conifer PA1033, and there may well be others. These units
are spec'd from 2100-2700 Mhz, so there's no question about them working in
the ham band. They bolt right up to the Drake cownconverter with "N"
connectors in/out, and operate from +12V just fine.
The second part of the project was to tap out some 2160 L.O. out of the downconverter for use in a transmit section. The tapout part has been successfully completed. It's a VERY touchy job that is a challenge just from the space available aspect. In short... a two-hole flange SMA connector was added to the case to come out between the two 3-terminal regulators. Again drilling and tapping for this purpose must be done VERY carefully. Subminiature .085 semi-rigid Teflon coax was used to get over to the area of the L.O. PC trace that routes it into the converter section. (In the QST picture... the trace that is dead center in the picture of the overall converter, crossing from the VCO/PLL section, to the RF amp/mixer section.) I've done it with a probe type pickup, and a hairpin loop type. The only caution here is to stay away from the area close to the VCO device itself... and ALWAYS check for L.O. output level, AND VCO/PLL lock with the outer cover in place. Too heavy loading of the pickup loop can cause the VCO/PLL to lose lock, and it might not show up until you put the cover on. I've been achieving -10 dBm of output out the added SMA. No connection is physically made to the L.O. output, and no degradation has been noted in receiver performance using this tapout. This must be boosted to transmit mixer level. MMIC's are being used here, and that phase is still in progress as of this writing.
Miscellaneous comments:
In my experience, the post amp section simply isn't made to provide optimum performance for a 2M I.F. I've concentrated on 2M I.F. because all of my bands 903 and above use 2M I.F... and 2M rigs are by far the most common around. With the coil-cap mods... performance in terms of basic sensitivity is totally adequate. It might be even better using a 222 or 432 I.F., but there are other considerations, such as... can you make the VCO/PLL work at certain frequencies necessary for this or that I.F.? My complaint about the basic post amp mod is the very high level of noise and signal that bombards my I.F. receiver. (A Yeasu FT290R Mk II.) The S meter reads 3/4's scale with no signal. Adding a 20 dB pad in the 2M I.F. line tames this and only adds marginally to the overall N.F. of the system when using a 1.0 dB N.F. preamp. However... this is not the proper solution, and the result of so much broadband gain ahead of the 2M front end may result in problems when operating in high RF level environments on mountaintops where I normally operate. I have yet to hit on the answer, but I'm still working on it. In the meantime... I get good numbers, good tests with weak signal marker sources, and am probably making a mountain out of a molehill to push further.
Things that will get you into trouble:
Be very careful when removing the boards from the case. Be especially careful when removing the input/output connectors. Also be very careful in re-assembling. The "N" connector is easiest, but be careful not to break or disturb the input blocking ceramic capacitor. Make sure that the mounting screws are tight on the "N' connector before re-soldering. The traces on input and output are easily damaged, To remove the "F" connector... wick away as much solder as possible... loosen the "F" connector slightly with it's nut... and then withdraw it carefully. You probably won't be able to wick out all the solder and need to apply judicious heat as you withdraw the connector to clear the board. Bulkhead type BNC connectors will screw into the same hole, although on some types, the threaded section is too long. The center pin of BNC's is too large to clear the board... so you'll have to jumper from the cut off center pin to the board. (No... I have nothing against "F" connectors... but have changed all of mine to BNC.)
Material sources:
Comnet Wireless, Inc. Aurora, CO. This company has been a source for the wireless cable pieces used in this project. As of 3/16/98, their phones have been disconnected and their web site gone. It is presumed they have ceased to do business.
SCTV-Skycable, Valley Village, CA. Tel: (818) 973-3118 Bob, WB5OHH.
E-mail:
If you find other sources at reasonable prices, by all means, let us know.
There ought to be plenty of this stuff out there somewhere. If you know of
a ham friend in the wireless cable business... maybe he can help obtain
LNA's if nothing else.
Last resort: From Drake directly, for the 2880, at the higher $68 price.
As always, a first rate company with first rate products.
73 DE Larry (K1LPS/Vermont)
FN34wl, etc...