KA7OEI 10 GHz pages

Modification of the Ramsey FR1 kit for use as a 30 MHz WFM IF for Gunn transceiver use.
Abstract:
Figure 1 - The fully-modified (except for the addition of the 2nd IF filter) Ramsey FR1 FM receiver.
Click on the image for a larger version.
Ramsey FR1 circuit board with described modifications

One of the more popular modes of operation on the 10 and 24 GHz microwave bands is Wideband FM using Gunn transceivers.  Because frequency modulation is very easy to accomplish using Gunn oscillators, this mode is a natural choice with these devices - and using wideband FM with a relatively wide 300 kHz bandwidth (the same standard as is used on FM Broadcast Bands) makes it easier to find and track signals when the frequency stability of one or both ends of the 10/24 GHz link may be suspect.

Traditionally, a 30 MHz Intermediate Frequency (IF) is utilized, implying a difference 30 MHz between the Gunn oscillator frequencies of the two stations in QSO.  Why is 30 MHz used instead of a seemingly more-obvious choice of 88 MHz or some other frequency in or near the FM Broadcast band where receivers are extremely common?  While it may seem natural to want to use readily-available receivers designed for wideband FM, it's a simple fact that in most areas it is difficult to fund a span of 2-6 MHz or so of the FM broadcast band in which one may track a drifting signal that does not already have a high-powered FM broadcast station in it:  This is especially true for those operators atop mountains that are likely to be line-of-sight with a high-powered broadcast station that may be some distance away - or right next door!  It should be noted that no matter how well one attempts to shield their receive system, it is nearly impossible to minimize ingress of FM broadcast signals to allow a weak signal being received via the Gunn transceiver to be received even when the converted IF is atop a frequency of an active FM broadcast station.

30 MHz, on the other hand, is a relatively low frequency (simplifying design) and more within the design range of the Gunn transceiver's receive mixer diode and it is very unlikely that one will encounter a significant signal in the 30 MHz area that is likely to cause an ingress problem.  An obvious problem with this frequency range is that there is little equipment to be found capable of receiving wideband FM signals around 30 MHz - a notable exception being the Icom IC-706 - which is not particularly cheap, portable or low-powered.

While it is practical to construct a homebrew 30 MHz WFM receiver, it is also possible to modify a readily-available kit receiver to work as a 30 MHz WFM IF receiver, and one of the more easily-available receivers is the relatively inexpensive Ramsey FR1.


Figure 2 - Details of the modifications of the VCO that allow tuning through 30 MHz.  Top:  The new inductor, L1.  Center:  The 68 pF capacitor added to the bottom of the board across L1.  Bottom:  The schematic of the modifications.
Click on an image for a larger version.
The new VCO coil (L1) to provide tuning to 30 MHz
The added 68 pF capacitor on the VCO
Schematic of the modifications to the VCO circuit
In addition to operation at 30 MHz this page contains additional modifications that can improve the performance of the FR1.  Note that these additional modifications are intended to improve various performance aspects of the receiver and are not absolutely necessary for 30 MHz WFM operation.

About the FR1:

The FR1 (or "FR1C") is a "bare-bones" wideband FM receiver intended for use as a broadcast band FM receiver amd as such, certain design decisions were made to keep it simple and its price down:  It contains a voltage-tuned local oscillator that mixes the input signal down to 10.7 MHz where is filtered through a 300 kHz wide ceramic bandpass filter and then demodulated.  As it is, it is a fairly easy matter to simply modify the local oscillator to provide tuning around 30 MHz - but a few additional modifications are required to provide optimal performance.

Being a bare-bones receiver, the FR1 does NOT have any input filtering at all:  Without such filtering, the receiver responds to image signals with sensitivity equal to that of the desired signal.  Within the FM broadcast band, this lack of filtering may not be noticed as FM signals are typically very much stronger than those that would appear at the image frequency:  When high-side LO injection is used, the image responses land at a frequency of twice the IF above the desired frequency (e.g. 21.4 MHz) and are within Aeronautical Mobile band, but when weaker signals are used, this lack of filtering can degrade performance somewhat due to "Image Noise" - not to mention making the receiver somewhat more susceptible to spurious responses.

Constructing the FR1:

Even though the ultimate use of the FR1 will be as a 30 MHz wideband FM receiver, it is recommended that it first be built as originally designed to allow testing using FM broadcast-band signals! 

Once the unit is verified to work properly, the unit may then be modified to operate at 30 MHz, but keeping in mind that the receiver will eventually be modified, there are a few changes that should be made while constructing the unit.
Tune up and test this receiver as described in the Ramsey instructions.  The adjustment of L2 is the only critical one and is best done using a weaker signal.  Because it is an IF adjustment, its adjustment does not need to be changed with the change in operating frequency if properly adjusted in the first place.    If a 10.7 MHz FM signal generator and AC-voltmeter or oscilloscope are available, L2 may be "fine-tuned" for lowest distortion and highest audio output - a procedure best done at low volume level.

Modifying the FR1 for operation at 30 MHz:

Once proper operation of the FR1 has been verified on the FM broadcast band, we can now modify its VCO so that it can receive signals at 30 MHz using high side injection - that is, with the local oscillator at 10.7 MHz above the receive frequency.  This involves a few simple modifications that are detailed in Figure 2:
The goal of these modifications is to shift the local oscillator frequency such that it tunes from approximately 38.7 to 42.7 MHz, correlating with a tuning range (using High-Side injection) of approximately 28 to 32 MHz - a range of about 4 MHz.  Note that the exact range isn't important:  A tuning width of 3-6 MHz (or from +-1.5 MHz to +-3 MHz) is adequate.
Figure 3 - Details of the 30 MHz bandpass filter.  Top:  The as-constructed filter, wired "dead bug" on a piece of circuit board material.  Note that the new "input" uses the "SCA Out" connector (J3) and the board is grounded and secured using existing holes in the circuit board.  Center-top:  Version of the filter wound using air-core inductors.  Center-bottom:  Passband response of the air-core filter.  Bottom:  The schematic of the filter.
Click on an image for a larger version.
The 30 MHz bandpass filter
Hand-wound bandpass filter coil
Passband response of the 30 MHz filter
Schematic of the bandpass filter

 
After the above modifications have been made, set the tuning potentiometer (R12) to 3.33 volts on the wiper and adjust L1 for a local oscillator frequency of 40.7 MHz, +- 100 kHz or so, corresponding with a receive frequency of 30 MHz.  For checking/setting the local oscillator it is probably easiest to use a general-coverage receiver or a scanner to find the local oscillator, taking care to assure that one is really hearing the LO and not a spurious response of the receiver.  For verification of proper operation, a signal generator or even an HF rig tuned to the top of 10 meters can be used to generate a signal in lieu of fancy test equipment.

Additional modifications

The following modifications are not absolutely necessary for WFM operation of the FR1 on 30 MHz, but they may be done to improve its overall performance.

A 30 MHz bandpass filter:

For best performance it is recommended that a 30 MHz bandpass filter be added to the receiver to remove "image noise" from the receiver input.  Note, however, that the addition of this filter is most efficacious if there is amplifier gain prior to the bandpass filter - such as the use of a low-noise MMIC preamplifier on the mixer output of the Gunn transceiver:  Without such preamplification, the receiver's intrinsic sensitivity may not be adequate to be able to detect the thermal noise of the Gunn transceiver's mixer diode and thus it may not be able to "hear" the image noise in the first place.  (This is to imply that optimal weak signal performance may be had with the installation of the MMIC preamplifier at the Gunn transceiver itself.

Details of this filter are shown in Figure 3:  The components of the filter are mounted "dead bug" on a small scrap of circuit board material (either single or double-sided.)  As it turns out, it is possible to wire the "SCA Output" connector (J3) as the bandpass filter's input (note in the picture how the center conductor of J3 is bent straight out and connects to the input of the filter) and mount the filter in the "empty" portion of the circuit board where the 9 volt battery would have original gone.  The original Antenna Input connector (J1) should be removed and a short piece of small-diameter coax cable (such as RG-174) is used to connect the output of the bandpass filter to where J1 (the "Antenna" connector) was originally connected.

To properly tune this filter it is necessary to use a sweep generator with diode detector and oscilloscope, a network analyzer, or a spectrum analyzer with either a noise generator or tracking generator.  The two inductors are adjusted to yield a flat bandpass of at least +-2 MHz (at the 3dB points) centered at 30 MHz, (see the center-bottom picture in Figure 3) dropping by at least 20 dB at 40 MHz with the insertion loss being between 4 and 6 dB at 30 MHz.  In the "image passband" (above 49 MHz) this filter should provide well over 40 dB of attenuation - more than necessary to provide highly effective rejection of image noise as well as minimizing spurious responses in the FM broadcast band.  The prototype showed about 50 dB of attenuation at the image frequency and 5 dB of insertion loss within 1.5 MHz of 30 MHz.

Winding your own coils for the bandpass filter:

While the prototype 30 MHz bandpass filter used some slug-tuned coils from a junk box, it is practical to simply wind some air-core coils that work equally well as long as one keeps in mind that they are mechanically rather fragile - See the center-top picture in Figure 3.

Suitable coils are wound using 9 turns of #22 enameled wire on the shank of a 1/4" drill bit.  If you look carefully at the picture you might notice that the two coils are wound in opposite directions:  This is done so that the ends of the coils are easily oriented such that they face the "input/output" terminals on one side and the other ends face the "100pF-22pF-100pF" junction in the middle.  Note also that the coils are very slightly elevated above the ground plane to minimize interaction between the body of the coil and the ground plane.

In initial testing, the close-wound 9-turn coils (in their original "un-stretched" state where all of the turns are against each other) resulted in a filter with a center frequency of about 27 MHz.  As can be seen in the center-top picture in Figure 3, the coils were spread slightly to lower the inductance, thus moving up the center frequency slightly.  With a bit of tweaking while using a tracking generator, spectrum analyzer, or some other means for "sweeping" the frequency response it is possible to "fine-tune" the bandpass to the designed response.

Once the filter is tuned, it is important to make sure that the coils can't move or stretch/compress and detune the filter.  If the filter has already been mounted to the receiver's circuit board, one source of coil movement (the flexing of the input/output leads and subsequent moving of associated components) has already been eliminated.  To make the filter even more rugged, it is suggested that the coils themselves be stabilized with RTV (silicone adhesive) to prevent movement.  If this is done, it is recommended that the filter be tuned up first (to make sure that it will tune properly) and then flow RTV over, under, and through the coils, and then re-check tuning before and after the RTV cures.

Improving the input preamplifier:

Figure 4 - Top:  The replacement of R5 with the network consisting of two resistors and one capacitor.  Bottom:  The schematic of the modifications.  Added components are indicated with arrows.
Click on an image for a larger version.
Picture showing the modification to R5 on the preamp.
Modifications of the preamplifier

It is possible to improve the input preamplifier somewhat over the original with the addition of a few components.  While the original preamplifier works, its gain is somewhat excessive for our purposes and the input impedance is not well-controlled.  The described modification will increase the stability of this amplifier, reduce the gain somewhat, and provide a consistent input impedance to better-terminate the bandpass filter.

The modifications, depicted in Figure 4, are as follows:

Improved discriminator operation:
The original data sheet for the ULN2111 shows coupling capacitor from the output of the quadrature tuning network (L2) connected to pin 9 - and that is the way that the FR1 is, in fact, wired.  The data sheet for the pin-compatible Motorola MC1357's that better performance may be usually obtained by using pin 10 instead of pin 9.  Why is pin 9 used, then?  Apparently, in some situations, some sort of overload or distortion could occur - particularly at higher deviations at lower frequencies.  At 10.7 MHz and using 75 kHz deviation, this isn't really an issue so one can benefit by making the following modification:

What "better" performance is gained by doing this?  The audio output is slightly higher, the limiting sensitivity is better, and the audio distortion on the received signal is lower.

Improved IF filter matching:

On the FR1, the single 10.7 MHz ceramic IF filter (FL1) is placed between the mixer and Q1, a simple IF amplifier.  While the output impedance of the mixer, an NE/SA602 is fairly high and a reasonable match to the 330 ohm in/out impedance of the filter, the input impedance to Q1, the IF amplifier is not.  The result of this significant mismatch is that the bandpass response of FL1 has significant ripple.

This deficiency may be easily remedied as follows:
Although this modification results in slightly lower overall IF gain, no appreciable change in sensitivity was noted on the receiver:  This is especially true if the "Discriminator Modification" (above) is also done.

Improved IF filtering:

The FR1 has only a single 10.7 MHz ceramic filter with a bandwidth of somewhere in the range of 280 to 350 kHz.  Typically, these ceramic filters have a stopband rejection of only 30-40 dB in the range of 9-12 MHz:  Outside this range, they can have a number of odd responses with less rejection:  It is these other responses that can contribute to spurious responses of the receiver.

Practically speaking, with no other signals likely to be present nearby, a single bandpass filter is adequate - but the presence of just the one filter and its lack of good ultimate off-frequency rejection can contribute to the presence and energy of miscellaneous spurious responses.  Because the LO itself is very strong (80 dB or so above the weakest detectable signals) it is understandable how even a very low-level mixing can appear on-frequency at 10.7 MHz.

Fortunately, it is pretty easy to add a second filter to this receiver.  Because these filters are rather ubiquitous, they are cheap (less than $2.00 each in single quantities) or they may be found for free in junked FM receivers.  For best performance it is recommended that the filters be placed in different stages rather than simply cascaded - and the second filter is easily added between Q1 and U1 and is done as follows and are shown in the pictures in Figure 5:
Figure 5 - Top and bottom views of the circuit board with the added ceramic IF filter.
Click on either image for a larger version.
Topside of FR1 with added IF filter
Bottom side of FR1 with added IF filter

As it turns out, U1 terminates the new filter fairly well with about 350 ohms or so.  The "output" of Q1 isn't quite as good a match, but it isn't severe enough of a mismatch to cause appreciable ripple on the added filter.

One noted change was that the no-signal noise (the "hiss") was notably quieter - probably due to less thermal noise reaching the discriminator than before from the FR1's preamp and IF amplifier, causing this noise to be below the limiter threshold:  If an antenna (or an operating Gunn transceiver) is connected to the FR1 with these modifications, much of the "hiss" will return - especially if the Gunn transceiver has an onboard MMIC IF amp.

Another notable change was a significant diminution of receiver spurious response:  The main spurious response (at about 31.8 MHz) is significantly reduced by the addition of the filter, no doubt due to less "blow-by" through the filter.  Even though the filtering adds a couple of extra dB of loss in the IF, no decrease in ultimate sensitivity was noted (provided that the "Improved Discriminator Operation" modification mentioned above was done) and tuning seemed to be "crisper" (due to "steeper" sides of the filtering resulting in a better shape factor) - especially with stronger signals.

Additional comments:

Receiver Sensitivity:

In testing with the above modifications, this receiver seems to have a sensitivity of 10-15 uV for a signal with 12 dB SINAD (as  measured in a 3 kHz audio bandwidth with 20 kHz deviation) at 30 MHz.  In the original Ramsey notes, a sensitivity of about 1 uV was claimed - a value that is probably optimistic (especially considering that the thermal noise alone in a 350 kHz bandwidth with a 3 dB noise figure and an antenna noise temperature of 300K would be over 1/3 of a microvolt alone) and certainly would not represent a signal of good quieting.

One minor problem with this receiver in its unmodified state is that it has somewhat of a tendency for spurious response - particularly at frequencies near multiples of the 10.7 MHz IF.  Fortunately, there are no such significant responses within 1.5 MHz of the 30 MHz center frequency.  The worst response in the unit that I constructed occurred at about 31.8 MHz - but it was completely overridden by a signal of about 30 uV.  As mentioned above, the addition of a second 10.7 MHz bandpass filter significantly reduces the magnitude of these responses.

Addition of an "S-Meter":

One further modification to this receiver would be the addition of a signal strength meter using an LM3089 or LM3189:  This modification may be described at a later date on this page.

"Narrowband" operation:

Less common at 10 and 24 GHz is so-called "narrow FM" operation.  Unlike VHF/UHF operation, however, this is not the traditional +-5 kHz deviation in a 15 kHz IF bandwidth, but rather a +-15 kHz deviation using an IF bandwidth of 30-50 kHz.

By replacing the existing IF bandpass filter with a narrower-bandwidth version (or simply placing it in series with the signal path - something that can be done with a selector switch or by using switching diodes) one can obtain superior (6-10 dB) weak-signal performance.  The disadvantage of this narrower bandwidth is that tuning of the 10/24 GHz units is far more touchy and is really only practical if an AFC circuit is used to compensate for drift!

Unfortunately, these narrower filters are relatively rare and somewhat expensive ($5-$20 each, depending on the model and source.)  It is possible that a few of these filters may be obtained via surplus channels:  Contact me if you are really interested.

AFC Circuit:

Originally, the FR1 has a rudimentary AFC, but in the above instructions, I recommend not installing R9 - effectively disabling the AFC.  Why is this done?  In the original FR1 operation, the tuning range of receiver is over 20 MHz - more than enough to cover the entire FM broadcast band.  In converting to 30 MHz, the tuning range is much more limited - typically in the range of 3-6 MHz.  This also means that the AFC lock-in range is also more restricted as well.

One can experiment with having R9 installed - but note that having an AFC on 10/24 GHz may be a two-edged sword:  While it is nice to have the receiver automatically tracking the signal, with AFC, the received signal will suddenly disappear when it gets out of the AFC circuit's tracking range rather than gradually drift out when no AFC is being used.  In many ways, having a gradual warning of drifting is preferable because you, as the operator, are more likely to know what happened to the signal and has had time to re-tweak the tuning.  If the signal abruptly disappears, on the other hand, you may not know if it simply drifted out of AFC range, or if something else happened.  If AFC is desired, it should be possible to implement it - but the user should know the potential pitfalls of its operation.

Note also that the AFC on this receiver affects only the receiver's operation:  On the fancy, commercial Gunn Transceiver units (such as the ARR) the AFC does not affect the 30 MHz IF receiver's tuning (which is, in fact, often crystal controlled and cannot be adjusted) but rather the AFC adjusts the Gunn oscillator's frequency - which means that the transmit frequency is tracked as well.  When this is done, the user on the other end must be aware that AFC is being used and shut his/her AFC off or else they may "fight" each other:  If configured properly, the "other" person (without AFC) also benefits as the AFC keeps the stations locked to each other - no matter who is drifting.

Another point of concern has to do with AFC polarity:  Because the Gunn Oscillator's frequency may be either above or below that of the other station's frequency, this affects AFC operation in that if the polarity of the AFC circuit is set improperly, it will always "push away" a received signal rather than lock onto it.  Of course, this does not happen if the AFC is only used in the 30 MHz receiver's LO (rather than on the Gunn oscillator) but also note that in this case the "other" person in the QSO does not benefit directly from the AFC either.

Power supply concerns:

The Ramsey documentation suggests powering of the receiver with a 9 volt battery - but it is strongly recommended that a 12 volt supply, regulated down to 9-10 volts, be used instead with the use of a 3-terminal regulator such as a 7809 or a simple transistor-Zener regulator.  The use of a regulated supply will improve the stability of the receiver as its tuning will drift as battery voltage drops, but the use of a "stiffer" supply will improve stability of the receiver as the current consumed by the audio amplifier would "modulate" the power supply voltage:  Such supply voltage modulation will cause distortion or feedback to occur in the audio or, even worse, an annoying frequency drift in response to audio - particularly if a weak 9 volt battery were used.

Additional comments:


Misc. comments on Gunn transceivers:

The use of a MMIC preamplifier at the Gunn transceiver's mixer diode:

Additional details of the MMIC circuit will be added here in the future.

It was mentioned above that better receiver system performance may be obtained by installing a MMIC preamplifier directly on the Gunn transceiver module itself to amplify the signal from the detector diode.  A good choice for this MMIC is the MAR-6 (a.k.a. the MSA0685) MMIC which may be obtained from DownEast Microwave.  This MMIC has a fairly low noise figure (under 3 dB) and moderately high gain (about 20 dB at 30 MHz) and is useful for overcoming cable losses between the Gunn transceiver and the IF receiver.  An added benefit is that it also offers some protection to the fragile mixer diode, protecting it from static discharges that might enter through the IF line connecting to the receiver.

This MMIC consumes only about 16 mA and one typically constructs the (very simple) circuit on a small piece of circuit board mounted directly on the Gunn transceiver right at the mixer diode connection.

The use of a MMIC preamplifier on a Gunn oscillator without a mixer diode:

It was recently noted that there are some surplus Gunn units on the market that do not have a mixer diode, but these units were still used as microwave motion detectors - a task that implicitly requires a means of detection.  After some puzzlement, Ron Jones, K7RJ, determined that the Gunn diode itself was being used as the detector diode:  This feat is possible only because there is relatively heavy coupling between the Gunn oscillator cavity and the feedhorn (e.g. a very large iris) and through careful decoupling of the power supply and the detected signal.

In some very crude testing, Ron determined that this "dual use" of the Gunn diode also worked at IF as well.  In his tests Ron did not make any attempt to optimize either the Gunn diode's power supply decoupling or IF signal coupling/matching but was able to make successful QSOs over a distance of a dozen miles.  While it is expected that the ultimate sensitivity of this detector scheme is less than that of the "traditional" mixer diode, it is also strongly suspected that Ron's first attempts to couple the received signals from the Gunn diode working as a detector could be greatly improved upon - notably through proper impedance matching and low-noise amplification- using a MMIC.

At the present time, further experimentation to determine an optimal means of coupling to the "Gunn diode as a detector" has not been done.

Disclaimers:

The reader is solely responsible for the modifications made to the FR1 and there are no guarantees that the receiver and/or modifications described will be suitable for your applications.

This page is not meant as a product endorsement of Ramsey's products, but rather as a resource to direct interested parties to a potential source of equipment for the described use.

Go to the KA7OEI main page.
 

This page last updated 20110518

Since 12/2010: