Clifton Laboratories 7236 Clifton Road  Clifton VA 20124 tel: (703) 830 0368 fax: (703) 830 0711



To search within the Clifton Laboratories site, enter your search term below.


Software Updates
Softrock Lite 6.2
Adventures in Electronics and Radio
Elecraft K2 and K3 Transceivers



30 September 2006

More bandpass filters today, and writing.

The Z10000 buffer amplifier manual is finished and is posted on the Documents page. Or, you can view it by clicking here. It's a 2.8 MB PDF file.

When I started writing the Z10000 manual, I didn't plan on 45 pages. But, trying to cover all the bases turns out to be complex. And, I'm always reminded of Pascal's comment "I have only made this letter rather long because I have not had time to make it shorter." It turns out, from a bit of Internet research, that the thought was not original with Pascal, but it remains true regardless of who said it first.

A couple of photos showing some of the things I did today and then I'm going upstairs and see what has been happening in the real world today.


I drilled a few  filter enclosure top covers today to see how long it will take to do them all and to see if I can find a faster way to set up the job. I use a milling machine with a drill chuck for the holes. The drill bit is a "Unibit" stepped bit, which leaves beautiful, round, smooth holes in aluminum, unlike many standard drill bits optimized for other functions.
Second batch of 10 filter boards being assembled. At the stage shown, all have the trimmer caps installed and two have the fixed capacitors as well. I finished this batch through a rough tune-up late tonight.

No Swordfish DDS update tonight, but I should be able to add new material tomorrow.


29 September 2006

Today turned out to be K2 bandpass filter day. I built 10 bandpass filters in a quasi-assembly line process. I still need to prepare the enclosures and do final sweep tuning, but the first 10 circuit boards went well. Here are a few photographs.


The 10 completed filter boards. I've given each filter a rough alignment to verify that it is working.
I made a tuning test fixture by drilling four access holes in the normal enclosure. A tuning tool fits the holes and allows me to adjust the filter in the same environment it will experience in the real enclosure.
Holding through-hole parts in place for soldering always is a problem. Here's my solution--a small piece of foam (anti-static foam from an IC sample package), two spring clothespins and a sheet of 0.062" aluminum. The foam has enough give to hold the parts in place, and the clothespins apply just the right amount pressure. I hold the aluminum plate in a table-top vise.

The photo shows the first stage of assembly; installing the four variable capacitors. However, I use the same arrangement to install all the capacitors. I wish that I had thought of this years ago.

I'll also add a new installment to the Swordfish DDS page tonight.
28 September 2006

I've been busy today shearing the composite boards into individual boards and building a test assembly on each. Today, I've built and verified:

  • The Z90 soft switch board
  • Filter board (done yesterday)
  • Buffer amplifier board

I'm saving the DDS board until tomorrow when I can set aside a block of time to work on it without interruption.

The soft switch and filter boards worked first time, but the buffer amplifier test assembly gave me more trouble. I assembled the board (about 1.5 hours, but I was taking notes and photographs.) and made up a test cable assembly so that I could perform a swept frequency gain measurement. The response looked like it should, but it was 40 dB below the expected level. I thought that I must have either installed a wrong part (not good, since the parts all came out of a randomly selected kit package) or that I had a bad part. To make a long story short, after three hours of chasing non-problems, I found that the AD8007's power pin (pin 7) was not soldered to the pad. It would make connection intermittently, and I must have skipped soldering it during the assembly stage. The good news is that chasing the trouble gave me an excuse to expand the trouble shooting section of instruction manual to include a DC and AC voltage table.

Here are a few photographs of the test assemblies.


I selected a Z10000-K2 buffer amplifier kit at random. Here's the output cable and some of the parts envelopes.
Top view of the assembled Z10000-K2 buffer amplifier. Note that all connections are made from the board's top.
The board installed in my K2. (Equipped with 100 watt amplifier, noise blanker board and SSB board.) There's a piece of fish paper underneath the buffer amplifier to insulate the bottom components from contacting the top of electrolytic capacitors on the K2's main board.
Here's the swept frequency response of the test assembled buffer amplifier. The design gain is -14 dB and the result is -15.4 dB, which is close enough for my purpose.
Z90 soft key switch board with switches installed. Key caps are not yet in place.
If you have a Z90 or Z91 on order along with buffer amplifiers and wish to receive your buffer amplifier kit early, please send me an E-mail. I'll let you know the price, including shipping. When I receive your check, I'll mail the buffer amplifier kits.


I'll also add a new installment to the Swordfish DDS page tonight.
27 September 2006

The first batch of printed circuit boards arrived today. These are the DDS module, the soft-key board (Z90 only), the K2 4.915 MHz IF filter board and the buffer amplifier board.


To save money on the one-time setup fee charged with each order, all four boards are combined. I'll cut them into separate units, of course. The combined board is about 12" x 1.5."

The 4.915 MHz filter board is pictured below. I built this one today to verify the final filter board is functional. (It is.) I'm trying an experiment, attaching the filter inductors with neutral-cure RTV (Dow Corning #748) to prevent them from moving during shipment and thus potentially disturbing the filter's alignment. The tests I've done on individual inductors show an un-measurable effect on Q after applying a dollop of Dow Corning #748 across the winding ends. Normal RTV is not good for electronic assemblies, as it emits corrosive gas during curing, but neutral cure RTV is safe for electronic work. I don't know what chemistry is in #748, but it has a faint odor of peppermint.

I've aligned this filter and will check it again in a day or two after the RTV sets up.

The "missing" capacitors (such as C11) are optional parts, should additional capacitance be necessary to bring a stage into resonance. Since I've checked each inductor after winding for 4.1 uH +/- 5%, additional capacitance is unlikely to be necessary. Still, better to have an empty hole or two than bridging parts across PCB traces to bring it into alignment.


I'll assemble a buffer board and switch board tomorrow. I'm not looking forward to building a DDS board, as it presses the limit of my craftsmanship, but before sending the DDS boards out for assembly, I'll build one to verify the final PCB layout is correct. I changed the AD9851 pad footprint and although I made no other intentional changes, the new pad footprint required moving several traces, so there's always the chance of an error creeping in.

I've also worked on the buffer board operating an assembly manual today. I'll re-shoot the photographs tomorrow when  assembling a trial board, as the photos in the current manual are one PCB version behind. Assuming the buffer board assembly works, all that remains to finish kitting the buffer board is to cut apart the master PCB and package the individual parts envelopes together.

TenTec is ready to ship some of the cabinets within the next few days, with the balance in October. That should not be a pacing item, based on my conversations with TenTec's enclosure sales manager.

I've added a bit of new text to the Swordfish DDS page. It's going to be a few paragraphs a day as I can't take time away from the Z90 project at the moment.
26 September 2006

I've finished making cables. This morning, I finished the SMA stub cables for the buffer amplifiers and this afternoon I made the SMA-BNC cables that are included with the Z90 and Z91. I also finished the buffer amplifier kitting process. All that remains is to add the printed circuit boards to the kit packages. And, of course, to test build one buffer amplifier before shipping the first one. I also have more work to do on the instruction manual.

I've also ordered the cabinets from TenTec, with delivery consistent with the front panels. The main parts remaining to be ordered are the general purpose electronic parts for the main Z90 board. These should be available on relatively short order cycles and I'll start placing those orders next week.

Part of today's work--building SMA-BNC cables.
I made up a simple fixture to hold the SMA and BNC center pins while soldering the center conductor. It's a piece of 3/4 inch diameter Teflon rod with a hole in one end that's a snug fit for the BNC center pin and the other end a snug fit for the SMA center pin.

Since the fixture is Teflon, it does not melt under soldering temperature. You can see this effect in the photo, as the Nylon jaws of my assembly vice show the results of being hit with my soldering iron.

So, what can you do with Swordfish? Or, for that matter, a PIC? This is a question you might ask yourself, after reading my recommendation of the Swordfish BASIC compiler. Yes, Swordfish is a BASIC compiler, but it's not your father's BASIC. It's full featured, with scoped variables and many other improvements over other BASICs for the PIC microcontroller. As far a PIC microcontrollers go, my Z90 uses one, N8LP's LP100 wattmeter uses one and Elecraft's K2 uses several.

I've added a page with a simple project--an audio signal generator using direct digital synthesis with a PIC, a digital-to-analog converter and a bit of Swordfish code. 

I'll start the page with the schematic and an oscilloscope shot or two of the output. I'll add the code and some discussion over the next few days, as I have time. Click here to jump to the Swordfish DDS page. This project is suitable for Swordfish SE, the free version of Swordfish. (You must have a way of programming Swordfish's output into the PIC. I'll discuss the options for that as well.)

25 September 2006

Important news on the PIC compiler front. Swordfish, the compiler I wrote the Z90's microcontroller code in, and which is also used by Larry, N8LP, for his LP100 wattmeter, now has a free version, Swordfish SE, available for downloading at the Swordfish home site.

Swordfish SE is limited only by restriction on the amount of user RAM available, 200 bytes. Although this may not sound like much memory, Swordfish is so efficient at RAM recycling that Swordfish SE can be used for many "real" programs, not just for experimentation.

If you are a PIC programmer using 18F series devices, or if you simply wish to see the best PIC compiler I've used, visit Swordfish's home page and download Swordfish SE. I've been using Swordfish since the first of this year, initially in an Alpha version and more recently in Beta test. I recommend it highly.

On the Z90 front, I've been working on preparing the buffer amplifier kits. I have a couple of small parts due tomorrow that should let me complete the buffer amplifier kits, except for the printed circuit boards, due Wednesday. Today's project has been making cable sets for the buffer amplifier--approximately two feet of RG-178 and an SMA bulkhead connector. My 23 September entry has a photo of the first cable sets I made. Well, I'm almost finished making cable sets, 10 more to go and I can knock those out tomorrow morning.

Here's today's cable output. I've checked each cable for shorts and opens (measure the capacitance at the connector end--an open will show low capacitance; a short will read "overload" on my digital capacitance meter and a correctly assembled cable will read about 60 pF).

It might not look like much, but it required most of the day.

I also plan to work on the BNC-SMA cables tomorrow. At least those will not involve slippery Teflon coax. RG178 seems to take on a life of its own, and the cables almost slither off my workbench.
23 September 2006

My consulting project wrapped up Thursday, and I've had two days back at work on the Z90 project. I can report some useful progress, including winding the last filter toroid, a chore I'm more than happy to see completed.

Most importantly, I started packaging the Z10000 buffer amplifier kits today. Here are a few photographs of the kitting process.

"Kit central" at Clifton Laboratories. The only spare space I could find to set up a table is the second floor landing. It's out of the way and turns out to be a good working area with natural lighting most of the day.
Making up a resistor package. The small manila "coin envelopes" have a press-on label with the contents (schematic part ID and value). This envelope holds five different surface mount resistors (1206 size)and each tray bin has one of the resistor values. The process is:

-- Cut the parts from the tape and load the tray
-- Stick the labels on the envelopes
-- Stuff the envelopes with the parts
-- Touch the envelope flap to a wet sponge and seal

Today's progress. In about four hours (with help from my wife) I completed envelopes 1 through 8, representing more than half the envelope work. I can't complete the buffer amplifier kitting until the last parts come in, which should be the end of the week. I'll then grab one kitted package at random and assemble it. Assuming the process works, I'll be ready to ship the buffer amplifiers to those who have requested early delivery.

When you receive your buffer amplifier kit, you will be able to identify the envelopes my wife worked on, as their labels will be exactly centered, parallel with the envelope edge and just about perfect. The others are my work.

Part of the Z10000 kit is the output coax cable, RG178 Teflon, terminated with a SMA female bulkhead connector. You will also cut a piece from the free end for the buffer amplifier input as well as trimming the output cable to an appropriate length.

I went with SMA connectors as they provide higher quality than my original idea of an RCA connector. The downside, in addition to costing about 12 times what an RCA connector does, is that I have to assemble the connector. It requires a 0.128 inch hex crimp tool that I can't be sure kit builders will have on hand. I built four cable sets tonight and it took nearly two hours. I hope this speeds up with practice.

The Z90 will ship with a 3 foot length of RG174 with a male SMA connector on one end and a male BNC on the other, so don't worry if you don't have SMA cables around the shack.

20 September 2006

My consulting project is still running at maximum, and it has prevented me from updating this page as frequently as I wished. Fortunately, my part of the project should be wrapped up by the 22nd and I can get back to full time Z90 work.

I ordered the Z90 displays today and Crystalfontz, my supplier, is back ordered until October 14th. Assuming this date is met, it should not delay shipping the kits, as my target has been the last half of October. And, of course, it should not be an issue with the Z91's.

More parts have arrived, and all the parts are on hand for the filters, buffer amplifiers, DDS daughter boards and soft key switch boards, except the PCBs themselves, which should be on hand one week from today. I hope to have the last of the major part orders placed over the weekend.

I'm still winding toroids for the filters, and I'm becoming a bit faster at it. One thing that's speeded up the process is chemically stripping the magnet wire insulation and then using a solder pot to tin the leads. General Cement (now GC Electronics) used to sell a small bottle of magnet wire stripping gel but it's been many years since I've seen it on the shelf. However, a recent trip to the hardware store lead me to purchase a quart of Klean-Strip KS-3 "super duper paint stripper." The KS-3 has the same consistency as the GC did (if my memory of 30 years is still accurate) and a quart can ran a bit over $8. It works very well at removing magnet wire insulation. I dip the wire into a small container of the KS-3 and let it work for 15 to 30 minutes. The insulation then can be wiped away with a rag. I then clean the toroid with water to wash away any residue and tin the leads with a solder pot.

I hope to have the parts bagged and instructions revised for the buffer amplifier (both K2-specific and generic versions) early in October, or maybe even the last day of September. I can ship the buffer amplifiers before the main kit if anyone wants to get an early start. Assembling the buffer amplifier board takes perhaps an hour or a bit more. How long it takes to install in your K2 depends on whether you have the noise blanker board or not and whether your K2 has an extra hole on the back of suitable size for the SMA bulkhead connector. If the NB board is installed, the necessary modifications to it (removing insulation from three pins on the header and soldering on a 3-pin female header connector) takes perhaps 20 minutes to a half-hour. If you do not have the noise blanker installed, the kit will include an 8-pin socket that is to be installed on the K2 main board. Since I have not done that, I don't have a good estimate of time. No parts are removed; just install the socket. The last step is to mount the SMA bulkhead connector on the K2's rear panel. My K2 does not have the transverter board, so I used one of the spare holes. If your K2 does not have an open hole (I measured the SMA bulkhead diameter as 0.246" or 6.25mm) you will need to punch or drill a hole at a suitable location.

Speaking of punches, I'll recommend a hand punch to anyone doing much home brewing. I use a Roper-Whitney 5JR and it produces beautiful, no-burr holes in aluminum up to 0.062" thick, in diameters ranging from 1/16" up to 9/32." (Metric conversions left as an exercise to the interested reader.) You can see the details at These are not very expensive, and I believe I paid about $60 for mine, with a complete set of punches from MSC. is one source, but a 5JR is not hard to find. Mine is pictured at the right.

I'll probably purchase the 5JR's bigger brother, the XX model, to punch the chassis pan holes for the Z90 and Z91, as the 5JR's throat depth is right at the limit for one of the holes. (Not good planning on my part, but I didn't think about tooling when laying out the PCB and chassis.)

If you wish to receive the buffer amplifiers early (contingent on the PCBs arriving when expected) drop me an E-mail. I'll let you know when to mail me a check and the amount. I'll put the buffer amplifier kits in the mail when payment arrives. I don't believe separately shipping just the buffer amplifiers makes sense for international customers, because of the rather high cost of international money transfers. I have not forgotten my commitment to find a way to reduce the international transaction cost and will get to it as soon as possible.

15 September 2006

I've been tied up on a consulting project that unexpectedly went from high level review and comment into "all hands to the pumps" mode so I've been remiss in not updating the site to reflect progress on the Z90 kit.

I've written an instruction manual for the Z10010 4915 KHz K2 IF filter and posted it at the Documents page. Or, you can view it by clicking here.

By the way, yesterday I sent an E-mail message to all prospective Z90/91 purchasers asking them to notify me if they plan to use their panadapter with an Elecraft K2 transceiver. If you have not yet replied to that message, please do so. There are different part sets required for the K2 buffer amplifier and the generic amplifier, and the Z10010 filter assembly is useful only if you have a K2 and will not be supplied unless you use a K2.

A couple of responses to the E-mail message have asked for additional buffer amplifier sets. I have ordered printed circuit boards for a few extra buffer amplifiers, but the available quantity is rapidly diminishing. If you want an extra buffer amplifier, please contact me as soon as possible.

I've also ordered the production printed circuit boards for the main Z90/91, and all PCBs are now in the delivery queue. Shipments of parts from Analog Devices and Coilcraft are now here. I hope to start assembling some of the simpler parts packages in the next week or so.

I've also been winding toroids for the Z10010 filter. Each filter uses four T50-2 cores, wound with 29 turns of #22AWG magnet wire. So far, I've completed 10 sets of inductors.


Inductor winding progress to date--10 sets of 4 inductors.

After winding each inductor, I measure the inductance and Q on an HP4342A Q-meter. If the inductance is not in the range 4.0-4.25 microhenries, I remove turns. (So far, every inductor I've wound has either been within the tolerance or on the high side, requiring one fewer turn.) The minimum Q is 200, and no inductor has failed to meet this limit, with most being in the 210-220 range. After verifying the inductance and Q, I coat the windings with Q-dope to secure them in place. This slightly reduces the Q--an inductor measuring 220 before coating will read perhaps 215 after coating--but helps ensure long term filter stability by reducing wire movement. It takes 45 minutes for me make up four inductors, measuring from the time I start with a 21 inch length of wire until the last inductor of the set has been coated with Q-dope.

I find that winding more than one or two sets of inductors at a single sitting cramps my fingers, so I've been trying to wind one set in the morning and two in the evening.

Any prospective Z90 purchaser that wants to build their own filter as a kit, instead of being supplied with a pre-assembled one should contact me. The reason I decided to supply the filter as an assembled item is not related to the difficulty in putting one together, but rather the specialized equipment needed to align the filter once built. I've summarized the process I use in the Z10010 Instruction manual and before you decide that you would like to build and align your own filter, please carefully read this document and verify that you have the necessary test equipment.

Here's $1,500 worth of Analog Devices integrated circuits--AD9851 DDS chips, AD8307 log amplifiers and AD8007 buffer amplifiers.
10 September 2006

Continuing on the dynamic range discussion of recent days, I've decided to provide a pre-selector filter to purchasers that will use their Z90/91 with an Elecraft K2. There will be no additional charge for the filter.

The pre-selector filter is installed between the buffer amplifier output and the Z90's input. It is a four-pole Butterworth filter, centered on 4915 KHz and with a nominal 3 dB bandwidth of 200 KHz. Its purpose is to reject strong out-of-band signals that will otherwise reduce the panadapter's dynamic range.

Although the Z90's dynamic range is 60 dB, strong out-of-band signals can cause gain compression and baseband lifting, most noticeably when tuning the 40, 30 and 20 meter bands. In the 40 meter band case, I've measured shortwave broadcast stations in the 7400-7500 KHz range as strong as -40 dBm. In the 30 meter band, strong shortwave broadcast stations are found in the 9500-9990 KHz range, while in the 20 meter band, it's the international shortwave broadcast assignment 13570-13870 KHz. Although the K2 incorporates bandpass filters, it's possible to make a significant improvement in the Z90's out-of-band rejection via the pre-selector filter.

The filter will be supplied "wired and tested" as part of the Z90 or Z91 if you inform me that you will use it with a K2. I will also provide it as a kit if the purchaser has the necessary test equipment to align the filter. I use an HP8752B vector network analyzer. There is unlikely to be a detailed assembly manual with the filter.

I've added a new page with photos of the prototype filter and measured and simulated bandpass sweeps here.

I've tried to find a specification for the dynamic range of Kenwood's BS5 and BS8 panadapter modules for the SM-220 and SM-230 station monitors, but without success. The instruction manual says only that the display is 20 dB/division.

Heathkit's SB620 panadapter is stated as having a 40 dB dynamic range, with an optional 20 dB extension via a rear panel 20 dB switchable attenuator. Applying that methodology to the Z90, the dynamic range is 90 dB, comprising 60 dB of base dynamic range plus up to 30 dB of fixed attenuation.

06 September 2006

I've mentioned dynamic range, signal levels, antennas and buffer amplifier gain before, but the relationship amongst these deserves further discussion, illustrated with practical examples.

The Z90's dynamic range is about 60 dB. That means that for accurate amplitude representation, the strongest signal displayed should not be more than 60 dB stronger than the weakest signal displayed. If your receiver's S-meter's calibration follows the traditional 1 S-unit = 6 dB standard, this means the Z90 will accurately (within its overall 2 dB error budget) display the relative amplitude of one signal at S1 and a second signal at S9 + 12 dB, as displayed on your receiver's S-meter. This statement assumes that the Z90 is operated such as to maximize its dynamic range, i.e., in our example, the S1 signal produces a barely discernable pip on the Z90's display, and the stronger signal will be about 6 graticule divisions higher.

In practice, operating the Z90 this way may not produce a "lively" display--one in which you can see the band noise on the screen. When listening to 80 meters on a hot summer night, of course, the residual noise level may be well over S1. Conversely, on 10 meters, the noise level may be below S1.

What happens when the strongest signal is more than 60 dB above the Z90's noise level? Gain compression and filter leakage sets in and, to the user, the entire display appears to shift upward. The relationship between the strongest displayed signal other displayed signals remains more or less accurate, but weaker signals will be suppressed and disappear below the base line.

Let's see how this works in practice. Our sample signal is the shortwave broadcast station at 7465 KHz. The receiver is a K2, with a prototype buffer amplifier with -18 dB gain, so that the net relationship between 7465 KHz signal input at the K2's antenna and the 4915 KHz level is defined by the following table (repeated from yesterday's discussion.)

Condition Measured Net at Buffer Amplifier Output
Pre-amp ON +15 dB
Pre-amp OFF +1 dB
Attenuator -10 dB

In other words, a 7465 KHz signal at -60 dBm level measured at the K2's antenna input appears as a -59 dBm signal at 4915 KHz presented to the Z90's input, assuming the K2 is set for normal receive gain (pre-amp off and attenuator not enabled.)

I measured the 7465 KHz signal at -40 dBm, as received on my M2 log periodic antenna at 100 ft above ground, here in Northern Virginia at 6:40 PM tonight. That's a signal strong enough to scorch the paint on the side of your house, as we used to say. -40 dBm is 2.2 mV into 50 ohms, S9 plus 33 dB on a true S-meter.

Applying the conversion gain figures above, we can determine the signal level the Z90 sees at my K2's IF sample port output:

Condition Signal to Z90's Input
Pre-amp ON -25 dBm
Pre-amp OFF -39 dBm
Attenuator -50 dBm

Reference to the Z90's specifications show the recommended signal level input is -40 dBm. Hence, we expect to see reasonable performance with the K2's pre-amplifier off, or with the K2's attenuator enabled, but we also expect the Z90 to be significantly overloaded should we engage the K2's pre-amplifier.


First, let's see how it looks with no antenna connected to the K2. The preamplifier is on, and we see the BFO leakage signal. Background noise is around 0.3 graticule divisions.
Here's the 7465 KHz signal with the log periodic antenna connected to my K2. The K2 is operated in "normal" gain mode, pre-amplifier off and attenuator not used.

This combination, with the gain setting I'm using in the buffer amplifier, presents the Z90 with a signal level at the maximum recommended level. There's a bit of base band lifting, but all in all, the screen presentation appears normal and reasonable.

If we switch the K2's attenuator in, the Z90 is presented with a signal level around -50 dBm and we see no indication of base band lifting.
With the K2's pre-amplifier enabled, the Z90 is presented with a signal that causes it to be grossly overloaded. The baseband shifts up more than 20 dB. Still, the relative shape of the signals appears correct.
In fact, even with the Z90's input grossly overloaded, moving the displaced baseline down to a normal reference point restores a reasonable appearing display. Signals that are less than 60 dB below the strongest signal will, of course, not be visible.
Alternatively, the Z90's attenuation may employed. With 20 dB attenuation selected, the Z90's display returns to normal display range.
05 September 2006

I received an acceptance notice from the ARRL today for both articles I submitted for consideration for QEX. The main article describes the Z90 and the second is an in-depth discussion of the Gaussian crystal filters used in the Z90. The review and acceptance process ran just over two months. The letter provides no indication when the articles will be published, but Larry, N8LP, found about 18 months between acceptance and publication for his LP100 wattmeter article.

I've been working more on the K2 interface board over the last few days, which has turned out to be a more difficult task than I envisioned. The main issue centers around getting the isolation amplifier's gain correct. The K2 does not apply AGC to stages ahead of the RF amplifier, which means that if the buffer amplifier gain is too high, the Z90 will be overloaded by strong out-of-band signals, requiring the user to add attenuation via the Z90's attenuator switches. Too little gain, and the display isn't "lively" enough. Complicating the factor is that, when engaged, the K2's pre-amplifier boosts the gain 14 dB and that the appropriate gain depends on whether the K2 is used with a poor antenna or a good antenna, as the signal levels will differ considerably. A final complicating factor is that the K2's RF gain control does not change the signal level fed to the Z90, as the RF amplifier and post-mixer amplifier are not gain controlled.p;

I've been running a prototype buffer amplifier in my K2 with nominal 18 dB loss, and it seems close to correct for my antenna and operating conditions. Stan, W5EWA, is building a second prototype to test with his K2 and I've recommended he set the gain for around -15 dB net.

I measured the net gain from antenna connector to IF sample point in my K2 as follows:

Condition Net Gain
Pre-amp ON +31.6 dB
Pre-amp OFF +17.6 dB
Attenuator +6.6 dB

The measurements compare the signal output at Q22's 4.915 MHz output into a 50 ohm load with a 14.1 MHz RF input.

In my case, the net effect of the buffer amplifier's -18 dB gain is that signals at my K2's antenna input are translated about 1:1 in amplitude to the Z90's input when the K2's pre-amp is off. Given the noise level and antenna gain at my station, that works well on 40 meters. On 20 meters, the noise level is lower and operating with the K2's preamplifier on also works reasonably well.

Condition Measured Net at Buffer Amplifier Output Minimum Discernable Signal on Z90
Pre-amp ON +15 dB -120 dBm / 0.23uV
Pre-amp OFF +1 dB -110 dBm / 0.72 uV
Attenuator -10 dB -100 dBm / 2.3 uV

These are quick measurements, and I see one apparent inconsistency--the difference between pre-amp on and pre-amp off is 10 dB, not the 14 dB pre-amp gain. This is likely due to the noise figure of the pre-amp, as it adds noise as well as signal gain.

Even the reduced gain I'm using may still be excessive. Certainly with the K2's pre-amp engaged, it will be necessary to engage the Z90's attenuators under conditions of strong signals on the band.

Since gain is not required for a K2 connection, it would be possible to make a resistive pad type connection. I've tested a simple resistive connection and it worked reasonably well. However, this provides no isolation from stray signals out of the Z90 that would possibly leak back into the K2's IF system.  I've measured the leakage out of the Z91 prototype as being -78 dBm or lower, and the additional isolation provided by the buffer amplifier reduces this to negligible levels.


Typical leakage signal from Z91. Z91 is set for K2 IF frequency, 4915 KHz. The strongest leakage signal is approximately -78 dBm, at 30 MHz and is likely the DDS's 30 MHz master clock oscillator.

Here's the reverse gain (isolation) of a prototype buffer amplifier. This particular prototype is configured for wideband operation, so it represents the worst case for isolation. At 4.915 MHz, the amplifier's reverse gain is -76 dB.

When combined with the Z90's low leakage, this degree of isolation is more than adequate.

As a point of comparison, I measured the K2's leakage out the antenna port. The illustration at the right shows the leakage when the K2 is receiving at 3.8 MHz. The strongest signal is -62 dBm, about 16 dB stronger than the Z90's leakage. Of course, leakage from the Z90 is more important, as it is potentially fed into the K2's IF, which is why, even though the Z90 is low leakage, I do not recommend a straight resistive connection to the K2's IF.
Here's how the BFO leakage looks with the new buffer amplifier gain settings.


This is pretty much the worst case. 10.1 MHz, antenna connected, K2's RF attenuator turned on. No strong (or even weak) signals heard on the band. The BFO leakage is about 1.5 divisions.
Same conditions, except receive gain mode is normal (no pre-amp, no attenuation). The BFO leakage is the same, but the noise masks it to some degree.
With pre-amp ON, the background noise is about the same as the BFO leakage and it is hard to distinguish from noise.

I've also made a small change in the Z90-Control software to add an optional automatic baseline setting. When the box is checked, the software automatically places the trace so that the lowest signal coincides with the bottom graticule. At the moment, this option only affects the computer screen. I'm considering whether to add it to the LCD display as well.


01 September 2006

I've completed measuring my K2's performance "before and after" installing the most recent IF buffer amplifier. I've posted a brief summary of the test methodology and a sample before and after photo at a new web page, K2 Measurements. I found no significant differences after installing the buffer amplifier.

I've extensively re-written the K2 Interface page to reflect the version 2.0 buffer amplifier and to add a discussion of BFO leakage in Elecraft's K2. If you plan to use your Z90/91 with a K2, please read this page.

I've also moved the August 2006 Updates page to the archives. You may access it via the link at the top of this page or by clicking here.