Softrock Lite 6.2
Adventures in Electronics and Radio
Elecraft K2 and K3 Transceivers
I've been studying your Z90 pages, and wondered about the
sweep widths. I see that over 250 kHz is not allowed. Is this due to a
hardware limitation, some programming necessity or what? I can see that a
receiver's front end filtering would set some limitations to the bandwidth.
I'm thinking about the Z90 used as a spectrum analyzer for sweeping
Yes, one practical limit to sweep width is set by the bandwidth
of the receiver circuits ahead of the IF connection point. That's why is is
desirable to make the IF connection ahead of the receiver's selective
filters, early in the RF chain. For example, the 455 KHz IF output in my
Racal RA6790/GM receiver is after the IF filtering, so listening in the SSB
mode limits the usable part of the Z90's span to 3 KHz or so,
regardless of whatever span setting I've entered.
The 250 KHz sweep limit is set be the number of pixels
available, the maximum panadapter selective filter bandwidth and the desire
to keep the display as "lively" as reasonably possible.
With 1 KHz maximum bandwidth and 240 available pixels for display, the
actual usable span is 240 KHz, and I probably will make that the maximum
Even that means there is a 3 dB error potential in amplitude, with regard
for other types of error (where the spectral line is half-way between the
two tuned frequencies). For ham purposes as a band activity display, that
error is not critical and certainly acceptable (and much better than
available than in my SM220 with BS-8 adapter card, for example) And, there
are things I could do in software to improve that, such as scanning in
smaller steps and taking the peak reading or average or whatever, but that's
really pushing the design beyond its intended purpose.
Since the command interface is open, you could, of course, write your own
BASIC program to command the frequency to a specific point and grab the
reading. (That specific command is not in the current firmware load but it's
on my list of things to add to it.) If you have sufficient time, therefore,
it would be possible to scan as many MHz as you wish, in as fine a frequency
step as is reasonable (1 Hz increments).
For a 1 KHz filter, the minimum dwell time should be somewhat over 1 ms, so
with a 200 KHz span, the actual refresh is around 3 traces / second when the
rest of the overhead is factored in.
If you are interested in filter analysis, you should really look into the
N2PK VNA. There's a Canadian ham selling PCBs for it. I have one (one of the
W8WWV group purchase) and it's an
excellent piece of gear--within its frequency range fully as accurate as any
of my HP network analyzers.
N2PK's VNA home page:
N2PK VNA PCB set:
Warning--this is a complex project. I bought mine as
an assembled unit when Greg, W8WWV, had a small quantity of the VNAs
professionally assembled. Looking at the assembled PCB, I seriously doubt my
ability to construct one.
My preference would be for a black enclosure...just my vote
What size is the cabinet?
Both the Z90 and Z91 use TenTec's BU/BK series enclosures, the
Z90 uses a BU or BK-959 and the Z91 uses a BU or BK-929 enclosure. BU
indicates a blue top and bottom and BK indicates a black top and bottom.
I will have custom front panels, laser cut, powder
coated and silk screened, but the basic enclosures are from TenTec. It may
turn out that the initial build is so small that there is no quantity
discount for all cabinets of a common color. If that's the case (no pun
intended), I should be able to give purchasers a choice of black or blue,
assuming TenTec has your color choice in stock. (They run the cabinets in a
batch and if they sell out of a particular color/style you may have to wait
until the next run.) Otherwise, my preference is for the large cabinet to be
blue and the small cabinet black, but if there's a consensus one way or the
other, I'll follow suit.
I've measured the cabinet dimensions as follows. Note
that there are screws on the side that stick out a small distance. Also,
suitable clearance at the back is necessary to provide access to the
connections and to select the input attenuation switches, if necessary. I've
shown the figures to 3 decimal places, but I made the measurements with a
standard machinist's scale, so the actual accuracy is less.
||2.75 inches (70 mm)
||9.375 inches (238 mm)
||9.50 inches (241 mm)
||5.75 inches (146 mm)
||9.375 inches (238 mm)
||9.50 inches (241 mm)
|The Z90 is shipped with a bail kit
that allows it to be angled upward by flipping the bail leg down.
Installing the bail kit increases the overall height approximately 0.5
inches (12 mm) above the figures in the table.
Will a Z10000 Buffer Board work with a <add your favorite radio make and
model here> to provide the necessary
signal for the Z90? [This question was asked concerning an FT-920, with a
68.9 MHz 1st IF, but because it is of general applicability, I've
intentionally changed the question.
I've added an extensive discussion to the draft Z10000 Buffer Amplifier
Assembly and Operating Manual addressing how one might connect to a generic
receiver or transceiver. This Manual will be available at the Documents page
when it's a bit closer to completion, likely around August 10th.
From a technical prospective, the Buffer Amplifier
works well beyond 68.9 MHz and is thus compatible with your Yeasu. Likewise,
the Z90 will work with an IF of 68.9 MHz, although the technical
specifications are very slightly degraded from operations at lower
frequencies. (I've successfully tested it with simulated IF frequencies up
to 73 MHz.)
The more difficult questions relate to whether there is enough room inside
the Yeasu to install the board (it's about 1.25 inches x 1.5 inches) and
whether there's a place to attach the buffer board to that will not disturb
the Yeasu's operation.
I'm confident that the answer to these two questions is yes, but since I
don't own an FT-920 I can't provide you an answer with 100% certainty. It's
also the case that it might be necessary add a couple of impedance matching
parts to the buffer amplifier, depending on where inside the FT-920 you make
the connection. I've tried to address these topics in the Z10000 Manual and
I can provide more specific information if necessary.
Also, I can provide an assembled buffer amplifier for a test installation
before you decide to purchase a Z90, but in order to make an assessment of
whether the buffer amplifier is correctly installed and working you will
need some test equipment. [This offer is open to anyone; you pay the postage
to return the buffer amplifier and agree to return it within 10 days or so.
Contact me via E-mail at
Jack.Smith@cliftonlaboratories.com if you wish to borrow a buffer
amplifier for this test.]
How complicated is the kit; does it require special tools or test equipment
to build or align and how long does it take to construct?
I've designed the Z90 and Z91 to be simple to build and not to require test
equipment beyond a digital voltmeter and ohmmeter.
Based on the time it has taken me to assemble
prototype boards, and Dario's experience as an independent prototype board
assembler (see N5QVF Build page for his
experience) I believe than an experienced kit builder should be able to
assemble a Z91 in about 8 to 9 hours, and a Z90 in about 9 to 10 hours. This
time is from opening the box until checkout completion.
These estimates are likely on the generous side, as
the board Dario assembled was not silk screened and solder masked, and he
had an extra module to assemble that not required in the release version, as
the DDS Daughter board will be provided as a wired and tested module. This
removes an hour or so assembly time, as well as simplifying the
The Z90/91 have about 25 surface mount components and
the rest through-hole parts. All the passive (resistors and capacitors)
surface mount components are 1206 size or larger. Although small in
comparison to most through-hole parts, these are considered "large" in the
surface mount world. I've also designed the PCB to be significantly larger
than necessary in areas around the surface mount components to make assembly
easier. For suggestions on the tools and techniques I use to work with
surface mount components see my Surface
Mount Assembly page.
The test equipment required during construction is
simple -- a digital voltmeter and ohmmeter. During calibration stage the ADC
reference voltage may be set using either a digital voltmeter, or the Z90's
built-in facilities. It is also desirable (not strictly necessary, but
certainly useful) to calibrate the DDS time base. This can be accomplished
with a frequency counter or by zero beating against WWV at 10 MHz using a
As far as complexity is concerned, I'll rank it as
"moderate." I would not recommend it as the first kit to give a complete
beginner, but previous kit builders should have no problems.
How do the span, RBW, skip, dwell, video averaging relate to each other?
What should I see on the screen?
This is far too complex to be part of the FAQ, so I've added a new page
Display to address these questions.
My computer does not have a serial port. Can I use the Z90 or Z91 with a
Yes. The first six months I worked on the Z90 hardware and software, I used
an older 500 MHz Gateway laptop computer running Windows 2000. That computer
had one hardware RS232 port and USB ports.. I found no difference in
performance using either the built-in hardware port or a Belkin
F5U109 USB-to-serial adapter. (But see later comments on the F5U109.)
A couple months ago, I updated to a Gateway Duo-core
laptop equipped with Windows XP MultiMedia Edition with no "legacy hardware"
ports -- it has only four USB ports and no RS-232 or Centronics printer
port. I use a Keyspan
USB "port replicator" that provides an RS-232 serial port, a Centronics
parallel printer port and two USB expansion ports to communicate with the
Z91 I use for firmware development.
When I tested the Belkin F5U109 with my new Gateway
computer, I found that it worked normally with the Z90-Control software.
However, I found that it failed when I used the Swordfish loader program to
upload new firmware to the Z90. It's important that you have a way to update
the Z90 firmware using the loader (supplied with the kit) and accordingly,
I cannot recommend the Belkin F5U109. It might work for you, as it did
with one of my computers, or it might fail in updating the Z90's flash
firmware code, as it did my new computer. I do recommend Keyspan's
UPR-112, based on actual use. I imagine--but have not tested--other
Keyspan adapters should work as well as its UPR-112.
[16 Aug 2006] Stan, W5EWA, reports that his
USA-19H USB-to-serial adapter works well with the prototype Z90
he is testing. Stan says it cost about $40 at the local Fry's Electronics.
Incidentally, looking at Keyspan's product listing, I
see a USB adapter with four RS-232 ports, model
This could be a useful if you need multiple serial ports. I have not tested
it with the Z90, but based on good results with two other Keyspan
USB-to-serial adapters, it should be a good bet for Z90 compatibility.
I noticed your product on the Elecraft website and I have been looking at
your website regarding the Z90/91. I was wondering if the displays for both
variants will contain your call letters or will you have the provision for
When I get around to it (next couple of weeks, I hope, but certainly before
the kits ship) I will add code to allow users to program in their own call
sign or name (or whatever, but not more than 10 characters). In the case of
the Z90, this will have to be entered via the RS-232 interface using
eitherZ90-Control software or a terminal program. The default entry will be
something other than my call, maybe CALLSIGN or something like that.
I may also add callsign personalization to the Z90-Control software, so
that screen copies and prints will show the owner's callsign. Probably at
the top left corner below the parameter string.
When connected to my K2, on the higher bands, signals below the tuned
frequency display to the left of center, but on lower bands signals below
the tuned frequency display to the right of center. Is this normal?
[This topic is not limited to the K2. JRS]
The short answer is "yes" this is normal. Let me provide some detail,
The Z90's current firmware always shows a lower input
frequency to the left of center and higher input frequencies to the right of
center. This is how a real spectrum analyzer works; the frequency sweeps
left to right, with the starting, lower, frequency at the left.
However, the relationship between the RF input frequency and the IF
frequency depends on the receiver architecture.
Let's look at an example with a double conversion,
up-converting receiver design using the arrangement shown below.
The receiver is tuned to 14.100 MHz and
there's a second, weaker, signal at 14.050 MHz. In this example, the
receiver has two Z90's connected, one at the 1st (45 MHz) and one at the
second IF (455 KHz).
The receiver architecture inverts the frequency band
with each conversion stage. Thus, the 1st IF frequency band is a mirror
image of the frequency to which the receiver is tuned; signals lower in
frequency than that to which the receiver is tuned are translated to a
higher 1st IF frequency. A Z90 connected to the 1st IF will therefore show
the lower frequency (considered with respect to the 14 MHz band) as to the
right of center.
If the Z90 is connected to the 2nd IF, the 2nd mixer's
inversion cancels the 1st mixer's inversion so the Z90's display will have
the correct display-to-frequency relationship.
Some receivers invert their IF output and others
don't. For example, my Watkins Johnson HF-1000's 455 KHz IF output is
inverted with respect to the original RF input. My Kenwood TS940's 8830 KHz
IF output, however, provides correct sense; lower RF frequencies correspond
to lower IF output frequencies. However, an upconverting receiver always
inverts the 1st IF (A bit of working with the math will convince you this is
necessary to avoid spurious signal problems,.)
Now, let's look at the K2, which is a single
conversion receiver, with high side injection on some bands and low side
injection on other bands.
We'll start with the 10 meter band. The K2's local
oscillator runs below the tuned frequency (low side injection). The IF
output tracks the RF input; frequencies below the one the receiver is tuned
to are translated into lower IF frequencies and the Z90 displays the visual
image of the signal correctly with respect to left/right of center.
Now, suppose we switch to the 80 meter band.
Again, we'll suppose the K2's receiver is tuned to 3.6 MHz and there's one
other signal on the band, a weaker one at 3.550 MHz.
On the 80 meter band, the K2's local
oscillator runs above the received frequency (high side injection) and hence
the IF frequencies are mirror imaged about the tuned frequency. Signals that
are below the tuned frequency appear on the Z90 right of center.
Now to the user, all of these frequency conversion
schemes are totally invisible--the K2's microprocessor adjusts the BFO and
frequency readout and tuning steps to make the receiver act consistently,
regardless of whether high-side or low-side injection is used for any
particular frequency band.
But, the Z90 displays the actual IF frequency and
hence will show a normal display on some bands and an inverted frequency
sense on other bands.
I have on the list of things to do to add a software
setting to the Z90 and Z91 to reverse the sense of the display, so that a
frequency lower than the center will appear at the right side of the screen,
not the left, so that someone with an inverting receiver can see the correct
relationship on the screen. If the relationship between input frequency and
IF output frequency is consistent, such as for an up-converting receiver,
this option makes sense.
When I try to generate an output signal in Signal Generate mode, the output
drops off above 60 MHz. Is this correct?
This is intentional. The Z90's Direct Digital Synthesis signal source uses a
60 MHz low pass filter as an anti-alias filter and will sharply roll off
signals above 60 MHz. An anti-alias filter is essential in the
digital-to-analog conversion process to control spurious signals.
As I exclusively use USB SSB for all my 50 through 1296 Mhz contacts, the
BFO sits below the center of screen, but moves to around 1.5kHz Higher if I
change to LSB. I note that you have 4916 kHz set as center frequency in your
explanation of the leak through of the BFO...I'm trying to get my head
around what I'm seeing, with my BFO leak through BELOW the center frequency
of 4914 kHz...I think everythings right ??
Yes, this is correct. The K2 has a single IF filter (by that I mean it does
not have separate USB and LSB filters; I know it has two filters) and is a
single conversion receiver, so in order to receive both USB and LSB, the BFO
must shift a total of approximately 3 KHz. Ignoring inversion on some bands
for simplicity, in USB mode, the BFO is positioned at the lower side of the
filter passband, actually a bit below the passband is typical. In LSB mode,
the BFO must be positioned at the upper edge (or a bit beyond) of the filter
This illustration shows why the K2's BFO frequency must
shift when receiving different sidebands.
Note that the BFO shifts from the low side of the
filter for USB to the high side for LSB.
With only a single IF filter, fixed in frequency,
the BFO injection frequency must move from one edge of the filter to the
other in order to correctly substitute for the not-transmitted carrier.
The illustration assumes there is no extra frequency
inversion in the K2. If there is, such as on the 3.5-4 MHz band, USB and
LSB are flipped.
Commercial receivers sometimes accomplish via separate
USB and LSB filters. My Racal RA6790/GM, for example, has symmetrical USB
and LSB filters, with their edges at the BFO frequency. This also allows you
to simultaneously receive USB and LSB (called ISB or independent side band)
if the receiver is equipped with a second product detector and audio chain.
With two symmetrical filters, the BFO stays at the same frequency for USB
To avoid the cost and complexity of two SSB filters,
the K2 designers use a single filter, and move the BFO from one side to the
other—the K2's microprocessor takes care of this for us when the mode switch
is pressed, so the process is transparent to the user. If you have a
panadapter connected to the K2's IF, you will see the BFO leakage signal
move. I believe the shift should be 2.5 to 3.0 KHz in theory (it should
equal the bandwidth of the SSB receive filter, plus a bit if the BFO runs
outside the filter bandpass as is commonly the case).
This then causes a question as to what the actual IF
frequency of the K2 is. I started out setting the stock IF frequency setting
to 4914 KHz, and then moved it to 4916 KHz at the suggestion of a K2 user
who thought it would be closer to the actual IF frequency. That's one reason
I added three "custom" IF frequencies, so feel free to define your own IF
that matches your particular K2. The custom IF frequencies are settable to 1
Hz resolution. Depending on whether you have downloaded the Z90-Control
program updates, your computer may show 4914, while the actual Z90 firmware
sets it at 4916. If you have not already done so, upload the most recent
file changes to the Z90-Control program.)
There's also a question about what is the "center" of
the IF frequency. For example, in SSB mode, it should be half the filter
bandwidth from the BFO frequency. In CW mode, it will be the same, but the
BFO offset shifts as well as the filter bandwidth changes. So, there's
really no single "center" frequency for all modes and filter bandwidths.
Hence, I decided to set the stock frequency to 4916 and let users program in
a custom IF frequency that will match their own favorite mode and filter.
Finally, further complicating things, I believe that when the IF filter is
tuned to different bandwidths via the varicaps, the center frequency of the
filter also changes to some degree, and that is compensated for by shifting
the BFO slightly, so as to maintain the same beat note and keep it centered
in the filter passband. Again, this is done by the microprocessor so it is
transparent to the user, but if you look at it with the Z90, there will be
Now back to your question as to why the BFO is below
the center frequency for USB...I think that is normal; to receive USB, the
BFO should be at the lower edge of the passband, if the K2 has not inverted
the frequency response. On 28 MHz, the K2's local oscillator is on the low
side, so there is no inversion. On 3.6 MHz, the K2's LO is on the high side,
so there is frequency inversion and the BFO will be on the reverse side,
i.e., on 3.6 MHz USB, the K2's BFO should be on the high side of the filter.