Here is the problem:
For the capacitive element in the LC circuit we have
essentially two 660 pF caps in series.
This results in a total capacitance of 330 pf. I measured 362 pF.
To get a resonant frequency of 7.0 MHz with 362 pF we need
1.428 uH.
To get 1.428 uH on the PTO coil form we need about 21 turns
of wire.
21 turns on our coil form yields 1.440 uH and resonates with
362 pf at 6.9708 MHz
That’s pretty close to what we need, but the problem
arises when we screw in the brass tuning screw. This reduces the inductance and raises the frequency. Putting the screw
all the way in reduces the inductance to 1.138 uH resulting in a resonant
frequency of 7.8414 MHz. So with a coil
this large (that we must use if we want to tune down to 7.0 MHz) we end up with
a tuning range that is far too large. We
only need 7.0 to 7.3. In effect, this
means that we end up using only a small portion of the tuning range: We can turn the screw approximately 34 times,
but only 6 turns keep us within the range of 7 to 7.3 MHz (the 40 meter
band). There is about 50 kHz per turn of
the dial. This makes tuning
difficult. It becomes more difficult to
separate stations and tune them in. It
would be better if we could tune across the band using more turns of the dial. At least 15 turns of the dial would be
nice: That would mean about 20 kHz per
turn. But how can we do this?
Just using a steel screw slows the tuning rate down. In a normal PTO we increase the inductance
(and reduce the frequency) by gradually introducing a ferrous material that
increases the inductance of the coil, pushing the frequency of oscillation
down. But our brass screw is
non-ferrous. This means that putting it
into the core does not change the permeability of the coil. The permeability of brass is the same as that
of air.
What does happen, however, is that introducing the brass screw
into the coil causes currents to flow in the screw. These are called eddy currents. In effect they become shorted secondary coils.
And they have the effect of lowering the
inductance of the coil – this is why the frequency of the oscillator increases
as we screw in the brass screw.
When you use a steel screw you get both effects: As you
screw it in, eddy currents flow in the screw, reducing the inductance and
increasing the frequency of oscillation.
But you are also introducing ferrous material – this pushes in the
opposite direction, increasing induction and lowering the frequency of
oscillation. I think the eddy current
effect dominates, but the increase in permeability pushes in the opposite
direction. This means that with a steel
screw you have to use more turns to cover the same frequency range. And that is what we want.
For example, using the same coil, with screw of the same
thread pitch (the same nuts), with both screws ten turns in, one turn of the
brass screw moved the inductance .014 uH.
The same single turn of the steel screw only moved the inductance .005
uH. So just because of metallurgy, the
steel screw will lead to a lower (better) tuning rate. I used a Hillman 45479 screw
that is steel with a Zinc (anti-corrosive) coating.
But there is more: steel
screws are also available with tighter (#28) thread pitches. The Hillman 45479 uses this tighter thread pitch. This too means that more turns are needed to
move through the same tuning range. Again, that is what we want.
I found that using a steel screw with #28 thread pitch allowed
for the coverage of the 40 meter band in approximately 11 turns of the dial. That is much better than what we got with the
brass screw: About 27 kHz per turn
instead of the 50 kHz per turn that we got with brass. But it is not quite good enough. It would be better if we could use the
entire range of that PTO coil form.
Solution Two: Add
a fixed inductor in series with the PTO coil.
After some noodling, I decided to split up the
inductor: A portion of it would remain
fixed, the other portion would continue to be tunable.
I estimated that I was starting out with a coil of about
1.428 uH. So I just put a 1 uH choke in
series with the variable inductor and reduced the variable coil to about .428
uH (about 9 coil turns). This worked,
but it worked a bit too well! It would
not tune the entire 40 meter band. So I
figured I needed less fixed inductance and more variable inductance. I found an air-cored coil in my junk box and
cut it so that it measured about .650 uH.
I added turns to the variable coil, going to a total of 15 turns. This REALLY worked well and yielded the 26 or
27 turns to tune across 40 meters that you can see in the video.
TWEAKS:
Later, I tweaked it a bit more: With 15 turns of #22 wire on the variable inductor, a steel screw tuned from .791 uH (screw out) to .662 uH (screw in). I put one additional turn on the fixed inductor, making it .749 uH, or about 8 turns of #22 (wound tighter on a cardboard tube from a coat hanger than was the coil on the variable inductor). With these coils I could tune from 6.9772 to 7.386 MHz. That's a bit more than we need but this allows us to keep the tuning away from the ends of the coil where tuning is more likely to become non-linear. I am able to go from 7.0 to 7.3 MHz in 23 turns of the dial. And the tuning is quite linear: The first turn from 7.0 MHz moves the frequency 12 kHz. At the mid-point of 7.150 MHz, one turn of the dial moves the frequency 12 kHz. At the high end, going down from 7.3 MHz, one turn of the dial moved the frequency 11 kHz. That, for me, is VERY linear tuning. You probably will have to adjust the coils a bit (just squeezing the turns together or spreading them apart) to get the tuning range where you want it.
YMMV – Keep it simple!
Like they used to say in the commercials: Your Mileage May Vary. There are many ways of doing this. The objective is smooth tuning across the 40
meter band. I think that by varying the
pitch of the variable coil turns you could get a more linear tuning response (please let us know if you have any luck). You might also be able to get similar results
by changing the amount of capacitance in the feedback network (which is also the
frequency determining element in this simple Colpitts oscillator). But remember that simplicity and a low parts
count were also our objectives in this.
This mod adds only 1 part (the fixed inductor), requires the removal of
some turns from the main tuning cap, and perhaps the replacement of the brass
screw with a steel #28 screw and nuts.
http://soldersmoke.com/soldersmoke244.mp3
Video version at:
(118) SolderSmoke - YouTube
.jpg)
We are starting to see similar efforts in different parts of the world -- Andreas with university students in Germany, Daniel with high school kids in Canada. We hope there will be others.
I'd really come to like this old signal generator. The construction is superb. It was built to be repaired. As you open it up you find all kinds of useful diagrams and pointers. It is very solidly built - it looks like something that was built for the Apollo program. And it was given to me by a friend: Steve Silverman KB3SII gave it to me in 2017 -- he had it in his New York City shack. Dave Bamford W2DAB picked it up for me just before Steve moved out of the city.
I've already done one complex repair on it -- one of the tines on one of the selection switches fell of and I had to replace the tine. That was difficult, but it was a very satisfying repair.
But lately, the HP8640B started acting up again. It developed an intermittent problem that caused both the signal generator and the frequency counter to just shut down.
I was thinking that this might be the end of the road for the HP8640B. I even started looking for alternatives. But they were all very unappealing. They come in plastic boxes with names like Feeltech and Kooletron. The boxes are filled with flaky wiring and boards hot glued to the plastic. Yuck. The contrast with the HP8640B could not be stronger.
So I started to think about the problem. This was the first part of the troubleshooting process. I asked myself: What would cause several different systems (counter, frequency generator, and display) to all shut down? The power supply was a leading candidate.
I started reading the power supply section of the HP8640B manual. There was a line in there that caught my eye: The power supply boards had on them LEDs that glowed if the board was functioning. Thank you Hewlett Packard! I opened the top of the signal generator and found the power supply boards. Sure enough, there were the LEDs. I turned the generator on, and found that one of the lights was out. Bingo. (Trevor takes a look at the power supply boards in the video above. I have it cued up to the 12:57 point at which he talks about and shows these boards.)
Here was the other clue: The problem was intermittent. It kind of seemed like a loose connection. So I just unseated the board and took it out. I put some De-Oxit on the connector and popped it back in. Boom: The LED came on and the HP8640B came pack to life.
There is a whole bunch of great info and videos on the HP8640B on the internet. It is almost as if a cult has developed. This signal generator is worthy of a cult following. Count me in.
I especially liked the video below. Kevin really captures the admiration that many of us feel toward the way this piece of gear was built. He also kind of hints at the way this sig gen could become a pirate transmitter on the FM broadcast band (at 8:44):
I know that eventually the problematic plastic gears in this device might fall apart. I am prepared for this: I already have the metal replacement gears from India.
Thanks again to Steve Silverman KB3SII and Dave Bamford W2DAB for bringing me into the HP8640B cult.
Our high school direct conversion project made me realize that I really need to upgrade my digital multimeter. I've been using an old Radio Shack DMM that I bought about 25 years ago. It is OK, but it is not auto-ranging and it is starting to physically deteriorate. So off I went to Bezos-land.
First I spotted the Fluke 101. I was enticed by the brand and the low price. But when it got here I was a bit disappointed. It is really small -- smaller than my cell phone. It is auto-ranging, and it does measure capacitance, but it doesn't measure hFe and the frequency counter only goes up to 100 kHz. I couldn't use it to measure the frequency of our DC receiver PTO. So, back to Bezos. (I'll keep the Fluke as a toolbox DMM.)
Next I found the AstroAI True RMS 6000 DMM. Obviously not as prestigious as the Fluke, but both the Fluke and the AstroAI are manufactured in China. The AstroAI was really inexpensive: Like 34 bucks. And Amazon would do same day delivery here. Soon it was on my front porch.
I've only been playing with it for a day or so, but I really like it. It is auto-ranging, it has automatic shut-down, the frequency counter goes up to 60 MHz, it measures hFe and even has temperature sensor. The frequency counter had no problem measuring the output frequency of our DC RX PTO. The screen is big and bright. And I think the True RMS feature will be very helpful when I try to measure amplifier gain.
I like it. And you can't really go wrong for the price. 34 dollars!
I have the Astro AI DMM in the Amazon ads on the right-hand column of the blog. I should have bought the package with the additional test probes. Click over there on the right for more info.
Posts
All Comments
Posts
All Comments
