Reducing RFI from a Samlex SEC1223 power supply ⚡
This is a pretty good power supply, but it had RFI at various frequencies, like 320KHz, 640KHz and more.
Measuring around S5 on my TS440SAT. Found this modification on the 'net
which turned out to knock the RFI down a lot, around four S units. The caps I used are poly films I got from the junk box,
and as you can see are not identical, but this doesn't matter here. I did have to drill a small hole above the
output terminals for a screw and grounding lug for these caps. This makes this power supply very good now.
The IEC line filter, which I did first, helped a little, but the caps above helped the most. This may be similar to plugging all but
the last hole in a leaky boat, biggest improvement is obtained when you plug that last hole. Do one of two "holes",
around 3dB improvement, do the second, a lot more dBs improvement.
The power switch on mine went bad, and I replaced it with one that fit the hole, DPST (switches both powerline wires), but without a pilot light. So I got
an LED pilot light from a dying Radio Shack, and changed the red LED in it to a blue one from a Xmas light set. The ones that have an inverted
cone molded in the plastic that houses the LED element, to make its light spread widely. Used a 33K resistor off the 12VDC (these LEDs are
much more efficient than older ones). Looks pretty,
and easily descerned from the other indicators in the shack.
If you keep the switch with the neon bulb pilot light, and you want to set this power supply for
250VAC in the USA or elsewhere that the 250VAC has a grounded centertap (what American power companies supply
to houses), move the lead of C5 from ground to the node "point A" as seen in the above detail of the
schematic of this power supply. Else the pilot light will glow because of a sneak path from ground thru C5 to the node "point A"
(which is still hot even when the switch is off). Thus be aware that the primary side of the switching circuits
will still be "hot" with 120VAC when the switch is off. Shouldn't be an issue once you put the
cover back on.
The Samlex power supply has an internal fan that should turn on when
the supply gets hot. But the thermostat must be set
rather high, the supply gets warm. So I figured I'd make the fan be dual powered.
Normally the fan is controlled by a transistor, Q5, a 2N2222. The fan positive lead
goes to the 13.8V output, and the negative lead goes to this transistor's collector. If
the thermostat goes open, the transistor turns on, turning the fan on. I added a 68Ω
power resistor from ground to this transistor's collector, to run the fan at a low power
level, and low audio noise, when the supply isn't so hot that it triggers the thermostat. This
will keep the supply cooler, and should help make the caps last longer. If the supply does get
hot enough to trigger the thermostat, the transistor Q5 will still turn on hard, essentially
shorting this new resistor, and the fan goes into high powered mode.
Converting a CATV splitter for 50Ω VHF receive work:
Using an air conditioneer shielded power cord on my rig's power supply
I salvaged a shielded power cord off a junked air conditioner to use it on my rig's power supply..
More recent manufactured window air conditioners have a power cord where both current carrying wires
are individually shielded. And includea a third ground wire. The shields are not grounded, and if you do connect
the shields to ground, the existing "smart" plug turns the power off, as it must think the power cord
is being cut by something grounded? Anyway, I removed that smart plug and replaced it with a regular 3 prong grounding
Above I grounded the shields to the ground wire, and the idea is that the powerline is now
shielded from RF fields. To
keep your power supply from getting RFIed. Or to contain RFI from a switching power supply.
It's fine to have parallel ground paths (the shields and the green wire),
but you want to avoid doing that with the normally current carrying wires. I
acquired a female IEC plug at a local hamfest you can wire to a power cord, as my power supply uses
an IEC power connector. (IEC is the sort of connector you find on desktop PC power supplies).
Above right is what a round cord looks like. Be aware of the hot and neutral wire guages, so you don't
draw too much current thru them.
Reducing RFI from ethernet routers
Many ethernet routers and switches have internal switch mode buck voltage reducers/regulators (separate
from the wall wart, which may have its own switcher).
These circuits can cause a fair amount of trash to pollute the AM broadcast band and HF. I found
that using one of those dual winding RFI coils salvaged from computer, VCR, DVD player and such
switching power supplies can reduce the RFI way down. You need to put it inside the router
housing like I did in the picture below
(or really close, in back) and have the DC power from its wall wart go thru this dual coil.
Look at how it was hooked up in the
switching power supply you scrapped and see where the inputs and where the
outputs are taken. It won't matter if you feed the outputs and get power from
the inputs, but you want to avoid getting the two sections out of phase.
A 0.1uF cap across the wall wart side of this dual coil also should help.
You shouldn't need to worry about RFI getting out along the ethernet cables, as
most if not all ethernet jacks have small isolation transformers associated with them. These
provide around 1000VDC HiPot isolation, and have maybe a dozen picofarads capacitive coupling,
common mode around -25dB in HF. This also means you could power the router off you ham shack's 12V power
supply bus, thus avoiding a switching power supply wall wart. The ehternet isolation transformers in the router
will prevent "ground loop" issues. An aside: I run another of these old ethernet switches off one of my computer's USB ports, using only the 5V
supply, drawing around 250ma. The signal wires (white and green) of the USB cable are left unconnected to anything.
DWI (DXing While Intoxicated)
Upper right: They caught me attending a hamfest!
Here I converted a Linksys router/BBHN node to be passively powered over Ethernet (POE).
Two of the 4 twisted pairs of ethernet cables are not used, and can be used to deliver
DC power (POE). The positive goes on the center pair (blue wires) and the return on the
brown pair. On the bottom of the board you can see which pins are the signal lines, as
there's a fine wire pair going from those pins to the ethernet transformers (black rectangles
topside). Note, if you don't find separate transformers, the jacks may have them internally, and
this means you can't get at the lines that connect the blue wires and the brown wires, and this trick isn't usable.
But if you see separate transformers (usually black rectangles about 7mm to 1cm wide and 2 to 4cm long, and
around 6mm tall) you can do this mod. You then should be able to use an ohmmeter to find
zero ohm conductivity between the brown and the blue wires of an ethernet
cable and traces on the circuit board. These nonsignal pairs are shorted together, and between the blue pair and the brown
pair are a pair of 75 ohm surface mount resistors. Here RA8 and RA11. Center point of these
resistors go to a cap that in turn goes to the router's ground. To avoid burning power, remove
one of these resistors per ethernet jack, else you have 150 ohms across the POE per jack. I used
needlenose pliers to crush them.
I used some pink nail polish stolen from the YL to mark the POE ethernet jacks.
Passive POE is a system where a power supply just injects some set voltage between the
brown pair and the blue pair, some at 12V and others at 24V. No handshaking like in regular POE
(which can be as high as 50VDC). The
Linksys I have uses a DC-DC converter chip that takes 12 to 24V and makes it 3.3V, higher input
voltage means less current draw. I looked up the converter's part number at
http://www.digchip.com/ to find its datasheet to see the specs. And check the
electrolytic cap on the input side to be sure it can take 24V as well. Be sure to label
the node so you remember that it's now a passive POE device, and what voltage it will want
This lets you place the node where there isn't powerline around. Like atop a tower.
I did a pair of jacks so I could drive an extra POE device.
. COAX TABLE
Here I took out the guts of a CATV splitter and put it in a small box with BNC
connectors, changed R1 from 150Ω to 100Ω, L1 and L2 are just a couple of loops of the resistor lead
wrapped a mandril like a 1/16 inch thick round rod (reproduce what the 150Ω resistor had),
and added some caps to make C1 about 1 1/2 times bigger (C1 compensates for the leakage inductance of T1). If you wanted to make a splitter that
could accept say 75Ω input and split to a pair of 50Ω outputs, change the turns ratio of T1 from
say 3 turns (N2) and 1 turn (N1), to 3 turns (same N2) and 2 turns (new N1). To do 50Ω in to
a pair of 75Ω outputs,
N1 would be 1 turn and N2 be 6 turns. R1 is twice what the outputs are. C1 I think is determined by what half the output impedance
will be, as it sees the impedance of the tap on T1, for 75Ω outputs it says the same, for 50Ω outputs it's 1 1/2 bigger. CAUTION There's supposed to be
isolation between the two outputs, but I wouldn't transmit into one output if the other output is
connected to a receiver. Below is a splitter that is 75Ω input (the F connector) and a pair of
50Ω outputs (the BNC connectors). I sawed off the two old F connectors and then drilled out
holes to accept the BNCs. I was going to thread these holes to accept the threads on the chassis mount BNCs,
but I didn't have the tap. So I just swedged the BNCs in with a vise, they are not going anywhere.
To hopefully protect my IC7300 transciever from nearby lightning strikes (not direct hits!) I did a mod
to my Heathkit HM102 SWR/power meter. The BNC connector replaced the original "input" UHF connector.
The antenna connector is still a UHF connector. I did this as an additional
modification. Having two different style connectors makes it easier when groping around for
cables behind stuff to get the correct connections done.
I wanted to try to protect my IC7300 transceiver from nearby lightning strikes (not direct hits!). I did a
mod to my Heathkit HM102 SWR/power meter. In case I forgot to switch the antenna switch to ground, I
placed a double throw single pole relay at the "input" to the meter where the rig connects to ground
the rig's antenna port when powered down. The meter was a convenient place to place this relay and to
package the modification.
How does it work?
When the rig power is off, the relay becomes unpowered, and connects the antenna "input" to ground,
and disconnects from the antenna. When the rig is powered on, the relay becomes powered and connects
the antenna "input" to the anyenna via the meter circuitry which is a toroid sampling circuit.
The relay is a 12 VDC type you find in a flat screen TV set or similar device. The relay is powered
through an #18 AWG wire connected to the rig's 12 VDC (actually 13.8 VDC) power supply. I used a 2A
fuse on the wire near the power supply, as this supply can produce 23 or so amps and I don't want
electrical fires! I used a bypass cap 0.01uF at the relay on this wire to avoid RF sneaking a
path. The relay's other 12 VDC connection is to ground. I used a rectifier diode across the relay
coil to suppress back EMF: cathode to +12V, anode to ground. I tucked the relay between the meter
circuitry circuit board and the coax connector for the "input".
I ran the relay 12 VDC wire thru a hole in the metal housing and then to the rig power supply.
The relay current returns thru the coax thru the rig and then to the power supply seems to not
effect anything, but I added a return 18 AWG wire to the relay anyway. This should divert most
of the return current off the coax.
The extra RF path thru the relay seems to have little impact on SWR across all HF bands and 6m.
I added a 1 megohm resistor across the antenna (output) coax connector, to bleed off static charge of
antennas that don't have a DC path to ground. This way, the radio front end won't see a
sudden static charge when the relay is activated.
L, C, Reactance and Frequency Calculator
Enter any two known values and press "Calculate" to solve for the others.
For example, a 1000pF capacitor or a 25.3 μH inductor will have 159Ω
of reactance at a frequency of 1 MegaHertz. Fields should be reset to 0
before doing a new calculation.
Adapted from http://ourworld.compuserve.com/homepages/Bill_Bowden/XLC.htm which is no longer there.
Inductive Reactance (Xℓ) = 2πFL
Capacitive Reactance (Xc) = 1 / ( 2πFC )
Resonant Frequency (Fo) = 1 / ( 2π√LC )
Modifying old CB radios that used the PLL02 chip, this mod steps every 5KHz
instead of 10KHz steps
An Inexpensive CB to 10 Meter Conversion by Jerry Coffman, K5JC some of which I quoted here. First do the mods
as described there, before doing mine for the 5KHz steps. Once you tune all the coils and transformers as he
described, you don't need to do it again after doing my mods. You only redo the channel switch wiring.
"In acquiring radios for my experimentation, I soon discovered there were very similar
radios to the three crystal, PLL02A PLL based ones I was looking for, but these radios had
only two crystals, without a 11.0866 MHz crystal to be changed! The PLL02A was still there,
but no suitable crystal to change. So, how do you move a radio up about 2 MHz in frequency when you
do not have a crystal to change? To begin, I downloaded a copy of the service manual5 from
www.cbtricks.com for the Hygain 2702 model radio, which was a typical PLL02A PLL radio, with only
the 10.240 and 10.695 MHz crystals.
These radios use the various pins on the PLL02A to apply or remove 5 VDC to change frequency,
using the channel selector switch. By modifying connections directly to the PLL02A, the frequency
produced can be changed. Pin P0 adds/subtracts 10 KHz; P1 20 KHz; P2 40 KHz; P3 80 KHz; P4 160 KHz;
P5 320 KHz; and P6 640 KHz. P7 was hardwired to always have 0 VDC and P8 always had 5 VDC applied to it.
In studying the PLL02A specifications 6 shown in Table 2, I discovered P7 should add/subtract 1.28 MHz and
P8 should add/subtract 2.56 MHz. In these 2 crystal radios, if a pin has 5 VDC, it does not add frequency;
if has 0 VDC, it adds frequency. A little work with the voltmeter showed P7 is always 1 (0 VDC) and P8 is always 0 (5 VDC) .
Assuming I could get the VCO to work at the higher frequencies, by removing 5 VDC from pin P8, I
could raise the frequency of the radio from 26.965-27.405 MHz to 29.525-29.965 MHz.
Now, that would be a great start, just a little high in frequency!
I located P8 on underside of the circuit board and quickly cut the traces on both sides. A short jumper was soldered around pin P8.
Now, would it work, or do I now have another radio for the parts box? First step is to adjust the VCO
voltage. TP8, located near L1, and ground would have to be between 1.5-3.6 volts, on all channels.
Note: Do not use chassis ground, use the -13.6 VDC connection. And be sure to use the proper adjustment
tool, as the ferrite slugs break easily. Slowly adjust the slug in L1, while monitoring the voltage on TP8
for about 2 volts. Turning the slug clockwise, brought the voltage down to 2 volts when on channel 1!"
I found I needed to remove L1 and change the little 24pF cap inside it (it's like an IF transformer) to a 20pF cap.
Then I could adjust this coil so the ferrite slug wasn't sticking out so much. I mounted this new cap under the board, as it
would not physically fit inside the L1 can. But it still forms the needed LC circuit.
"In studying Table A in the service manual, channels 10-38 could be lowered in frequency by 640 KHz by hardwiring +5 VDC to pin P6.
Channels 1-9 already have +5 VDC on this pin. This would have channel 10 on 28.995 MHz and channel
40 on 29.325 MHz, covering the 10 meter AM band perfectly! So soldering a wire on the circuit
board from the area where pin P8 had gotten +5 VDC to the connection P6, should put the radio into
the 10 meter AM band. This little jumper was soon in place. In theory, channel 1-9 should go from 29.525-29.625 MHz, part of the 10 meter FM
band, 5 KHz off frequency, and channels 10-40 should go from 28.995-29.325 MHz, covering the 10 meter AM band. These channels and frequencies are
shown in Table 1. I checked the VCO voltage and adjusted L1 so TP8 varied from 1.5-4 VDC on all channels.
Note: if the VCO will not lock, get it working before going any further. Now for the
receiver and transmitter adjustments.
These are the steps I followed: Set the handy service monitor or signal
generator to 29.125 MHz. Set the channel selector to channel 20, and open the
squelch. Increase the signal, until it can be heard in the radio's speaker. Standard
adjustment procedures were used: decrease signal strength as the signal becomes too strong,
as adjustments are made. T1, T2, T6, and T5 were adjusted, in that order.
How did it work? Less than 1 uv sensitivity! In fact, 0.5 uv or less.
Now for the transmitter. Hook the radio up to a power meter and a dummy load. You will not have any
output power, yet. Set the radio to channel 20 and tune a receiver across the room to 29.125 MHz and
adjust the volume so you can hear it at the radio. A large S-meter on the receiver is helpful, also.
Adjustments L5, T3 and T4 are critical here, as they form a filter, to only let a narrow band of frequencies
through to the final amplifier. As you key the transmitter, you should hear the transmitted signal.
If not, either move the receiver closer, or use a better antenna. Slowly adjust L2, listening for a change
in tone on the receiver across the room and watching for the S-meter increase. Unkey the transmitter between
adjustments. If you do not notice an improvement, move the slug back to where it originally started. You may
need to use only a short wire for an antenna on the receiver, as you will overload it, if not careful. Again
key the transmitter and adjust L5 slowly, listening carefully. About 1/4 turn clockwise should be close. Then
move to T3. Slowly adjust the slug clockwise, listening for a stronger signal in the receiver, and watching the
power meter for any movement. Again about 1/4 -1/2 turn should be all that is required. No power output may be
obvious on the power meter yet, but you might already see some power output. T4 is next. Only about a 1/4 turn
is all that should be required.
As you adjust T4 clockwise, at some point, the power meter should show measurable power. If not, go back
to L5, T3, and T4 and tune slightly until you have measurable output and adjust for maximum signal output,
as measured on the power meter. These filter adjustments are quite narrow and may require readjustment.
Then repeak L2, L5, T3, and T4 for maximum power output.
Now for the PA adjustments. Adjust L7, L11 and L12 for maximum output, in that order.
They should tune counterclockwise. Power output should now be about 4 watts. Use your
frequency counter and verify that the radio is transmitting on 29.125 MHz and not elsewhere.
Check for output and frequency on channels 10-38. If power drops off on some channels, you may
want to adjust L2, L5, T3 and T4, until power output is uniform from 28.995-29.325 MHz, channels
10-40. Low output on channels 1-9 is not a real problem, since you do not want to transmit AM on the
FM portion of the band, anyway.
Let's start looking for a suitable radio and break out the solder iron. A quick search of the internet showed several radios use this board.
A Google search revealed page 91 of? "Screwdriver Experts Guide to Peaking Out and Repairing CB Radios" by Lou Franklin lists several late 2-crystal AM
CB radios using the PLL02A PLL chip. If you find a radio with a PLL02A chip, and only two crystals,
10.240 MHz and 10.695 MHz, it is probably a candidate for this conversion. The circuit board used by all these radios is essentially the
same, and suitable for conversion to 10 meters. Some radios have more options than others, but they use
the same basic circuit board, with the major components in the same locations. On some models,
L5 was not there, but the other adjustments were the same. I have converted models
Midland 77-857, Kraco KCB4020, J.C. Penney 981-6204 and several others, using the techniques I have described."
Above quoted in case the link becomes broken.
Okay, now you got it working fully, you can then pregress to my mod:
I want to get most of the 40 channels in the AM subband of 10 meters (29 to 29.2MHz), and doing mostly 5KHz steps
will do this. Oh, some frequencies will be skipped, just like before, and as when the radio was a CB set.
But all but one skip is 10KHz instead of 20KHz.
And renumbering by repositioning the knob:
Note that the top and bottom of the
subband happens, in the left chart, between channels 9 and 10.
Undo the channel selector knob's setscrew, reposition
it to put channel 1 where channel 10 was, tighten the setscrew. Then the chart on the right becomes the valid one.
The XOR gate below will make this happen, instead of putting
channels 1 thru 9 (left chart) outside the AM subband if you didn't use the XOR gate.
First, locate C61 (electronically between Q2 and the PLL02 chip) and remove. We will wire up a flip-flop, a 74LS74,
here. Q2's emitter will feed the flip flop's clock pin, and its Q output will feed the PLL02 chip, pin 3. Wire the
flip flop's D input to the flop flop's
inverted Q output. 5V and ground as well. This will divide the 10.24MHz reference frequency in half. I used a SMD 74LS74 chip, soldering thin stiff leads
to it and mounting it in the air above where the cap was. In theory, the PLL02's FS pin should
also do this, but it didn't work for me; it gave me a divide by 1139 instead of 2048, yielding steps of 8.89KHz, not
useful. This flip flop combined with the PLL02's divide by 1024 will give me the 5KHz steps I want.
Once you do this, you'll find a new frequency on channel 13 (left chart) shifted up to 29.65MHz vs 29.035MHz we had before.
That's to be expected, as the next step is to redo the channel selector switch wiring.
Cut P7 (pin 8 of the PLL02) free from ground, and tie it high to the 5V supply. This should make channel 13 (left chart) be on
29.01MHz. Channels 11 thru 40 (left chart) should be as on the above left frequency table. Channel 29 (left chart) should now be 29.1MHz, and you may want to tweak
L1 to get the VCO voltage, TP8, to midrange, about 2.5V or so.
Now to take care of channels 1 thru 9 (left chart). Disconnect the PLL02's P6 (pin 9) from the channel selector switch.
I connected this disconnected selector switch's P6 to one input of the XOR gate (a 74LS86). You will need a 1K
resistor to ground to make sure a low is actually low (the PLL02 chip has pull down resistors inside it).
Cut the P4 trace between the channel selector switch and the PLL04 chip (pin 11). The selector switch side feeds the
other input of the same XOR gate (and use another 1K pulldown resistor), and the XOR gate's output now feeds the P4 of the PLL02 (pin 11). Connect the
gate's ground to ground, and Vcc to the 5V supply (which may start to sag a little, so find R5 and parallel a 200 ohm
resistor to it, this should bring the 5V supply back up). A better mod would be to use a pass transistor
and a 6V zener diode to create a simple regulated 5V supply. See below diagram. Remove R5 and the old zener. Channels 1 thru 9 (left chart) should now be around 29.2MHz, as per the above left table.
As 29.000MHz is the most popular frequency, I used some "glue" logic to make the new channel 2
create 29.000 instead of 28.995MHz. I used a triple input triple AND gate 74LS11 and the previously unused XOR gates
to make the logic. From the selector switch, if P0 = low, and P1 thru P5 are all 1's, then I use a pair of XOR gates to
invert P0 and P1 before feeding it to the PLL02 chip. If not, P0 and P1 are not inverted. Gettin' ugly... As the phase comparitor inside the PLL02 is now running at 5KHz instead of 10KHz, I doubled the caps in the loop
filter (C1, C2 and C3). Didn't seem to make a difference, but did it for completeness.
Build an RFI sniffer:
To track down exactly where RFI is coming out of a computer or
other equipment, use this RFI sniffer. It's a ferrite toroid ring
with a gap cut into it. Several turns of wire connect to coax that
then feeds into a receiver or spectrum analyzer (I'm sure you have
one handy!). The gap is the sensitive area. It will pick up RFI
magnetic fields. Thus you can identify wires or other leakage areas
with RFI on them.
It takes a long time to file the gap into the toroid; your needle
files will get dull. Don't push too hard, as ferrite is fragile.
You want a narrow gap on the outside part of the ring, so file from
the inside. A gap about the thickness of a fingernail is good.
Depending on the ferrite material, this sniffer should be good for
HF and VHF.
For low frequency work, use a tape head from an audio cassette machine.
The center of the front part of the head (where the tape used to pass over)
will be the sensitive spot.
HF antenna tuner:
A 50Ω low pass filter, cutoff at 50MHz
No longer using this filter on the HF rig, as the Icom has 6 meters...
My father WB2JIA (SK) used this for code practice.
Clock radio modified for GMT 24hr "zulu" time.
A dual band HT: .
No, heard that it's likely just regular tank radios on VHF FM. In the
30-70 MHz region. That these antennas are broad-band half-wave with tuners.
Received U2MIR on packet, as you can barely make out in this blurry photo of the dumb terminal I used back in the day:
My ten watts into a groundplane vertical couldn't connect to it.
Truckers should never use 28.085MHz as that is in the CW/RTTY 10 meter ham radio band. Hams never use voice there, you'd stick out like a sore thumb.