The NooElec Upconverter
|
|
I've been designing and
building my own test equipment and various useful items over
the last umpteen years but, having a number of e-cheques that
I needed to cash before they expired, I decided to buy some ready-made
things including the upconverter I'm describing below. Also,
partly so I could compare my results with professionally made
stuff. My own upconverter is described
here. Why did I choose to make an upconverter? Well... a
few years back I purchased what I thought was the ultimate SDR
capable of transmit as well as receive, but discovered to my
dismay that it had been described in a way which I considered
to be entirely misleading and totally dishonest. In order to
use the thing for reception and transmission in the short-wave
amateur bands you'd need to add an upconverter for reception
and a downconverter to transmit. My homebrew upconverter worked
admirably and I was interested to see how the Nooelec compared. |
|
 |
|
Above is a view
of the Nooelec upconverter (less its case). Click
the picture above to see the circuit diagram. My example
appears to be Version 1.30b9 dated 20th July 2020.
What attracted me to this particular
upconverter was that it included a noise source, another
one of my home brew experiments. In fact the general design
of my own noise source turns out to have been based on this particular
Nooelec design.
At the top right of the circuit
board is the "optional" noise source, which in this
example, I'm informed, requires only an SMA socket and a pair
of jumper pins, compared with the earlier version which is missing
various chips as well as the connector.
However, all was not as expected
and, having come across a video describing the addition of extra
parts to activate the noise source, plus some puzzling test results
I examined this circuit more closely and spotted something amiss.
Read on. |
|
 |
 |
|
If you compare the circuitry on
the left with the diagram above you'll see it's much the same,
with a pair of amplifier chips (IC9 & IC10) marked "LSt",
the code for the Malaysian manufactured BGA2817.
The noise diode D8, marked "WC"
is fitted below the SMA connector holes and I suspect this is
a 7.5 volt SOD323 zener type BZT52-C7V5. If the diode is a 7.5
volt device, how can it produce avalanche noise when the power
supply for the board is derived from the USB port? The answer
is to be found if one studies the circuit diagram (click
on the picture to the left) where you can see there's a 12
volt power supply (below). This uses a DC-DC converter chip,
IC3 (SC4503 or now an NCP1403), which is missing. Top left are
holes for inserting a pair of pins used to link the 5 volt rail
to the noise source. When linked the LED adjacent to the holes
illuminates and power connects to the noise source, but without
IC3 a little under 5 volts passes through to the noise diode
D8. In fact I measured around 3.6 volts across D8. |
 |
|
|
|
Below is a view of the other
side of the circuit board. I've fitted a pair of brass standoffs
used to mount the board into its case. |
 |
|
Here's the upconverter
mounted in its case. The board and case were priced at £41
and £17 respectively.
As I was essentially cashing
a couple of e-cheques my choice of supplier was limited, but
I bought the complete upconverter for only £8 plus two
e-cheques.
How did the upconverter perform?
Below is a picture of the 80
metre band using
the Lime SDR and a long wire aerial and operated close to
my PC.
The Lime and the upconverter
are powered via USB leads. Click
the picture below to see more detail of the results. The
Lime gain setting was 55dB because any more than this resulted
in overloading and loads of spurii. |
 |
|
 |
|
I then decided to see
how the noise source performed so I found a temporary connector
which I soldered to the output pads and added a pair of pins
at the enable jumper location then connected the noise source
to the upconverter input.......
Having previously built a similar
noise source I was disappointed because the amount of noise was
very very low, giving an increase to the noise baseline of a
mere +4dBm at LF and -4dBm at HF ! and surprisingly un-noisy,
so I connected the output to my HP power meter and surprisingly
this recorded +10dBm. Compared with my home brew noise source,
which gives me an increase in noise above the baseline of around
40dB and 0dBm read by the power meter, these figures were baffling
(I should mention that at this point I had been unaware of the
missing chip at IC3 which provides the 12 volt supply to the
noise diode ). Touching the connection at the noise source output
produced a large increase in the noise baseline which I initially
thought may point to instability but later I realised it was
akin to the effect of touching the top cap of an RF amplifier
valve. I measured the noise diode voltage and found it was only
3.6 volts and wondered if D8 had been swapped to a lower voltage
type, but re-checking its code letters "WC", I reckoned
that unless a new manufacturer had entered the field and used
identical code letters the diode was in fact a 7.5 volt type.
I found the schematic for the upconverter, looked at the pcb
and noticed IC3 wasn't fitted.
Below: Left 198KHz (-122dBm)
and 10MHz (-115dBm) with noise source connected to the upconverter
but switched OFF.
Below: Right 198KHz (-118dBm)
and 10MHz (-119dBm) with noise source ON and a clean output.
Click pictures to see them full
size (note that the centre spikes are software artefacts). The
Lime gain was set to 55dB. |
|
|
|
I contacted Nooelec
and was informed that IC3 should have been fitted and my best
option should be to purchase and fit an NCP1403SNT1G.
I have three preferred suppliers
two of whom (RS & Farnell) compete on prices and the third
(CPC) often much cheaper but with poorer or variable delivery.
Over the years I've noticed
the search facilities on RS and Farnell give unreliable results
and sometimes one will show something in stock but the other
doesn't show the item at all. The trick is to note the manufacturer's
part code given by one supplier and type this at the other's
website. Usually you'll find it and often much cheaper than those
resulting from a search. My listing for parts needed to activate
the noise source was as follows:
Connector: RS 185-9271 @ £2.71;
DC-DC converter: RS 464-182, Five @ £0.67 each = £3.36;
On/off switch RS 734-7062 @ £1.68. Total = £7.75
How much from Farnell? Their
total came to £9.17 but included an alternative DC-DC converter
because the other wasn't in stock until June (5 months away).
The next morning the parts arrived
(in two parcels because they'd come from different RS warehouses).
The first step was to fit the
new SMA socket and I discovered its centre pin was too large
to fit in the circuit board hole so this needed drilling out
to 1.3mm. As the only connection to the centre pin is under the
board it doesn't matter if the plated-through hole is compromised.
Once fitted I marked the front panel for a new 8.5mm hole to
fit the threaded part of the connector. Next was to fit on on/off
switch for the noise source. The Nooelec design dictates that
the noise source is turned on at a jumper but, as this is hidden
once the outer case is fitted, a panel switch is necessary. The
existing Upconvert/Through switch is soldered directly on the
circuit board but the new switch has to be fitted to the panel.
The best position is above the existing switch, but taking care
that the outer case fits properly as clearance is tight. The
fixing hole was 6mm and once fitted the switch was wired to the
pair of jumpers previously fitted at the Enable position. |
|
|
|
Once the mechanical work
had been completed I fitted the tiny 12 volt power supply chip
in location IC3. This is a very small chip and has the part number
NCP1403SNT1G. It's function
is to boost the voltage from the USB connector (around 4.0 volts
after protection diode drop etc) to 12 volts. When I first looked
at the spec (click the number above) I was surprised to see a
high current drain, but then noticed it was uA not mA... As the
chip makes use of an oscillator it's not a good idea to leave
it running when using only the upconverter.
I was advised that the chip
should have been fitted during board assembly but for some unexplained
reason it was missing. As the minimum order was five it cost
me £3.36 and negated most of the savings I'd made by buying
the case and board separately.
Once fitted I assembled the
Lime with the upconverter and repeated the tests, the results
of which can be seen below. I noticed noise spikes on the displays
so reduced the gain of the Lime SDR from the setting of 55dB
used in the previous tests to 30dB and this helped.
Two frequencies were tested:
198KHz and 10MHz. I chose 198KHz because on this frequency is
a strong long wave broadcast station and its complete absence
indicates to me that no external broadcast signals are leaking
into the measurements. |
 |
|
|
|
|
Above: Left 198KHz (-142dBm)
and 10MHz (-141dBm) with noise source connected to the upconverter
but switched off
Above: Right 198KHz (-107dBm)
and 10MHz (-84dBm) with noise source on.
For this setting of the Lime
SDR gain the noise at 198KHz is 35dB and at 10MHz is 57dB (not
very linear using the Lime SDR but I'll need to check on my Rigol
SA to see exactly how it's performing).
Click the pictures above to
see them full size (note that the centre spikes are software
artefacts). The Lime SDR gain was (reduced to) 30dB.
Having tested the upconverter
it was time to fit it into the case which had arrived with end
panels (one of which I drilled for the noise output connector
and the noise on/off switch) and ten screws plus a couple of
6BA brass posts. Two of the screws (the two shorter ones) were
6BA but the others were something like M3. Unfortunately these
wouldn't fit the case. I examined the holes in the case and found
they were tapped so tried a selection and discovered the best
fit was M2. I had a selection from an old laptop computer so
used 8 of these. |
|
Now some tests using the
SDRPlay SDR connected and powered by its USB lead with the Nooelec
noise source turned either OFF (left pictures) or ON (right pictures).
Gain set at MAX and IF gain at -30dB Auto. Click each to see
larger picture (10KHz, 50KHz, 198KHz and 10MHz). |
|
|
|
Frequency |
dBm baseline |
dBm noise |
Increase dB |
Notes |
|
10KHz |
-132 |
-58 |
74 |
2 spurii |
|
50KHz |
-130 |
-49 |
81 |
> 10 spurii |
|
198KHz |
-126 |
-50 |
76 |
lots of spurii |
|
198KHz (Lime) |
-142 |
-107 |
35 |
spurii |
|
10MHz |
-133 |
-67 |
66 |
Clean |
|
10MHz (Lime) |
-141 |
-84 |
57 |
Clean |
|
Here's a summary of results
with the SDRPlay (with the Lime SDR added). Slight spurious noise
is being picked up via the SDR USB cable, and a lot more appear
when the noise source is switched on, possibly from the upconverter
or the 12 volt power supply but I need to carry out tests using
my Rigol SA to determine which. Note that the Lime results use
the Nooelec upconverter and the SDRPlay do not. The differences
in noise levels seems to indicate the superior performance of
the SDRPlay at 198KHz at the gain settings used. |
|
|
I then checked the output
of the noise source on my power
meter and it recorded +16dBm. This was slightly worrying
because the probe is marked "30mW max" which represents
a tad under+15dBm and the meter shot up to full scale before
I inserted a 20dB attenuator. Care needs to be taken when using
a spectrum analyser as these often have a +20dBm max rating and
switching spikes might exceed this level.
The power meter probe is marked
as covering 10MHz to 10GHz and, as my home brew noise source
registered no more than 0dBm, the implication is that perhaps
the Nooelec has a much higher frequency range? Later I checked
see the next page for the results... |
|
|
|
|
|
|
