The Nooelec Ham-it-Up
Noise Source
|
|
|
|
Below are pictures from
tests made by connecting the noise output of the Nooelec upcoverter
to my Rigol DSA815TG via a 20dB attenuator which was needed in
order to limit the total power fed into it to less than 2dBm.
Power was supplied from a local PC via a USB cable and, apart
from one test noted below, the upconverter switch was in the
Passthrough postion. Apologies for the poor definition. |
 |
Scan 0-300MHz
Noise source OFF. |
 |
Scan 0-300MHz
Noise source ON. |
 |
Scan 0-30MHz
Noise source OFF. |
 |
Scan 0-30MHz
Noise source ON. |
 |
Scan 0-2MHz
Noise source OFF. |
 |
Scan 0-2MHz
Noise source ON. |
 |
Scan 0-500KHz
Noise source OFF. |
 |
Scan 0-500KHz
Noise source ON. |
 |
Scan 0-200KHz
Noise source OFF. |
 |
Scan 0-200KHz
Noise source ON. |
 |
Scan 0-200KHz
Noise source ON.
Upconverter ON. |
 |
Scan 0-50KHz
Noise source OFF. |
 |
Scan 0-50KHz
Noise source ON. |
|
|
Looking at the results
above it's clear that something needs to be done to improve the
perfromance if the noise source is to be used at low frequencies.
I noticed that with the noise switched off and the upconverter
switched on there was no descernable change in the traces, however
as you can see in the 200KHz scan, if the upconverter is turned
on when the noise source is turned on the position improves slightly.
Presumably, if the USB voltage drops slightly, the LF noise pattern
changes.
The culprit appears to be the
12 volt power supply circuit. Obviously the avalanche noise from
the zener diode is pretty small so that any power supply noise
that's about the same order as this will affect the output. Looking
at the published circuit. This shows C24 (4.7uF) for input decoupling
and C26 (also 4.7uF) for 12 volt output smoothing. Looking at
the picture of the corner of the printed circuit board carrying
the noise source I can see that the circut is different to that
shown in the published data because the DC-DC converter has been
changed from an SC4503 to an NCP1403 which has different pinning.
The NCP1403 spec calls for a low ESR output capacitor and suggests
this to be 33uF. The frequency of oscillations within the NCP1403
depend on a few factors. The value of the choke used to isolate
the input and output can typically be say 22uH or 47uH, but the
higher the value the lower the oscillation frequency. I checked
the frequency of the spikes and the lowest was 6.583KHz with
strong harmonics ranging up to 100KHz and more. The theoretical
impedance of a 33uF capacitor at 6KHz is about 0.8 ohm and drops
to about 0.05 ohm at 100KHz. The question must be.. what value
smoothing capacitor is fitted? I've asked the manufacturer and
awaiting their reply.....
I received Nooelec's reply in
the form of a couple of pictures of the Upconverter which I've
added below... |
 |
 |
|
Note that the component
designations and choice of values differs from earlier published
schematics. Specifically "C26", smoothing for the output
of the 12 volt power supply chip, is now "C9". The
schematic above shows this to be a 1206 size capacitor of 22uF,
which is at the low end of the NCP1403 maker's recommended range
of values. So why is there severe interference on the scan
0-200KHz for example? One reason is 22uF is too low for this
application which requires a very high degree of amplification
of a very low level of avalanche noise. Add low level spikes
to the noise and this will be faithfully amplified too. That
means the 12 volt supply must be very clean and therefore the
value of the smoothing capacitor needs to be adequate. The maker
suggests 22-47uF but in this application maybe even 47uF is too
low?
I removed the circuit board
from its case and measured C9 and found it was 10uF and 0.52
ohms. So its got a fairly low ESR but only half the capacity
of the designer's intended part. As there are plated-through
holes adjacent, and connected, to C9 it was easy to try additional
capacitance and observe the effect. 100uF eliminated the spikes
but introduced a low frequency noise hump. Anything less failed
to clear the lowest frequency spikes, 330uF cleared the 6KHz
spike but appeared to upset the operation of the chip by forcing
it to oscillate below a couple of KHz. Rather than add a high
value I fitted 100uF. The two scans below show before and after
adding the extra 100uF capacitor. Click either picture to
see a larger detailed view. |
 |
 |
|
The pictures show that
the noise is cleaned up at the expense of slightly increased
amplitude variation, particularly below 10KHz which is anyway
not commonly required in radio alignment anyway. Whether operation
of the NCP1403 is adversely changed by using 100uF I don't know.
The datasheet on the chip doesn't go into any detail on this
but perhaps the chip needs to work harder as evidenced by the
low frequency humps?
The noise diode is D8 on the
layout above. This was reported to be a BZX384-C7V5, but as that
type is coded "WD" and the diode does indeed carry
7.5 volts across it, a BZT52-C7V5 is more likely because that
is coded "WC". The diode I selected for my home brew
noise source was a SOD123 MMSZ5V1T1G
(because I had lots left over from a repair job). |
|
pending
|
