Perseus Blog

some unsorted tricks and hints about the Perseus software defined receiver.

Antenna setup comparison

28 MHz:
noise level is the same (8dbµV). The ALA 1530S shows better results probably because of its height. (SNR 3 dB against 1.5 dB receiving the local beacon DM0AAB, 70 km from here)
remarkably low noise level on the of antenna on the field (9 dB vs 21 dB), also less signal (23 dB vs 31dB) on 1602 KHz (Radio Seagull, 1kW, 300km). But the improved SNR by 4 dB already makes a significant difference. On 603 KHz noise and signal levels of both antennas are almost equal to each other.

USB issues

It's a strange thing: the Microtelecom Perseus software (V4 and 5) runs well on both USB hubs of my PC (front and back side). But external software like Spectrum Lab or SDR Console only accepts the front ports. On the rear ports, no data will flow (Acquisition 0%, attenuators still click on change) or perseususb.dll won't be loaded. It's not a USB 2.0/3.0 issue. Both ports have the same controller type (Intel 8 series/C220). I spent a few hours analyzing this without a final result. But here are some conclusions that might for other purposes:

Driver update

It did not help to update the Perseus driver.

Spectrum Lab Perseus files

registry diagnostics

FM+ RDS decoder vs RDS Spy

The FM+ software already offers excellent RDS decoding for weak DX signals. But is it possible to get a little further? I've tested some recordings with both variants. The result: RDS Spy is a little ahead. Don't expect miracles. Nevertheless, both SDR solutions are a great leap forward compared to RDS decoding by my old-but-excellent Onkyo T4970 tuner. RDS decoding of weak signals starts at about 90 KHz bandwidth, best bandwiths are 125-140 KHz.

advantages Perseus FM+ software

advantages SDR# RDS Spy

Setting up RDS Spy with SDR sharp software (SDR#)

In SDR# you can play any WAV recorded by Perseus or FM+ software. It's recommended to use the File Player plugin. It adds a waterfall timeline to the recording and the ability to skip / rewind / forward and loop (no loops between selected parts unfortunately, only full length). The MPX output requires a virtual audio cable with 192 KHz capability (a real 192 KHz sound card is not neccessary). I did not manage to do it with VAC. But VB-Audio is an alternative. Live performance is not possible. I've tried routing the Perseus output to the virtual VB-Audio. But the audio spectrum is completely cut above 15 KHz. Not even the 19 KHz MPX is visible, let alone the RDS carrier at 57 KHz. Nico, if you read this, is there any trick to change this?

Waterfall compression


just found out that there must be a spectrum compression in the waterfall in the Perseus software. This is important to know when you interpret signal levels. I've listened to KVRI on the Masset recording (3-Jan-2009, 17 UTC) on 1600 KHz. Until 16:59:56 there is Asian music which suddenly stops. After that a station ID is spoken, followed by 2 seconds of silence. Finally, music starts again.

Perseus waterfall

Watching the waterfall, I first wondered why the carrier signal seems stronger during the station ID and the silence part. Simultaneously, the weaker 1602 KHz carrier becomes visible strongly. When the music starts again, it disappears completely. If the waterfall was linear, it should remain. But actually it's always there. It's just masked by the modulated signal. And the waterfall follows to the overall level. I guess Spectrum Lab wouldn't do that.

Exact frequency measurement

Theoretically it is possible to measure frquencies by 0.03 Hz. This is the highest resulution possible. 0.8 button How to get there: use 0.8kHz for the secondary spectrum, use Zoom. Now you see a highres spectrum in the first window. It only refreshes about every 30 seconds. The calculation overlaps. So if you've justed tuned to one frequency it will take two or three refreshings. The span is indicated on the left: 24Hz. So the vertical lines are 2 Hz apart in the highest resolution. The AMLIST used by FMSCAN now lists more than 1000 exact MW frequencies (use the "exact frequency" option in the experts menu).

1602 waterfall

This is a spectrum of 1602 KHz in North Germany (calibrated). The labels of the x-scale are misleading. The center is 1602.000 KHz, the vertical lines would be 1602.002, 1602.004 and so on. You can estimate the carrier frequencies by 0.1 Hz here. There are 9 carriers between 1601.990 and 1602.007 KHz. Most of the MW carriers are relatively stable over a few weeks here in Europe. From February 2009 to September 2009 I've observed that many carriers have drifted about 1 Hz. So does it make sense measuring 1/10 Hz? Well it does if you compare recordings from different parts of the world from a limited time span. Then you can identify weak MW signals even beyond 10.000 km of distance.


But when you work as exact as this, calibration becomes extremely important. The manual describes calibration with the secondary spectrum. But there's an easier way. 10 mhz spectrum

Beware of doppler effects. The smaller peaks on the right side could be reflected signals from airplanes. Sometimes they're as strong as the main signal.

Unfortunately it is impossible to calibrate the scale as exact as it could be possible theoretically. The minimum VFO step is 1 Hz while you could determine the peak by 0.03 Hz. The shift is proportional to the frequency: If 10 MHz has 0.5 Hz, then 20 MHz would have 1 Hz, 1 MHz would have 0.05 Hz. There is no other way than calculating.

Check the reference frequency (usually 10 MHz) every once in a while: when the Perseus is turned on, it will take one hour until the calibration is relatively stable. During the first 10 minutes the peak of my receiver drifts by about 30 Hz (from 9999.945 to 9999.975 with CalClr for example).

Recordings are affected by calibration but not exactly the way they should be. If the receiver is calibrated correctly, the center frequency of the recording will also be. But the border frequencies will be detuned.

Example: peak found on 9999.980 KHz, calibrated, then started recording with center frequency 1000 KHz.
This means the calibration factor is 20 Hz per 10 MHz, thus 1 Hz per 500 KHz.
If you tune to 1500 KHz in the playback (500 KHz off the center), the peak of an exact signal will be found on 1499.999 KHz (1 Hz less).

The good thing: when you deal with recordings with unknown calibration status, you will only need two refence offsets. The rest is proportional.

Example: The playback contains France Inter on 162 KHz with a peak found on 162.0017 and DLF on 549 KHz with peak found on 549.0030 (both really transmit on exact reference frequency). The offset in Hz for any frequency in KHz will be (frq-162)/(549-162)*(3-1.7)+1.7
I've compiled a list of the detuning on the recordings distributed by the former Five Below Blog. It wasn't always easy to find reference frequencies but I think the result is quite good.

PRS - the Perseus record scheduler

First of all you need to install com0com to make it work (see release notes). But I still had errors saying "one device is not working" ("ein angeschlossenes Gerät funktioniert nicht" in German). The cause: COM10 and COM11 were already used by the virtual Bluetooth Serial Port. In fact Bluetooth uses all the COM-Ports from 3 to 13 - I don't know why so many. In such a case you need to


Here are a few bugs of the current Perseus software. is it possible to navigate around them? Please help.

2000 KHz recordings

Recordings always have the same span as according to the sample rate. It's not only 1600 KHz but even 2000 KHz if you tune via the secondary window (drag and drop) or enter the frequency manually. But keep in mind: Beyond the span the recorded range is mirrored to phantom frequencies.
Handewitt Perseus site
main index
Questions? Corrections? Ideas? Comments?
This site is based at: