More on Ground Loops and Audio Settings

SDRs like the Softrock use the audio card to convert the received signals from analogue to digital. Because the SDR utilises the maximum sensitivity of the audio card it is susceptible to minute currents  that may flow due to ground loops, bad screening or just bad grounding.

An SDR should be constructed with ground loops in mind. Tony KB9YIG designs his Softrock range in such a way as to make that easy. [BUT see VHF Ensemble, at bottom] However the constructor MUST bear this in mind when connecting it to the outside world. Some models have a position for a link to connect the antenna ground to Softrock ground. DO NOT FIT THIS. Note that a small, maybe 100pF, capacitor between the two grounds may reduce interference caused by common-mode RF currents in the antenna coax.
Connections to antenna, power supply, maybe other devices and computer will all be grounded. A loop is formed if by any means two ground connections are made between the SDR and computer.

When mounting the SDR in a metal box the best way would be to mount the antenna socket on the box. This would ground the box to the antenna ground, yes, the antenna co-ax SHOULD be connected to a good ground! The SDR's ground connections should all be insulated from this box. If the SDR box is instead  connected to SDR ground then should this box touch another grounded object a loop is formed. Look here for my recommendation about the KM5H box.

I have connected many Softrocks in various ways. Sometimes I have connected carelessly and seen enormous ground loop effects on the SDR program display. In all cases I have found the best results are obtained by making the screen of the audio connections the only connection between the SDR and the computer. This means ensuring Tony's antenna isolation is not bypassed. The power supply negative is usually grounded so only the positive is connected to the SDR. This is not an ideal situation, the power return is via the audio cable, there is a possibility of damage if the cable is unplugged with power applied. In most cases I have found that connecting a ground from the computer  to the SDR following the same route as the audio cable does not show any increase in noise on the display. It has been suggested that multiple ground connections directly from the SDR to computer may be made if they are all the same length, bundled together following the same route.
One way of reducing the effect of ground loops is to connect everything to a good ground with thick wire and well terminated connections. This will help to equalise the ground potentials. Leif's logigal method for eliminating loops.

This PDF gives a full description and method of eliminating ground loops. For soundcards with a balanced input look at page 29.

Some use transformers as a way of avoiding ground loops. These may actually pickup noise if placed near a source of radiated energy. I have seen pickup from transformers, monitors and low-energy lamps. In addition transformers may introduce phase shift or other distortions. I have known a situation where a DRM signal was decoded with a direct connection but could not be decoded when the SDR was connected through transformers. In my opinion good results may always be obtained without transformers.

 Below I show some HDSDR Winrad screenshots. Lower down, a PSDR shot.

First, a shot from a low-end card. Very obvious is a "spike" at the centre. This is not unusual with cheaper cards. It does not really affect performance and may be considered normal for cheaper cards. It is considered to be a deficiency of the card's Analogue-to-Digital-Converter (ADC). It actually represents zero Hertz at the audio card, nothing can be received here, this area should be avoided on all soundcards.

This also shows a small "mound" around the spike. This is due to a ground loop, often it can be far more obvious. This is a very cheap USB device. But it produces adequate results for initial testing. 



Below is how a fairly bad loop may look, but it could be much worse!
The ground loop is basically formed by power line noise getting into the audio. It is always at the exact centre of the unzoomed display. Because of the way SDR works the exact centre of the display represents the local oscillator frequency. This is at zero hertz audio frequency. The powerline harmonics spread outwards from this point. 
This is at 48KHz sample rate, no zoom. The Creative XFi only has a small spike that is not often obvious. See the shot below.


This shows four different configurations, Winrad was stopped between each change. 
This is at 96KHz sample rate. No zoom.
The two cards are both in the same computer. I will have to checkout the Delta 44. shows some noise in the middle and a few spurii with no audio connected. I will have to investigate this. The other is a Creative XFi Xtreme PCI-E card. It is not too expensive and seems to be of reasonable quality.



This is a close-up of the centre. Only 3KHz shown.

Do not trust audio cables and connections. I have seen audio cables from a retailer who many would say is respectable. The cables look very nice but when slit with a knife I saw the ground was just run as a parallel wire. It was not screened! 
The picture below shows the effect of what may have been a few ohms resistance in the contact area of the plug. Is this one reason why professionals do not use the small stereo plugs?
48KHz no zoom.


PSDR

It is not recommended that beginners try any version of PSDR until they are certain their SDR is working. PSDR is sometimes tricky to setup so you may have an extra unknown.
Below, a screenshot of PSDR-IQ. This uses a continuous tuning system, the ground loop effect, and centre of the tuning range is always separated from the actual tuned signal by the "IF Frequency". The default is 10KHZ, but may be altered in "Setup" 

Here PSDR-IQ is tuned to 7148KHz LSB. The centre, local oscillator frequency, is where the mouse pointer shows. The ground loop products are seen on each side of this point. It should be possible to reduce this significantly.



VHF Ensemble.
Tony has not included antenna ground isolation in this model.
Depending on your configuration a ground loop may be formed.
Look here http://sites.google.com/site/g4zfqradio/vhf_ensemble Isolating the antenna socket one way of removing this loop.




Some may consider that there is no need to take too much effort to reduce the effect of ground loops. If you are prepared to ignore it and keep operation away from the centre area then provided it is not excessive I would say there will be no problem.

Audio settings to follow but this will just show Winrad instead of Rocky on the Soundcard Faults page below.


August 2011

Comments