Audio isolation transformers

[richard]

I am trying to electrically isolate a radio from a PC to prevent RFI. A cheap miniature audio transformer placed between mixer and soundcard doesn’t appear to offer good isolation. Maybe the capacitive coupling between the two coils in the transformer is too great.

It appears that some radio amateurs use isolation transformers from modems. Does any one know whether these are likely to offer better RFI isolation and will they provide sufficient bandwidth for DRM?

Richard

[aru]

Hi Richard,

I have used a Bourns LM-NP-1001-B with very good results.

http://www.sintesinet.it/manuali/lmnp1001b.pdf

aru

[dk8cb]

Does any one know whether these are likely to offer better RFI isolation and will they provide sufficient bandwidth for DRM?

I have used such transformers, they work well, I haven't even spent too much effort in properly matching them, originally these transformers are usually intended to be matched to 600 Ohms (the impedance of a telephone line) on both sides. If have connected them to a low impedance output and loaded them directly with the sound card, this worked very well.

But don't expect much RF isolation, the main purpose of such a transformer is to prevent low frequency (mains hum) ground loops. With regard to this purpose, they work very well. To get additional RF isolation it is absolutely essential to wind the audio cable onto a low permeability toroid as shown in the attached picture.
I have repeated this over and over again on the forums but I think I'll have to repeat it even more often... :o
Also have a look here (http://www.arrl.org/tis/info/rficomp.html). Snap-on ferrites that can be found on monitor cables etc. do not provide sufficient RF isolation unless you manage to wind many turns through them, which you can't because the hole through such a ferrite is just not large enough.

Using a ferrite as shown in the attached picture once made the SNR on a 31m-band transmission jump from 16 dB without ferrite toroid to 22 dB with it.

Roland

[dk8cb]

Here is a plot of the spectrum at my soundcard's input with and without isolating transformer.

It may be much worse than it is in my case, I could certainly do without the transformer but I have also seen much worse effects in the past. The effects that a ground loop causes, may depend on the individual case, even going so far as to in which way the power connector is inserted into the mains plug.

Roland

[mitajohn]

Hi all,

I used the attached bifilar choke (2x6mH) at the IF output of DRT1 to the soundcard's line input, but I didn't notice any significant improvement in SNR. The combination of soundcard+DRT1, using a CD burn "Dream -t" + upconverter at 1MHz and DRT1 reaches up to 47.5 dB SNR. I think this is a very good figure without using any transformer or choke.

[richard]

Thank you for your comments. I already have some chokes fitted which appear to help a lot. Probably not worth adding one of these transformers as my radio is battery powered, so no problem with ground loops. Just wondered if this type transformer could be used to provide similar isolation.

Before using chokes I was thinking of buffering a balanced line to the soundcard using a balanced amplifier. The two systems (radio/PC) would then be isolated by virtue of the high input impedance the amplifier presented in the signal path. However I think both systems would require a common ground, so complete isolation would be lost. The amplifier could not just be connected to the two balanced lines.

Richard

[mitajohn]

Hi all,

I made some experiments using a transformer between the RX IF output and the Line input of PC’s soundcard that may be useful to members using an isolation transformer.
I used a Bourns 1:1 transformer type LM-NP-1001-B1 with the secondary loaded with a standard 1.5m pc audio interconnect (~250pF capacitance/channel) and a 10kOhm resistor which represents the line input impedance of the sound card.
The primary input signal was 1 V/1 kHz/square wave with 470-Ohm source impedance.
In pic_1 you may see the Input and Output graphs, the output graph has a ringing. In order to compensate for the ringing I added a zobel network at the secondary in parallel to the 10kOhm resistor. The zobel network consisted of a 100nF MKT capacitor in series with a 10kOhm multi-turn pot set at ~5.6kOhm for optimum results, as seen in pic_2.
I recommend adding a ~1uF MKT capacitor in series to block the DC at input of the soundcard, which appears in a lot of them, to prevent for possible magnetisation of the transformer.