Isolation of Ground Differentials and USB Interfaces

[Rabbitz]

Hi Gurus,

I am working on a bit of a left field project. I have the need to interface the audio to and from a piece of software on a laptop to a radio transceiver.

Now it's a given that the computer uses a switchmode power supply and the radio a linear power supply. this means there is a rather good chance of quite varied ground potentials between the two devices.

To protect both I ideally need to isolate the two and the simplest way (I thought) is to put a couple of 1:1 transformers in the audio circuits. A bit like a passive DI without the impedance change. So this was done. Strangely this simple thing has been very difficult to get working. Faulty components and very low signals have plagued the project. To add a further complication I used to borrow test kit from work to investigate this sort of thing (CRO, signal generator, spectrum analyser, etc) but where I work these days doesn't use that sort of gear.

Anyhoo, after a few frustrating days of dicking around with a very simple circuit and getting stymied in every direction I was about to give up when I spied my little Behringer 3 channel mixer. I cobbled some leads up and bingo audio now flows in both directions and radio happily sends its messages out into the ether, well almost happily. The little mixer id quite susceptible to RF and chucks a wobbly every so often.

This situation has me wondering if a mixer would generally have galvanic isolation as a part of its design. I do wonder because of the existence of DI's.

Further, what types of USB interfaces are there for use with DAW's and the like?

Using the simple transformer isolation in front of the audio I/O of a simpler USB interface may overcome the low signal level issues I am facing.

The RF issues are something else I need to sort but getting a reliable interface is the first step.

In short, what USB interfaces are there for use with DAW's?

And what isolation would I expect to find in a little USB capable mixer, if any?

Cheers

[dave.king1]

I have both a Peavey PV6 and a Behringer 6 ch mixer, both USB connected, I reckon the Peavey was cleaner than the Behringer and preferred it.

Was because I very cleverly let the smoke out of the Peavey and must get around to fixing it one day

[Rabbitz]

The current PV6 has a bluetooth audio input - now there's some fancy "air gap" isolation

[blinddrew]

Well, interfaces are available from as little as £25 in the UK depending on what inputs you need. Is it just a single stereo input at line level? But I don't think you're going to get much in the way of isolation at the budget end of the market. However something like this might help at the input stage:
https://www.juno.co.uk/products/art-...RoC_dIQAvD_BwE
It'll take an unbalanced input and give you an isolated balanced or unbalanced output.

[Marcel]

Many passive DI boxs are just small 1:1 transformers with appropriate connectors...

But your RFI problem puts lots of questions in my head, however I'll limit to just a few.. Are we talking a real transmitter (say for FM narrowcasting) or a digital thing like Bluetooth or Wifi ? Are you using the correct antenna for the transmitter? For earthing purposes how are you powering things, all off the one power board or via various GPO's around the building? Have you rack mounted all that can be rack mounted?

Usually the suffering of RFI can be due mostly attributed to a bad earth plan or an antenna that has poor VSWR. Most new gear these days has quite good RFI protection and will only present symptoms when other gear is having problems. As an example, at one time I had a 20W VHF transmitter in a rack on test in my workshop and a Behringer UMC 204HD USB audio interface operating fine without any RFI into my Reaper loaded PC all located less than 2m away, though I did have to pay special attention to grounding and avoidance of ground loops. So it can be done, though even simple things like that 'old faithful' or that one 'cheap' power supply can bring a whole lot of unique grief to your party..

[Simon Barden]

Many passive DI boxs are just small 1:1 transformers with appropriate connectors...

To be technically accurate, if using 1:1 transformers they are line isolators and not DI boxes. A true passive DI box is designed to turn a line level (or greater) signal into a balanced mic level signal, so needs to reduce the signal level quite significantly. The ART line isolator box Drew linked to is a very good little unit, and the transformers that ART use in its products punch well above their weight.

Your standard low-cost mixer won't have any galvanic isolation. You need to spend quite a lot of money before you get input and output transformers built-in to a mixer, as good quality audio transformers aren't cheap and solid-state electronics do a very good job of providing a clean, quiet signal path.

The main cause of noise issues between different items of powered audio equipment is ground loop noise, especially where at least one item is digital, so has high frequency circuits that can radiate noise. In some cases, lifting the shield on connecting balanced cables can cure it.

Balanced to unbalance cabling noise can normally be cured with a pseudo-balanced cable,

Tip to hot wire (red)
Sleeve to cold wire (black)
Cable screen: easiest solution is to leave it disconnected and insulated. Best solution is to link it to the sleeve through a 100 ohm resistor, ideally with a 100nF capacitor in parallel.

At the Balanced input end, wire as normal:

tip to hot wire (red)
ring to cold wire (black)
sleeve to screen

But if it's an unbalanced connection throughout, then a line isolator box like the ART one is about the only way to do it.

[Simon Barden]

There are all sorts of USB interfaces, from $50 to $5000 or more, all depending on features and quality of components. However, even cheap USB interfaces are extremely good these days in terms of A/D and D/A capability, and it's really the surrounding components that determine the price and desirability. Almost all cheap interfaces will provide very good quality uncoloured sound. All the highly-rated esoteric pre-amps are more about what the signal sounds like when pushed hard as an effect, and less about the 95% of the time they are used in a clean, uncoloured mode.

Obviously USB interfaces vary with the amount of gain they can add to a signal before it hits the A/D converter and the level of noise added during the amplification stage. So looking at the adjustable gain range of a USB interface is important if you want to get a low level signal up to a reasonable one. Some cheap interfaces, the latest Behringer ones for example, have a limited amount of gain, around 40dB, so are designed for use with capacitor mics with strong outputs, or dynamic mics on loud guitar amps. The quoted 'gain range' is larger because they include the signal levels when the 16dB pad is used. So for quiet signals - dynamics on voice or ribbon mics, you want to look for an interface with 60db of actual gain (and some interfaces have 70db).

A lot of these gain stages can be a bit noisy in the last 10dB or so, so it's often better to record at 24-bit depth (you should do this as a matter of course anyway) and add the last bit of gain digitally, which won't add extra noise (just amplify any existing noise).

If the radio has unbalanced inputs and outputs, then using a short length of unbalanced cable into a line isolator, with the majority of the run in balanced cable into a balanced differential input, will minimise noise pickup by the cable and will stop any ground loop issues.

Plugging all the equipment into the same multi-way plug strip will also help to minimise any ground loop paths.

If you want to have monitors connected as well as having outputs to and from the radio, then the minimum you'd need is a 2-in, 4-out interface. Something like the Focusrite Scarlet 2i4; http://uk.focusrite.com/usb-audio-in...s/scarlett-2i4

If you want to have 'better' pre-amps with the ability to colour the sound more, then the Focusrite Claret 2PreUSB is a decent product. https://uk.focusrite.com/usb-audio-i...arett-2pre-usb

Of course, many other interfaces are available, and the NI Komplete 6 is another very good interface with lots of recommendations. https://www.native-instruments.com/e...plete-audio-6/

[Rabbitz]

<<TAKE 3>>

Thanks for the well reasoned replies. The ART device looks interesting, about $105 here in Oz.

We really are lucky to have such knowledgeable folk who are willing to share and help. (Without the "GOD MODE" attitude found on most specialist forums).

Marcel, as for the RFI, I can't really blame the little Behringer for having the occasional conniption. The environment isn't *exactly* what the designers had in mind.

The station is my amateur radio (ham) station. Hi everyone I am VK2CSW and I am a sad-sack radio geek .

The interface is for digital modes such as FT-8, JT65, WSPR, etc. The radio is a FT-857D (@ ~100 watts) into a 20m wire dipole above the roof.

Most of the RF getting into the mixer was via the headphones that were being used to assess the 'cleanness' of the digital tones - ironically this was being done to avoid causing RFI. A toriod ferrite on the headphone lead sorted a lot of it. The set up is not really conducive to avoiding RF in the (presumably) un-screened mixer (see attached).

The mixer is right next to the transmitter and directly below the antenna. Once I cleaned up the input signal (was being overdriven) there is no obvious RFI in other devices around the house (TV's, AM radios and the like).

As I have said I have been playing with a simple 1:1 transformer interface which I have been struggling with due to the lack of test tools. Well lack of test tools and issues with the super deluxe quality of $5 USB sound cards - thus the idea of bypassing the sound card by using a USB Audio device. It also has the advantage of not accidentally transmitting any windows noises, on-board mic or the thrash metal MP3 I happen to be listening to. I have ordered a $30 TFT oscilloscope kit - cheap and nasty but as I am losing sleep trying to work out why a simple circuit isn't working, I think is a good investment.

I have ordered some USB extension cables and will knock up some interface cables to move the mixer further away - originally, in good ham radio fashion, I made the cables as short as practical (to avoid RF issues, yes irony again).

There isn't any appreciable ground or 50Hz hum being picked up, so the ground isolation isn't quite as important right now. If it does raise its head I might experiment by bodging up some cables and chucking a DI into the circuit.

[dave.king1]

Couldn't agree more about the helpful nature of this forum, long may it remain so.

Looking at the pic, that's the same Behringer mixer I'm using in place of the smokeless Peavey

[Marcel]

Tut tut Rabbitz... naughty of you bringing another hobby of mine into this guitar building forum....de VK4TMH .... lol

The 20W transmitter I mentioned in my earlier post was a 2m AM beacon on 144.584MHz in the same workspace that I do all my guitar and amp builds, so I'm reasonably familiar with the kinds of RFI you're suffering and then some. The best two things I did to minimise my RFI was get a decent ATU such as a MFJ-945E, and install a decent star wired station earth with an equipotential bond to the mains electrical earth. So many problems simply went away on 40m and 20m when I did it, such as RF Rx noise floor fell by tens of dB and WSPR works surprisingly well and RFI became almost non existent.

http://kc.flexradio.com/KnowledgebaseArticle50426.aspx is a great article which gives good pointers on building good RF earths. It is from a USA perspective so all I will add is in AU you need to include the MEN electrical earth stake in your design but do not disturb any of the existing electrical wiring. Basically run a decent sized wire of any length from your wonderful and possibly elaborate station earth peg/stake system to a separate 2nd clamp on your mains earth stake/peg (it works and complies with AS3000 as a 'equipotential bond').

Admittedly I went a bit 'industrial' on my station earth system in my non-rental home. Three 6' stakes in the garden spread over 8m tied together and back to my common point in the shack using 3x25mm copper strap, but testing using 4mm wire worked and proved the layout and design, and the 4mm run to the house earth is still in service after near 14 years of regular use.

That's enough of that radio jargon... I'll shut up now.... ;-)