ProSoundWeb Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: 1 2 [All]   Go Down

Author Topic: Confusing Isolation Transformer Traces  (Read 4204 times)

Mike Karseboom

  • Full Member
  • ***
  • Offline Offline
  • Posts: 230
Confusing Isolation Transformer Traces
« on: April 03, 2014, 03:46:02 am »

I like to use Systune to test out any new devices I get to see how "linear" they are and to sort of explore their limitations.  I recently got an Allen Avionics AGL-600  "ground loop eliminator" which is basically an isolation transformer.  I wanted it for line level signals like those that might come out of an electric keyboard.  This device is supposed to present 600 Ohms as it's input impedance and wants to see 600 ohms as the input impedance on the next device in line.  This might be the source of the problem as modern audio gear typically has much higher input impedance, but please read on.


I wanted to test the magnitude and phase response of this device to confirm that it would not "color" the audio sent through it.  So I ran a dual channel test with this signal path:


This was the signal source:
    Ipad Pink Noise generator to mixer channel 1 line in > mixer main out left set to about +4dBu = 1.23Vrms >


From here I split the signal with a simple "y" mic splitter  to get a reference and measurement channel.


Reference channel:
    -30dB pad  > Audiobox USB mic input 2  > USB cable to Systune on laptop.


Measurment Channel:
    -30dB pad > Audiobox USB mic input 1 > USB cable to Systune on laptop.


The -30dB pad is actually a whirlwind Director passive DI box set to "amp".  This is necessary to get the +4dBu out of the mixer down to a level that works with the Auiobox USB pre-amps.  The Audiobox USB pre-amp gains for channel 1 and 2 are set  and matched to give good level using the input meters in Systune. 


So with that signal path and using channel 2 as the reference channel  everything is good and flat.  The dual fft magnitude response is 0 for the entire frequency spectrum and the phase response is flat as well.  OK that all makes sense!


Now for the Device Under Test, the AGL-600.  Iput this downstream of the -30dB pads on channel 1 and repeat the test.   So in this case it is getting more like a mic level signal through it (about -26dBu).  A nice flat trace is the result with magnitude just a hair under zero and phase still flat.  Great results - the device is "transparent" for all audio frequencies from 20-20kHz.


Now I move the DUT upstream of the -30dB pad on channel 1.  Now the DUT is seeing a line level (+4dBu) signal.  Here is where I start getting confused.  Instead of a nice flat magnitude and phase response, Systune is telling me the is a -25 or so low frequency roll off starting around 1kHz and ending  down below 50hZ or so.  There is also a 90 degree phase shift over that same range.  See trace attached.


If this is a real response, then the device would be a very poor choice as an isolation transformer for a line level signal.  Yet that is how it is marketed and it is supposedly a reputable brand.  I suspect I am screwing up the test setup or misinterpreting the results.  Help! What is going on here?



Logged
--Mike
"If you're not confused, you don't know what is going on"

Live Sound for the Mt. Shasta area
http://www.shastalivesound.com

Jason Lavoie

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 896
Re: Confusing Isolation Transformer Traces
« Reply #1 on: April 03, 2014, 10:34:21 am »

Looks like maybe a high impedance output from your mixer that gets translated to a lower impedance at the other end of the whirlwind DI?
It should be pretty easy to slowly lower the output level from the mixer and watch if/when it smooths out (if you're right about it being level dependent)
Otherwise it's interface dependent.
Logged

Mike Karseboom

  • Full Member
  • ***
  • Offline Offline
  • Posts: 230
Re: Confusing Isolation Transformer Traces
« Reply #2 on: April 03, 2014, 11:17:40 am »

Looks like maybe a high impedance output from your mixer that gets translated to a lower impedance at the other end of the whirlwind DI?
It should be pretty easy to slowly lower the output level from the mixer and watch if/when it smooths out (if you're right about it being level dependent)
Otherwise it's interface dependent.


Jason - thank you for the answer.  To clarify - when the AGL-600 unit is not in the signal chain I get flat traces for both magnitude response (at 0dB) and phase (at 0 degrees) for all levels of output from the mixer up until the mixer clips.  So I was thinking that the whirlwind director DI's were not causing the problem. 


When the AGL-600 is after the DI box, so getting a "mic" level signal and feeding a mic-preamp, the traces are flat.  It is only when the AGL-600 is ahead of the DI-box, getting a line level signal, that I see the shifts in the traces.


Is  the impedance of the "next" item in the chain the deciding factor here?.  The input impedance of the mic pre-amps on the Audiobox USB is supposed to be 1200 Ohm.  Normally the DI box would not be in the actual chain but I needed something here so my measurement system is not over-driven.


But in actual use, my intent is to put the AGL-600 between a line level source such as a keyboard or sub-mixer and an TRS aux in on my mixer.  The TRS aux in on my mixer will have a high impedance of about 20K.


So I was trying to verify that the AGL-600 would work this way without changing the signal.  And in general I am trying to understand and verify that a 1:1 isolation transformer spec'ed for 600 ohm operation is a correct tool to use for isolation in my setup.  The trace results confusing me into thinking there are dramatic changes to the signal.



Logged
--Mike
"If you're not confused, you don't know what is going on"

Live Sound for the Mt. Shasta area
http://www.shastalivesound.com

John Roberts {JR}

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 16115
  • Hickory, Mississippi, USA
    • Resotune
Re: Confusing Isolation Transformer Traces
« Reply #3 on: April 03, 2014, 11:30:27 am »

there are a couple possibilities. A) the transformer can not handle that much voltage and is saturating, or B) your test set up can not drive a 600 ohm load.

I did not find any mention of how much voltage the transformer can handle on their data sheets, and lots of hype about 300kHz response, but nary a mention of LF signal handling.

Good luck

JR

Logged
On the internet people tell you everything "they" know, not the answer to "your" question.....  http://circularscience.com/

Mark McFarlane

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1865
  • Middle East
    • Arkose Records
Re: Confusing Isolation Transformer Traces
« Reply #4 on: April 03, 2014, 01:08:23 pm »

I would be surprised if the Whirlwind Director doesn't introduce its own distortion and phase shift at line levels.... but it's in your reference and measurement traces.

What make and model mixer?

What happens if you turn down the level on the mixer so its not +4 (-6db, -12db, -24 db)?  If the 'crap' goes away as you turn down the mixer then I think transformer saturation is more likely (John's A choice).

Logged
Mark McFarlane
ARKOSERECORDS
Turn down what's too loud.

John Roberts {JR}

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 16115
  • Hickory, Mississippi, USA
    • Resotune
Re: Confusing Isolation Transformer Traces
« Reply #5 on: April 03, 2014, 02:07:35 pm »

If it cleans up at low level it could be A or B...

If it cleans up wide band it's B, if it cleans up more at HF but still distorts at LF it's A.

A roll-off at LF could be a cap coupled output not sized for 600 ohm.

JR
Logged
On the internet people tell you everything "they" know, not the answer to "your" question.....  http://circularscience.com/

Mark McFarlane

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1865
  • Middle East
    • Arkose Records
Re: Confusing Isolation Transformer Traces
« Reply #6 on: April 03, 2014, 02:09:51 pm »

If it cleans up at low level it could be A or B...

If it cleans up wide band it's B, if it cleans up more at HF but still distorts at LF it's A.

A roll-off at LF could be a cap coupled output not sized for 600 ohm.

JR
Thanks JR.
Logged
Mark McFarlane
ARKOSERECORDS
Turn down what's too loud.

Mike Karseboom

  • Full Member
  • ***
  • Offline Offline
  • Posts: 230
Re: Confusing Isolation Transformer Traces
« Reply #7 on: April 03, 2014, 05:38:04 pm »

Thanks JR & Mark.  I got a chance to try some different test set ups and it seems everyone was right in suspecting that the DI in the circuit was causing some problems, although I am not sure why.  To answer a few questions first:


-When the AGL-600 was upstream of the DI the traces had the same slopes at mixer output levels from -30dB on the mixer main out  meter (~ 0.43Vrms measuring pink noise with an rms DVM) up to +10dB on the mixer main out meter (~2.49Vrms).  So low or high "line" levels all showed the same "problem".

-The mixer supplying the signal is a Mackie 1202VLZ and I don't think would have a problem driving the test devices.



More testing:  I realized that I could remove the DI's from the signal path if I kept the mixer output lower than -10dB on it's meter (~.125Vrms) and turned the trim on the Audiobox USB preamps all the way down.  When I did this I could get a nice flat response curve and phase plot from the AGL-600. 


Ok so that was a good result but it is still going into a 1200 Ohm load and it is not that much voltage.


Next I tested the AGL-600 more directly how I would use it, by plugging directly into the X32 compact mixer aux in port.  That is a 20K+ imput impedance.


The new test signal path:


-the same iPad pink noise > mackie mixer > main out L > Y split for  test and reference.

-The reference channel 2 went directly into Aux In 2

-The test channle 1 went first through the AGL-600 then into Aux In 1.


So now I can turn up the signal out of the mackie to about +4dB  (~ 1.13Vrms measuring the pink noise) before clipping the levels in Systune.  The result is good - flat magnitude response of 0dB from 20Hz-20kHz and flat phase response at 0 degrees.


So I conclude that the AGL-600 will work well as a line level isolation transformer, at least up to 1.13V, without "coloring" the signal.  It would be nice to drive it with a higher voltage to see what happens but I cannot figure out how to keep the levels shown in Systune from clipping.  The X32 aux inputs can take more voltage, it is something to do with Systune's interpretation of what comes from the X32.


I still don't understand why having the DI boxes in the reference and test signal path would cause the sloping traces. 

Logged
--Mike
"If you're not confused, you don't know what is going on"

Live Sound for the Mt. Shasta area
http://www.shastalivesound.com

Mark McFarlane

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1865
  • Middle East
    • Arkose Records
Re: Confusing Isolation Transformer Traces
« Reply #8 on: April 05, 2014, 03:03:12 am »

...
I still don't understand why having the DI boxes in the reference and test signal path would cause the sloping traces. 

It is theoretically possible that your two different Whirlwind Director boxes have different responses due to parts tolerances, quality control,...  I don't know if this is likely, but it is possible since you are inserting different devices in the measurement and reference paths.

You can test this buy running two tests where all you are trying to measure is the Directors and nothing else.  iPad to USB Pre is the reference for both tests, iPad through Director 1 is a measurement, iPad through director 2 is another measurement, compare results.


Also, you didn't mention how you are measuring Aux 1 and 2 from the X32 in your new test, you show the input chain ...
Logged
Mark McFarlane
ARKOSERECORDS
Turn down what's too loud.

Mike Karseboom

  • Full Member
  • ***
  • Offline Offline
  • Posts: 230
Re: Confusing Isolation Transformer Traces
« Reply #9 on: April 05, 2014, 08:45:33 pm »

Mark - every time I have measured a DI box at "normal" line levels from a mixer, maybe up to about +4dBu, they have displayed a very flat frequency response. Of course they show about a -30dB gain reduction. This includes the Directors, a countryman, and active DI's like the whirlwind hotbox and even the lowly Behringer DI-100.  I will try and get some time to re-test those in a variety of situations and double check the phase response also but have to get back to the regular job for a while here.




[
Also, you didn't mention how you are measuring Aux 1 and 2 from the X32 in your new test, you show the input chain ...



I am not sure if this is the answer to your question but here goes... >> The X32 is an A/D converter and there is a driver for Systune so I can read Aux1 and Aux2 input signals on the desk directly in Systune by routing those inputs to "Card out 1" and "Card out 2". A USB cable from the "card" interface on the X32 to the laptop with systune completes the connections.  So those two aux inputs are acting just like the two mic inputs on the audiobox USB except they have a high impedance and can take a hotter signal before clipping. 


This whole investigation is being driven by two things:


1) my desire to insulate my gear from the musician's equipment.  I have had some pretty flaky and poorly maintained keyboards, ungrounded amps with DI outs, and even home made sub-mixers show up where they hand me a cable and say "here you go".  Some times I am thinking I want to be electrically isolated from these things.  I can just imagine a good voltage spike from some sort of equipment fault frying my whole digital board.


2) I want to extend my input channel count by using the Aux inputs on the X32.  I have plenty of DI boxes that would work to separate me from the muso's gear but the DI boxes have to plug into a regular mic channel.  If I use a 1:1 line level iso transformer I can plug a line level source into the aux inputs and not use up a regular mic channel.


It seems after some of this testing that a line level transformer will typically work very well - hey that is what they are for right?  But also that you if you get to creative with your  signal chain weird things can happen.  I am sure all this is old hat to the seasoned pros.
Logged
--Mike
"If you're not confused, you don't know what is going on"

Live Sound for the Mt. Shasta area
http://www.shastalivesound.com

Simon Byrne

  • Newbie
  • *
  • Offline Offline
  • Posts: 1
Re: Confusing Isolation Transformer Traces
« Reply #10 on: April 06, 2014, 11:20:52 pm »

Mike,

I think you nailed it in your first post when you commented on the source impedance design of the Directors. That is, they are designed to take much, much higher impedance sources.

Unfortunately I cannot find specifications, let alone a schematic for the Directors, but I am guessing you have inadvertently created a gentle high pass filter with the Allen Avionics AGL-600/Director combination. Both devices are are inductive and the source impedance to the Director is massively different to what it is designed for. It makes sense that a inductive/resistive response could change over frequency.

Regards,

Simon.
Logged
Pages: 1 2 [All]   Go Up
 


Page created in 0.058 seconds with 22 queries.