Digital Interface - transformer seems to drastically weaken signal


Rob French (KC4UPR)
 

Hey there,

I've prototyped myself a variant of KK5JY's AnyRig digital interface ( http://kk5jy.net/AnyRig-v1/ ).  The Teensy microcontroller piece of this is very straightforward to me, as I have previously embedded a Teensy in my uBITX as a DSP. In this case, I'm externally interfacing a uBITX to computer audio (trying to keep the uBITX relatively stock). 

Anyway, the issue I have is with the 1:1 isolation transformer for the RX audio.  It drops the RX signal level drastically!  If I drive the RX line out straight to the Teensy audio input, I get a solid signal level of ~0.5V peak-to-peak, based on observing the signal level on my computer using Audacity (I have an op-amp preamp driving the line-out).  If I put the isolation transformer inline, the signal level barely even registers!  I have to amplify it significantly in software to hear the signal, when using the transformer.

I tried it with two different transformers.  About the same on both.  The transformers are cheap Amazon ones ( https://www.amazon.com/gp/product/B07HTBQ7FY/ ), but given that I really only needed them to cover ~300-3000 Hz, I figured they'd be sufficient... I wouldn't think they'd have that much loss!  What am I missing?  What should I look at?

(The RX line-out is driven by an op-amp pre-amp per this URL:  http://rason.org/Projects/opamps/opamps.htm )

Thanks,
-Rob KC4UPR


Gordon Gibby
 

Couldn't quite find a real schematic in all that stuff which explained exactly where the RX transformer feeds into the Teensy.   As such, cannot know the input impedance, or normal signal level expected of the input for the Teensy.   I see lots of mention of "line audio" --- far higher voltage than the speaker output....
so you might have to dig more into exactly what input impedance you are trying to feed with that isolation transformer, and what signal level is required.

I can tell you that I have used those type 600:600 ohm transformers to feed normal $4.75 Adafruit 1475 USB sound dongles and there is PLENTY of audio....I always have to add a trim pot to cut it down, on one or the other sides of the transformer.   The USB sound dongles are apparently set for a relatively LOW input voltage....

Gordon


On Sun, Jul 26, 2020 at 8:45 PM Rob French (KC4UPR) <kc4upr@...> wrote:
Hey there,

I've prototyped myself a variant of KK5JY's AnyRig digital interface ( http://kk5jy.net/AnyRig-v1/ ).  The Teensy microcontroller piece of this is very straightforward to me, as I have previously embedded a Teensy in my uBITX as a DSP. In this case, I'm externally interfacing a uBITX to computer audio (trying to keep the uBITX relatively stock). 

Anyway, the issue I have is with the 1:1 isolation transformer for the RX audio.  It drops the RX signal level drastically!  If I drive the RX line out straight to the Teensy audio input, I get a solid signal level of ~0.5V peak-to-peak, based on observing the signal level on my computer using Audacity (I have an op-amp preamp driving the line-out).  If I put the isolation transformer inline, the signal level barely even registers!  I have to amplify it significantly in software to hear the signal, when using the transformer.

I tried it with two different transformers.  About the same on both.  The transformers are cheap Amazon ones ( https://www.amazon.com/gp/product/B07HTBQ7FY/ ), but given that I really only needed them to cover ~300-3000 Hz, I figured they'd be sufficient... I wouldn't think they'd have that much loss!  What am I missing?  What should I look at?

(The RX line-out is driven by an op-amp pre-amp per this URL:  http://rason.org/Projects/opamps/opamps.htm )

Thanks,
-Rob KC4UPR


Evan Hand
 

Rob,
As with Gordon, I cannot get to all of the links that you posted.  Regardless, my first thought is that you are possibly shorting out any DC bias with the transformer unless you have a capacitor on the input to your op-amp/Teensy.  There is a capacitor on the output of all of the uBITX versions, which may be why it works when directly connected.  I would try a 1 uF or higher cap for DC isolation.

73
Evan
AC9TU


Evan Hand
 

Rob,
You can ignor my response.  There are caps on the input and output of the audio adapter board (I went back and found the schematics).

73
Evan
AC9TU


Evan Hand
 

Rob,
One question: Why do you need the AnyLink interface?  I have been using the EasyDigi (really just the transformers and capacitors for audio isolation) and CAT control over the uBITX USB connection.  Reduces parts count and complexity for WSJT-X and JS8Call.

Just curiosity on my part.
73
Evan
AC9TU


Rob French (KC4UPR)
 

Hey Evan,

Really I think the only difference between EasyDigi and my setup, is that the Teensy + Audio Adapter takes the place of the sound card.  The added capability of the Teensy is that I plan to (and have previously) implemented standalone DSP filters for transmit and receive in the Teensy.  I'll have a mic input option to the Teensy, with S/W speech compressor, as well as variable width receive filters.  Right now, the part that I'm having issues with is in fact the "EasyDigi" part (i.e. the transformers).

As you noted, yes, the Teensy Audio Adapter is capacitively coupled with 2.2uF capacitors on all of the audio input/output lines.  If I can't figure out this transformer thing, I'll probably just directly connect to the Teensy.

I noted recently that, when transmitting into a dummy load, everything seemed fine (without the transformers installed).  However, when I transmitted through my antenna, I had a terribly distorted audio waveform.  I traced this to the unused wire in the stereo audio cable I was using... when I grounded that unused wire, the distortion went away.  I have a suspicion that issues with RFI are contributing to my woes.

I'm going to try prototyping a quick setup where I (a) add a couple 0.01 uF bypass capacitors at each side of the transformer, and at each end of the audio cable, and (b) use a good, shielded mono audio cable for each of the RX and TX audio connections.  If that doesn't work, then I'll try swapping out one of the El Cheapo 1:1 isolation transformers I'm using, for an ostensibly higher quality Xicon transformer I have lying around from Mouser.  That one isn't a 1:1 transformer (I think it's 4k:600 ohm), but maybe I can use the transformation ratio to my advantage as well (e.g. step down the transmit audio).

Rob KC4UPR