I have a Digirig Mobile Rev 1.10 and Baofeng cables for connecting it to a UV-5R. I am using direwolf on debian 13 on a Raspberry Pi Zero 2W on the USB side of the Digirig Mobile. I am attempting to use this on 2m FM APRS. Although I have tried every alsamixer and direwolf.conf setting that I can think of to attenuate the audio getting to the Baofeng, the Digirig Mobile is wildly over-driving the mic input of the radio. I can tell this by looking at the spectrum of the transmitted signal with a quality SDR and observing the overdeviation. I tried substuting a TIDRadio TD-H3 and got nearly identical results. From what I have read here on the forum, it seems that the ATT pads on the Digirig mobile affect only the received audio, not the transmit audio, so cutting the ATT pad is likely to do nothing to improve this situation. From what I gather, the Digirig Mobile is outputting line-level audio, but the mic inputs on these HTs are expecting electret mic levels, so some form of attenuation is likely necessary between the Digirig output and the mic input, yet I don’t see this on any of the wiring diagrams provided on in the Digirig blog. I am wondering if I am the only one having this issue, or if there are a plethora of users out there overdriving their HT’s with Digirig Mobiles? Is this a known issue? (I haven’t discovered it being written about in this forum yet, but maybe I am looking in the wrong place?) Does anyone know what specific level of attenuation is needed here? (I will start experimenting with it now.)
I’m sorry my response doesn’t have a solution:
I have 2 digirig mobiles, using a couple windows installs as well as an android smartphone.
i have no trouble over driving:
UV-5R
UV-3R
IC-T7
IC-T90
HTX-202
In other threads, @K0TX , has explained/detailed the drive levels.
other than a hardware fault with the dr mobile, ‘good’ drive range should be achieved through the os ‘mixer’ (I saw that didn’t work for you) without issue. mixer preamp/output boost etc off.
hopefully someone with more experience with the Pi will jump in and help.
(i almost started a digipi build but really don’t need it)
good luck
kb0wlf
Thanks for the feedback. I’ve been investigating this issue further by running my spectral analysis through AI and switching back-and-forth between the UV-5RE and the TD-H3. (TD-H3 is actually better for debug because it has a variable mic gain that the UV-5RE does not.) Here is what it said:
“You’re using:
-
TIDradio TD-H3
-
Digirig Mobile
-
Mic input only (no data port)
That combination is notoriously hot on transmit audio.
What your latest screenshot proves
With a single 1200 Hz tone and:
-
Mic gain = 0
-
ALSA ≈ minimum
…and you still get:
-
~±5 kHz deviation
-
Dense sidebands
That means:
The Digirig output + TD-H3 mic input sensitivity is far too high—even at minimum settings
Why this is happening (no mystery now)
1. TD-H3 mic input is very sensitive
Typical HT mic inputs expect:
- ~10–50 mV audio
2. Digirig outputs line-level-ish audio
Typical USB audio output:
- ~0.5–1.0 V
That’s a 10×–100× mismatch
3. The radio still amplifies it
Even at “mic gain 0”:
-
There is fixed gain
-
There is pre-emphasis
-
There is likely a limiter
So your signal becomes:
Overdriven → clipped → wide → ugly
Exactly what your SDR shows.
Why software will never fix this
You already proved it:
-
ALSA at minimum
-
Single tone
-
Still too wide
You’ve hit the floor of digital attenuation
After that point:
The DAC output voltage is still too high for the radio
The correct fix (hardware attenuation)
You need to pad down the audio before it hits the mic input.
Bottom line
You didn’t do anything wrong.
You just hit a classic limitation:
Modern USB audio + HT mic input = massive overdrive without attenuation
Final answer in one line
You need ~20–30 dB of analog attenuation between the Digirig and the TD-H3 mic input—nothing in software will fix this.”
I am about to implement said 20-30 dB of attenuation and see what happens. I will report back at conclusion.
Bottom line is that I bet that there are many operators who are blindly hooking up their HTs to the Digirig with direct cables without looking at their spectral output. Maybe it works(?) but it would probably work much more cleanly and produce better decodes if they matched and leveled their audio to produce the proper deviation levels.
73 de KC7X
i gotcha, i’m lucky enough to have an HP8924c to ‘check’ my gear, most don’t
to be honest, FM packet gets really close with just an ear on a receiving radio.
like you, i’ve had my ticket for awhile so i remember elmering rather than simply ‘vending’ answers.
any os should be able to turn the output down so far that there is NO audio passed to the ‘mic’ of the HT. so there must be something wrong that volume setting ‘0’ and volume setting ‘1’ blows out the ‘mic’ input
we were doing soundcard packet long before dedicated interfaces with built in serial ports.
just SPEAKER out from the pc, some audio transformers from the junk box.
hopefully @K0TX can give you the engineer/designer/creator’s information on why the digirig mobile is the way it is.
i’m just a user who likes the tiny interface compared to my homebrew stuff the size of a shoebox
kb0wlf
Results: I placed an inline voltage divider between the Digirig and the HT. It consisted of a series 22K resistor on the mic line between the Digirig and the HT, and a 680-ohm resistor from the HT-side of the 22K resistor to ground, resulting in about 30.5 dB of attenuation (around 32 dB if you take into account the input impedance of the HT mic input.) It took some experimentation to find the right resistor values. I then re-tested, captured the spectral output for both single- and two-tone signals, and had AI analyze it. Here is what it said:
”That’s a solid result—you’ve essentially landed it.
What your latest spectrum shows
-
The signal is now mostly contained within about ±3 kHz
-
Sidebands are clean and discrete, not smeared
-
The “shoulders” fall off much more naturally
-
No obvious clipping/flat-topping
In other words:
You are now in the correct deviation regime for 1200 baud AFSK
You’ve achieved:
-
Proper deviation
-
Out of limiter/compression region
-
Good spectral containment
-
Compatible with narrowband receivers
What your two-tone (1200/2200 Hz) plot shows
-
The envelope is symmetrical around center
-
Energy is well-contained (~±3 kHz)
-
No “tilt” or obvious dominance to one side
-
No flattening or splatter
In plain terms:Your mark (1200 Hz) and space (2200 Hz) tones are well balanced
-
Final verdict
“
I could audibly hear the difference on the received signal as well. It is now clean and pure without distortion.
Hopefully this will provide useful information to others who are attempting to use their HT’s with Digirig Mobile - and help to clean up our airwaves. Just remember that the cable alone (in its current form, at least) is not sufficient.
I plan to make this adjustable and DC block the electret mic bias voltage by using a fixed resistor in the input from the Digirig of 2.2k plus a 25k variable resistor (potentiometer) in series with a 10uF DC blocking capacitor in the HT mic lead. (Digirig → 2.2k resistor → 25k pot → 10 uF capacitor → HT mic input.) The 680 ohm voltage dividing resistor then goes from the junction of the 25k pot and the 10 uF DC blocking capacitor to ground. The 2.2k pot ensures a minimal attenuation and gives the 25k pot more resolution in dividing the audio voltage.
Maybe adding an optional output attenuator pad to the Digirig is something that the makers might want to consider in the next release.
73 de KC7X
I have used digirig 1.9 and Digirig Lite with BaoFeng UV-5RTP, BTECH UV-5X3, Kenwood TH-D72A, Kenwood TH-F6A, Kenwood TM-V71A, Elecraft KX2. No audio level problems so far.
I see this thread calls out digirig 1.10. I havena compared schematics. We’d have to be told if there are audio level changes from 1.9 to 1.10.
I use several different methods to look at my TX audio.
- Scope. These are audio levels, so even an inexpensive USB scope will do.
- I use fldigi’s SIG display (waterfall, FFT, SIG). I set the TX to 500 Hz (audio, of course). I TX from one radio, and RX on another. I look at the RX waveform for distortion as I adjust all the bits in the TX and audio streams, including fldigi, CODEC levels, sound card level, if any, and radio. All of those affect signal quality.
- I use Zeitnitz (Windows) software scope to look at both TX and RX signals.
- Please see TM-V71A sanity check before I cut ATT - #7 by TheDustyKey for the inline attenuator I built. That was for RX, not TX, but it is convenient, and takes advantage of digirig components, to minimize components.
@KC7X: I am interested in specifically how you used AG (the data analysis tool known as AI). By that I mean specific files that you used, and exactly how you used them, and what “AI” tool(s) gave you those replies. I’d like enough information (and data files) to be able to duplicate your steps, in the interest of learning. I am quite disappointed in any AG responses I have gotten, and am looking to improve my use of the tools. Feel free to contact me off list, if that seems more appropriate.
The whole issue of audio distortion when using software is present for every station. When I first got into fldigi years ago, I was told many myths about using fldigi. In fact, I was offered the chance to be “trained” to listen to incoming signals, so that I could determine if they were distorting. But my history includes using instruments, and not relying on individual hearing, so I turned to those mentioned above. At an earlier ARES meeting, when we were all learning how to use fldigi, one of the ops could not RX, whilst many others in the same room were 100% copy. Looking at Zeitnitz and fldigi SIG on that op’s computer, it was clear that some simple adjustments to radio VOL out and CODEC gains would solve the problem. QED.
The test arrangement that I used was to setup the APRS rig (with Digirig and Pi) remotely on low power, keying it with single tone and two tone AFSK test signals, and receiving those signals with an SDRPlay RSP1A SDR connected to my PC, capturing screen shots of the resultant received spectrum, then feeding those screen shots into AI for spectral analysis. At this point, I don’t recall with certainty if the AI tool I used was Gemini, ChatGPT, or Grok, as I often use all 3.
Thanks for the reply. It’s time for me to do some tests here anyway, so I’ll add this to my list. I have an SDR, though I havna used it this way. I am going to try my usual and the SDR, and I’ll report my results if I am successful. This, for me, is only a test of audio signal quality, as send to the radio, and as transmitted, so it will be a simple test. If I send 500, 1500, and 2500 Hz tones, those should be adequate.
73 ~R~