For the last few weeks I have been debugging some noise issues in “analog mode”, and testing the SM1000 between a couple of HF radios.
The SM1000 needs to operate in “analog” mode as well as support FreeDV Digital Voice (DV mode). In analog mode, the ADC samples the mic signal, and sends it straight to the DAC where it is sent to the mic input of the radio. This lets you use the SM1000 for SSB as well as DV, without unplugging the SM1000 and changing microphones. Analog mode is a bit more challenging as electrical noise in the SM1000, if not controlled, makes it through to the transmit audio. DV mode is less sensitive, as the modem doesn’t care about low level noise.
Tracking down noise sources involves a lot of detail work, not very exciting but time consuming. For example I can hear a noise in the received audio, is it from the DAC or ADC side? Write software so I can press a button to send 0 samples to the DAC so I can separate the DAC and ADC at run time. OK it’s the ADC side, is it the ADC itself or the microphone amplifier? Break net and terminate ADC with 1k resistor to ground (thanks Matt VK5ZM for this suggestion). OK it’s the microphone amplifier, so is it on the input side or the op-amp itself? Does the noise level change with the mic gain control? No, then it must not be from the input. And so it goes.
I found noise due to the ADC, the mic amp, the mic bias circuit, and the 5V switcher. Various capacitors and RC filters helped reduce it to acceptable levels. The switcher caused high frequency hiss, this was improved with a 100nF cap across R40, and a 1500 ohm/1nF RC filter between U9 and the ADC input on U1 (prototype schematic). The mic amp and mic bias circuit was picking up 50Hz noise at the frame rate of the DSP software that was fixed with 220uF cap across R40 and a 100 ohm/220uF RC filter in series with R39, the condenser mic bias network.
To further improve noise, Rick and I are also working on changes to the PCB layout. My analog skills are growing and I am now working methodically. It’s nice to learn some new skills, useful for other radio projects as well. Satisfying.
Testing Between Two Radios
Next step is to see how the SM1000 performs over real radios. In particular how does it go with nearby RF energy? Does the uC reset itself, is there RF noise getting into the sensitive microphone amplifier and causing runaway feedback in analog mode? Also user set up issues: how easy is it to interface to the mic input of a radio? Is the level reaching the radio mic input OK?
The first step was to connect the SM1000 to a FT817 as the transmit radio, then to a IC7200 via 100dB of attenuation. The IC7200 receive audio was connected to a laptop running FreeDV. The FT817 was set to 0.5W output so I wouldn’t let the smoke out of my little in-line attenuators. This worked pretty well, and I obtained SNRs of up to 20dB from FreeDV. It’s always a little lower through real radios, but that’s acceptable. The PTT control from the SM1000 worked well. It was at this point that I heard some noises using the SM1000 in “analog” mode that I chased down as described above.
At the IC7200 output I recorded this file demonstrating audio using the stock FT817 MH31 microphone, the SM1000 used in analog mode, and the SM1000 in DV mode. The audio levels are unequal (MH31 is louder), but I am satisfied there are no strange noises in the SM1000 audio (especially in analog mode) when compared to the MH31 microphone. The levels can be easily tweaked.
Then I swapped the configuration to use the IC7200 as the transmitter. This has up to 100W PEP output, so I connected it to an end fed dipole, and used the FT817 with the (non-resonant) VHF antenna as the receiver. It took me a while to get the basic radio configuration working. Even with the stock IC7200 mic I could hear all sorts of strange noises in the receive audio due to the proximity of the two radios. Separating them (walking up the street with the FT817) or winding the RF gain all the way down helped.
However the FreeDV SNR was quite low, a maximum of 15dB. I spent some time trying to work out why but didn’t get to the bottom of it. I suspect there is some transmit pass-band filtering in the IC7200, making some FDMDV carriers a few dB lower than others. Note x-shaped scatter diagram and sloped spectrum below:
However the main purpose of these tests was to see how the SM1000 handled high RF fields. So I decided to move on.
I tested a bunch of different combinations, all with good results:
- IC7200 with stock HM36 mic, SM1000 in analog mode, SM1000 in DV mode (high and low drive)
- Radios tuned to 7.05, 14.235 and 28.5 MHz.
- Tested with IC7200 and SM1000 running from the same 12V battery (breaking transformer isolation).
- Had a 1m headphone cable plugged into the SM1000 act as an additional “antenna”.
- Rigged up an adaptor to plug the FT817 MH31 mic into the CN5 “ext mic” connector on the SM1000. Total of 1.5m in mic lead, so plenty of opportunity for RF pick up.
- Running full power into low and 3:1 SWR loads. (Matt, VK5ZM suggested high SWR loads is a harsh RF environment).
Here are some samples, SM1000 analog, stock IC7200 mic, SM1000 DV low drive, SM1000 high drive. There are some funny noises on the analog and stock mic samples due to the proximity of the rx to the tx, but they are consistent across both samples. No evidence of runaway RF feedback or obvious strange noises. Once again the DV level is a bit lower. All the nasty HF channel noise is gone too!
Rick and I are coordinating our work with a change log text file that is under SVN version control. As I perform tests and make changes to the SM1000, I record them in the change log. Rick then works from this document to modify the schematic and PCB, making notes on the change log. I can then review his notes against the latest schematic and PCB files. The change log, combined with email and occasional Skype calls, is working really well, despite us being half way around the planet from each other.
One open issue for me is what enclosure we provide for the Beta units. I’ve spoken to a few people about this, and am open to suggestions from you, dear reader. Please comment below on your needs or ideas for a SM1000 enclosure. My requirements are:
- Holes for loudspeaker, PTT switch, many connectors.
- Support operation in “hand held” or “small box next to the radio” form
- Be reasonably priced, quick to produce for the Qty 100 beta run.
It’s a little over two months since I started working on the SM1000 prototype, and just 6 months since Rick and I started the project. I’m very pleased with progress. We are on track to meet our goal of having Betas available in 2014. I’ve kicked off the manufacturing process with my good friend Edwin from Dragino in China, ordering parts and working together with Rick on the BOM.