The bhi Dual In-line Module
Clint Gouveia has the bhi Dual In-Line Noise Reduction Unit on his test bench
Clint Gouveia has the bhi Dual In-Line Noise Reduction Unit on his test bench, looking at the capabilities and performance of this unit, principally from the point of view of a short wave DXer.
As a kid back in the 1980s, I would listen to the world on my analogue Amstrad communications receiver from my bedroom or in the garden. I remember copying All India Radio in the evenings and Radio Australia in the mornings, whilst getting ready for school, with stonking great signals, just on the telescopic antenna. Those were the days; I don’t remember local QRM ever being a problem.
Fast forward 30 years and for many of us, it’s a nightmare. Our modern way of life, so much of which is defined by noise-producing technology – from Wi-Fi adapters and digital TVs to PLTs that bounce the noise-floor up and down by 10 dBs or more.
I only have to take a look at the HF spectrum on one of my SDRs and that 1980s noise floor of -130dBs is now -110dBs on a good day and, perhaps, very exceptionally, -115dBs. On a bad day, it’s as high as -90dBs and that really doesn’t make for good listening.
The cheap solution to improving signal-to-noise was described in my previous article (RadioUser, October 2018: 10-13): Get outdoors with a decent portable! Otherwise, like me, you’re probably stuck in the shack, employing magnetic loop antennas, galvanic isolators – anything, in fact, that will help mitigate the omnipresent ‘electrosmog’.
A New Way to Mitigate Noise
When asked to test the bhi Dual In-Line module, I jumped at the chance. Here, potentially, was another weapon in my arsenal, ready to engage in the constant fight against local QRM and one, what is more, that didn’t involve sitting in a quiet wood at 2 am!
The unit is essentially a box containing some DSP wizardry that takes the output from your radio (or another audio device), via the headphone, extension speaker or line-out socket and processes-out any noise present, to improve audio clarity. With dual channels, it is possible to combine two input signals and/or employ two output channels for monitoring one input source.
The bhi module can also handle mono or stereo signals; bhi claim that their DSP algorithms deliver superior noise suppression with fewer artefacts and thus clearer processed speech, even when signals are weak.
One important point readers should be aware of; the bhi module was designed primarily to improve noisy speech signals, rather than music or data. However, as you will see, at lower levels I was able to obtain some quite good results at lower filtering levels.
First Impressions and Setup
The Dual In-Line module arrived with the necessary audio cables (standard 3.5 mm jacks) and a fused DC power lead, but no power supply. I decided to drive it via a home-brew 12V battery pack to completely isolate the unit from any potential mains noise.
The device itself is well constructed, made from high impact polystyrene, the quality of which is more akin to a consumer electronics product than a construction kit. The printing on the front panel is also of good quality and is better than some consumer electronics I have owned.
The controls are well laid out: a single potentiometer for adjusting the audio input level from either an audio or line-in source and two regulators for the output: one each for the audio output and line-out output.
There are three switches on the front panel. A two-way switch to alternate between separate/ combined input channels and a three-way switch to turn the unit on/off and engage/disengage the DSP filtering. A radial multi-position switch is utilised to select the DSP filtering level, from a minimum of 1 to a maximum of 8.
I decided to use my Sangean ATS-909x portable receiver for initial testing (Fig. 1).
It has a line-out socket and is one of the best portables on the market for audio fidelity, thus providing an excellent comparison and measure of the performance of the Dual In-Line Module. For monitoring the processed output audio, I used my Bose SoundLink Mini 2 speaker in all tests (Fig. 2).
Having connected the Sangean ATS-909x and speaker to the Dual In-Line module, I switched everything on and turned up the audio input until both channel input overload LEDs illuminated. Following the instructions, I wound the control anti-clockwise – less than a quarter of a turn – until both LEDs went out.
Then, by turning up the Line-out level, I could hear audio on the speaker.
Thus, I was ready to test the unit.
On Long and Medium Wave
I tuned the Sangean ATS-909x to 153kHz, Radio Antenna Satelor, right at the bottom of the LW band. Their familiar Romanian folk music could just be heard above the noise from our plasma TV in the living room, and from our neighbour’s PLT adapter. An excellent first target then!
Toggling the DSP filtering to ‘on’, I clicked through the filter levels and found that at a setting of ‘3’ the audio from Romania was now largely noise-free. An excellent result. Interestingly, however, by selecting the next level of filtering, the resultant audio immediately became rather muffled and artificial-sounding.
The cut-off point for the filtering is therefore rather sharp.
In this case, the Dual In-Line Module was either working appropriately to improve signal-to-noise, but with one more click of the filter level switch, the signal has been completely over processed.
No matter though, it worked very well on this S7 signal.
I tried the same test on MW, starting with Spirit Radio from County Monoghan, Ireland on 549kHz, which has been reaching my shack at S9 +20 since early October. Again, the presence of local QRM could be heard as the usual ‘buzzing’, which increases the noise-floor to around -100dBs, and, above that, the PLT adapter switching on and off.
The Dual In-Line really shone in this instance. The stronger signal from Spirit Radio, with superior signal-to-noise as compared to Antenna Satelor, seemed to allow for greater DSP intervention (set at 4) before the audio fidelity became too processed. As a result, there was a significant improvement in the overall quality of the audio, making the station much easier to spend time listening to, than with just the radio output alone.
I thought it worthwhile to subsequently tune to a much more ‘difficult’ target:
Attaching my Wellbrook ALA1530 magnetic loop to a Sony ICF-2001D, I tuned to 1560kHz WFME Family Radio in New York who were playing their usual style of music. With this very weak signal, the quality of the audio fidelity could not be improved with the DSP filtering, and the clearest audio was without any DSP intervention.
The device was then tested for performance on HF, using a Kenwood TS2000 transceiver (Fig. 3) coupled to a 104-foot G5RV antenna, via an MFJ-948 ATU. Various SSB amateur radio signals on 40 and 20m, with strengths varying from S0 to S9+20, were copied; interestingly, I found that irrespective of signal strength in SSB mode, the optimum DSP filtering level was always either 2 or 3.
Above 3, I found the audio in all cases to be over-processed and no further improvement in signal-to-noise audio was apparent. The primary effect of this device on the ham bands was, therefore, to make the audio more easily discernible. However, I wasn’t able to copy a signal with non-discernible audio, intervene with DSP filtering, and ultimately reproduce audio that could be understood.
Tuning the 19m broadcast band, I copied the VOA relay from Botswana on 15260 kHz, using the SDR Play RSPDUO SDR and a Wellbrook ALA1530 magnetic loop antenna, indoors. Their signal was strong (around S9 +20). However, despite the indoor antenna, the presence of noise was not what I would regard as particularly intrusive, and again, the optimum DSP filtering was achieved at a setting of 2.
Those of us who use SDRs will be used to audio fidelity from a computer that often leaves much to be desired. The audio from the Bose SoundLink Mini 2 with the DSP filtering was superb, by comparison to the tinny speakers built into my laptop.
Further testing on the broadcast band confirmed that weak DX signals were either difficult or impossible to improve with DSP filtering. A typical example is the daylight signal from North Korea on 12015kHz with a signal of S3 at 16:27. Modulation was weak and thus audio fidelity poor; the SNR was too low for the SDRuno software to calculate, so you get the picture. I couldn’t improve the audio fidelity with the Dual In-Line when the Voice of Korea played their usual operatic style of music. The clearest audio was always without DSP engaged. However, when the announcer took over, the discernibility of the speech could be improved, although, as would be expected, only marginally given this was a very weak signal.
This is not an uncommon scenario for those seeking hard-core DX. Any SDR device employing signal conditioning, noise suppression, noise blanking and so forth, results in unavoidable processing of the modulation with very weak signals and the best results are usually obtained by turning off the AGC and all noise suppression. Nevertheless, I was still impressed with the bhi unit because even marginal improvements in the discernibility of modulation/audio can prove useful in identifying a signal that is too weak for the content to be enjoyed by the listener.
The bhi Dual In-Line Noise Eliminating Module is very effective in cleaning up the audio on MW, LW and HF AM signals, particularly if those signals are at around S5 and higher. Whilst the unit was designed primarily to remove noise from speech signals, I obtained decent results from stronger signals playing music. Don’t expect this for weak signals, however.
I have no doubt whatsoever that those of us who spend many hours trying to enjoy broadcast content, but who suffer from omnipresent electrosmog, would benefit from this device (Fig. 4).
Similarly, amateur radio operators would certainly benefit from the noise suppression this device offers, delivering clearer audio on SSB, which is much easier to understand.
My only word of caution would be simply that, hard-core DXers, or ham operators for that matter, cannot expect this device to process out the noise on very weak signals and effectively ‘uncover’ audio from what was otherwise indiscernible.
[Many thanks to Graham Somerville MD of bhi Ltd., for the loan of the review unit – Ed./ CG].
22 Woolven Close
This article was featured in the December 2018 issue of Radio User