| Author |
Message |
|
|
 Posted: Dec 16, 2011 - 09:30 PM |
|
Joined: Apr 13, 2008
Posts: 236
Location: New Zealand
|
|
Hi all:
I am working for a company that produces mobile / portable radio. We frequently need to go onto a field trial to test functionality of a portable radio. What we do are:
We listen to one talking person, see if there is any voice distortion, drop-out, poor audio quality and etc.
Along with us, each one has a sheet of paper which we use to write down any things that are not supposed to happen and time stamps when they happen.
Currently, the downside of the system is:
1: an operator may forget to write down a time stamp or get a time stamp wrong.
2: everyone has to describe what happened using words, such as "a large electronic noise", "a distortion". In engineering, those words do not mean much.
What I am thinking is to produce a embedded system orientated detachable device that goes with each portable radio, actively recording what has happened and at the end of a trial pass all data to a PC for analysis. The advantages of this project are:
1: All events will be time stamped with high accuracy.
2: A voice will be analysed and presented in visual form, e.g. applying a FFT onto all speeches.
It is just a concept. Any comments?
Thanks[/list] |
|
|
| |
|
|
|
|
|
|
Posted: Dec 16, 2011 - 11:37 PM |
|
Joined: Sep 12, 2009
Posts: 2406
Location: Sacramento, CA
|
|
| Sounds reasonable to me, with the right hardware. In addition to voice, send different test patterns (single-tone, two-tone, different frequencies, different modulation levels) and preceed each test with some identifier like a DTMF code. |
|
|
| |
|
|
|
|
|
|
Posted: Dec 17, 2011 - 12:48 AM |
|
Joined: Aug 04, 2002
Posts: 1706
Location: Brisbane, Australia
|
|
What kind of modulation do You apply?
Direct audio? nodigital stage between the microphone and modulator?
If simple analogue system then what You are doing is probably good enough. |
|
|
| |
|
|
|
|
|
|
Posted: Dec 17, 2011 - 12:58 AM |
|
Joined: Jun 15, 2008
Posts: 1762
Location: North Carolina USA
|
|
| Time stamps are automatic if you record one long audio file. This can be very compact if the radios have good squelch. Lossless compression methods that use run length encoding can represent hours of silence in less than 32 bits, accurate to 20 microseconds at 48KHz sampling. |
|
|
| |
|
|
|
|
|
|
Posted: Dec 17, 2011 - 01:58 AM |
|
Joined: Dec 11, 2007
Posts: 6856
Location: Cleveland, OH
|
|
Ah, the memories...
I can recall maybe 25 years aqo, or so, when cell phones were still a novelty, towers were scarce, and my phone was 300mW while hand held, 3W when in the car adapter.
I had just installed cell phones in the rescue squads for communications with the hospitals. The squads then had ~ 154 MHz VHF radio, ~460 MHz UHF voice and telemetry radio, and ~ 800 MHz (analog) cell phones for communications.
I built a small sin wave tone generator to manually hold up to the microphones for all three systems, and a big audio processing board for the hospital. One could route any of the three sources to my box. It used a Basic Stamp to sweep tune a series of switched capacitor filters to find the "peak" frequency, and then measured the energy in the tone to the full bandwidth's energy, to get a rough Signal to Noise ratio for each system, measured from a number of different locations and distances from the hospital.
One had to tune the system for the received tone for each transmission to correct for the various small drifts in the signal's exact frequency and insure one was getting the maximum signal from the narrow band pass filter.
I hadn't thought about that system in years!
I'd do it a bit differently these days!
JC |
|
|
| |
|
|
|
|
|
|
Posted: Dec 17, 2011 - 06:40 AM |
|
Joined: Jun 22, 2004
Posts: 3849
Location: South West Utah, USA
|
|
| Add a constant amplitude test tone that is frequency swept through the entire audio bandpass of your transmitter and receiver. Detect and record this transmitted sweep from the receiver. This will identify your audio bandpass frequency characteristics (as in flat audio, emphasized audio, etc.). Then if it does not match your expectations find out what is wrong. |
|
|
| |
|
|
|
|
|
|
Posted: Mar 23, 2012 - 09:36 PM |
|
Joined: Apr 13, 2008
Posts: 236
Location: New Zealand
|
|
Hi all:
Just to clear up two ideas, what single tone and two tone tests do? What are their objectives? I could not find any information on line.
Thanks |
|
|
| |
|
|
|
|
|
|
Posted: Mar 24, 2012 - 12:56 AM |
|
Joined: Jun 22, 2004
Posts: 3849
Location: South West Utah, USA
|
|
First you never said if your radios are digital or analog voice or the type of modulation used.
Sweeping and measuring the radio system audio frequency response tells you how well the analog modulation and analog demodulation works over the radio system audio frequency range.
https://en.wikipedia.org/wiki/Emphasis_ ... cations%29
https://en.wikipedia.org/wiki/Deemphasis
If these are not correctly balanced in an analog Tx/Rx voice radio system then you get distortion and/or reduced signal to noise ratios. The radio spectrum tends to have higher audio frequency base line noise levels. Emphasis increases the high frequency modulation levels where there is more natural background noise (it improves the Tx signal-to-noise level). Uncorrected emphasis sounds distorted in a receiver. De-emphasis reverses the emphasis inside the receiver to remove the distortion. Measuring this will tell you if the modulation/demodulation Tx/Rx process is fundamentally flawed or not.
In any analog/digital radio system (even without emphasis/de-emphasis) measuring modulation/de-modulation distortion will give you information about how well your system works. |
|
|
| |
|
|
|
|
|
|
Posted: Mar 24, 2012 - 05:22 AM |
|
Joined: Nov 22, 2002
Posts: 12055
Location: Tangent, OR, USA
|
|
Quantitative measurement of voice quality is very hard. You could have your device generate a series of tones at a sequence of frequencies and a sequence of amplitudes at each frequency. This would catch both frequency response issues and distortion issues.
But, this is only one of several tests you ought to be doing. One is generation and detection of CTSS tones. Another might be generation and handling of DTMF tone-pairs.
There are also other important tests. RF sensitivity and quieting level. Does the squelch work properly? Any squeaks and squawks when push-to-talk is activated or released? Is transmitter output power correct? Or, maybe more correctly, is radiated power correct?
You might need to connect into the "guts" of the radio to do some of the things like CTSS. Not all would be possible only using the mic and speaker jack(s).
Jim |
_________________
Jim Wagner
Oregon Research Electronics, Consulting Div.
Tangent, OR, USA
"The only thing standing between us and victory is defeat" P.G.Wodhouse in Wooster & Jeeves series
|
| |
|
|
|
|
|
|
Posted: May 03, 2012 - 09:40 PM |
|
Joined: Apr 13, 2008
Posts: 236
Location: New Zealand
|
|
Hi all:
I have finished writing my K-brief for senior engineers to look at. Please feel free to read it and provide feedbacks.
The objective is to present my concept and convince them to give me a "go ahead". |
|
|
| |
|
|
|
|
|
|
Posted: May 04, 2012 - 02:25 AM |
|
Joined: Dec 30, 2004
Posts: 8789
Location: Melbourne,Australia
|
|
| Ever thought of using an iPod/phone/pad or similar? They have the codec, non-vol storage and enough cpu grunt to do your dsp with a GUI to boot. If you have a 3G connection, you'll get GPS as well. That takes care of the hardware. Since most of your work is already done in the o/s as in audio record/replay, text to speech, gps going down this path is going to be much quicker than your approach. |
|
|
| |
|
|
|
|
|
|
Posted: May 04, 2012 - 02:27 AM |
|
Joined: Apr 13, 2008
Posts: 236
Location: New Zealand
|
|
To Kartman:
The cost will be too high. For each portable radio, we need a device attached to it. 10 portable radios need 10 iphones. Not really practical. Besides, an iphone is too big and heavy for easy handling. |
|
|
| |
|
|
|
|
|
|
Posted: May 04, 2012 - 02:51 AM |
|
Joined: Dec 30, 2004
Posts: 8789
Location: Melbourne,Australia
|
|
| How much do you think your design will cost in regards to engineering? I would guess at around $50k+ That buys a few iphones. |
|
|
| |
|
|
|
|
|
|
Posted: May 04, 2012 - 03:16 AM |
|
Joined: Apr 13, 2008
Posts: 236
Location: New Zealand
|
|
To Kartman:
Very good point. I will do a cost estimate. |
|
|
| |
|
|
|
|
|
FAQ
Search
Register
Log in to check your private messages
Log in
Maximize
RSS
