[VARA-MODEM] VARA FM Connection problem


Hi: San Diego Red Cross DST group

(To Jerry and Bob - I would like to post this to the "Red Cross Volunteer Connection" - but not sure the best way to accomplish this - as a PDF in the DST "files section" ?) Thanks for your help in this!

In the meeting today, there was mention of a bulk purchase for Red Cross of the Signalink modem. Before any of you lock yourself into spending $200+ for that unit, here is some information, so you can make an informed decision. The Signalink due to its bandwidth of 2.7 KHz design limitations, will limit your Vara thru-put speed to around 10 Kbps.  Vara-FM-Wide requires a band width of 8 KHz!   The Masters Communications modems are designed to pass signal band widths up to 20 KHz! The best radio's to use for VARA-FM are ones that have a separate data port with a 9600 pin. The standard that is now used - is the "Mini-Din 6" cable. This will allow receive audio to be taken straight from the discriminator, and transmit audio to be sent, straight to the FM modulator, - thus bypassing the filters that limit audio bandwidth. Most of the newer radio's use this standard with the "Mini-Din 6" cable, which makes the setup "plug and play"   Also this setup leaves  the the radio microphone functional, so voice communications and VARA data can be handled with out having to unplug the mic and plug in the data cable. Also the speaker audio can be turned down during data transmissions and not effect he constant receive audio levels going into the modem from the discriminator. Most all of the newer Yaesu radio's have this data configuration, also some Icom's.  For those interested in HF - the DRA modems will outperform the Signalink when on VARA 4.0.  Beware of the FT991 - its "data port" is set up to pass AFSK RTTY only and will NOT work for VARA! I learned this from: VARA-MODEM@groups.io 

As some of you know, I use an FT7900 with a Masters Communications DRA50 for both my Base/Portable and my Mobile. I have two Signalinks and was frustrated with the slow thru-put speed of around 10-12 KHz even with a close and strong connection. From the: ("VARA-MODEM@..." ) site I learned about the speeds possible with the Masters Communications modems, and ordered two of  them. I am SO GLAD I did!  I am now getting 29+ KBps on a good path!!! Don't make the mistake like I did,  when I purchased 2 Signalink's before learning of the DRA50  !!   The DRA50 cost was $65 each, which is far cheaper than the Signalink!.    Yes - they are a kit to assemble, but the only SMT part comes soldered to the PCB. The rest of the parts are easily inserted into the clearly marked locations. The assembly takes about 1 hour. For those older, It could take 2 hours. I suggest a bulk purchase of the DRA50 be made - for the Red Cross DST members. The assembly could be done  by multiple members, that have kit building skills. The customer support for Masters Communications is excellent and there is on line documentation and video's to show the builder the way to success, and level setting. For VARA-FM  on the display - the "VU Meter" should be set at 3/4 scale. This points out another problem with the Signalink, in many cases, its transmit output level has been shown to be too low - and not enough to compensate for the radio bandwidth roll off.  Below is all the design info that went into the DRA modems, to make them so superior to other modems!   

73 Bruce  WA6DNT@...  WA6DNT@...

-------- Forwarded Message --------
Subject: Re: [VARA-MODEM] VARA FM Connection problem
Date: Fri, 6 Nov 2020 01:38:59 -0500
From: Kevin Custer <kuggie@...>
Reply-To: VARA-MODEM@groups.io
To: VARA-MODEM@groups.io

Sometimes I get a request for the audio bandwidth specifications of the DRA Series radio adapters.  These specifications aren't on our website, because the DRA Series drastically exceeds the audio requirements of any mode designed for two-way radio - analog or digital.  This is mainly because the heart of the RA and DRA adapters is a C-Media chipset that was originally designed for stereo headset applications where people would be listening to high-fidelity audio, and the need to process audio from a microphone or other high-fidelity audio source.  The CM119A is superior to other chipsets, including its proposed replacement the CM119B.  While you'd think a simple letter revision doesn't mean much - in the case of this integrated circuit, it's a bigger difference than one would think - at least when the chip is used as a radio interface.

Masters Communications has been producing radio adapters (sound cards that connect to two-way radios) since 2014.  Our original RA Series was designed for AllStar Link - an analog linking and repeater controller software designed for ham radio.  The C-Media CM Series is the only chipset that would work with AllStar Link, and is why many of the higher quality radio interfaces uses one of these chips.  The CM108 was the first one, then the CM119 (no suffix), then the CM119A (the one that we currently use).  This chip is still in production, and anyone purchasing a DRA series radio adapter will see a date code that is comprised of 4 numbers stamped on the chip by the manufacturer.  These numbers are arranged like this "2009", where the "20" is for the year 2020 and the week manufactured "09" for the 9th week.  This date code relates to a manufacturing date of somewhere between February 24 and March 1 2020.

I'll address the frequency response of the RA and DRA-Series for everyone, since it was mentioned here a couple of times.

When I hear someone mention about frequency response limitations, the first place I look is at the radio.  This is especially true if the receive audio is taken from anywhere except the discriminator.  Almost every two-way radio I know of has CTCSS filtering to the speaker and most auxiliary audio connections.  This filtering is to eliminate PL buzz in the speaker.  Likewise, for the transmitter, frequencies below 300Hz are intentionally filtered in the MIC path so that any voice audio that includes fundamental frequencies in the CTCSS range doesn't "talk off" the PL decoder on the far end.  Any sub 300 Hz voice characteristic that would cause the PL tone to cancel will degrade CTCSS performance, and therefore are filtered in the transmitter.

Proper deployment of digital communications using lower and higher fundamental frequencies (than human voice) requires the connection of the radio adapter directly the the receiver discriminator absent from any such filtering or de-emphasis circuitry.  Likewise, the radio adapter needs to be connected directly the the radio's FM modulator, avoiding any limiting, filtering, and pre-emphasis.  Many people just use the MIC input and Speaker connections because it's easy, or that is all that's available.  Even the "9600" or wide-band (data port) audio connections on most radios are filtered (to some degree) to keep people from doing stuff they shouldn't and splattering over into the next channel or worst yet, out of the band.  It's these filters in the radios that limit the maximum throughputs.  Read that last sentence again....

Let's now look at the design of the RA and DRA Series adapters.  This gets pretty deep technically, and we're going to do some math, so I'll apologize for its content and length now.

The following specifications are used in these calculations:
The stated frequency response of the CM119A DAC (transmit audio path) is 20Hz to 20kHz when driven by a 48kHz sampling rate program.  This information is taken directly from the products data sheet.
Its ADC input (the receive audio path) is somewhat less at 20Hz to 19.2kHz.
The DAC output impedance is 32 ohms (in the case of the DRA-30 and DRA-36).
The LM386 output impedance is 16 ohms (in the case of the DRA-45, DRA-50, DRA-65, and DRA-70).


The input impedance of the CM119A MIC input (our receiver audio path) is somewhere around 100k ohms.  In addition, there's a 100k potentiometer pot going to ground.  Worst case, the input impedance seen at the radio adapters input is around 50k ohms, because these two resistances are in parallel.  The coupling capacitors inline to the MIC input are both 4.7uF.  So worst case the series capacitance is slightly over 2uF (2.35).  When you do the math, this creates a high-pass filter at 1.35 Hertz (Yes - less than 2 cycles per second).  As such, the receive audio chain of the RA-Series adapters have a frequency response well below the 20 Hz cut-off specification of the CM119A.

The parallel capacitance on the MIC line is 47pF.  Considering the source impedance is 50k ohms or less (many radios are between 600 ohms and 10k) the worst-case high end frequency response would extend to almost 68kHz.  This is well above the 20kHz specification of the CM119A.

For DRA's without the LM386 TX audio amplification (DRA-30 and '36), each CM119A DAC output (left and right) go through a pair of 100uF coupling capacitors and a 1k potentiometer.  This provides a low frequency cut-off of about 100 Hz assuming the load is 32 ohms.  In reality, most transmitters have an audio input impedance of 600 ohms or more.  At 600 ohms, this equates to a low frequency cut-off of 5.3 Hz, well below the 20 cycle low frequency specification of the CM119A chip.  For DRA's that have the LM386 TX audio amplification (DRA-45, '50, '65, and '70) each CM119A DAC output (left and right) feed the inputs to the LM386 IC's.  Those coupling capacitors are sufficiently sized to exceed the frequency response of those IC's.  Each output of the LM386 goes through a 100uF coupling capacitor.  This provides a low frequency cut-off of about 100 Hz assuming the load is 16 ohms.  In reality, most transmitters have an audio input impedance of 600 ohms or more.  At 600 ohms, this equates to a low frequency cut-off of 5.3 Hz, again, well below the 20 cycle low frequency specification of the CM119A chip.

For all intents and purposes, the frequency response of the DRA radio adapters are no less that 100Hz to 20kHz worst case scenario, and likely to be 20Hz to 20kHz the specification of the CM119A, depending on the connected radio.

Here's a audio spectrum analyzer sweep of a CM119A with 100uF coupling capacitors being tested for maximum frequency response:

This testing was done using a Windows based audio spectrum analyzer program called TrueRTA.  I own the 1/24 Octave license.  This testing was done at a 44.1kHz sampling rate.  It's actually a little better at 48kHz.

In terms of dynamic range - it's 16 bits (96 dB theoretical) through the chain, which is way way more than the noise-limited dynamic range of two-way radio.  Even if you write off the additional digital headroom requirements in dealing with preemphasis and deemphasis, there's still more dynamic range available than would be realized in two-way (about 40dB at best). 

In terms of noise immunity, the USB input to the CM119A can be a limiting factor where signal to noise is concerned.  We've worked very hard on the design of the circuit board to insure the USB inputs are properly routed and terminated and deliver the lowest possible noise.  The use of a high quality circuit board with a massive ground plane and rounded traces on the USB lines insure high signal to noise ratio.  47pF capacitors on all input and output lines help to insure noise immunity is high as well as providing protection against RFI.

At first, Jose was a little hung-up on the radio adapters that have the added amplification of the LM386's, IE: the DRA-45, '50, '65, and '70.  He felt it was somehow limiting the audio bandwidth or adding distortion or other artifacts which would limit the throughput.  I can guarantee you that the coupling capacitors and added amplifiers in the DRA Series aren't causing any throughput limitations.  As I said earlier, limitations are because of the connected radios and their individual bandwidth and distortion characteristics.  If you're not convinced - buy radio adapters without the added amplifiers.  But - realize certain radios require more P-P audio voltage than a naked CM119A can output.

How do I know for sure these are as good as I claim?  Here are some real world examples:

There are several guys that are using highly modified UHF GE Delta radios that were highly modified.  These radios had their I-F filters replaced with wider ones. They're using the optional FL-10 filter at 15 kHz on the transmit audio path and feeding the FM modulator directly.  The modulators were modified to provide flat frequency response from 50Hz out to nearly 8kHz.  They're using stock DRA's that have the LM386 amplification, as the TX audios injection point (after modification) requires a good bit of P-P voltage for proper deviation.  They are purportedly running a modified version of (some software package) that uses audio from 50Hz to about 8kHz on wide-band UHF channels.  They are reporting 50+ kbps throughputs, so I can assure you there are no bandwidth limitation concerns with the DRA Series.  The FL-10 filters are not needed in 99% of the deployments, so please don't think there is something to gain by adding one to your existing installation.  They are only needed when used with highly modified radios for extreme throughputs where the original filtering was removed.

Others are using the Yaesu FT-8800 and FT-8900 for gateways.  These radios perform up to their inherent filtering limitations and folks are getting in excess of 30kbps throughput, and as Bruce mentions, the FT-7900 is also an excellent radio for VARA FM.  No modifications of these radios are required for exceptional throughput rates.

In summary, the DRA Series radio adapters are not a limiting factor in exceptional throughput digital communications using two-way radios.  They will work well beyond the capabilities of the radios we're connecting these to. Remember, it's the filtering in the radios that limit the maximum throughputs.

I recently wrote an article that explains this a little differently:

Comments and questions are welcomed.

Kevin W3KKC

On 11/5/2020 7:35 PM, Bruce wrote:
Also the Signalink audio band -pass rolls off around 2.7 KHz which limits the tru-put to around 10-15 KBps.
Vara-FM- wide needs an 8 KHz B.W.  With the Masters Communications DRA50 and my FT7900 (set to 9600)
I get 29 KBps on a good P2P path.  And YES VARA is a GREAT mode with the right equipment!
73 Bruce WA6DNT2JUNO.com   WA6DNT@...

On 11/4/2020 9:32 PM, Bradley Glen via groups.io wrote:
Vara requires a flat audio reponse for HF and FM- look on the waterfall and if there is any signs of band edge filtering like seen when taking off the speaker- not a good practice .Every radio has taylored audio designed for that radio .

1-Flat audio you will get the better data throughputs.
2-Transmit - as wide as possible and use gain control .
3-Make sure your tx and RX points do not LOAD that inout/output , use sound interfacing practises .

With these simple points above  great speeds can be seen .
My highest on HF is over 21000 bytes /min , in  ideal conditions..

When you see speeds increase and then suddenly you lose a block - reduce drive level.
You will find a sweet spot .

good luck and vara is a great mode 

Join main@ARC2-SDICC.groups.io to automatically receive all group messages.