[RTTY] UPDATE Re: FSK Keying

[Joe Subich, W4TV]

On 2/26/2018 1:35 PM, David G3YYD via RTTY wrote:
 > Robert
 >
> Yes I have measured it and timing jitter is a real problem using 
 > windows to key a COM port line.

The "timing jitter" is only an issue when using software generated
FSK (e.g., EXTFSK or "Software FSK").  If one uses the serial port
in its intended UART mode, the bit rate is set based on a hardware
divisor of a crystal oscillator on the UART chip.  The bit rate will
not "jitter".

Unfortunately, modern USB to serial UART (or "Serial converter")
chips are limited by the bit rate divisor in Windows (a 14 bit
integer).  Since the chip designers have to push the output data
rate to 1 Mb/s, 5 Mb/s or 10 Mb/s to meet USB 2.0 and 3.0 specs,
the bit rate divisor limits the slowest supported data rate to
100, 300 or 600 bps leaving amateurs to resort to software
methods to reach 45.45, 50, 60 and 75 bps data rates.

*PROPERLY DESIGNED HARDWARE* will provide jitter free operation
at amateur data rates.  The issue is the lack of proper hardware
as "standard" in modern computers and software developers who
would rather write a "hack" than use what hardware that is available
to address the issue (e.g. write software that keys the TxD line
of a real UART using software timing rather than write a properly
paced UART driver).

73,

    ... Joe, W4TV


On 2/26/2018 1:35 PM, David G3YYD via RTTY wrote:
> Robert
> 
> Yes I have measured it and timing jitter is a real problem using windows to key a COM port line. The logging software has nothing to do with it. EXTFSK and others which include 2Tone make use of a timing register in Windows and that is accurate in the long term. What is not accurate is that Windows needs to give EXTFSK and others a slice of the CPU at the right time. It is this that has the timing jitter as often Windows will give another thread higher priority and will delay giving CPU time to EXTFSK.
> 
> Further the driver that is used to drive the COM port will also have the same timing issue so overall what happens is the COM port is not accurately timed so some bits are longer and other shorter than the 22mS required for 45.45 baud RTTY.
> 
> The best way of getting the best result at the decode end of the link is to use sound card based system as the timing comes from using a sound card as the timing for that comes from the crystal oscillator on the sound card. This timing is independent of windows timing.
> 
> 73 David G3YYD
> 
> -----Original Message-----
> From: Robert [mailto:woodr90 at gmail.com]
> Sent: 26 February 2018 18:20
> To: 'David G3YYD'
> Subject: RE: [RTTY] UPDATE Re: FSK Keying
> 
> I use rttyrite to transmit with and have been told several times - it's EXT FSK and rttyrite have it fixed right
> 
> Have you decoded or scoped such and seen issue (understand the other you're saying / MMTTY but...)
> 
> 73 Robert W5AJ
> 
> -----Original Message-----
> From: RTTY [mailto:rtty-bounces at contesting.com] On Behalf Of David G3YYD via RTTY
> Sent: Monday, February 26, 2018 3:55 AM
> To: rtty at contesting.com
> Subject: [RTTY] UPDATE Re: FSK Keying
> 
> Tim
> 
> AFSK will outperform serial port FSK at the receive end of the link. No doubt about it because Windows is not and never will be a real time operating system.
> 
> Windows internal timing causes the transitions between mark/space COM port line to jitter in time from the 22mS per bit timing required for optimum decode. As a result this increases the error rate of the decoder.
> 
> With AFSK the timing is set by the crystal clock that is driving the sound card DAC( Digital to Analogue Converter) and therefore has no timing jitter.
> The decoder error rate will be as low as it can be.
> 
> Further most radios FSK keying is unfiltered creating a very wide transmitted signal that will cause avoidable QRM to adjacent frequencies.
> Using AFSK the filtering of the signal (in MMTTY the default should not be used - TX filtering should be set to 512 taps) means that the TX will take up a lot less bandwidth.
> 
> Better still is to use 2Tone DOOK (a version of AFSK) that is very narrow at <350Hz at -80dB. Fldigi uses a similar scheme and while not quite as narrow as 2Tone DOOK is very similar and narrower than MMTTY AFSK.
> 
> By the way do you think that RTTY is FSK? Many do because of what has been written in the past. But look at how the decoders work and then match the transmission to the decoder. All decoders treat the signal as two tone on different frequencies. They use a narrow filter on each tone frequency - in the case of 2Tone they are just 45.45Hz wide at -6dB. Then they combine/compare the amplitude of those tones to decide if it is a mark or space tone that was transmitted. Note this means each tone is treated as a on/off keyed signal. In other words RTTY is really ASK (Amplitude Shift
> Keying) or if you like OOK (on-Off Keying) So it is not FSK. Or as per 2Tone Differential OOK (DOOK) as the tone are differentially keyed one going off as the other comes on.
> 
> The best transmission system to use is the one that matches the decoder.
> 2Tone and FLdigi do this.
> 
> 73 David G3YYD
> 
> -----Original Message-----
> From: RTTY [mailto:rtty-bounces at contesting.com] On Behalf Of groupsrichart at gmail.com
> Sent: 26 February 2018 03:29
> To: KenwoodTS-590 at yahoogroups.com; rtty at contesting.com; TS-590 at groups.io; TS-590S at yahoogroups.com
> Subject: [RTTY] UPDATE Re: FSK Keying of TS-590SG
> 
> Just to give everyone who has commented a little update:
> 
> 1. I located another serial card and installed it in my computer. It gives me a full -/+12VDC swing on the data lines.
> 2. The optoisolators in my interface are not pulling the PTT and FSK inputs of the rig low enough to work. When the optoisolator is turned on, I am still seeing +0.54VDC on the ACC2 inputs to the rig.
> 3. I tried different values of current limiting resistors from 100Ω to 1.5KΩ on the inputs to the optoisolators with no change.
> 4. I checked my junk box for some NPN transistors and found none (I could have sworn I had some 2N2222s), so I have a few on order for the coming week. I plan to build the simple resistor/transistor interface that many have suggested.
> 5. After I get FSK working, I do plan to also try AFSK. There seems to be a spirited debate among many as to which is best. I have no doubt that either method properly used can give good results so I want to find out operationally which method I prefer.
> 
> My thanks to everyone who has taken the time to respond with your comments and suggestions. I found them all very helpful.
> 
> 73,
> Tim WS4V
> 
> On 2/23/2018 2:49 PM, groupsrichart at gmail.com wrote:
>>
>> I would like to hear from anyone who is successfully keying the FSK
>> and PTT lines via the ACC2 connector on the rear panel of the
>> TS-590SG. What serial adapter and what hardware interface are you
>> using to do so?
>>
>> I have tried two different interfaces using optoisolators to attempt
>> FSK and PTT keying from a serial card to the rear panel ACC2
>> connector. Both attempts have failed. I am using MMTTY software for
>> keying and I do get the requisite signal change on TXD pin 3 (FSK) and
>> RTS pin 7 (PTT) lines. I have also verified proper setup of the '590
>> by manually grounding the inputs PKS pin 9 (PTT) and RTTY pin 2 (FSK)
>> on the ACC2 connector.
>>
>> However, I *think* the problem lies with the serial card not providing
>> enough voltage/current for the optoisolators to completely turn on and
>> provide a hard ground to the ACC2 inputs. I have tried two different
>> serial cards and one swings from -4.9 to +5.0 VDC on the data lines
>> and the other card swings from -3.8 to +3.9 VDC. The cards I am using
>> are a two port generic PCI Express SI-PEX15037 and a four port Perle
>> Fast 4 PCI card. Does anyone know of a serial card that has a full -12
>> to +12 VDC swing?
>>
>> One of the hardware interfaces has 4N25 optoisolators and the other
>> uses Fairchild FOD852 optoisolators. There are 1K series resistors in
>> the input line to the optoisolators and I have tried various values
>> (100, 220 & 470Ω) with limited success but nothing consistent.
>>
>> While my objective was to keep the rig isolated from the computer, I
>> am thinking about building an interface that uses NPN transistors for
>> keying. Is anyone using NPN transistors to key their '590? Are you
>> tying the ground from pin 5 of the RS-232 port to the pin 4, 8 or 12
>> ground of the ACC2?
>>
>> Any input is appreciated...I have been off RTTY since purchasing this
>> rig in November and I want to get back on the bands.
>>
>> Thanks,
>>
>> Tim WS4V
>>
> 
> _______________________________________________
> RTTY mailing list
> RTTY at contesting.com
> http://lists.contesting.com/mailman/listinfo/rtty
> 
> _______________________________________________
> RTTY mailing list
> RTTY at contesting.com
> http://lists.contesting.com/mailman/listinfo/rtty
> 
> 
> _______________________________________________
> RTTY mailing list
> RTTY at contesting.com
> http://lists.contesting.com/mailman/listinfo/rtty
>