eHam.net - Amateur Radio (Ham Radio) Community

Call Search
     

New to Ham Radio?
My Profile

Community
Articles
Forums
News
Reviews
Friends Remembered
Strays
Survey Question

Operating
Contesting
DX Cluster Spots
Propagation

Resources
Calendar
Classifieds
Ham Exams
Ham Links
List Archives
News Articles
Product Reviews
QSL Managers

Site Info
eHam Help (FAQ)
Support the site
The eHam Team
Advertising Info
Vision Statement
About eHam.net



[Articles Home]  [Add Article]  

Making the Most of your PSK31 Signal

KF6VSG (KF6VSG) on July 27, 2003
View comments about this article!

Making the Most of your PSK31 Signal

I've been hooked on PSK31 every since I gave it try several years ago. Its one great little mode if you like doing things the digital way, enjoy keyboard-to-keyboard QSOs, and knowing that 5 watts can take you anywhere in the world. I've enjoyed both contesting as well as evenings with one long QSO getting to know how life is for the other guy.

My problem was always pausing to "re-tune" my signal. Not the transmitter, but the audio level setting of the computer sound card. At first I was tempted to crank up the Windows audio mixer's sliders to the max, in order to get the most out of my transmitter. I found out from the signal reports I was getting that wasn't such a good idea--overdriving the rig is going to create a ton of splatter and a 31 Hz wide signal can be hundreds of hertz, stepping on other QSOs. I brought my scope home from work and looked at the RF output when I was overdriving and it was a classic case of clipping a sine wave.

The solution is to keep those sliders down, but how far down? To little audio and little (or nothing) comes out of the rig. So my big expensive HP scope sat with my rig where I could watch the RF output and tweak the sound mixer sliders until I got big beautiful sine envelopes with no clipping. It also meant I could compute my average power output (P = E**/2R, remember?), and of course I got good signal reports.

The only downside was that my scope was being tied up for this single purpose. So I got the idea of replacing the scope with a PIC microprocessor. Many of the PICs have built-in ADCs, although nothing fast enough to sample a 14 MHz signal. But I was interested in knowing only the shape of the envelope, which means I just have to rectify the signal and go through a low pass RC filter on the way to the ADC. I then programmed the firmware on the PIC to make exactly 64 samples of the RF envelop at 1 msec intervals--exactly two PSK31 bits. This buffer is then uploaded to the PC (and if I'm running PSK31, I'm guaranteed I'll have one of those around!) via the serial port. From there I wrote software to compute the Fourier transform, analyze the harmonics, and calculate the IMD (intermodulation distortion). At first the application would just show the wave form and advise me to increase or decrease the sound card's audio to provide perfect modulation. I was happy to oblige the program, but then it hit me--my program has access to the sound card through the Windows API, the application can tune up the audio for me. When I put feedback into the program, it worked. My PSK31 signal was perpetually perfect without my having to fuss with anything.

The circuit for this little psk meter looks like this:

Its pretty simple: D1 and D2 provide the rectification, and R3 + C2 provide the filtering. The PIC uses a couple of 2N2202 NPN transistor to drive the RS-232 connection to the computer. Although using transistors to switch between +5V and ground is not technically RS-232 (which likes switching between +12 and -12), it works just fine (and is much less expensive than using a MAXIM interface chip).

The hex file with the firmware is at this link: http://www.ssiserver.com/info/pskmeter/pskmeter.hex so you can build the circuit and program the chip, and that's the hardware part of it.

The Windows application looks like this:

The green part is the RF envelope, and the yellow line is the ideal PSK31 sine wave. You want your signal to match the yellow curve as closely as possible. In the screenshot above, its a pretty good match. The software tells me the RMS deviation from a perfect signal, and more importantly tells me the IMD (distortion) in dB. As you can see, this signal will be received on the other end with an IMD of -28 dB which is pretty darn good. The display also shows the percent of full scale audio from the sound card used to achieve this result, and the peak voltage going into the ADC. You can download the program at this link: http://www.ssiserver.com/info/pskmeter/pskmeter.exe

I ended up giving a talk to some of the local radio clubs and then at Pacificon 2002 and hams kept asking me if I would kit the meter. I was reluctant to do it because I am used to breadboarding everything I need to make, but a kit involved a lot more: designing and fabricating a printed circuit board, quality control, an assembly manual, and so on. What convinced me to try it was the amazing software at http://www.expresspcb.com. Their free software lets me design circuits and draw out the printed circuit board. A click of button and the files are uploaded to their production facilty and two days later the boards arrive in the mail. It's very much like sending a file to your printer, except that the printer is in another city. Very cool.

After several iterations and revisions, here was the board now looks like:

I have a lot more information on PSK31 theory, firmware source code, application software, and all that stuff at http://www.ssiserver.com/info/pskmeter. So far hams who have tried it seem to really like being able to watch their signal and knowing their signal is the best it can be. If you have questions, I'd be happy to try to answer them.

Member Comments:
This article has expired. No more comments may be added.
 
Making the Most of your PSK31 Signal  
by VK2GWK on July 27, 2003 Mail this to a friend!

Looks like a very nice project. But as you are used to breadboarding you may understand that I would like to copy the whole thing without ordering the PC board. Can you tell me which processor? Looks like the 16F872.... This essential piece of information is missing....
Is the source of the software available? There are a few things I might like to change to suit my own situation.
 
RE: Making the Most of your PSK31 Signal  
by W8JI on July 27, 2003 Mail this to a friend!
>>The software tells me the RMS deviation from a perfect signal, and more importantly tells me the IMD (distortion) in dB. As you can see, this signal will be received on the other end with an IMD of -28 dB which is pretty darn good.<<<

Perhaps this is one reason why PSK operators complain so much about the power level of other operators, and want everyone to use wet noodle antennas or the transmitter power equivalent?

-28dB IMD doesn't sound very impressive. It is 1950's IMD.

It is horrible (in terms of communications performance) if we want to crowd multiple stations together that have more than several dB difference in signal level.

Aren't there any digital systems where total IM is 40 or more dB down, and where the receiver does filtering long before going into a computer sound card?

Then a mode could truly be narrow, and would have dynamic range like CW or SSB systems offer. It would be compatable with operation near strong signals, weak signals, and even other modes.

Good dynamic range would be a solution to being "narrow but dirty". I wonder if anyone is working on clean generation of PSK signals?

73 Tom
 
Making the Most of your PSK31 Signal  
by KG6OYK on July 27, 2003 Mail this to a friend!
Excellent article! I have really enjoyed PSK31 over the past few months and was just yesterday having a local 2mSSB PSK QSO with some friends to test and tweak each other’s settings for clean modulation. I have noticed that I have to adjust my audio card output frequently, especially if I change where I am on the waterfall due to the audio pass filters in my rig. Also, move too far up or down the 5K waterfall and you will get nothing out – the rig wants to contain your signal within ~2.8K.

My method to date has been to increase the audio out into my rig until the ALC meter just moves (or the first segment in my display flickers as is in my case), and back it off a little. I notice that when it is set this way, my reports are good and the power setting on the rig matches or closely approximates what I am putting out. If I am under modulating, I might only see 10 watts out when I have it set to 20. When I am over modulating, my ALC meter is generally moving. This method is based on reading the MixW instructions and my limited experience, but seems to work reasonably well.

Unfortunately, this is not a “set and forgot” mode – if I jump around the waterfall (keeping in mind the cutoffs around my 2.8K pass band) responding to CQ’s, I have to re-adjust the output to get my modulation in line.

So what serendipity when I saw your article this morning! I really enjoyed the PDF file on your website of the presentation on this project (http://www.ssiserver.com/info/pskmeter/pskmeter.pdf) – the detail was exactly what I was looking for! I will be building a kit real soon!

Again, Thanx for the excellent and timely (at least in my case) info – see you on the waterfall!

Steve

KG6OYK
 
Making the Most of your PSK31 Signal  
by N0TONE on July 27, 2003 Mail this to a friend!
I agree with W8JI. -28dB IMD is terrible. If your intended signal is 10dB over S9 into someone's receiver, and your IMD is only 28dB down, then your splatter is S5-S6 into someone else's receiver, assuming similar propagation. S5 is a high level of QRM to someone trying to work a weak signal. This is unacceptable IMD.

What I would like to know is, just how accurate is your meter? In other words, if you fed it a good clean signal, with IMD down 40 to 50dB as it should be, would the meter read 40 to 50dB down, or is 28dB the cleanest it can read? Also, how are you measuring IMD? Are you performing the complex division of the measured signal by the theoretically perfect signal and then doing FFT, or are you just FFTing the detected envelope, and looking at spectral components outside the intended PSK31 envelope? If you're doing the latter, then you're missing the in-band IMD, and the actual IMD generated is worse than 28dB.

The envelope picture you showed makes it look as if there's maybe some pretty serious sampling granularity happening, which would limit your ability to measure true IMD.

In your FFT, what did you use as a window? Possibly the high IMD you're reading is caused by aliasing on the window edges, or spectral overflow from one bin to another?

This is a great start - we need the PSK31 operators to start thinking about how clean their signal is, because PSK31, when operated close to CW stations, really messes up the CW because operators are generally set all wrong. Most PSK signals I hear are many kHz wide, and there's no excuse for that.

Now that you have the start of a good system, it shouldn't take much more effort to get it helping to create proper PSK31 signals. 28dB down is better than most PSK ops achieve, but it's still too high to avoid QRMing others when you don't have to.

am
 
Keeping it Simple  
by W7KID on July 27, 2003 Mail this to a friend!
Ever since getting into PSK I've been hooked. For the most part, at least compared to how many people are out there sometimes, people keep their signals under control. However, at least once a week I'll see someone in there with a signal that's at least 4-5 times as wide as it should be. The biggest problem is overdriving the signal. That's very easy to keep under control. Just make sure you can see the ALC level on your rig, and either adjust the audio output from your computer (if you're using a fixed af in on your rig) or adjust the mic gain (if using the mic plug for the af in) and turn it down until you just barely see any deviation. About the only other problem I've seen with signals is people pumping out too much power. It takes very little power to get to anywhere you want to go. I've never run over 50 watts even in the noisiest conditions.

Another note for anyone wanting to get into PSK. You don't need to rush out and buy a $100 or even a $50 interface. For most rigs you can spend about $10 at RS and make your own interface. I've had about 10 different rigs that I've made interfaces for. Never spending more than $15, and that was for my current 847 that I also incorporated the CAT interface into the box as well.

73s Ken
W7KID
 
RE: Making the Most of your PSK31 Signal  
by W8JI on July 27, 2003 Mail this to a friend!
>tweak each other’s settings for clean modulation. I have noticed that I have to adjust my audio card output frequently, especially if I change where I am on the waterfall due to the audio pass filters in my rig. Also, move too far up or down the 5K waterfall and you will get nothing out – the rig wants to contain your signal within ~2.8K.<<

There is a second problem with PSK that is almost always ignored. Harmonic distortion will cause harmonics of the audio tone to appear in RF output.

Even if IM distortion is acceptable, harmonic distortion can be present and causing QRM. It will not show on IM displays.

The best way to avoid harmonic distortion problems is to operate only with tones above 1.5kHz, then any harmonics generated in the soundcard, audio, or modulator system (harmful harmonics can't occur in anything after the signal is converted to RF from audio) fall outside the passband of the SSB filter.

Harmonic distortion is a big problem, because it doesn't show on IM displays and it cause a problem at multiples of the audio frequency + or - dial display (depending on LSB or USB.

For example, a transmitter on 10.000 MHz USB with a 500Hz tone could have audio harmonic spurs on 10.001, 10.0015, 10.002, 10.0025 and/or 10.003.

If the tone was moved to 1500Hz, only one spur would appear and it would be at the slope of the SSB filter on 10.003.

Using a narrow CW filter for receiving, if your rig allows it, will eliminate most of the "he is running too much power" complaints. Receivers are not designed to nor are they good at allowing a window of twenty stations through, and then have them sorted out in (of all things) a sound card.

73 Tom
 
Making the Most of your PSK31 Signal  
by N7XB on July 28, 2003 Mail this to a friend!
Good article. I have logged more than 1000 psk31 qso's since January 2002 (WAS PSK31, BARTG PSK-40, 54 confirmed DX). I have used the method discussed elsewhere here: adjust the ALC control (easy on the FT-920) to where it just barely tickles the meter. I've never had a complaint about overdriven signals.

Two major problems with new psk31 users:

1. Signal from sound card is so overdriven that we can copy you Here . . . and Here . . . . and Here . . . across the waterfall.

2. UPPERCASE USE - psk31 was written with an algorithm that decodes lower case better and faster than upper case. Perfectly OK to type in "normal typing mode." But you can usually spot the RTTY convert to psk31 BECAUSE EVERY CHARACTER IS TYPED UPPERCASE, WHICH IRONCIALLY IS HARDER FOR PSK31 TO DECODE.

73 de Bruce N7XB
 
RE: Making the Most of your PSK31 Signal  
by W4EF on July 28, 2003 Mail this to a friend!
W8JI wrote <<<<<<<<<<<<Perhaps this is one reason why PSK operators complain so much about the power level of other operators, and want everyone to use wet noodle antennas or the transmitter power equivalent?

-28dB IMD doesn't sound very impressive. It is 1950's IMD.

Aren't there any digital systems where total IM is 40 or more dB down, and where the receiver does filtering long before going into a computer sound card? >>>>>>>>>>>>>>

With the new DSP rigs like the Kenwood TS-870, Ten-Tec Orion, and Icom 756 PRO, etc, it should be relatively easy to implement direct digital modulation in the transmitter IF without the need to pass audio tones from a sound card into a mic jack. This would solve all the problems of hum and harmonic distortion in the TX audio chain, as the TX audio chain would be eliminated. Combine this with a class-A PA mode like that found in the FT-1000MP MK-V and you could have 100 watt PSK31 with 40 to 50 dB IM performance.

If you took it a step further and used a constant envelope modulation technique like some of those developed for the satellite and cellular industries instead of shaped symbol BPSK which has lots of AM on it, you could pass the RF thru high-efficiency non-linear PAs with minimal spectral broadening which would allow for clean 1500 watt HF digital transmission.

On the receive side, it should be possible to do matched filter detection and carrier tracking at IF using the DSP in the modern rigs. Of course you may still have dynamic range limitations unless you use narrow crystal roofing filters or really expensive A-D converters with lots of bits.

73 de Mike, W4EF...........................
 
RE: Making the Most of your PSK31 Signal  
by N5ZVP on July 28, 2003 Mail this to a friend!
Utilizing a high end radio's DSP would be nice, but that doesn't help for older rigs.

Something that would interesting to try would be a 16 bit DSP like the Motorola 56F805, which has a faster A/D conversion time and more horsepower than the PIC.

Chris
WA5TT
 
RE: Making the Most of your PSK31 Signal  
by W8JI on July 29, 2003 Mail this to a friend!
The problem is two-fold.

First, the generation of the PSK signal depends on good IM perfromance with a slow amplitude repetition. The problem is most radios are already very poor at high frequencies, and get worse at low frequencies because filtering and bypass capacitors time constants are far above the repeating load variations.

Generating something in a DSP won't help if the PA system has -25 to -35 IM, which is what most radios have.

The second factor is the receiver. Virtually all receivers have very limited performance when the filtering is all done after the audio stages. That means every single stage from the antenna to the audio output can degrade or limit adjacent channel interference.

Audio generated and audio detected digital modes will never work in the mainstream unless they are designed to be independent of transmitter and receiver defects. We can't even get manufacturers to eliminate keyclicks, or improve the poor SSB IM performance. Good luck on having them do all that, plus make receiver filters work at 100Hz or less.

Just something to consider if a digital mode is going to replace CW as a poor propagation mode.

73 Tom
 
-28 IM is better than 90% of the operators out the  
by KF6IIU on July 29, 2003 Mail this to a friend!
Don't forget the IM figure in George's software includes all IM introduced by his hardware and software. (Someone with more theory than me can probably calculate the IM introduced by diode clipping alone.) Still, I'll bet his signal is cleaner than 90% of the operators out there, and the device is a cool idea.

There are still lots of ops cranking the sound card up to 11......

I occasionally use an FT-101E for PSK - I leave the mic gain at the same setting I normally use for a mic, turn compression off of course, attenuate the signal with a SignaLink, and use the Windows volume sliders to produce 20W power. The sliders are generally between 1 1/2 and 2 1/2.
 
RE: Making the Most of your PSK31 Signal  
by W4EF on July 29, 2003 Mail this to a friend!
W8JI wrote >>>>Generating something in a DSP won't help if the PA system has -25 to -35 IM, which is what most radios have. <<<<<<<

That's why I mentioned the FT-1000MP MK-V in Class A mode. Seems like that is the only radio on the market that is addressing the IM issue. Of course, I don't know how well its IMD holds up at very small tone spacings. Have you tested it at close spacings? I've heard people say that the TS-830S is good, but don't imagine you will find any DSP in there.

>>Just something to consider if a digital mode is going to replace CW as a poor propagation mode.>>

PSK mostly finds application in the satcom world where most of the signals are close to the noise floor. When I worked for Scientific Atlanta, I did a dynamic range test on one of our BPSK modems. If the QRMer was inside the modem's SAW roofing filter, the best you could hope for was about 20dB to 25dB of headroom between an adjacent channel QRMer and the desired signal.


<<<Audio generated and audio detected digital modes will never work in the mainstream unless they are designed to be independent of transmitter and receiver defects. We can't even get manufacturers to eliminate keyclicks, or improve the poor SSB IM performance. Good luck on having them do all that, plus make receiver filters work at 100Hz or less. >>>>>>

Although all of these problems are readily solvable, I am not holding my breath. With a few notable exceptions, all the engineering effort these days seeems to be going into bells and whistles (memories, DSP filters, flourescent displays, etc). The Kenwood TS-2000 is a good example of this. Unless the bulk of the amateur customer base becomes educated enough to demand higher performance analog sections, this won't change much. Wish I had more spare time to mess with this stuff. Unfortunately it takes a lot more effort to come up with working hardware than it does dreaming up good ideas.

Later,

Mike, W4EF..................
 
RE: Making the Most of your PSK31 Signal  
by VK3HE on July 30, 2003 Mail this to a friend!
A FeedForward amplifier as used in the cellular phone industry will achieve a 3rd order IMD figure probably better than class A. Typically in a feedforward amp the 3rd order products are down -60 db with high efficiency, the 5th order products are almost non existent.

Looking at the current batch of radios i doubt there is a will amongst manufacturers to do anything about the shocking 3rd order IMD levels. Like Key clicks the 3rd order IMD issue is of concern.


Craig


 
RE: Making the Most of your PSK31 Signal  
by W8JI on July 31, 2003 Mail this to a friend!
That's because too many of us are willing to accept radios that are poorly made, and "pretend" things are as good as they should be.

It would be nice to see a push towards truly improved technology, increasing dynamic range of very close-spaced signals.

If you look at radio reviews, they all have 4-5 ratings. Even the very poor radios with problems. People who send radios back for refunds still give the radios an avarage rating or higher!!

Not much chance things will improve when we tell everyone everything is great!

73 Tom

 
Making the Most of your PSK31 Signal  
by W9SZ on August 1, 2003 Mail this to a friend!
Looks like a neat little circuit. Someone guessed it's a PIC 16F872, my guess was a 16F876 (more memory). At any rate, are you willing to share your source listing (.ASM file)? That way some of us could play around with mods in the firmware if we want to.
 
RE: Making the Most of your PSK31 Signal  
by W4TME on August 10, 2003 Mail this to a friend!
>>Looks like a neat little circuit. Someone guessed it's a PIC 16F872, my guess was a 16F876 (more memory).<<

I just finished the construction of the kit yesterday and it is a 16F876.

It works great. I used to clean up my signal, which was fairly good to begin with. It takes some playing with the software to get max power at lowest IMD. I eventually put it in manual mode and adjusted the signal levels (PC & interface) manually. Because of the lack calibration and other items noted above, I don't use it as a highly precise diagnostic instrument. Just tune for lowest IMD and don't worry too much about the absolute IMD value.

I recommend getting one if you use PSK often. The price is right and can be constructed in less than an hour. Documentation is excellent. Best I have ever seen.

As noted above I think it would be cool to allow folks to work with the firmware to see what evolves. A commercial version and a "community supported" version. Thanks for the neat kit.


73 de W4TME
-Tim
 
Email Subscription
You are not subscribed to discussions on this article.

Subscribe!
My Subscriptions
Subscriptions Help

Other How To Articles
Possible Cure for Relay Contact Cancer?