EMRFD Message Archive 4674

Message Date From Subject
4674 2010-05-11 11:32:05 Corey Minyard Questions about an SDR PSK-31 rig for a balloon beacon
I've been tasked with obtaining a rig for doing PSK-31 telemetry
(position, temperature) and perhaps some control functions for a
balloon. I've looked around and nothing available seems terribly
suitable. The closest I found was an ATS-3B or ATS-4B with some
modifications to the software, but that won't receive without an
external piece of hardware, plus, though I have an ATS-3B, I don't
really want to have the possibility of having it destroyed, as they are
hard to come by.

There are some PSK-31 kits, but they require an external PC, and power
and weight become an issue, as they use class A or AB amplifiers and are
not built for low power. I could build something instead of the PC, so
why not just build the whole thing?

So I've been looking at what it would take to build something like
this. I'd like something that's flexible, so I'm looking at doing an
SDR rig to do this, but I have some concerns.

My biggest concern has to do with A/D and D/A quantization noise. I'd
like to use one of the low-power TI microcontrollers with built-in
12-bit converters. That gives me about a 74dB signal-to-quantization
noise ratio, I believe, which seems a little marginal to me. I think I
can manage well over 1 million MAC operations/sec with the hardware
multiplier/accumulator, so that should be ok for the low sampling rate I
will need (probably 10k samples/sec each for the I and Q channels). I
haven't been able to find another low-power microcontroller that seems
better. Cirrius Logic makes an audio SOC that would be massive overkill
(24 bit A/D and 150MOPS processor), though it would draw a lot more
power (probably a hundred times more!) and it's more expensive all
around. I can't find much in between without going to separate chips;
if I could do it all on the TI uC that would be very nice.

I plan on using a Tayloe mixer for receive and transmit and a 4th order
active filter both driving a mixer and another one from the receiving
mixer. It looks like I'll need about 60dB gain on the receive side. I
plan on adding digital potentiometers for input volume control and
output power control.

For transmit, I'm hoping to use a class E amplifier for power
efficiency, since that's where most of the power will go, anyway. Since
a class E transmitter is not linear, my plan was to compensate in the
DSP with leveling filters. Have an automated procedure to first measure
receive non-linearities and calculate a leveling filter for the receive
side, then put the transmitter (at minimum power) into a closed loop
with the receiver with a lot of attenuation, sweep the frequencies, then
calculate a transmit filter to compensate for the nonlinearities of the
active filter and the class E amp. This should also let me make more
aggressive active filters; the compensation should help with
nonlinearities in the passband of the filters. This also makes the
12-bit range of the converters a little more of an issue, though.

I'd eventually like to end up with something that's flexible for
different purposes, perhaps even for use as a general purpose rig that
can do SSB and CW, too.

Does this sound reasonable? I'd like to get some feedback before I move on.

Thanks

-corey - AE5KM
4675 2010-05-11 12:08:42 Alois Stöckl Re: ***SPAM*** [emrfd] Questions about an SDR PSK-31 rig for a ballo
Hello Corey,

sorry poor english.

Balloon TX
down on earth RX

the simplest way to make a psk-signal
shift an xtal !
Programm a Mikrocontroller with PSK-31 and spend your oszillator puffer an PA
for a lightwight rig for your balloon.
Spend th

72 de alois dl8ram


> I've been tasked with obtaining a rig for doing PSK-31 telemetry
> (position, temperature) and perhaps some control functions for a
> balloon. I've looked around and nothing available seems terribly
> suitable. The closest I found was an ATS-3B or ATS-4B with some
> modifications to the software, but that won't receive without an
> external piece of hardware, plus, though I have an ATS-3B, I don't
> really want to have the possibility of having it destroyed, as they are
> hard to come by.
>
> There are some PSK-31 kits, but they require an external PC, and power
> and weight become an issue, as they use class A or AB amplifiers and are
> not built for low power. I could build something instead of the PC, so
> why not just build the whole thing?
>
> So I've been looking at what it would take to build something like
> this. I'd like something that's flexible, so I'm looking at doing an
> SDR rig to do this, but I have some concerns.
>
> My biggest concern has to do with A/D and D/A quantization noise. I'd
> like to use one of the low-power TI microcontrollers with built-in
> 12-bit converters. That gives me about a 74dB signal-to-quantization
> noise ratio, I believe, which seems a little marginal to me. I think I
> can manage well over 1 million MAC operations/sec with the hardware
> multiplier/accumulator, so that should be ok for the low sampling rate I
> will need (probably 10k samples/sec each for the I and Q channels). I
> haven't been able to find another low-power microcontroller that seems
> better. Cirrius Logic makes an audio SOC that would be massive overkill
> (24 bit A/D and 150MOPS processor), though it would draw a lot more
> power (probably a hundred times more!) and it's more expensive all
> around. I can't find much in between without going to separate chips;
> if I could do it all on the TI uC that would be very nice.
>
> I plan on using a Tayloe mixer for receive and transmit and a 4th order
> active filter both driving a mixer and another one from the receiving
> mixer. It looks like I'll need about 60dB gain on the receive side. I
> plan on adding digital potentiometers for input volume control and
> output power control.
>
> For transmit, I'm hoping to use a class E amplifier for power
> efficiency, since that's where most of the power will go, anyway. Since
> a class E transmitter is not linear, my plan was to compensate in the
> DSP with leveling filters. Have an automated procedure to first measure
> receive non-linearities and calculate a leveling filter for the receive
> side, then put the transmitter (at minimum power) into a closed loop
> with the receiver with a lot of attenuation, sweep the frequencies, then
> calculate a transmit filter to compensate for the nonlinearities of the
> active filter and the class E amp. This should also let me make more
> aggressive active filters; the compensation should help with
> nonlinearities in the passband of the filters. This also makes the
> 12-bit range of the converters a little more of an issue, though.
>
> I'd eventually like to end up with something that's flexible for
> different purposes, perhaps even for use as a general purpose rig that
> can do SSB and CW, too.
>
> Does this sound reasonable? I'd like to get some feedback before I move
> on.
>
> Thanks
>
> -corey - AE5KM
4676 2010-05-11 14:59:47 Corey Minyard Re: Questions about an SDR PSK-31 rig for a balloon beacon
Alois Stöckl wrote:
> Hello Corey,
>
> sorry poor english.
>
> Balloon TX
> down on earth RX
>
> the simplest way to make a psk-signal
> shift an xtal !
> Programm a Mikrocontroller with PSK-31 and spend your oszillator puffer an PA
> for a lightwight rig for your balloon.
> Spend th
>
> 72 de alois dl8ram
>
Poor english is ok, though I might have to work a little to understand.

I understand you correctly, you are saying to use some method to
directly phase-shift the signal. This works ok for PSK-31, and you
could adapt something like
http://kd1jv.qrpradio.com/PSKTX/SIMPLEPSKTX.HTM though you could
directly drive the XOR to modulate the signal from the uC and not have
to deal with all the phase change detection and power detection, which
is about half that circuit.

Looking at the receive side, I don't think the circuit there rejects the
opposite sideband, so that's a problem, though perhaps not a huge one.

However, it does not give the flexibility to do anything but PSK-31
(well, and CW). I would like the flexibility to do different digital
modes and perhaps SSB. Getting a quadrature signal makes those things
possible.

-corey - AE5KM
>
>
>> I've been tasked with obtaining a rig for doing PSK-31 telemetry
>> (position, temperature) and perhaps some control functions for a
>> balloon. I've looked around and nothing available seems terribly
>> suitable. The closest I found was an ATS-3B or ATS-4B with some
>> modifications to the software, but that won't receive without an
>> external piece of hardware, plus, though I have an ATS-3B, I don't
>> really want to have the possibility of having it destroyed, as they are
>> hard to come by.
>>
>> There are some PSK-31 kits, but they require an external PC, and power
>> and weight become an issue, as they use class A or AB amplifiers and are
>> not built for low power. I could build something instead of the PC, so
>> why not just build the whole thing?
>>
>> So I've been looking at what it would take to build something like
>> this. I'd like something that's flexible, so I'm looking at doing an
>> SDR rig to do this, but I have some concerns.
>>
>> My biggest concern has to do with A/D and D/A quantization noise. I'd
>> like to use one of the low-power TI microcontrollers with built-in
>> 12-bit converters. That gives me about a 74dB signal-to-quantization
>> noise ratio, I believe, which seems a little marginal to me. I think I
>> can manage well over 1 million MAC operations/sec with the hardware
>> multiplier/accumulator, so that should be ok for the low sampling rate I
>> will need (probably 10k samples/sec each for the I and Q channels). I
>> haven't been able to find another low-power microcontroller that seems
>> better. Cirrius Logic makes an audio SOC that would be massive overkill
>> (24 bit A/D and 150MOPS processor), though it would draw a lot more
>> power (probably a hundred times more!) and it's more expensive all
>> around. I can't find much in between without going to separate chips;
>> if I could do it all on the TI uC that would be very nice.
>>
>> I plan on using a Tayloe mixer for receive and transmit and a 4th order
>> active filter both driving a mixer and another one from the receiving
>> mixer. It looks like I'll need about 60dB gain on the receive side. I
>> plan on adding digital potentiometers for input volume control and
>> output power control.
>>
>> For transmit, I'm hoping to use a class E amplifier for power
>> efficiency, since that's where most of the power will go, anyway. Since
>> a class E transmitter is not linear, my plan was to compensate in the
>> DSP with leveling filters. Have an automated procedure to first measure
>> receive non-linearities and calculate a leveling filter for the receive
>> side, then put the transmitter (at minimum power) into a closed loop
>> with the receiver with a lot of attenuation, sweep the frequencies, then
>> calculate a transmit filter to compensate for the nonlinearities of the
>> active filter and the class E amp. This should also let me make more
>> aggressive active filters; the compensation should help with
>> nonlinearities in the passband of the filters. This also makes the
>> 12-bit range of the converters a little more of an issue, though.
>>
>> I'd eventually like to end up with something that's flexible for
>> different purposes, perhaps even for use as a general purpose rig that
>> can do SSB and CW, too.
>>
>> Does this sound reasonable? I'd like to get some feedback before I move
>> on.
>>
>> Thanks
>>
>> -corey - AE5KM
>>
>
>
4678 2010-05-12 07:37:25 Vojtech Re: Questions about an SDR PSK-31 rig for a balloon beacon
HI Corey.

I am not quite getting it. You need a TX for your balloon, then you need a separate RX to decode it. For RX you could use anything. For TX I suppose you are limited by weight and power consumption.

The simplest PSK31 TX could be built with a crystal oscillator, limiting amplifier, XOR gate for flipping phase of the RF wave and AM modulated Class-E PA. You can do that by combining Steve Weber's unique simple PSK31 TRX and his ATS-3 TX part.

Interestingly, one can achieve 24dB IMD with Steve's ATS-3b CW shaping circuit by modifying the ramp up/down slope and playing with the timing of the trapezoid timing. I spent some weekends on theoretic derivation of the formulas, which I plan to publish in QEX one day.

In regard to your 12bit ADC. If the input signal is already frequency limited and you have some AGC in place, 12bit may be ok. 50dB dynamic range for CW or PSK31 signal centered around 600Hz seems adequate to me. You may get more bits with oversampling. If you have a frequency limited signal with some random noise, you get additional bit of resolution by 4x oversampling. For example, in my standalone CW/PSK decoder I am samping at 40ksamples / second and summing each 8 samples to get 5ksamples / second. I am gaining >6dB with this process.

73, Vojtech AB2ZA
4681 2010-05-12 14:28:06 Corey Minyard Re: Questions about an SDR PSK-31 rig for a balloon beacon
Vojtech wrote:
> HI Corey.
>
> I am not quite getting it. You need a TX for your balloon, then you need a separate RX to decode it. For RX you could use anything. For TX I suppose you are limited by weight and power consumption.
>
I was looking for RX on the balloon.

> The simplest PSK31 TX could be built with a crystal oscillator, limiting amplifier, XOR gate for flipping phase of the RF wave and AM modulated Class-E PA. You can do that by combining Steve Weber's unique simple PSK31 TRX and his ATS-3 TX part.
>
> Interestingly, one can achieve 24dB IMD with Steve's ATS-3b CW shaping circuit by modifying the ramp up/down slope and playing with the timing of the trapezoid timing. I spent some weekends on theoretic derivation of the formulas, which I plan to publish in QEX one day.
>
> In regard to your 12bit ADC. If the input signal is already frequency limited and you have some AGC in place, 12bit may be ok. 50dB dynamic range for CW or PSK31 signal centered around 600Hz seems adequate to me. You may get more bits with oversampling. If you have a frequency limited signal with some random noise, you get additional bit of resolution by 4x oversampling. For example, in my standalone CW/PSK decoder I am samping at 40ksamples / second and summing each 8 samples to get 5ksamples / second. I am gaining >6dB with this process.
>
Ah, I had forgotten that you can oversample and get another bit. I've
spent some more time poking around, though, and I think I'm going to go
with a separate CODEC and uC. Unfortunately, the TI uC chips don't have
any CODEC interfaces on them. I'm looking at the dsPIC processors. The
dsPIC draws more power when running (80ma vs 10ma), but it is easily 10
times faster so it can spend more time in low-power mode. It might be a
wash in the end.

Plus, I discovered that all the audio CODEC chips have built-in low pass
filters, oversampling, etc. Some of them have DSP filters built in.
With that I don't have to have as much external filtering. And they
have headphone drivers, speaker drivers and microphone inputs, things
that will come in handy if I ever do a general purpose rig from this.

Thanks

-corey - AE5KM