[Amps] Solid State Amps

[Manfred Mornhinweg]

Roger,

> Small, generic LCD screens could 
> be produced for pennies, or at most, a few dollars as you can now 
> purchase a 40" smart HD TV receiver for $300-$400. 

It's a matter of checking what a simple cellphone costs these days. Two years 
ago I got my latest one, an old-style dumbphone in clamshell style. That one has 
two displays, a monochromatic one outside and a pretty decent color display 
inside. The whole phone cost like $30, including shipping from China, and that's 
a price that wasn't subventioned by a phone company! So the displays really must 
be cheap, when bought in the quantity a cellphone make buys them.

 > How many
> transistors used in SS rigs over the last two decades are no longer 
> available?

That's not a problem, because essentially any transistor can be replaced by a 
different one. But the problem does exist with specific ICs. A famous (or 
rather, infamous) case is the DDS chips of the Kenwood TS-450, 690, 850, that 
seem to rot from the inside out due to some fabrication mistake. Those ICs are 
specific, discontinued, and presently unavailable except perhaps from someone 
who has a few stockpiled, and sells them at a very stiff price. And there is 
nothing that can replace them in a plug-in fashion. One would have to cook a new 
board with a different DDS chip and a MCU or other circuit to translate the 
commands, to revive any of these radios whose DDS chips went bad.

Mine went bad in time, when I still could get replacements without the 
fabrication mistake, so I'm fine in this regard. But I know of several hams who 
have dead radios of these models, and no way to fix them.

> However, HRD and other Graphical interfaces can replace most of the 
> front panel, let alone the individual  displays on today's rigs.  If 
> using HRD you might not even notice a failed, non essential display.  

It's true, but when you tie a radio to a computer, it's getting twice as 
expensive, 5 times larger and heavier, and the power consumption is 10 times 
higher. So it's really good to be able to operate a radio without an external 
computer!

And now, let's go to really interesting matters!


Warren,

> ** It seems to me that it's probably much easier to optimally implement 
> pre-distortion in a DDC/DUC (digital up-conversion / digital 
> down-conversion) software defined radio than in other radios.  There are 
> simply very few places that errors can creep into the required 
> processing.  The required precision and accuracy increase exponentially 
> as the desired reduction in IMD increases. So, having few error sources 
> and the benefit of double-precision floating point become important.

What does "double-precision" mean in this case? 32 bits, 64 bits, or how much?

To have errors down 60dB, even 16 bit should be OK, for a single operation. But 
I wonder how many math operations are involved in a radio's whole processing. 
That surely calls for more bits of resolution in each operation!

I saw that a DSP chip with integrated CODEC, aimed at mid level HiFi 
applications, has a 56 bit processor! They strive to put any errors about 100dB 
down from the peak signal.

> ** 50V LDMOS amplifiers are probably the "most correctable" solid state 
> amps.  Tube type amps also correct very well.  13.8V amps are more 
> difficult due to "memory effects."

What memory effects do you mean here? Dielectric absorption, thermal effects, or 
what?

> ** No problem to correct an entire amplifier chain, at legal limit. No 
> extra hardware or software is required.  One just feeds back a sample of 
> the output from the last stage to calculate the correction.

That's the nicest thing about SDRs. Once you have the interfaces to the analog 
world up and running, and a good chunk of processing power, you can do pretty 
much _anything_ at no additional cost.

> ** You cannot successfully correct by pre-distorting the MIC input to a 
> radio UNLESS you have VERY wide bandwidth from the MIC through the 
> entire transmitter. The IMD frequencies that you want to correct must be 
> within the bandpass of the correction path.

Here you caught me in a shaky position. It's where I touch the limits of my 
present theoretical knowledge about the matter. If I set up an amplifier that 
corrects amplitude nonlinearities, with a loop bandwidth of 30kHz, does this 
mean that any amplitude-error-caused IMD more than 30kHz away from the center 
frequency will not be corrected at all? But that could still be quite useful, 
given that IMD so far away from the center frequency should anyway be pretty low.

> At Ham Radio Friedrichshafen this past June, I gave a talk (approx. 30 
> min.) on the openHPSDR solution for pre-distortion.  It's posted on the 
> openHPSDR web site.  If any are interested, here's a link:
> 
> "*Warren Pratt,*NR0V, 2014.Digital Predistortion linearizes RF 
> amplifiers <http://video.openhpsdr.org/HRF2014/PureSignal1.2.mp4>(MP4)"

Is the contents of that video available in written form? Due to internet 
connection limitations, I cannot download videos...

> I hope this helps explain a little more about this technology and sparks 
> more interest in finding ways to reduce IMD!

It does! For several years I have been wanting to get seriously into DSP and 
really do something myself in the area, but never have found the entry door. 
Warren, can you perhaps suggest a practical, inexpensive way to set up a kind of 
DSP breadboard, that would allow me to mount a processor, suitable A/D, D/A 
sections, and then go along getting my feet wet programming it? So far my only 
attempts at incorporating DSP into my own projects has been by basic digital 
filtering functions in low frequency applications. Mostly power line frequency. 
And implemented in integer math, on 8-bit PICs!

Years ago I tried my hand with a DSP-93 kit that I got for free, but never found 
out how to even start doing something useful with it.

Roger seems to be in the same boat I am.

Manfred

========================
Visit my hobby homepage!
http://ludens.cl
========================