Semi OT: History of consumer radio designs

On Tue, 9 Dec 2014 20:18:01 -0500 (EST), bitrex Gave us:

Ever heard of Joe Walsh of "James Gang", and "The Eagles" (and solo) fame?

He has a pretty nice collection. Maybe these are not the radios you are looking for...

In the 1980's, we had synthesized local oscillators and digitally tuned stereo receivers. Progress for most of the 1970's was getting rid of mechanical parts involved in the receiver tuning. That meant microprocessor push button control of the PLL. However, the IF and discriminator circuits were largely unchanged throughout the 1970's and 1980's. What also changed was the inexorable drift from discrete to integrated designs. Large sections of the circuitry were size and cost reduced down to individual IC's.

During the 1990's, most of the RF, IF, demod, and audio circuitry was crammed into a single IC, surrounded by a few discrete components, and run by a micro. However, the receiver was still fundamentally an analog design full of LC and ceramic filters.

Fairly recently would be about 2005. That's when DSP receivers changed everything. For example: At the same time, the audio section went digital with a Class D switching power amplifier. The only part that's still linear was the added AGC (automatic gain control) at the front end required to keep the input level stable enough to use all the available bits.

Somewhere along the time line, sub-carrier data, HD Radio (IBOC), RDS-TMC arrived.

I'm sure I left out quite a few details...

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

The transition to stereo decoding in the 1970's was also fairly interesting with the early chips and discrete modules having a lot of very finickitey coils and tuning capacitors that had to be just right. ISTR MC1310 was the first easy design to use in homebrew kit.

In the UK Ferranti made modules that various high end makers like Roberts plugged together in a chassis to make consumer radios. Then there were the ubiquitous miniature Japanese transistor radios.

Clive Sinclairs Micro-6 was also notable way back in 1964 as the smallest radio in the world. Fun webpage about it at:

Interestingly modern FM tuners do not seem to have improved their performance much beyond state of the art in the 1980's. The front end sensitivity might be a bit better but the distortion is sometimes worse.

What we have now at least in the UK is DAB broadcasting at a bitrate that produces a sound quality inferior to FM (although to be fair it does reproduce the inter programme gaps with a lower noise floor).

The fun bit is in borderline areas of DAB reception voice breaks up so that the newsreader sounds like the subterraneans on Stingray gargling whilst trying to read the news. Every now and then the bitstream crashes the decoder chip completely. I have given up on it at home in favour of internet radio where the signal quality and bitrate is high enough for studio quality on Radio 3 and way better than DAB on Radio 4.

One fun feature of modern DSP is that you can repurpose a cheap digital TV dongle as a wide band radio scanner by using alternative drivers and mutilating its firmware a bit. Become a new hobby for some folk.

--
Regards, 
Martin Brown

On a sunny day (Tue, 09 Dec 2014 18:47:49 -0800) it happened Jeff Liebermann wrote in :

One or 2 years ago I bought a Tecsun radio:

It is based on the Silabs S-4730 chip, BROADCAST AM/FM/SW/LW RADIO RECEIVER dataheet here:

It is an amazing receiver, use it everyday. Look at the block diagram.

Thanks for the reply. I'm still curious how in these later designs the input filter was tuned to be offset from the LO by the IF frequency; I don't think they're using variable caps or diodes. I think I'm missing something - when you have have a PLL synthesized LO is there a way that the need to tune the front end filter can be eliminated

--


----Android NewsGroup Reader---- 
http://usenet.sinaapp.com/

Jan,

Do you know what CPU drives the Si4734 chip?

Asking because I have a couple of Si4735 chips here, including a software patch from the manufacturer for SSB reception (under NDA), with the intention of building an 80m direction-finding receiver with them. I'm not sure that the RSSI/AGC steps will be small enough. Even 0.1dB is a big enough step to be annoying for DF when using "whoopee" mode, so I'll probably have to rectify the audio and produce my own RSSI. Luckily the AGC limits, step size, and delays are software-settable, so I could just use that for auto-ranging, and the post-AGC RSSI for whoopee. Otherwise the AGC can be put under full software control anyhow.

The PL360 is a nice format, and with a modified antenna would suit DF use very well, as long as the CPU is reprogrammable/replaceable and there's room to add the whoopee circuit. Whoopee is essential, no-one wants to be looking at RSSI numbers while waving an antenna around in the scrub.

Clifford Heath.

Thats easy You use the same control voltage that the pll creates to drive the vco varicaps, to tune the front end filter varicaps. With proper design you can use the exact same voltage, the tuned circuits are scaled such that it all tracks pretty well.

Mark

On a sunny day (Thu, 11 Dec 2014 12:38:08 +1100) it happened Clifford Heath wrote in :

Well, as you asked I tried to have a look:

The CPU (or FPGA??) is under the metal cover that is soldered, did not want to rip that apart. Seems like it could be a PIC:

Yes, for sure it is a pity this radio does not have an SSB mode. For that I have an Tecsun PL600 (longwave through shortwave + FM). I have the circuit diagram for that, but quite different, does not use the Si4730 chip:

Anyways, you could control the Si4730 with a PIC I think, you can even make PWM audio with a PIC:

But the PL360 has an analog volume control, some DA must be in it somewhere. What I do not see in the Si4730-31-34-35-D60.pdf data sheet is the registers to program the chip. Without that data it will be next to impossible to program it from some micro. (Well almost impossible, we have ways of finding out). Do you have that register data under non-NDA?

That seems fairly likely. thanks for looking. A friend has used a PIC32 for his, which does triangulation from GPS measurements.

If I can get the patch I'm sure you can too. It's a encrypted data file to program the internal DSP, so knowing the registers is not enough; you need to be able to produce the DSP code and encrypt it.

We got the SSB patch because for DF use, you tune to maybe 1KHz off the fundamental CW signal to get a heterodyne whistle.

The DSP uses a very low IF; 45KHz I believe. Keeps the power consumption down :)

Yes. I have several possibilities I'm thinking of: MSP430 (cheapest/low-power), STM32 ARM7 (to implement IMU and drive a colour LCD), or an FT245 USB chip in bit-bang mode (for prototyping from the desktop computer). I was thinking of trying the last one first, because I have a couple of alternate front-ends to try.

Also under NDA I believe, but just ask, they're pretty helpful. As I said, not useful without the DSP and encryption data anyhow. It would be excellent if they opened that up.

Clifford Heath.

If you use a much higher than usual IF you can get image rejection quite a bit easier. Thus (little or) no RF side tuning at all.

?-)

Which later designs? I guess you mean PLL design. The front end was always broadbanded from 88 to 108 MHz. With a 10.7 MHz IF frequency, the 20 Mhz band pass is just a little smaller than 2 * 10.7 MHz, which would be the frequency range most like to produce receiver image reception. A tracking filter front end would be nice in high signal strength areas, where some additional off frequency rejection would be useful to reduce intermod. However, with FM capture effect, it doesn't buy much for consumer FM listeners, who rarely listens to weak signals, where it would most beneficial.

On the other foot, it's not that difficult to build a tracking filter RF amplifier, that tracks the LO frequency. All that's needed is a lookup table that relates the RF tuning frequency with the varactor tuning voltage. If I had surplus CPU cycles, I can do it by sweeping across the approximate receive frequency, stopping at the varactor voltage with the best SNR, and storing the voltage in RAM or FLASH. Extrapolated to its logical extreme, I helped with the design of a fairly conventional VHF/UHF NBFM receiver and PLL LO, that had no tuning adjustments. However, it never went into production because the savings in production tuning labor and lower component costs were negated by the increased cost and complexity of the design.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

It's a very good radio (due to the SiLabs chipset) for a different reason. The IF bandwidth is sufficiently narrow, and DSP demodulator sufficiently noise immune, to remove the overlapping HD Radio sidebands from the receive signal. Most other non-DSP receivers will pickup the HD Radio junk, and all you hear is distorted garbage. With these receivers, it's clean.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

On a sunny day (Fri, 12 Dec 2014 13:07:27 +1100) it happened Clifford Heath wrote in :

I did receive SSB with it, used my Raspberry Pi as signal generator

to beat against the station, so say if station is at 14 MHz, set raspi signal to 14 MHz, and keep it close to the radio.

That other Tecsun I have with SSB has no USB LSB selector, took a bit getting used to it.

Aww, and here I was hoping for a discussion of the "All-American 5" - the classic 5-tube design for a superheterodyne AM radio.

John Savard