Stretchyman 40 W TX Reliability Modifications

[Charlie_Dont_Surf]

The very first ever design I made was an unashamed copy of the venerable LuLu design. Fat better than any other design to date and I must emphasise not my design. All I ever did was make a PCB to make replication a simple matter.

It uses a Xtal which I had to get made at considerable cost (£14 ea.) and managed to get quite a few units made (100 approx).

That was five years ago and all designs since 2016/17 have an AD9833 DDS.

...So as I must say again it's not my design, however  there are other designs out there that are and Yes you've slagged those off as well....

Your name and email address are on the PCB silkscreen. You probably put your "STRETCHYMAN PIRATE RADIO" stickers (a ripoff of Breaking Bad) on the case too. But now you are claiming because the original design came from someone else that's it's not your fault?

I'm sorry, but you own that problem.

The rate of occurrence of this weird drifty oscillator issue is more than some sort of random, chance occurrence. You shipped "several" of these units but apparently never put them through sufficient validation and or any sort of burn-in to confirm that they stay stable over time.  If you had, then you would have found this issue. Just like you would have found the famous high-voltage snap that takes out either the drain filter cap or the FET or both (and caught on fire in front of me), but you didn't, because you don't know what you are doing and can't be bothered.

Designs get shared and tweaked all the time. There's a fellow named Paul Brokaw, whose Brokaw Bandgap Cell (https://en.wikipedia.org/wiki/Brokaw_bandgap_reference) is used as a reference in nearly every IC from all manufacturers since the 1970s. There are billions of these things around the world now. By your logic, if someone makes a Brokaw Bandgap and fucks it up, then we should complain to Brokaw and not the person who fucked it up? Riiiiiiighhhhhhht.

You want all the accolades and none of the responsibility.

I still suspect most 'simple' Xtal controlled designs to exhibit some form of FM that's visible on an SDR.

Cool, except that a sudden drift of 1 KHz that Chris showed is not FMing.

I look forward to seeing any design from particularly yourself or anyone else for that matter that you mass market and build 100+ units.

I doubt (know) that this will ever happen.

I look forward to your reply.

So very nice of you to invite me to play the "my balls are bigger than yours" game. Explain to me how many you've shipped has anything to do with your ability to solve problems before they are created? Apparently none. And shipping lots of defective product is the hill you've chosen to die on? And the barrier to entry for whether something is "worthy" or not is not whether it's good, it's just that you've shipped many of them, warts and all?

Also, what good would it be for me to put a schematic out when most of my problem with you is that you are absolute crap at validating and testing? You aren't going to learn those skills from a schematic.

(Shaking my head)

My professional life is in the design, validation and manufacturing and post-sale support of electronics that sells in somewhat higher orders of magnitude than what you are talking about, but please do tell me more about the the 100 you made over the course of a many years. It's amusing to me in a very haughty sort of way.

Since I have been doing this for decades, I've give you a couple of lessons for success:

1) If you don't know something, don't pull some bullshit text off the internet and use it to justify your preconceived notions. Look into yourself. Figure it out. If you don't have the answer, say, "I don't know but I'm looking into it." The smart people I know admit when they don't know something and don't make shit up.

2) Test your shit out extensively. Build 10 and put them through their paces. Leave them on for a long time. Examine design marginalities. Push and pull the design before you ever ship any product.

3) Despite your best efforts (which you have not made, see above), there will be post-sale problems with your products. Don't sweep them under the rug. Fix it or make it up to the customer.

I'll say it again: You want all the accolades and none of the responsibility. You don't know what you are doing, and don't care to know how to do things better.

As for my designs, my employment is somewhat engrossing and I'm not interested in becoming a BIG TIME PIRATE SUPPLIER MOGUL like you apparently are. I get my jollies by shipping good product that doesn't catch on fire right in front of customers.

[redhat]

Boys,

Can we do with a little less hostility?  There is already enough negativity on the web.  If this must continue, please take it elsewhere.

Thanks,

+-RH

[Stretchyman]

Yeh, he's being such a TW*T

Just ignore...

[Charlie_Dont_Surf]

Just ignore...

There you go. Keep your head in the sand and keep thinking that your shit don't stink. At least you are on brand.

[Dare4444]

Use DDS 9850 board with Arduino and 2n7000 for very high quality professional sounding AM signal. I'll paste the code if needed. Its output is 10mW. Feed it to two parallel 2n7000 each with 1ohm source resistor and set the standing current to 15ma each with a variable pot connected to their gates. It's drain can then be connected to the final linear amplifier. No crystals or complex synthesizer needed. A simple Arduino code sets the TX frequency!!

https://ibb.co/3TMm2gb

Arduino Sketch.

#define W_CLK 13       // Pin 13 - connect to AD9850 module word load clock pin (CLK)
 #define FQ_UD 8     // Pin 8 - connect to freq update pin (FQ)
 #define DATA 10       // Pin 10 - connect to serial data load pin (DATA)
 

 #define pulseHigh(pin) {digitalWrite(pin, HIGH); digitalWrite(pin, LOW); }

 // transfers a byte, a bit at a time, LSB first to the 9850 via serial DATA line
void tfr_byte(byte data)
{
  for (int i=0; i<8; i++, data>>=1) {
    digitalWrite(DATA, data & 0x01);
    pulseHigh(W_CLK);   //after each bit sent, CLK is pulsed high
  }
}

 // frequency calc from datasheet page 8 = <sys clock> * <frequency tuning word>/2^32
void sendFrequency(double frequency) {
  int32_t freq = frequency * 4294967295/125000000;  // note 125 MHz clock on 9850
  for (int b=0; b<4; b++, freq>>=8) {
    tfr_byte(freq & 0xFF);
  }
  tfr_byte(0x000);   // Final control byte, all 0 for 9850 chip
  pulseHigh(FQ_UD);  // Done!  Should see output
}

void setup() {
 // configure arduino data pins for output
  pinMode(FQ_UD, OUTPUT);
  pinMode(W_CLK, OUTPUT);
  pinMode(DATA, OUTPUT);
 

 
  pulseHigh(W_CLK);
  pulseHigh(FQ_UD);  //
}

void loop() {
  sendFrequency(1e6);  // Enter freq here, Right now Set to 1000,000 Hz ,1000KHz, 1MHz . Example, for 1540KHz enter 1.54e6 in the sendFrequency bracket above. I hope this is easy to understand
  while(1);
}

[Stretchyman]

https://www.analog.com/media/en/technical-documentation/application-notes/an-423.pdf

Yes, we've known about this for sometime!

25 years!

However generating low power A M. and using a linear isn't very efficient. Plenty of folk still using that method as they don't know any better!

Class E is the way to go, Simple, Efficient, Perfect!

+ DDS's are generally used to generate frequencies these days and are a single chip and as simple to program as an Arduino. Complex synths and PLLs are no longer nessarary.

No point searching the web for any decent designs as all you generally find is a load of old rubbish designs from 30+ years ago.

Modern design are usually confined to datasheets from the manufacturer or others in the know...

Str.

[Charlie_Dont_Surf]

No crystals or complex synthesizer needed. A simple Arduino code sets the TX frequency!!

No disrespect to you personally intended but I consider myself to be comfortable enough with microcontrollers and synthesizers to be "dangerous" with them and I don't feel this to be a terribly simple solution. You need a microcontroller, you have had to write software for it, then the output is pretty weak and it will need a decent linear amplifier (which you have not discussed here) to boost the output to be heard farther than 10 meters away. Also, I don't consider the AD9850 to be a particularly simple synthesizer. Perhaps it appears that way if you treat it like a black box and don't know what's going on inside but I don't recommend doing that. (I actually don't feel that any synthesizer is "simple".)

To be clear, there is a need and a place for synthesizers and I don't know how you do a AD9850 (and the like)-based synth without something providing the digital words to them (here, it's a microcontroller) but let's not pretend that this is a simple solution.

So not my cup of tea, but to each his or her own I suppose.

Finally, I'm not quite sure why this ended up here in this thread. It's completely unrelated to the original topic.

[Stretchyman]

Indeed, somewhat off topic.

However once you have the code written it's a no brainer.

I have 64ch code for the AD9850 PCBs that are readily available. The opposing squarewave outputs will drive FET drivers directly. Commonly use for class D/E push pull.

Also have single channel software for the AD9833 (<$2) PCBs and use a discrete 3 tran amp to get to 5V p-p to directly drive FET driver for single ended transmitter.

Saves all that xtal nonsense!

Str.

[Charlie_Dont_Surf]

I don't know how many versions of your synths you have gone through but the ones that I have used have always ended up a bit lower in frequency than I would like - usually about ~100 Hertz low. We're not expecting WWV accuracy here and the consequences of this are not the end of the world but I think that we can agree that they could be closer to target.

I suspect that a number of the loggings on this website indicated as "xxxx.9 KHz" (more or less) come from your synths, but I can't prove that. Perhaps this is yet another way to identify one of your transmitters on the air. 

[ChrisSmolinski]

I suspect that a number of the loggings on this website indicated as "xxxx.9 KHz" (more or less) come from your synths, but I can't prove that. Perhaps this is yet another way to identify one of your transmitters on the air. 

There may indeed be a statistical anomaly regarding frequency offsets from integer values in pirate loggings. I will add it to the list of things to investigate 

[Stretchyman]

Well..They aren't GPS locked or anything fancy just plain DDS which rely on the internal clock being exact. Quite easy to reprogram them to a higher (offset) frequency to negate the error.

If anyone would like such, get in touch!

Who cares anyway I wonder?

More window watchers?

Str.

[Brian]

I don't know how many versions of your synths you have gone through but the ones that I have used have always ended up a bit lower in frequency than I would like - usually about ~100 Hertz low. We're not expecting WWV accuracy here and the consequences of this are not the end of the world but I think that we can agree that they could be closer to target.

I suspect that a number of the loggings on this website indicated as "xxxx.9 KHz" (more or less) come from your synths, but I can't prove that. Perhaps this is yet another way to identify one of your transmitters on the air. 

I run 2 rigs here. One Xtal and the other a very cheap Chinesium DDS. The DDS is cheaper than an Xtal (about 20 bucks plus a few more for connectors and the rotary encoder), goes up to 50 MHz and is adjustable in 1 Hertz steps. Less drift than the Xtal rig.
Problem is not getting it to be xxxx.ooo but having the way to measure this accurately.
With the DDS, I'm probably more on channel than many legal stations.

[Charlie_Dont_Surf]

Quite easy to reprogram them to a higher (offset) frequency to negate the error.

Someone asked me privately last year if there was a way to fix the offset. I was very bored over Xmas and finally got around to reverse engineering several of your .hex file codes for 43 meters to figure out corrections and get them a lot closer to the intended target (~ +/-20 Hz off, instead of -100 Hz).

The other thing to note is that there are multiple sources of drift or frequency shifts. Most of them are pretty small and not worth worrying about. The largest one is that the synth takes a big jump of ~ -70 Hertz when the final transistor is turned on. That could be due to many reasons but I'm guessing it's due to supply voltage shift. My corrections take this into account but because of self-heating and temperature drift with time, it can't be better than ~ +/- 20 Hertz over the long term, which I feel is acceptable.

[Stretchyman]

I've made the offer and no one has asked.

I'll ask again.

If you have a programmable TX and the frequency is offset by xxx Hertz please let me know the frequency +/- XXX and I'll send you the revised .Hex file to put you bang on frequency.

Very simple for me to do.

Please get in touch by email.

Str.

[Albert H]

I got really annoyed trying to get true frequency accuracy and stability for Medium Wave from the three or four DDS modules I tried....  I gave up, and built a PLL that's locked to the 198kHz BBC Radio 4 transmitter.  The 198 kHz rig is run from the National Physical Laboratory's frequency standard source, so the Medium Wave boxen I built were rather more accurate than I could show on my frequency counter!

Obviously, there are standard frequency transmitters all over the world - I went for 198 kHz because the rigs were for the UK and Netherlands, and it's easy to derive frequencies in 9 kHz steps from that source with the minimum number of CMOS logic ICs!  Incidentally, I didn't mind adding proper gate driver ICs, rather than trying to cane a logic IC for the drive, and generated twice the frequency I wanted, then used a final bistable to guarantee an absolutely 1:1 M/S ratio.

The advice given on Steve Cloutier's "ClassEradio" website about component ratings is pretty useful, and has helped me avoid the fireworks displays that others have had!  For higher powers, I've been using ex-telecom SMPSUs for high current 48V supplies, and PWM modulators for efficiency.  It's got to the stage that a 200W carrier / 800W peak rig can be built for pretty cheap these days.  The expensive parts it seems are the damned toroid cores and the mechanical stuff.

Now.... back to designing a little (80W PEP) PWM rig for as little cost as possible.....