DIY OpAmp AM Modulator

[Zazzle]

Hey,

I'd like to share this. It's an AM Modulation Driver. So far it's just a simulation. I've to re-calculate some values to minimise component stress and power loss. I tested the simulation from 8-48V. All it needs it to adapt R6 (voltage feedback to full swing). I think I'll build a H/W version in November.

Just some words on the Bakckground: I have this 10W CW-Beacon that I took on a vacation last year in the shelf. The PA exists of a push-pull configuration. I thought about modulating the PA feed with Audio, turning the Beacon into an AM Transmitter. First I thought about using a TDA2030 with transistorfollower (more current). But, eh, the voltage swing isn't that nice. And I also only need a positive swing. An yeeeeess. I could use an rail2rail OpAmp and give it some transistorfollower to get the current I want. But buying isn't as much fun as building. And, you can build this thing here with simple components you have in your junkbox.

Kind greetings,
Zazzle

[redhat]

On paper this looks pretty similar to Dave Martin's design, which works well, I might add.  The only issue I see arising when this is committed to copper is going to be the differential stage.  Historically, you would like to use a long tailed pair here to ensure good balance, but if you have a transistor tester that can show hfe, you can hand match two that will play nicely.  I had something similar to this running at 50W carrier using some big TO-3 darlingtons I found surplus and the circuit was well behaved.

Good Luck!

+-RH

[Stretchyman]

Hmmm, a series modulator......

You could use PWM perhaps?

Linear have the LTC6992 and as LTSPICE is available you could sim that driving a single FET with LPF.

You would then have the most efficient, cheapest and simplest series modulator in the world.

Now wouldn't that be something?

Moving on........

 

Str.

[Zazzle]

Hi,

On paper this looks pretty similar to Dave Martin's design, which works well, I might add.  The only issue I see arising when this is committed to copper is going to be the differential stage. (...)

Well, I guess I stuck to the basics that'll required to make it operate. I'm not surprised that other people came up with such before, too.
Can you link me the curcuit Dave came up with?

Any yeah, a matched pair is required.  I'll solve that by using combo-transistors. Those for high class amps where two transistor are on the same die. Have some flying around. But good point mentioning it. I'm so familiar with that stuff that I forgot that it's worth being mentioned.

Kind greetins,
Zazzle

[Zazzle]

You could use PWM perhaps?

Hahaha. I *knew* someone would come up with that.

In fact, yes, I could use PWM but I decided against. It's an emotional reason. Analog stuff has it's charm and with just ~30W it's okay to turn some Watts into heat. I mean, It's a fun project and no comptetion for the most high tech and efficient projekt. Less is more sometimes.

In my old workplace I've engineered multi-KW SMPS for (fast) charging forklifter (etc). And I know that, when I start building a PWM driver now, I'll end up with overengineering it to the point where it's not fun anymore. I' ll have enough opportunities to overengineer stuff when I'm building that 500W Class D Transmitter next year.

Greetings,
Zazzle

[Stretchyman]

PWM is analogue, just that it's analogue that works efficiently!

Class D O/P stage rather than class A

I will trouble you no more, you like your heat, I have no problem with that!

Let us know how you get on re sim vs real build.

I'm also a big fan of LTSpice, fantastic and free!

 

[redhat]

I can't find his website anymore, maybe it moved?  Anyway, here it is.

+-RH

[Pigmeat]

I saved all of that stuff that Dave designed and had it go "pffft" when my old station email got hijacked about a month ago. Of course, I was saving it for questions like this.

BTW, if any of you guys receive an email purporting be me through the pigmeat_voab at the yahoo place, it's not me. I can't access it.

[Stretchyman]

Yeh that's the Class A (Classic) circuit.

You can now replace the 741 with a LTC6992 and the 3055 etc with a SiC FET, add a filter and you have moved on!

Will it ever happen?

I hope so.

one day.....

 

[redhat]

One thing about the SiC fets though, their power dissipation is comparatively low verses other fets I guess due to the way the die is attached to the back plate.  In linear applications you have to watch this closely, as you can easily exceed the dissipation derating limits.

Saturation voltage will be higher with a fet due to the gate threshold voltage, a darlington is typically 1.4V, a fet can be as much as 3-5V which may impact your modulation overhead.

Just a thought...

+-RH

[Stretchyman]

The FET is PWM'd it's either ON or OFF.

I can push 500W thru' a pair at $4 ea.

On Res is circa 180mR, so with 2 that's ONLY 90mR on res

SiC (appropriately!) is most definitely where it's at the mo' and the technology is advancing rapidly as they're so rugged and used in military and space apps.

Cheap as chips, why bother with anything else?

I ask you!

Anyway I'll leave this one.......

 

[redhat]

Note the afforementioned problems with fets relates to their use in a series pass modulator in linear mode.  Your preaching to the choir here, I'm on the PWM bandwagon too, but it is fun to build simple things with little regard to circuit layout, ect.  I wouldn't want to build a PWM generator on a breadboard as circuit strays would hamper performance.  The linear series mod OTOH has no issues due to its limited response bandwidth.

+-RH

[Zazzle]

Hi again,

It's like with all Forums on the net. You post something and people say "yes, but xyz works better". Had that on a Linux fourm yesterday. Someone asked a Debian specific question and someone else said "but, but... Ubuntu is better!". Some people just can't let it be.

Anyway. Of course there always something more state-of-the-art. Like PWM in this case. Btw: I deem PWM pretty much digital because it converts analog with a reference to something digital. Not with a bit-pattern but width length. And we ned a circuit (low pass) to integrate to convert it to analog again. Of course we ca say "a free variable width length is analog". We could have a pages long discussion now.

Mind, PWM isn't that "just take some parts". If you want to use the controller referenced you need to come up with:

• A circuit that adapts the input signal to the 0-1V signal (or a lesser frame, see Data sheet).
• The input circuit also needs to perform Level-Shift since pos.-peak equals 100% Mod and neg.-peak equals 0% Mod. Also, peak sensing would be nice to not run into clipping.
  
• You ned to pay attention to a PWM frequency highter 10 times the max expected audio frequency to prevent aliasing.
• Inductors must be choosen properly regarding core saturation and material.
• For best performans and low distortion it's important to calculate the proper values for the Lowpass filter.
• A FET-Driver is needed, like IR2xxx something (too lazy to look the exact number up).
• Oh, and that FET-Driver needs be be capable of doing level shift since we wanna have the best efficiency and want to use a low Rs(on) N-Ch FET.
• Given that we work with high emergency pulses we also need to pick capacitors that can handle those.
• Layout may become critcal.
• We need to deal with RFI suppression or otherwiese we may build a broad-spectrum neighbour annoyer.

And all the other stuff that may inexperienced people stumple upon their way to build "something such simple".

The intention, why I posted that, was to leave it for people who wanna get started with a simple modulator with standard parts. And yep. Even an rail-to-rail Op-Amp would have made the thing even more simple. As for myself: i just need a simple modulator that can run with a 30W TX. PWM is just overdone in my case since testing comes before building a solid thing.

But to do something good: I'll post some Quad-OpAmp based 'simple' PWM-Modulator in future.

I hope that sets your troubled soul free so it can move on.

Kind gretings,
Zazzle

[Stretchyman]

I like to work to the 'Best Engineering Solution'

as I'm an Engineer, I guess?