The AM Forum

I am about to submit the Rev D PWM board set for manufacturing.  Looking for any possible input
on new features before the boards are submitted.

The PWM generator board may be used for multiple purposes, including vacuum tube
and other non-class E projects, as well as traditional class E and other solid state modulators.

This board set includes the PWM generator/aliasing filter/neg peak limiter board,
The Overload Sense and Shutdown/Efficiency Meter board and
The PWM output board suitable for RF amplifiers up to 1200W DC input at carrier.

I have made some minor changes since Rev C.  The PWM board now features 2 PWM outputs -
both low impedance (50 ohms); one 12V output and one 5V (TTL compatible) output.  These outputs
allow the board to drive relatively long interconnects (usually shielded cable) that are properly
terminated.  Of course, short interconnects can still be driven as in previous revisions.

The overload/efficiency meter board has expanded external inputs so that an overload shutdown
can be triggered by off-board events (such as lack of drive, SWR, etc.).  A number of RF filtered
test points have also been added, making board set-up and calibration much easier.

The PWM output board now employs a terminated, low impedance PWM input.

Here is a picture of the PWM generator board.  Let me know if there are things you need in these boards.
(http://www.classeradio.com/pwm_generator_rev_d.jpg)

Steve

Balanced, line-level audio input would be a nice feature...think I mentioned this before, though.

Is there a new schizmatic available yet? Please put me in line for a set of boards and thanks for your efforts!

(With apologies to Tom Payne)

Mr. Class E
by Bill Horne, W1AC

Would you put this station together for me?
some I got through the mail and some I got free
The pieces all work, except two or three
Please, Mr. Class E

Now for a transmitter, all I want to do
is run legal limit, or maybe times two
to get a manly sound, like I hear on TV
and have the gang think that the sound comes from me

I'd like to work mobiles, no matter how small
amoung all the ships, make me the most tall
since I can't afford to install a three-phase
I'm looking to you, to make me all the rage

A software receiver, DC to Daylight
so I can hear Osama sneeze at midnight
Including, of course, the digital mode
in fashion while I was learning the code

Don't forget Baudot, and EBCD too
I want to send telegrams from my PS/2
Lay optic fiber, shaped in a token ring
add a peak LED -- have I forgotten anything?

Put it together -- take an hour -- or two
hide all the cables, and when you are through
write me a contract where you guarantee
it'll all work forever, or you'll fix it for free

Would you put this station together for me
some I got through the mail and some I got free
The pieces all work, except two or three
Please, Mr. Class E

Steve,
It looks like you added an emitter follower on the pwm drive. I think it would be a good idea to add a local resistor to ground to terminate the line when the pulse is turned off since there is no sink at the output. The line could ring without a termination. I'm surprised you needed the transistor since the pwm chip has a low Z output totem pole.
I thought you could just change divider resistor values or grab the output upstream of the resistive divider. That was my plan when driving two output boards.
My only other comment is. It would be nice to flash a LED when you hit the negative peak limiter. I think you had a spare OP amp section in the TL074.. It would be just a few traces  fc

The all new self assembling feature??

               _-_-bear

This is good feedback, thank you !  Link to the schematic is at the end.


Done !  Please check the new schematic, particularly the input section and email me comments.  I would like the 2nd set of eyes to make sure it comes out right the FIRST time  :D


Termination of the output lines?  Done !! (please check the schematic and email me with comments about this!!).

As far as the indicator light - I would do that, but there are no free gates or diff. amps left in the circuit !  I was trying not to increase the board area, although it would be a nice little extra.

On the PWM outputs... it is unclear (at least to me   ;)  ) that the UCC25701 can in fact deliver CONSTANT current to a loaded output (like a resistive load).  The spec shows that series resistor for the shunt regulator (R401 in my circuit), cannot be too low due to the limitations of the shunt regulator itself (internal to the chip).  So, to err on the safe side, I buffered the output with the emitter follower.  The TTL compatible output needed the emitter follower anyway, so I figured - what the heck !!

Schematic: http://www.classeradio.com/pwm_generator_rev_d.pdf (http://www.classeradio.com/pwm_generator_rev_d.pdf)

Thanks and Regards,

Steve

And, we have this one too !!!!!  If you need an ASSEMBLED (and tested) board (or boards) or a whole rig for that matter, we can accommodate.  I believe Green Mountain Radio Electronics will do the assembly at a reasonable cost.  Try emailing bg@greenmountainradio.com to find out more  8)

Regards,

Steve

Steve
 The pwm spec sheet says Iout 100 ma test case source or sink and gives saturation voltage for each state. Later words say 1 A peak. So I would think it safe to run 1/2 that current at room temp operation.
Also made a mistake on op amps that was an unused cap, not section. Sorry about that. fc

I fully agree on the PWM spec sheet, but the sheet is contradictory to an extent.  The VDD pin (pin 3) of the UCC25701 PWM chip maximum current (from the sheet) is 20mA, and is a shunt regulator.  So, the chip's output section will probably sink a lot of peak current, but the average current cannot exceed what is essentially the shunt regulator current of 20mA (and 10mA is recommended) - at least that's my read on the data sheet.

So, if I take the 12V output and put it into 50 ohms to drive a line, I'm looking at 240mA for the high.  At a 50% duty cycle, that's still 120mA average current - way over the 20mA max shunt regulator current.

It may be that the chip might function properly with a "hard" 12V supply, but in my experiments, the chip was unreliable as to whether it would start up or not with a hard 12V connected.  With the shunt regulator keeping the VDD at 14V, everything works every time.

In fact, I bet it WOULD work properly if the chip started up with the 14V supply, and a "hard" 12V were subsequently connected.  The application notes suggest this in their schematics, but don't come right out and say so.  It would make sense, based on the stated output current of the PWM output section.

It sure would be nice if manufacturers produced unambiguous spec sheets - and most DON'T !!  :P

Anyway, for this one we have a solution with the emitter follower.  At least that spec if a known entity  :)

Regards,

Steve

I went back and looked again. I suppose the shunt is to protect the FET hanging off the output. I see what you mean. Here is a thought, how about taking the two output transistors and set them up as a totem pole with a jumper selectable VCC. Might reduce parts count. Rare that you would need both outputs. You would need a series resistor off the pwm output. NPN on top PNP on bottom. or better yet a little 8 pin driver chip with selectable VCC and eliminate a bunch of parts. fc

I like it  ;D   It will reduce the parts count by 1 capacitor and 2 resistors.  With the space saved, I will be able to put in a switch to select the output voltage.  I want to keep the transistors operating at their output voltage (+12 or +5) to avoid any heating of the pull-up transistor.  This is easily accomplished by selecting the VCC as well as the input voltage at the same time (and I have lots of DPDT PC board mini switches)!

I prefer jumpers myself. once you set it there are no contacts to fail. I would also eliminate the audio phase switch. I almost didn't install it in my board.
Tracking selection is a good idea though to get the drive right for both voltages.

Agree with Frank...can live without the phase switch. Maybe the space saved could be used for a neg peak limiter LED driver ckt. An LED on the front panel (not the ckt board) would be a neat feature.

Eliminate the 1N34 and add an opto in the NPL circuit. A  nice slow one should have a good diode. I've always used LEDs as diodes in my NPL circuits. The output of the opto drives the LED. I think the 1n34 is normally conducting current so the opto would need to short out the led until the diode stops conducting. (I'm also not a fan of board mounted pots since it isn't allowed in mil electronics but it might be hard to eliminate from the board)

Hey Jay, did you look at the input circuit?  Just want a sanity check on that  ;)

The phase switch is, believe it or not, used by a lot of folks !  Many times, the audio system in use does not have phase switching capabilities, and the phase switch in the transmitter is the only thing available.  I even use it myself on occasion, depending on what audio system I'm using at the time (like when I bring the transmitter to Deerfield).

In order to put a front panel LED for the negative peak limiter, there would have to be 2 op-amps avilable - one used as a comparator, and the 2nd used as a peak detector and driver for the LED.  This would provide sufficient isolation between the audio circuitry and long wires subject to RF, PWM output secondary coupling and other nasties which could cause audio anomalies.

I have this circuit implemented in the class H modulator board.

But, I may yet put such a circuit into the system anyway.  Make the TL082 a TL084 and expand the size of the board slightly   8)  Still pondering, although there would have to be some R & D and testing of the circuit before committing to PC board.  I have a lot of people banging on me for these boards now  :o

What's that famous saying?  "There comes a point in the life of every project when you have to lock up the engineers and go into production!".  (actually, I think it was shoot the engineers)

Regards,  Steve

I know it probably doesn't matter for this application, but the old, venerable TL072/4 opamps are crap opamps by today's standards. The dual is easy to replace with an op amp of your choice (if you wanted to or needed to, like in a pinch) but the quad makes things very limited. Don't go for a Quad, imho. There ought to be enough space to wiggle in a pair of duals, if you really need an extra pair, since the footprint is pretty close when you go to the quad anyhow... ymmv.

                     _-_-bear

PS. I'd suggest the same thing for the antialiasing filter's opamps, btw... looks like there is some realestate just to the south east of that section, where the power comes in...

Bear the problem is these cool new devices are surface mount. I don't have a problem with surface mount but most hams are afraid of it. You really need a good microscope to play SM.
Steve, I would think an opto would provide isolation working against the op amp reference voltage? Yup sometimes you need to get on with it.
A balanced input could be configured in either phase with a front panel switch. Set it up so it will also work single ended. 
Also delete the input level pot. A pot for single ended and phase switch could go on the front panel. Diff audio would be line level and controlled externally. This deletes a pot and switch. You have enough caps and resistors to do this. Heck maybe you could rearrange traces and pads to do either. Then make it a user choice.

Bear, BTW I have seen a fair amount of cross talk on a quad op amp die. So I wouldn't mix audio and comparator functions on one die.

...I suggested the use of duals rather than a quad in any location... in the main because it permits the swapping of a very wide range of opamps... which may be of value off in the near future (or not).

Surface mount or not, that's the way things are now... if you want to be in on the latest you need a fine fine tip on a temp controlled iron and/or a hot air reflow system... that and a wide field stereo microscope? :)

In about 20 seconds you won't be able to find DIP chips at all, they'll be commanding big prices for rebuilding and maintenence of that "vintage" IC gear!!

                  _-_-bear

I agree it won't be long before homebrew will be all surface mount. I actually prefer it myself. 

Well, I don't about 20 seconds :-)  The DIP has a HUGE installed base, and many are still in current production.  If some day everything has to go surface mount, then probably the way to go is a "built" board (like in a computer) and swap things out as a unit.

But, I think we're a ways away yet....

Tubes were essentially obsoleted in mainstream electronic equipment during the early '70s (actually, probably during the late '60s), but some 35 to 40 YEARS later, you can still buy tubes - and there are even certain specialty HIFI equipment builders who actually look for them  ;D

Not to mention a bunch of ol' AM freaks, antique radio nuts and even vintage SSB restorers heh, heh.

"H..," I love everything from smokium pubes to veeery hollow state LSI SMT devices.

Some of these new op amps have impressive specs.

Hi Steve:

Thanks for doing all this work. Now for some comments from a non design engineer (FWIW)...

I think the way you have the input gain pot wired may affect common mode rejection depending on where the pot is adjusted. I use a dual pot at the input of my op-amp (not the best either but simple). The best way is to use an instrumentation op-amp circuit to unbalance the signal, then have a single pot adjusting the level feeding another op-amp (i.e your clipper).

A small item that will not be much of an issue and is not worth correcting...your low pass filter will create overshoots (tilt) of the clipped signal. I added another clipper after the filter to shave off the few overshoots (at the expense of a very small amount of splatter), or I could have built a phase corrector circuit. This will only make about a db of modulation difference so I did not correct for it with any phase circuit.

On my breadboarded UCC25701 (based on your circuit from about 3 years ago) I added a switch and capacitor (3-30uf?...I can find out if needed) to pin 14 (the SS or soft start pin) that allows a ramping up of the PWM generator when I flip the switch to transmit. It takes about a 1/4 second to ramp up to the normal level from no output. The advantage here also for receiving is there are no birdies generated that may be picked up in your receiver. Also, a shut-down signal can be fed here to turn the output off during any failure. I also had to invert the input to the modulator so that when no PWM signal exists, the power transistors of the modulator are turned off.

I fed my UCC25701 with a regulated 15v from a 7815. I had to because of power supply hum issues. This makes me think that the PSRR is low on this chip. Looks like you are getting your power from the unregulated 18v?

I think (??) the 78xx series of regulator is most happy with about 10uf of capacity at it's output pin. I think I read it someplace, but cannot find it again. I'd consider bypassing the op-amp power with 0.1uf at the op-amp pins also.

I got away with feeding about 6 feet of RG59 style coax from the generator chip output to the modulator...I know I'd do better with a buffer here.

Thanks for all your good work. I enjoy seeing what you do and have learned alot.

As you say..."Good Stuff!"

73,
Dan
W1DAN

Good stuff  :D

This has produced many useful suggestions - some of which have been incorporated into the upcoming release   8)
I very much appreciate all of the feedback, information and suggestions, as much of this will most likely be incorporated into some release, some day, of some PWM board set or modulator  ;)

My own final goal is to use a DSP filter for anti-aliasing - no phase shift, no overshoot and no passband ripple !!!  Won't THAT be nice??

The current release, Revision C of this board set works beautifully, and produces low distortion, high quality audio.  The boards are easy to build and set up, and appear to be fairly foolproof!  This next revision, Rev. D, is just an update to Rev. C - with some relatively minor (but very important) additions which will make the boards more useful in a greater variety of applications and environments.

The PWM boards have been used in a wide range of implementations and power levels from 100 watts to 1kW, with as much variation in the implementations as there are folks who have built the transmitters - the boards worked, regardless.

For me, this continues to be a challenging, educational, fun and rewarding project, and hopefully others who have ventured into the waters of solid state, high power AM have had similar experiences.  The work and experiences of many have created the body of knowledge we all enjoy today.  And, it just keeps getting better and better  ;D

Regards,

Steve

Where is the USB input. Or Toslink? Or S/PDIF? I want to run digital right into my TX. I've modified all my rigs with RS-232 but I want to upgrade.

Steve,
A DSP could do all the audio processing and generate the PDM all in 1 part. So maybe an op amp at the input and mos driver at the output. Heck you could add a serial interface to say USB for SW upgrades. OR imagine a computer interface to a spectrum monitor with slders to taylor audio. Then imagine all the audio processing gear that could go on the scrap heap.
Getting even more nutty you could include an audio output to drive alalog gear. I bet the board size could remain the same.
I had a lot of fun playing with a DSP eval board when I started playing with SDR. You could have some real fun with the software.

Steve,

I'm not trying to argue for the sake of arguing...

I'm not sure how a pair of DIP-8s side by side takes up that much more realestate than a single quad?? The comparison being valid between std size parts as well as between surface mount duals vs. quads...which is what I was getting at. Yeah you might need some jumpers to make the PS connections, if you need to, but beyond that, not many more pins or size?? I'm confused on this...

And, what is the need to make everything surface mount in the first place?
Why does it need to be "small"??

Perhaps because the parts are harder to find in large these days?
More expensive in large?
The board is more expensive in large?
All of the above?

Now, as far as tubes? Yeah, tubes were once ubiquitous, not not more!!

Try to find me a bunch of 6AQ8s today... for a simple example.

Now try to find me some 845 op amps. Not new ones but earlier pre ROHS production (the current production happens to be not so good and have noisy front ends for reasons that are unclear). Not quite so easy.

Just getting your hands on TTL chips can be a slight pain.

None of these technologies will absolutely become unavailable in our lifetimes, but they have become harder to find, so more expensive, and not as likely to be found residing in a parts drawer as they once were.

So, as I said, there were and are far more duals made than quads, making a swap out much more likely to happen out of the parts drawer in a pinch... or in the first place when building the board up virgin... that's all.

                       _-_-bear

Bear there is still a lot of ttl around. Rochester Electronics for one.
But why bother when HC, AC or HCT is pin for pin and better specs. oh and a lot lower power.

Steve:

I see that some of my suggestions could cause a major re-draw of the PCB artwork. So maybe down the road.

I know your board will work great the way it is. This thread caused me to go to your web site after a long time away. Good updates.

Thanks for asking for suggestions...often people bringing up ideas can make a product better....until you need to shoot the engineers and build the darn thing!

73,
Dan
W1DAN

I know what I mean, but I guess it's not being understood as I expect?

What I am saying is that the older packages and older types no matter what the part is are fading into the sunset as we go along in time...

A better example is that some of the best low noise Toshiba FETs are no longer in production. So, if you wanted to do complementary discrete FETs in a new analog audio front end, you're SOL...

Looks to me like the "heyday" of ubiquitos parts availability is passing quickly, and we're moving quickly away from discrete parts being available and used at all, and those that are now being used are increasingly low voltage and low power, smaller and smaller SMT stuff... the fewer the number of years that a part is in production, the less likely it is that there will be sufficient "surplus" supplies for future years use... then as the availability goes away, the price of the remaining stocks go up and up... it seems to me that we've turned the corner on this part of the cycle and that prices of older stocks will go up and up from here on in, and availability will drop more and more.

But what do I know?  ::)

            _-_-bear