Feed Forward

[K9ACT]

I now have my 100V supply up and running . I see about 1V of ripple and would like to get rid of this so it's time to get into the Feed Forward bit.

I can't find any explanation of exactly how to implement this so I ask here.

Schematic says to connect FF pad to 47k resistor to power supply.

It's not obvious how the Generator board can control the totally separate HV supply but elsewhere it is implied that I can get rid of ripple on HV using this.

So, I connect the 47K resistor to the +side of the HV supply.... then what?

What do I do with the pot? Do I need a separate return from the HV supply?

JS

[kf1z]

Connect the 47kohm resistor to the positive of the HV DC supply, and to the pad on the board.
Connect the ground pad next to where the 47k resistor connects to the board, to negative of the HV DC supply.

The feed forward does not "control" the actual HV DC supply.

the ripple voltage is basically used as one form of regulation in the output PWM square wave pulse-train.

On a positive going peak in the ripple voltage, the output voltage would be lowered by that same amount (or maybe by 1/2 ? that makes more sense), by varying the pulse-width of the square wave... and a negative going peak, would result in raising the output voltage...

That may not be the best explanation... sorry... the negative and positive going peaks, are probablt referenced to an averaged value of the dc supply voltage....
I hope you get the idea anyway.....


So the actual HV supply will STILL have the same ripple voltage ...always...that won't change.
But the resulting DC on the output of the PWM filter will be varied to compensate for the ripple...
Thus "nulling it out"..... or be "smoothed"

------------------
Looking at the output of the modulator or even the RF out of the transmitter, slowly rotate the Feed forward POT until the ripple is gone.
Or at least minimized as good as you can get.

Rotate slowly, because as you turn that pot, the output of the modulator will JUMP up, or down depending on which way you turn the pot.
(It will jump, and settle back to it's origional voltage)

-------

I wonder how much filter capacitance you have?

10,000uF 20,000uF?

[kf1z]

I'm sure eventually Steve will pop in and explain that in more eloquent, technical terms.

He had explained it to me several years ago.... and is very simple and straight forward...

But I'm sure it looses something in my translation!

The important thing to remember, is that the FF will NOT take the ripple off the HV DC supply, it only cancels out the ripple in the output of the modulator.
So when you hook up the FF and try to adjust the pot, your scope has to be connected to the output of the PWM filter, or the output of the RF deck. NOT to the HV DC supply.
And 'obviously' in either case, the modulator must be operating, and producing output..

Oh.. and the audio gain pot on the pwm generator board should be turned all the way counter-clockwise, and audio gear unplugged from the modulator...

Bruce

[K9ACT]

Thanks for the explanation. Good enough to at least try it without worrying about frying something.

Also would have been frustrated watching the ripple on the supply do nothing as I diddled.

Next question is on the peak limiter and how to set it.

As I reduce the limiter, the neg peak goes down until it reaches the base line which then gets very bright. At this point, I go the other way till I get a nice fat line at the base but it's not obvious how fat is the correct setting or if this is the correct procedure at all.

js

[kf1z]

The negative peak limit should be set so the modulation never gets to 100% negative. (99%)

I set it while talking, and the modulation turned up a little more than what I would normally have it set...
And simply adjust the NPL pot until I "never" see more than 100% negative modulation.... JUST flat trace.

Probably the way to do it correctly would be with an audio generator... but not sure you'd get the results you want due to
symetry/asymetry between a sine wave, and the voice...


Maybe you've noticed...
But it is also possible to set the neg peak limiter, so there is NO negative modulation... all positive!
Looks fun on the power meter and scope... but sounds horrid of course..

[K9ACT]

The negative peak limit should be set so the modulation never gets to 100% negative. (99%)

How do you resolve 99% on a scope? Best I can do is just make it fatter than the base line trace.

>Probably the way to do it correctly would be with an audio generator... but not sure you'd get the results you want due to
symetry/asymetry between a sine wave, and the voice...

Maybe it's my voice but I have never seen any difference in symmetry so I just nod off when people talk about it.

>Maybe you've noticed...
But it is also possible to set the neg peak limiter, so there is NO negative modulation... all positive!
Looks fun on the power meter and scope... but sounds horrid of course..

I was about to ask as it's not obvious why it would not sound normal.

So, I hooked up the Feed Forward and can't see much of anything happening looking at the PwSdr display but there is quite a bit of low freq fuzz around the carrier that I do not see with the regulated supply. I can't see any hum on the scope mod monitor and don't hear any so maybe it's in the noise and the stuff on the SDR is a gremlin.

I also hooked up the external 5k ext carrier level pot but it's pretty useless as is. I really need it because the one on the board is very noisy and jumps all over the place when I change it.

No matter where I set the one on the board, I only get about a 20% change with the external pot.

Any ideas?

js

[kf1z]

Well.... 100% modulation is 2x the carrier pk-pk voltage right...
So if you wanted to get really picky and hit 99%
Then you would expand the scale so your carrier takes up say 8 full divisions (8cm) (full display).
That's 80 mm.
So 99% would be just less than 1mm .


If you eliminate all the negative modulation...you 've removed 1/2 of the modulation for starters...
Then beyond that, may just be distortion caused by a diode detector?
All I know is I tried it a few years ago...sounded hideous... no low end... real tinny...


Not sure what you mean by, or why you wouldn't see anything happening with the feed-forward....

The external carrier level control is only meant to "trim" the carrier level, not to widely vary it.
If you want full carrier level control you need to change the value of the pot.

In most cases with the PWM rigs, people are not changing the power levels much.
If it's designed to run 400 watts, they run 400 watts.
With a class-E rig, changing power levels means retuning the RF deck..
And with PWM, you must maintain the proper impedance load for your filter.

Not saying it can't, or shouldn't be done... but most don't bother.

[steve_qix]

Hi !

The ripple output from the power supply will not change with feed forward, but the hum in the output of the modulator- and the carrier will (or should!). You cannot reduce it to 0, but you can decrease it by a fair amount. I don't actually use it in my big rigs because I have 64,000uF of filter capacitance, and there is very little ripple. I do use the feed-forward in my smaller rig, because I only have 8000uF, and that's not enough...

On the Neg Peak Limiter - because this is PWM, and there are analog filters involved, the NPL cannot be perfect across all frequencies and with all waveforms. Usually, I set it for about 90% negative and that catches pretty much everything. It is quite effective!!

Asymmetry - Jack, you should see LOTS of asymmetry in your natural voice. If you do not, there is some issue with the audio system, microphone or equalization. I have worked with literally thousands of human voices over the years in broadcasting, and they are all asymmetrical to some degree. Some are strikingly so.

Not only that, the asymmetry "phase" will change with frequency response characteristics. If you have a very flat system (I mean really flat - down to 10 Hz or so), you will be able to generate the classic "class E modulation pattern" (so-called, because only really flat rigs can produce it, so this is only seen on class E rigs and now Flex radios), also called the "shark fin" pattern. It is a distinctly asymmetrical pattern - I often get 150% positive and less than 90% negative at the same time!

You must also leave sufficient voltage for the positive peaks. With 40 volts DC for the carrier, the high voltage supply should be around 120V. This will allow you to modulate close to 200% positive. I generally use a 3:1 ratio - 45 volts on the carrier, 135V for the high voltage.

Here is what mine looks like at times: