PDM Class E IMD AM Tests

[W1DAN]

Tom:

That's weird!

On the 1k tone, when listening to an analog rx in SSB mode on only one sideband, do you hear a single tone, or one every 1 kc as in the image?

Maybe the 1k tones are mixing with each other in the receiver. I do not understand why the 7k tones would not. Just wonder if it is an issue with the SDR rx or software. Try other frequencies to see where things start to go bad.

Anybody else have ideas....?

Dan

[Steve - K4HX]

Yep. Or harmonic distortion.

What are we actually seeing here?  Is this a 1-tone test with the tone beating against the carrier or what?

[K1JJ]

On ssb, using low or high power I get a single tone on the SDR screen. It's looks just like an unmodulated carrier as it should.


A single tone on AM produces the spectrum as posted.  

I just tried it with a voice whistle at about 800 Hz. It also shows mutiple spectrum peaks at 20% modulation or higher.

Also, on the FT-1000D in the AM mode I see the same thing, though the higher order peaks are way down.  Of course the FT-1000D low level output is about -70 3rd IMD, so much cleaner than the E rig.  On the 1000D, the higher order spectrum peaks get drastically higher as I go into flat topping, so it does indicate distortion.

This appears to be normal.


Dan, I listened on a separate receiver and the tones are definately there, marching up the band.  They appear sensitive to how hard I hit the audio, just like IMD does when flat topping. Can someone else inject a single tone into their AM rig and see if you get this spectrum too?

It appears a single tone may be a good way to measure AM IMD -  whereas the a two-tone is better suited for ssb.
Maybe this is another example where the carrier comes into play, just like the class C AM biasing thing we talked about recently.

T  

[Steve - K4HX]

I did exactly this on the NE602 balanced mixer circuit alone and as installed in the FT-102. I got the same spectra use you are getting Tom. The sidebands were -55 dBc and better with the NE602 alone. As installed in the FT-102 they were more than -40 dBc.

[K1JJ]

I did exactly this on the NE602 balanced mixer circuit alone and as installed in the FT-102. I got the same spectra use you are getting Tom.

That's good to hear, Steve.

How far down is the "3rd order" tone from the carrier peak and the main injected tone when at 100%  or 150% modulation?   What happens when you flat top it?

T

[Steve - K4HX]

I was in the middle of adding the IMD numbers when you posted. See my previous post.

[K1JJ]

I did exactly this on the NE602 balanced mixer circuit alone and as installed in the FT-102. I got the same spectra use you are getting Tom. The sidebands were -55 dBc and better with the NE602 alone. As installed in the FT-102 they were more than -40 dBc.

A-HA!!  We're making progress.

Your AM sidebands are pretty clean as expected.

I just did a 800 Hz single tone test on AM with the FT-1000D 4 mW into the 4CX-350FJ.  I didn't have a lot of signal to work with, but am seeing at least -40 Db 3rd - and about -55 dB down from the carrier.  

The class E rig shows about -25 dB down 3rd order  and  -30dB down from the carrier peak in the same test.  This is a good baseline to try improvements on the E rig later on.

I would expect to see these kinds of differences, but it didn't show clearly until now.  After all, the e-rig is running 1KW and is high level modulated.  That's axing a lot.

T

FT-1000D driving 4CX-350FJ on AM - single 800 Hz tone: (Looks very clean)

[K1JJ]

To Steve / QIX -

Could you tell me the proper procedure to tune up a class E rig using C1 and C2?  

Does the E/drain waveform need to be looked at for optimization each time or can we do it using the wattmeter and current meters?


Tnx.

T

[Tom WA3KLR]

Tom,

Just out of curiosity, can you do a wide sweep to show the switching frequency sidebands?

[K1JJ]

I will do, Tom.


The plot thickens....  Frank / GFZ says that the rig will not generate IMD with only one tone unless we are overdriving the audio amplifier into clipping.  He says we are looking at the harmonics of the audio.

I thought I was careful to limit audio clipping on the various rigs, but will look into this today on the e-rig, since it seems the most prominent there. It may be as simple as adjusting the audio input pot on the PDM gen board.

Steve/ HX, you might recheck your audio going into the bal modulater and see what's happening too. But perhaps -40 to -50 Db 3rd IMD is indicative of pretty clean audio to begin with.


Well, at  least this one tone test is another tool for checking the overall health of a rig, since it turned up a possible audio problem.

More later -

T

[K1JJ]

Update on E-Rig tests:

I checked the audio input with three different audio generators and see the same results.  The input test audio  is clean.

I checked the 187 KHz switching sidebands and see them down about -65dB, but this may be normal  shack overload, since they should be down more. I've had guys check it and heard nothing on the air.

I did tests at  20%, 50% and 100% modulation at 500 w and 1KW.  At 20%, the harmonic distortion is quite good, but once it passes 50% it gets poor. I can hear it up the band.

C1 and C2 tuning had a small effect on the THD, but I need to look into that more. Maybe Steve / QIX will come in and describe the latest C1/C2 tuning. I might even try some different L tank coil values to see if that helps...

T

[K1JJ]

And, class E rig 100 % modulated with a single 1 Khz tone...

I notice that the harmonic frequencies are exact  X2 X3 X4, etc of the 1 Khz fundamental, which shows they are true audio harmonics.  This may indicate modulator / audio issues to look into.


Still testing.  Any ideas appreciated.

T

[K1JJ]

Trying to get to the bottom of this...

Here's 20% modulation, 50% and 100% using the FT-1000D driving the 4CX-350FJ amplifier.  I am using the same audio generator and test set up as when testing the class E rig.

We see that the THD is better by quite a bit.  This leads me to believe that I need to find what is going on in my class E rig... 

Can anyone else with a PDM  E-rig run these same tests at full power to compare results?  I'm hoping to see I have a problem in my rig or test set up so it can be corrected.

T

[N2DTS]

That looks nasty!

And, class E rig 100 % modulated with a single 1 Khz tone...

I notice that the harmonic frequencies are exact  X2 X3 X4, etc of the 1 Khz fundamental, which shows they are true audio harmonics.  This may indicate modulator / audio issues to look into.


Still testing.  Any ideas appreciated.

T

[K1JJ]

It pays to be persistant.... seems like a step in the right direction.

I decided to experiment with tank Q...  I reduced the numbers of inductor turns to 1/2 and added in large padder caps to C1 and C2.   Immediately the THD started to fall.  

I haven't optimized it yet, but at 60% modulation, I am seeing -40 Db  down from the main tone...   and -60 dB down from the carrier peak.   This is really significant when it comes to sideband splatter reduction.

This is the same thing I did with my linear amplifier chain - find the highest Q I could run. It coincided with best IMD.  It may be more about matching the output of a low impedance amplifier rather than Q.  It takes a LOT of tank C1 and C2  to match a big amplifier.

I also measured the HV supply and found it was stiff. This is important for low IMD and THD figures.

So, now I'll bolt in some fixed cap padders and try to make it even better.  I'll post more observations once I get everything dialed in.  I plan to check the e waveform for form and also be sure the efficiency is still up there.  There is also a risk of higher tank voltages that can blow FETS.  But I'll try anything for a cleaner rig.

T

Pic Below:

Class E rig at 60% modulation looking better. Notice the higher order peaks are better behaved. I found my optimized linear chain had the same pattern.. the higher order stuff tended to stay "pushed down" as the power increased. If there is a problem, it tends to climb with power.

[Tom WA3KLR]

Tom,

I noted how far down the (upper sidebands) 2nd, 3rd, 4th and 5th harmonics were relative to the 1 kHz fundamental sideband and plugged those numbers into a THD spreadsheet I did a number of years ago.

Your "original" plot in Reply 36 = 6.8 % THD

Your "Higher-Q tank" plot in Reply 39 = 1.2 % THD.

I do wonder what is going on there though; cud be a number of things.  I would tend to start by running the PDM into a resistor load only and see what distortion levels you get.  

Do you have way to directly get a measurement of THD of the audio from the PDM LPF output?

[K1JJ]

Thanks for the THD calculations, Tom.

Under 1% would be my goal.

I may try the PDM modulator into a load thing if I don't get good results playing around with the tank.  I'm cornvinced the rig should do well once set up properly.   It's just that I don't have a lot of experience optimizing class E rigs, so it's slow going.

I don't have a method to analyze audio-based THD yet, and will look into it if needed next.

T

[Steve - K4HX]

There is something wrong there Tom. With the NE602 and the FT-102, the 2nd harmonics were down 40-45 dB. The third was more than -55 and everything else was more than -70 dB.

It's not clear to me how your can put more than one tone (the carrier and the modulation sidebands) into and amp and there not be IMD.

[K1JJ]

Well, I know that my linear system checks out very well for THD using the same test set up.  So that's the standard.


The e-rig is working OK, it's just that the THD is not quite good enough.  I've found that by setting up the tank for higher Q (less L and more C)  the efficiency goes down, but the THD is improving.  Each e module wants to run at closer to 7 to 8 amps rather than 6 amps to give better THD by about -10 dB. This is almost acceptable.  The e waveform still looks decent.

This would mean setting the trip circuit higher, so there is more risk.

Still trying different stuff.

T

[KL7OF]

And, class E rig 100 % modulated with a single 1 Khz tone...

I notice that the harmonic frequencies are exact  X2 X3 X4, etc of the 1 Khz fundamental, which shows they are true audio harmonics.  This may indicate modulator / audio issues to look into.


Still testing.  Any ideas appreciated.

T

This looks like  A LEAK...

[K1JJ]

Well, I found the sweet spot and put it to bed.

Once the best L/C ratio was found which gives best efficiency, power out, and best general THD,  C1 and C2 need to be slightly tweaked further for best THD.  When the L/C ratio is right,  C1 and C2 are most sensitive to THD fine-tuning.  I can see the THD go up and down as C1 and C2 are fine-tuned. This affects power output very little.

At 100% modulation I was able to drop the first lower audio peak  (second harmonic) an additional  -10 dB from the main audio peak. This is about -35 to -40 dB down. The higher order harmonics also drop off faster than before.

This is acceptable.  When using voice, the rig seems pretty clean for side products.  

T

[Steve - K4HX]

Quite interesting. This might explain some things I see and hear on the bands.

[steve_qix]

Just got back from visiting with Tim and Larry in Maine.

In general, I haven't noticed a lot of difference in the modulation with final tuning at least with my own transmitters, but every RF amplifier is a bit different since they were all built by different people.

It seems as if you've got it figured out, in any event.

I made some distortion tests way back when I first designed the PWM modulator.  I was consistently able to get less than 1% distortion which I found perfectly acceptable for the application.  Unfortunately, I don't have access to the distortion analyzer anymore, else I would run a couple of more tests 

[K1JJ]

Hi Steve,

Yep, it's all sorted out for now. Was on 75M with the e-rig for a few hours today and the bamdwidth reports were good... just what I see on the SDR monitor in the shack.

In summary, after finding the best L/C tank ratio and tweaking C1 and C2 for best THD, I found that running a FAT carrier as high as possible and limiting modulation percentage to about 80% negative (using external processor limiting, not the internal NPL) produced the cleanest signal.   For my voice, 80% negative equates to about 100% positive.   Coupled with the external 5 kHZ audio filter, it's hard to beat.  It also seems the higher the current per stage, the better the THD.   IE, at the same power output, the THD was poorer at 5 Amps vs: 7 Amps per stage. (better THD when tuned for lower efficiency)   So I run the current up a little, have lower efficiency on purpose and see cleaner performance.  Something like idling a linear amplifer higher maybe.

Sure, we could run the modulation much higher and get no complaints, but this is the sweet spot where it runs the cleanest for me.  This may or may not apply to other rigs. We will just have to see if others try these tests.

T

[K1JJ]

It just dawned on me..

When modulation gets above about 90% negative, 110% positive, the THD starts to take off. I wonder if this is the internal negative peak limiter clipping action?   The IMD stays reasonably low, but that THD soars.  I'll check this out tmw.

Easy solution to fix if so.

T