Bug in DUP displayed power level--need someone else to test & confirm

[w-u-2-o]

The Bug:

Power levels shown on the panadapter/waterfall are not correctly scaled in DUP display mode during TX. Levels are showing a -24dB error.

The openHPSDR paradigm has always been to accurately display signal levels referenced to the rear panel input.

Note: scaling for S-ATT is working correctly. I.e., for a given input level the displayed level does not change with S-ATT.

To Reproduce:

ANAN-8000: place a calibrated signal generator output on the 8000 external PureSignal feedback input, enable PS, set S-ATT to 0dB, set mode to USB, and key the radio with 0 drive.

ANAN-7000 (any variant incl. Andromeda): place a calibrated signal generator output on the 7000 bypass input, select "Use Bypass on TX", enable PS, set S-ATT to 0dB, set mode to USB, and key the radio with 0 drive.

My results with an 8000 (accurate to about ±1 dB):

Input level -3 dBm = ADC overload indication.

Input level -4 dBm = panadapter level of -28 dBm

Input level -14 dBm = panadapter level of -38 dBm

Is anyone able to test this with a 7000, or any other hardware type?

[K1LSB]

Scott,

I tried your steps to reproduce on my early model 7000DEL MKII, here are my findings:

With 21 mV input Thetis shows -52dBm in the display.

With 67 mV input Thetis shows -42dBm in the display.

With 205 mV input Thetis shows -32dBm in the display.

With 265 mV input Thetis shows -30dBm in the display.

With 295 mV input Thetis shows -29dBm in the display.

With 305 mV input Thetis shows -29dBm in the display, flashing Yellow Overload.

With 330 mV input Thetis shows -28dBm in the display, flashing Red Overload.

The test signal is a sine wave output from the G1 signal generator in a Rigol MSO5074 digital oscilloscope. I can't testify to the accuracy of the signal strength of the output, but I'd assume it's within a couple of percent.

Mark

[w9mdb]

I used an XG3 which I have confirmed the -33dBm and 0dBm levels with an 8920A RMS meter.

My 7000DLE MKII with the latest checkout shows this:

0dB = -99.6dBm
-33dB = -133dBm

So the linearity is correct but the level is wrong. If I recalibrate using the 0dBm level I get the right answers at 0dBm and -33dBm but the noise floor runs at -45dBm which is wrong. Normal noise floor (without calibrating level) is at ~-143 dBm so everything is 100dB off when calibrated.

[Joe-W4WT]

I just did the test as Richie described he did; use internal coupler, set to 10 watts, press Tune, check level, and then key radio in CW mode both with DUP on. I see around -10db for BOTH of those tests in the panadapter reading. Both CW and Tune agree. I'm using a 7000 DLE MK II. Not sure what Richie is using for a rig.

This is a bit easier way to test since you don't have to drag out the signal generator (old and feeble here; both mentally and physically).

Joe, W4WT

Edit.. I'm using the latest version of Thetis, 2.10.0.

[w-u-2-o]

Thank you, Mark!

With 295 mV input Thetis shows -29dBm in the display.

Assuming that is peak-to-peak voltage, then 295 mV into 50 Ohms = -3.6dB, and that gives an error of 25.4dB. That's close enough to my results showing a 24dB error!

If you could confirm that is peak-to-peak voltage that would be most helpful.

[w-u-2-o]

I used an XG3 which I have confirmed the -33dBm and 0dBm levels with an 8920A RMS meter.

My 7000DLE MKII with the latest checkout shows this:

0dB = -99.6dBm
-33dB = -133dBm

So the linearity is correct but the level is wrong. If I recalibrate using the 0dBm level I get the right answers at 0dBm and -33dBm but the noise floor runs at -45dBm which is wrong. Normal noise floor (without calibrating level) is at ~-143 dBm so everything is 100dB off when calibrated.

That's substantially different than what others are getting, which is in the range of 20 to 25dB of error. You might want to look at it again.

[w-u-2-o]

I just did the test as Richie described he did; use internal coupler, set to 10 watts, press Tune, check level, and then key radio in CW mode both with DUP on. I see around -10db for BOTH of those tests in the panadapter reading. Both CW and Tune agree. I'm using a 7000 DLE MK II. Not sure what Richie is using for a rig.

This is a bit easier way to test since you don't have to drag out the signal generator (old and feeble here; both mentally and physically).

Joe, W4WT

Edit.. I'm using the latest version of Thetis, 2.10.0.

Without knowing the coupling factor of the internal coupler one can't be certain of these numbers. That's why I used a signal generator and set up for an external input during TX.

If someone knows, or wants to measure, the coupling factor of 7000 internal coupler that would be most helpful.

[Joe-W4WT]

Well, I couldn't stand not to know, so, risking a back injury that would likely be curtains for me, I lifted the Sig gen over to the radio and did the test as you described Scott. Here is what I saw:

-3dbm through -6dbm show Overload ADC1
-7dbm = No overload -10dbm reading
-14dbm = -17dbm
-20dbm = -23

Double checking relative calibration, with -14dbm into antenna input 1 I read -14dbm on RX1 meter. Calibration looks reasonable.

It looks like my radio is a lot closer than what you are seeing Scott, for whatever reasons. Mine seems to read about 3db low and appears linear.

Hope this helps,

Joe, W4WT

edit... if there is anything else you would like me to test since my radio seems to be different, let me know. I'll leave the generator where it is for a while.

[w-u-2-o]

edit... if there is anything else you would like me to test since my radio seems to be different, let me know. I'll leave the generator where it is for a while.

Did you follow this procedure, Joe?

ANAN-7000 (any variant incl. Andromeda): place a calibrated signal generator output on the 7000 bypass input, select "Use Bypass on TX", enable PS, set S-ATT to 0dB, set mode to USB, and key the radio with 0 drive.

And set display mode to DUP, of course...

[Joe-W4WT]

"Did you follow this procedure, Joe?"

sure did. Followed it exactly.

Joe

Update: I remembered a new (to me) calibration setting called Tx Display Cal on the calibration panel. I haven't looked into exactly what it did but thought I'd turn it and see what happened. Lo and behold, I set it to 3.8 and now my readings on your test Scott are spot on! -15 reads -15, and -20 reads -20.

[K1LSB]

Thank you, Mark!

With 295 mV input Thetis shows -29dBm in the display.

Assuming that is peak-to-peak voltage, then 295 mV into 50 Ohms = -3.6dB, and that gives an error of 25.4dB. That's close enough to my results showing a 24dB error!

If you could confirm that is peak-to-peak voltage that would be most helpful.

I can confirm that each signal tested was peak-to-peak voltage; i.e., the 205 mV test was a +205mV to -205mV sine wave. I visually verified each signal on the o'scope voltage scale prior to injecting it into the radio.

Mark

[w-u-2-o]

So Mark & I have matching results, and Joe doesn't. So now we have to figure out why.

I'm running P2 firmware version 2.1.18 and Thetis 2.9.0.8 (haven't had the time to go to 2.10.0 yet).

[w9mdb]

Re-did my test (helps when you plug the sig gen into the right port

0dBm = -2.7dB, ADC1 Overload
-33dBm = -36.7 dBm
-73dBm = -75.1dBm
-107dBm = -111.6 dBm

So not quite linear here...

If I put 0dBm in and S-ATT at 8 I get no overloads....peak shows -2.9dBm

[Joe-W4WT]

Interesting. So Mike, Mark, and I are running 7000 DLE MKII's of some vintage. My serial # is 0107. Mike and I seem to have a radio that is within 3db + or - of what it should be (before I adjusted the cal setting) and mostly track pretty good. Mark's 7000 seems to be different. Wonder why?

My firmware is also P2 2.1.18 and as I mentioned earlier, Thetis 2.10.0.

Joe

[w9mdb]

My firmware is 2.1.23 and Thetis 2.9.10

[Joe-W4WT]

Mike, what is your radio's serial number?

[w9mdb]

DX8300219

[K1LSB]

My serial number is 7000DLEMKII081.

Mark

[Joe-W4WT]

Well, now that is confusing. Mark and I have the same prefix before the number but Mike yours is completely different.

Apache really needs to get with the program and start labeling things with some kind of logic behind it. Or at least some kind of logic that makes sense. Maybe it makes sense to them?

Joe

[w9mdb]

Just to be complete here's my label.

IMG_0379.jpg (2.31 MiB) Viewed 3182 times

[Joe-W4WT]

Well, I guess we can assume, just going by the number, that Mike has the latest version, Mark yours is the oldest and I'm in the middle.

Maybe.

Joe

[w-u-2-o]

So Mark and I are getting 25dB errors, and Joe and Mike are getting almost no errors.

Version of Thetis should not matter, they all have the same code in this area. Firmware should not matter, calibration is done in Thetis. The hardware is all functionally the same. This does not make any sense.

The most likely answer is that the two pairs of results are because the test is not being performed the same way by everyone.

[Joe-W4WT]

Scott,

Here are the steps I took to do the requested test; let me know if I misunderstood your request and did it incorrectly and I'll do it again. I'm not sure how I could "do it incorrectly" yet still provide the results I provided but who knows.

1. I set the radio output to zero watts
2. I connected the radio to a dummy load "just in case".
3. I made sure DUP and PSA were ON and SATT was set to zero.
4. I disconnected my external coupler from the radio and connected my Marconi 2019 signal generator to the bypass connector.
5. I checked to make sure "bypass on TX' was checked.
6. I let the Marconi warm up for 30 minutes
7. I set the Marconi to various output levels starting with minus 3dbm down to minus 20dbm during testing.
8. After each dbm setting on the Marconi I pressed MOX on the radio and read the resulting level on the panadapter and compared it to the level set of the Marconi.
9. Finally, I plugged the Marconi into antenna 1 with it set to -14dbm just to check it's frequency and amplitude with the radio. Both agreed.

The fact that I reported getting an ADC1 overload on several of the highest settings pretty much says I had the generator plugged into the Bypass port and had the radio keyed up and PSA ON; you can't get that otherwise as far as I know.

I'm happy to provde screen grabs of the process so you can see what I see if you like.

Joe

[w-u-2-o]

Joe--what mode were you in when you ran that test? USB? CW? Perhaps mode makes a difference? I am using USB.

[w9mdb]

I did a Level Cal using the XG3 at -33dBm
I now get
0dBm -- with S-ATT at 9 I get 2.2dBm and no ADC overflow
-33dBm -32.1dBm
-73dBm -72.6dBm
-107dBm -106.5dBm

That's getting closer.....

[w-u-2-o]

Level cal has nothing to do with the TX DUP display. This has been confirmed by Richie. See this link:

https://github.com/ramdor/Thetis/issues/137#issuecomment-1626949307

Quoting from that post: "None of the rx display calibration offset values are used when in TX DUP mode."

If what you are measuring is changing with level cal then you can't be measuring the DUP display during TX.

[Joe-W4WT]

Scott, I was in USB; same as you.

Does your "display cal", if I remember the correct name, make a diffeence in the error amount if you adjust it?

Oops, looks like I missed your last post about the Cal. I'm not sure I understand then, based on what you said RIchie said why it definitely made a difference with me. As I mentioned, I adjusted out the 3 +- db difference and it certainly showed on the display. I'm going to play with that again later today and make sure I saw what I think I saw.

Edit: I just looked and it is called "Tx Display Cal" on the Calibration tab. It definitely affects the DUP display on my system. Not sure when that adjustment showed up as I don't remember it from numerous versions previously but I could have just not noticed it.

Joe

[w9mdb]

As far as I can tell there is no manual adjustment. I'd have to check the code to see where it's saving it.

[Joe-W4WT]

Scott,

Have you or Mark tried saving and then resetting your database to rule out any other settings that we may not be aware of interacting with the display? I just did this to see if anything changed on my end. I saved, then reset and reloaded a default database and only went in and checked the Rx Bypass on Tx leaving everything else at default and then set up for and did the test again. I saw just the 3db difference I noted when I first did the test prior to adjusting the "Tx Display Cal" to remove the 3db error so it appears I don't have anything set in the database that "makes it work" but that doesn't mean you guys don't maybe have something set that makes it NOT work.

Just a thought,

Joe

[w-u-2-o]

I did a Level Cal using the XG3 at -33dBm

TX Display Cal is not the same as Level Cal.

Precision of language is important with a system this complex!

I get the same behavior (25dB error) with a reset database.

It's worth noting that Richie also saw the same issue.