Severe problem with PureSignal

[K1LSB]

My early model 7000DLE MKII has a problem with PureSignal behavior on several bands, but it's so severe on 20 meters that it's totally unusable.

Here's a snapshot of the PS curves on 20 meters, captured with the internal sampler at 41 watts drive. The resulting TX audio is quite nasty!

20 meters 41 watts.jpg (61.44 KiB) Viewed 8469 times

It's my understanding that the design intent of PS is to apply an exact mirror image correction curve of the measured nonlinearity in the PA output, which is not at all what I'm seeing in the screenshot...the red "Correction" curve is oscillating wildly and is nowhere near a mirror image of the blue Measured curve coming from the PA.

Can someone with specific knowledge of how PS is supposed to work please help me understand what the problem is? In its current state, the problem with PS on 20 meters renders that entire band completely unusable on my radio.

Truth be told, I suspect there is such severe non-linearity in my PA that the PS algorithm is simply failing in its effort to correct for it.

For reference, here's a Drive folder of PS screenshots taken on every band at several different drive levels. As you can see, there is no drive level at all on 20 meters where PS functions as it should. All screenshots were taken with the internal sampler while the radio was at room temperature (cool to the touch).

https://drive.google.com/drive/folders/ ... sp=sharing

Also for comparison, I've included in that folder a snapshot of Juha NI2M's PS curves (I spent a weekend with his 7000DLE MKII running at my QTH). Man, I wish my radio was that clean!

Thanks in advance,
Mark

[w-u-2-o]

What happens if you try this into a dummy load?

[K1LSB]

Every one of those screenshots was into a cantenna dummy load.

Mark

[w-u-2-o]

Please post a screen shot of your Linearity > Advanced window.

What S-ATT value are you getting with 41W on 20M?

[K1LSB]

Please post a screen shot of your Linearity > Advanced window.

What S-ATT value are you getting with 41W on 20M?

Here's my PS settings, I've tried the full gamut of settings over the years but this is what works best for me:

PureSignal Settings.jpg (66.46 KiB) Viewed 8438 times

I'm at work so I can't see the S-ATT value but I'm sure it's what would be expected from the internal sampler at 41 watts drive.

Regardless, all of those PS curves look very nearly the same irrespective of whether I'm running barefoot with the internal sampler or from the external sampler behind my AL-80B while running 600-ish watts.

Mark

[w-u-2-o]

I'd recommend turning off MAP and making your amp delay at least 150nS, but those are nits. Other than that all those numbers look good.

The data coming in on 20M just looks plain old bad.

Is this also a problem when using the external sampler, either barefoot or with the amp?

[K1LSB]

Yes, all of those PS curves look very nearly the same regardless of whether I'm on the internal sampler or an external sampler.

I'm not smart enough to be able to imagine what might be causing the ugly PS curves, other than maybe a very bad non-linearity in my PA stage or a wonky feedback circuit. But those are straight-up wild guesses, and in either case would point to a hardware issue.

I was running Juha's machine with the exact same settings (in fact, using the same database) as my own rig and his rig behaved beautifully running PS on all bands, which further indicates it's something internally wrong with my radio.

I almost never operated 20 meters before so it wasn't until long after the warranty had expired that I discovered I had a problem. But it's abundantly clear to me that I can't operate clean on 20 meters phone at all.

Also, look at 12 meters running 85 and 90 watts. Those curves strongly suggest a PA issue somewhere:

12 meters 85 watts.jpg (61.7 KiB) Viewed 8433 times

12 meters 90 watts.jpg (61.89 KiB) Viewed 8433 times

Mark

[K1LSB]

Scott,

When I got home this evening I checked the S-ATT for each band at 41 watts:

Band....S-ATT
160.......13
80.........13
60.........13
40.........13
30.........13
20.........12
17.........12
15.........12
12.........12
10.........13
6...........7

I also made the changes to the PS settings you suggested but no discernible effect was observed.

So do you think I have a hardware problem?

Mark

[w-u-2-o]

I'm not sure what the problem is, yet.

1. The plots you show on 12 meters, when the blue line starts to go horizontal at the top right corner that indicates that the internal amp is badly overdriven and well into saturation. That's obviously bad. Have you calibrated the internal watt meter? Do you have an external watt meter you can use to double check that?

2. I'm assuming you are using two-tone to make these plots?

3. What sort of external coupler do you have? What is it's coupling factor? I'd suggest trying it barefoot in lieu of the internal coupler to see if a problem with the internal coupler can be ruled out. Don't forget to activate the Bypass on TX option.

[K1LSB]

I'm not sure what the problem is, yet.

1. The plots you show on 12 meters, when the blue line starts to go horizontal at the top right corner that indicates that the internal amp is badly overdriven and well into saturation. That's obviously bad. Have you calibrated the internal watt meter? Do you have an external watt meter you can use to double check that? I calibrated the internal PA and watt meter a few months after purchasing 3 years ago using the services of Mike N5INP, see the instrument bench on his QRZ page. https://www.qrz.com/db/N5INP

2. I'm assuming you are using two-tone to make these plots? No, PS is locking onto voice into the mic, the lock is very fast and strong, virtually instantaneous.

3. What sort of external coupler do you have? What is it's coupling factor? I'd suggest trying it barefoot in lieu of the internal coupler to see if a problem with the internal coupler can be ruled out. Don't forget to activate the Bypass on TX option. I have a homemade pure resistive sampler that's tapped directly into the center conductor of the coax coming out of the rear of the AL-80B and going to the antenna. I'm not smart enough to calculate the dB padding, it's an 18K-ohm resistor pack soldered to the coax center conductor.

Here are comparative snapshots of different bands sampled through the internal and external samplers, all captured this evening at 41 watts, with S-ATT readings embedded in the titles:

17M external 41W, S-ATT=9.jpg (61.37 KiB) Viewed 8396 times

17M internal 41W, S-ATT=12.jpg (61.87 KiB) Viewed 8396 times

20M external 41W, S-ATT=7.jpg (61.85 KiB) Viewed 8396 times

20M internal 41W, S-ATT=12.jpg (61.12 KiB) Viewed 8396 times

40M external 41W, S-ATT=5.jpg (59.3 KiB) Viewed 8396 times

40M internal 41W, S-ATT=13.jpg (59.95 KiB) Viewed 8396 times

60M external 41W, S-ATT=7.jpg (59.83 KiB) Viewed 8396 times

60M internal 41W, S-ATT=13.jpg (60.48 KiB) Viewed 8396 times

Mark

[vk1hx]

I've seen this on setups where RFI and issues in the firmware (P2 and P1) was generated internal crosstalk on the feedback.

I've had success in correcting these issues through applying several "tweaks" in the OS running thetis, Networking.

Reach out if you require assistance, happy to help.

[K1LSB]

Phil,

PM sent.

Thanks,
Mark

[w-u-2-o]

There is no software, firmware or network tweak that can fix RF crosstalk. If changes to network parameters fixed a problem all it fixed was bad or missing UDP packets. That's not crosstalk.

BREAK

Mark,

FWIW, this is what I'm seeing:

1. There is no way you should be seeing the internal amp overdriven into saturation like that at 100W. Therefore either your internal watt meter calibration is off (with all due respect to Mike's fine lab), or your ANAN is broken somehow. Has it always worked like this, or has something changed, and when did it change?

2. Because both the internal and external sampler are giving nearly identical results, it doesn't seem like it can be a cable problem or a problem with either the internal or external feedback path.

3. Grasping for straws: have you tried backing up your database and trying with a fresh, reset database? If not, definitely try this, especially if the problem just suddenly appeared out of thin air.

[K1LSB]

There is no software, firmware or network tweak that can fix RF crosstalk. If changes to network parameters fixed a problem all it fixed was bad or missing UDP packets. That's not crosstalk.

BREAK

Mark,

FWIW, this is what I'm seeing:

1. There is no way you should be seeing the internal amp overdriven into saturation like that at 100W. Therefore either your internal watt meter calibration is off (with all due respect to Mike's fine lab), or your ANAN is broken somehow. Has it always worked like this, or has something changed, and when did it change?

2. Because both the internal and external sampler are giving nearly identical results, it doesn't seem like it can be a cable problem or a problem with either the internal or external feedback path.

3. Grasping for straws: have you tried backing up your database and trying with a fresh, reset database? If not, definitely try this, especially if the problem just suddenly appeared out of thin air.

Unfortunately, I've not been operating on 20 meters before now so I can't really say whether the problem has always existed or not. Comparing the PA output on 20 meters vs all of the other bands, every power meter in my shack is in agreement that the power output on 20 meters is in accord with the power output on all of the other bands.

I've done multiple database resets, including deleting the database and starting from scratch.

Here's one very basic question that I hope someone can provide a definitive answer to: Should the Measured and Corrected curves in the PS graph be mirror images of each other? If the answer is Yes, then PS is clearly not functioning correctly at any power level on 20 meters, which is prima facie evidence that there's something wrong with my radio that needs fixing.

Mark

[w-u-2-o]

Unfortunately, I've not been operating on 20 meters before now so I can't really say whether the problem has always existed or not. Comparing the PA output on 20 meters vs all of the other bands, every power meter in my shack is in agreement that the power output on 20 meters is in accord with the power output on all of the other bands.

But the saturated amp example was (so far) only shown on 12M. You definitely have a problem with an overdriven amp on 12M based on what you posted. You never posted a 90 or 100W measurement on 20M.

Here's one very basic question that I hope someone can provide a definitive answer to: Should the Measured and Corrected curves in the PS graph be mirror images of each other? If the answer is Yes, then PS is clearly not functioning correctly at any power level on 20 meters, which is prima facie evidence that there's something wrong with my radio that needs fixing.

The answer is yes.

To put a really fine point on it, there are conditions under which PureSignal cannot obtain a good solution. But looking at the blue input curve that is not a challenging condition for PureSignal.

What version of Thetis, and what version of firmware, are you running?

[K1LSB]

I'm running Protocol 2 v2.1.18 for the early 7000DLE MKII. Currently running Richie's Thetis v2.10.1 but I've tried a number of early and late versions of Thetis from both Richie and Steve, with no difference in PS behavior between any of them, on any band.

PM sent.

Mark

[w-u-2-o]

At this point I can only suggest that you try to downgrade to P1 firmware and see if the problem continues. If it does, then we can rule out any firmware issues as well.

I'd also like to see you try using two-tone to make these measurements. Try two tone on 20M with both P2 and P1 firmware.

[K1LSB]

At this point I can only suggest that you try to downgrade to P1 firmware and see if the problem continues. If it does, then we can rule out any firmware issues as well.

I'd also like to see you try using two-tone to make these measurements. Try two tone on 20M with both P2 and P1 firmware.

I can certainly do that this evening, I'll post the results.

I don't even recall ever running Protocol 1, I think I loaded P2 as soon as the radio arrived.

Thanks,
Mark

[K1LSB]

At this point I can only suggest that you try to downgrade to P1 firmware and see if the problem continues. If it does, then we can rule out any firmware issues as well.

I'd also like to see you try using two-tone to make these measurements. Try two tone on 20M with both P2 and P1 firmware.

Here are a bunch of two-tone tests on the internal sampler. This first batch is 20 meters on Protocol 1:

20M P1 17W.jpg (60.53 KiB) Viewed 8222 times

20M P1 25W.jpg (59.29 KiB) Viewed 8222 times

20M P1 33W.jpg (59.53 KiB) Viewed 8222 times

20M P1 41W.jpg (61.34 KiB) Viewed 8222 times

20M P1 50W.jpg (61.4 KiB) Viewed 8222 times

20M P1 70W.jpg (61.49 KiB) Viewed 8222 times

20M P1 90W.jpg (60.75 KiB) Viewed 8222 times

More to come,

Mark

[K1LSB]

This next batch is 20 meters (still two tone) on early Protocol 2 pre4:

20M P2pre4 25W.jpg (61.09 KiB) Viewed 8222 times

20M P2pre4 33W.jpg (60.08 KiB) Viewed 8222 times

20M P2pre4 41W.jpg (61.05 KiB) Viewed 8222 times

20M P2pre4 50W.jpg (60.82 KiB) Viewed 8222 times

Still more to come,

Mark

[w-u-2-o]

That's good. More to follow when I get to a real keyboard.

[K1LSB]

This batch is 20 meters two tone on the latest Protocol 2 v2.1.18:

20M P2v2.1.18 25W.jpg (61.8 KiB) Viewed 8220 times

20M P2v2.1.18 33W.jpg (60.54 KiB) Viewed 8220 times

20M P2v2.1.18 41W.jpg (61.49 KiB) Viewed 8220 times

Yet more to come,

Mark

[K1LSB]

And this last batch is 40 and 80 meters two tone on Protocol 1 and Protocol 2 v2.1.18:

40M P1 41W.jpg (59.66 KiB) Viewed 8219 times

40M P2v2.1.18 41W.jpg (60.76 KiB) Viewed 8219 times

80M P1 41W.jpg (61.4 KiB) Viewed 8219 times

80M P2v2.1.18 41W.jpg (61.41 KiB) Viewed 8219 times

Mark

[K1LSB]

So on 20 meters there does appear to be some slight differences between P1 and early P2 and the latest P2 in the range of 33 watts drive, but notwithstanding that I'm not seeing enough difference to make an argument that firmware is the problem in any case. 20 meters is broken at any power level, on any firmware.

Mark

[w-u-2-o]

Mark--your work has ruled out a difference between internal and external couplers, and between P1 and P2.

The next most likely cause is some aberrant RF or RFI condition that is causing bad data. Although you are running into a dummy load such a possibility cannot be discounted. Warren saw your post and sent me a note suggesting the same thing.

What I'd recommend is to continue to gather more data in an effort to discover the problem:

- Concentrate on 20M only, that's the worst case.
- Redo the 25 and 41W plots without the low res button being checked--let's look at the full resolution plots.
- Use two-tone for all tests in order to maintain the most consistent input conditions.

Things to consider:

- A bad RF cable (internal or external).
- A bad dummy load (what does all this look like into an antenna?).
- Grounding and bonding issues.

You might drop the rear panel on the 7000 and check that all internal RF cables and connectors are secure.

[K1LSB]

Thanks Scott and Warren, I'll pursue those possibilities after work tomorrow, I'm off Friday so I have time this weekend to focus on that.

Mark

[K1LSB]

Here are some shots of 20 meters at 25 and 41 watts into the Cantenna, the internal 300W dummy load in my 989D tuner and a tuned wire dipole. Some are low-res and some are hi-res.

A fresh 4-foot LMR-240 cable was installed going to the Cantenna. A 12-inch LMR-240 cable runs from the 7000 to the AL-80B, a 3-foot LMR-400 cable runs from the AL-80B to the tuner, and 75 feet of LMR-400 runs from the tuner to the wire dipole.

Some of the shots look somewhat different than yesterday, no idea why, maybe because it's cooler in the shack this morning or maybe because the mains voltage is 125.2 this morning and was 118.7 yesterday.

Edit: I was confused, a 12-inch length of LMR-400 runs from the 7000 to the AL-80B (an 18-inch RG-400 is feedback from the external sampler).

20M 25W HR 989D Dummy.jpg (66.02 KiB) Viewed 8173 times

20M 25W HR Cantenna.jpg (65.65 KiB) Viewed 8173 times

20M 25W HR Dipole.jpg (64.43 KiB) Viewed 8173 times

20M 41W HR 989D Dummy.jpg (62.05 KiB) Viewed 8173 times

20M 41W HR Cantenna.jpg (62.07 KiB) Viewed 8173 times

20M 41W HR Dipole.jpg (61.43 KiB) Viewed 8173 times

20M 41W LR 989D Dummy.jpg (61.85 KiB) Viewed 8173 times

20M 41W LR Cantenna.jpg (62.13 KiB) Viewed 8173 times

Mark

[w4bf]

Look at linearity with PS off using spectrum display.

This is 100 watts - 3rd order IMD on my 7000 about -30 dB from each tone.

With PS ON, it drops to -52 dB.

Sounds like his internal HW has issues since he gets same results with amplifier.

[K1LSB]

Looking at my latest screenshots, I see no differences between the Cantenna and the 989D dummy load and the wire dipole, nor between the lo-res and hi-res graphs.

So I think we can eliminate all of the above (loads and graph resolution) as issues.

Mark

[w-u-2-o]

Agreed. So now it's grasping at straws time. I'd drop that rear panel and check all the RF connections you can reach to make sure they are secure.