Pocket-Sized 6 GHz 1 TS/s ET Scope

[joeqsmith]

Look forward to the software updates. 

Just watched the review - impressive job by Lasmux.  Loading and performance are outstanding for the price.

We think the probe might make a good accessory for the GigaWave. We've contacted Lasmux about this - there should be some way to benefit everybody (including users/customers).

Hopefully the two of you can work something out.  Maybe you could even get Lasmux to make all of your probes.  You mentioned your Z0 probe you were working on.   

One small thing, if you do work something out, it would be nice if the probe would interface directly with the scope without the need for adapters. 

[KE5FX]

One small thing, if you do work something out, it would be nice if the probe would interface directly with the scope without the need for adapters.

Or additional batteries.

[joeqsmith]

Would be nice if it were all powered off the PC but not at the expense of higher noise.  SJL had stated that the reason for using USB3 was for the higher currents but I thought they had a lot of margin.  Maybe there is enough left to power 4 additional probes at 250mA. 

[SJL-Instruments]

We had a chat with Lasmux earlier today.

In the short-term, we settled on offering $300 off any GigaWave scope when purchasing one of Lasmux's probes. Lasmux would hand out a coupon code which can be applied at our checkout.
For any customer who buys a GigaWave, it's then a no-brainer to also buy a probe (effectively for free). Lasmux gets extra sales at no cost. From our end, we know we missed out on several customers due to the lack of probes, so we'd get an increase in sales as well.

Lasmux is working on listing a version of the probe with an SMA-SMA cable, so no adapters are required. Once that's up, both of us will advertise the deal on our websites.
This offer will apply retroactively (for those who already bought the probe).

Would be nice if it were all powered off the PC but not at the expense of higher noise.  SJL had stated that the reason for using USB3 was for the higher currents but I thought they had a lot of margin.  Maybe there is enough left to power 4 additional probes at 250mA. 

It's possible to have a separate USB 3.0 breakout to supply power for up to 8 probes (60 mA @ 9V per probe, total 4.3 W).
We have a bit over 200 mA of extra margin on the scope, but exposing this would break compatibility with our existing units, so we'd prefer not to do this.

We didn't discuss this for our short-term plans. It could be a possibility in the medium-term, depending on the probe's power supply noise rejection. Lasmux knows much more about the details.

[SJL-Instruments]

During the review, you can how I am saving the waveforms and recalling them for a reference.  Something I commonly do when using a scope.  Note the number of steps required.   You have to manually select a file name, manually reload them, then manually set the gain and offset.   I like the idea of having a memory rather than a file and having the software automatically show the data, properly scaled like my other scopes.   The less buttons I need to press and typing I need to do, the better.   Something you may want to consider with future releases. 

As promised, this feature has now been implemented. Should help speed up the process considerably in the future.

[joeqsmith]

Very easy to use, works well.  I was thinking I may want you to adjust the scale of the reference if the sampled channel is changed, or automatically delete the reference if I want to resample but after playing with it, I think this is better.   It's simple enough to remove the old references. 

Good job and big thanks!

[joeqsmith]

The following was taken from:
https://www.eevblog.com/forum/testgear/show-us-your-square-wave/msg5285605/#msg5285605

GigaWave with Tektronix S-52 tunnel diode head (<25ps Tr)

https://www.sjl-instruments.com/

Nice, what I would like to see the real time behavior, as 20..80% or 10..90% rise time changes over time

On a LeCroy easy to graph over time as rise f(t).

Thanks for the suggestion  We'll add the ability to graph any measurement vs. time to the software. (Fastest update rate would be 1 datapoint per sweep.)

There's been some activity in that thread which reminded me about this request.    It's been about 5 months since the original post, so I thought I would set this up and show it.  I think I demonstrated this using my custom software back then. 

Using your latest release, I can not figure out how.   If it is a MATH function, the manual does not explain how to use it.  Looking in the Extended Measurement Information, I don't see any options to graph the data.

A few other comments, bugs?  If I select a custom scale, say 25mV/div.  Then I save that waveform as a reference.  When the new reference is displayed, us uses 20mV/div rather than the expected 25mV, requiring me to select Custom, then enter in the 25mV/div.    I thought these two would always track but it appears not.  Is there a reason why you made it work this way?   Personally, I would rather it always work the same way and IMO, less keystrokes is always better.

If I select Color Grading for a waveform with vectors disabled, then save the waveform as a reference, the new waveform is shown with vectors disabled but the color grading is deselected.  Is there a reason it does not match?

Have you ever considered adding a way to auto align the data?   So, for example, we want to plot the risetime.   Now, if the hardware drifts, the transition may move off the screen.   For a long term trend, have you considered having the softeware adjust the position automatically (as a option) to keep the transition centered? 

[SJL-Instruments]

Using your latest release, I can not figure out how.   If it is a MATH function, the manual does not explain how to use it.  Looking in the Extended Measurement Information, I don't see any options to graph the data.

Whoops, this slipped out of the pipeline at some point. We have a software branch with this feature but it was not merged. We'll release it by EOW.
For now you can export the entire sweep history from the Extended Measurement Information window, and plot it externally.

A few other comments, bugs?  If I select a custom scale, say 25mV/div.  Then I save that waveform as a reference.  When the new reference is displayed, us uses 20mV/div rather than the expected 25mV, requiring me to select Custom, then enter in the 25mV/div.    I thought these two would always track but it appears not.  Is there a reason why you made it work this way?   Personally, I would rather it always work the same way and IMO, less keystrokes is always better.

This is a bug and will be fixed in next release (EOW).

If I select Color Grading for a waveform with vectors disabled, then save the waveform as a reference, the new waveform is shown with vectors disabled but the color grading is deselected.  Is there a reason it does not match?

We originally had the color-grading setting copied over, but during QA caught ourselves clicking the "Save as Reference" button multiple times because the software appeared unresponsive. What was happening is that the new overlaid waveform looked exactly identical to the original. The current behavior ensures that the new trace is a different color.

Have you ever considered adding a way to auto align the data?   So, for example, we want to plot the risetime.   Now, if the hardware drifts, the transition may move off the screen.   For a long term trend, have you considered having the softeware adjust the position automatically (as a option) to keep the transition centered? 

We can probably have this in by EOW. The UI requires some thought - given how the hardware works, only one transition can be followed at a time. We would allow the user to select one channel to track, and adjust the timebase at the end of each sweep.

[joeqsmith]

The reason I brought up the tracking is when looking at these fast transitions, I know that vintage Tektronix pulser will drift a fair amount. Because the pulser's cycle time is so slow,  sweep times can easily be several minutes.  Using the setting I selected back when I posted the risetime, it requires about 3 minutes to collect a single sweep. 

I assume the original question was looking for more than a couple of sweeps, and running a long term test (guessing maybe a day or so) is currently not possible with what I have on-hand without the aid of some sort of auto tracking.  That or I need to try and maybe temperature stabilize it.   

***
Showing two sweeps after an 8hr warmup.  Drifts about 25ps (added delay).   

[joeqsmith]

It may be nice to be able to adjust (drag) the reference independent from the other channels.  For example, if I collect data from channel 2, then change the setup to collect on channel 3 as shown.  I want to manually align the two edges.   

Also, just an FYI, the setup is a bit different than when I first took that data but we are still well in the ballpark of the original 35.8ps measured.  Also, all below your published typical 40ps, which is good to see!

[SJL-Instruments]

It may be nice to be able to adjust (drag) the reference independent from the other channels.  For example, if I collect data from channel 2, then change the setup to collect on channel 3 as shown.  I want to manually align the two edges.   

On a reference waveform, you can open the channel settings and enter a timebase shift to adjust the horizontal position.
On a "physical" channel, this setting only relabels the timebase (due to hardware limitations). The behavior for a reference waveform differs in order to allow exactly the application you're describing.

The reason I brought up the tracking is when looking at these fast transitions, I know that vintage Tektronix pulser will drift a fair amount. Because the pulser's cycle time is so slow,  sweep times can easily be several minutes.  Using the setting I selected back when I posted the risetime, it requires about 3 minutes to collect a single sweep. 

I assume the original question was looking for more than a couple of sweeps, and running a long term test (guessing maybe a day or so) is currently not possible with what I have on-hand without the aid of some sort of auto tracking.  That or I need to try and maybe temperature stabilize it.   

***
Showing two sweeps after an 8hr warmup.  Drifts about 25ps (added delay).   

Thanks for clarifying. For a single rising/falling edge this is not too hard. For arbitrary waveforms it should still be possible using autocorrelations, but we'll see if this works in practice.

[joeqsmith]

Showing the bigger problem, which you can't address with software.    If we want to increase the resolution as shown for example, the signal can very well drift enough during a single sweep to throw off the measurement.   In this case a sweep is more than 10 minutes.  Imagine the errors at 128pts/div.   Having a way to collect the data faster with slower trigger rates and without the need for a pre-trigger is really what I would want.  It's obvious and I doubt causes users to publish bad data.

[joeqsmith]

Users have to try and balance the drift and resolution to get accurate data.  Here shown with 128pts/div.  20ish minutes for the sweep.  We know that time has drifted a lot during the measurement and appears to have added a few ps of error.    Still, we are below that typical number and we are certainly pushing your scope's limits.   

[joeqsmith]

It may be nice to be able to adjust (drag) the reference independent from the other channels.  For example, if I collect data from channel 2, then change the setup to collect on channel 3 as shown.  I want to manually align the two edges.   

On a reference waveform, you can open the channel settings and enter a timebase shift to adjust the horizontal position.
On a "physical" channel, this setting only relabels the timebase (due to hardware limitations). The behavior for a reference waveform differs in order to allow exactly the application you're describing.

While it certainly allows me to adjust it.  I have to select the advanced tab and type in a number.  Personally I would like the option to be able to select and drag the reference waveform with the mouse rather than having to resort to typing on the keyboard.   

[SJL-Instruments]

Users have to try and balance the drift and resolution to get accurate data.  Here shown with 128pts/div.  20ish minutes for the sweep.  We know that time has drifted a lot during the measurement and appears to have added a few ps of error.    Still, we are below that typical number and we are certainly pushing your scope's limits.   

With a 70 ktrig/s source, we were able to get sweep times of less than 30 seconds @ 128 pts/div by setting all the CDF settings (Nmin, Nmax, K) to 20.
Of course, this only gives good results in vector trace mode. The settings will need to be increased to accurately resolve the rare parts of the distribution. But consider that a traditional ADC-based sampling scope would also need to run for a minute or so to get a good sense of the tails of the distribution.

That said, the speed-quality tradeoff is inherent to the architecture. As long as the user is aware of it, they can choose settings appropriately for the application.

In the long-term, we agree that getting rid of the pre-trigger is ideal, and is something we are currently working on.

While it certainly allows me to adjust it.  I have to select the advanced tab and type in a number.  Personally I would like the option to be able to select and drag the reference waveform with the mouse rather than having to resort to typing on the keyboard.   

Got it. We can assign middle click + drag to this functionality. Currently, dragging horizontally with either the left or the middle mouse button do the same thing.
Figuring out if the user is trying to interact with a specific waveform based on mouse position is error-prone (especially in point mode).

[joeqsmith]

Got it. We can assign middle click + drag to this functionality. Currently, dragging horizontally with either the left or the middle mouse button do the same thing.
Figuring out if the user is trying to interact with a specific waveform based on mouse position is error-prone (especially in point mode).

I was thinking the user would click on far right area to select the waveform they want to work with.  Maybe change the box thickness or some other way to graphically denote it is the selected waveform.  Then just drag away.   

Indeed playing with CDF settings will improve sweep rates, but at a cost.  Shown with 202020.   

There does appear to be some strange interaction with the CDF settings and trigger that I could never explain.   I can reproduce it on this end and it has worked this way for some time.  I'll try and walk you through what I see using the Tektronixs pulser.

Power up unit using defaults.   Set the trigger level to 500mV.  Set the position to center the waveform (86ish ps).  Set the Base to 20ps/div.  Set resolution to 128 pts/div.  Set the holdoff to 150ns.    Sweeping should be very slow (>10 minutes).    Next select the CDF sampling options and use your 202020 settings and apply.   Change the resolution to 4 pts/div, sweep is maybe 2 seconds.   

Now the fun part, change the CDF sampling options back to auto and apply.  Now, select a resolution of 128 pts/div.   This should get us back to the initial state but the sweep rates will be much faster.   Here is the trick.  Set the trigger level to 100mV.  Now change it back to 500mV.   If all is right, I am expecting it to again sweep very slowly as it did in the start.   Now why this trigger level being set to something else and then returned to the initial value has any effect on the sweep time, you would need to chime in.  I'm sure it's a user problem and some misunderstanding I have.   Guessing the settings are not critical as I have ran into this before.  The order does not seem to matter.  Nor does changing the holdoff.  It seems tied to having to change the trigger level.  Like the act of doing that is somehow causing some state of the software to change. 

If this is too confusing to follow (I take no offense to that  ),  I can make a small video for you demonstrating what I am seeing. 

*******
It appears that when selecting the "Automatically determine from holdoff"  the software does not immediately calculate the new values.  Rather is seems to require some event such as changing the trigger level or the holdoff time.   Maybe this is by design but personally, I would rather it calculate the new values based on the current settings instead of having to change values to get it update.   

*******
Also, if I load settings that do not have the CDF settings set to auto, they will take effect immediately.   If I then load settings that use auto for the CDF, the software does not appear to update to the correct values.   Lots of clues. 

[joeqsmith]

Using your 202020 settings and a resolution of 8pts/div,  I place my hand on the Tektronixs pulser to cool it to give you some sense of the drift I am referring to.   Note how it shifts the phase about 60ps.   Zoomed into 10 ps/div, you can see how this bit of drift will easily cause the edge to go off screen.   

[joeqsmith]

I tried to save a reference for each sweep.  Doing so, I notice the maximum allowed is letter X.  More than enough.  What is interesting is if you select save reference again, it doesn't round robin.  Rather it saves to Y but it turns off the readable scales.    Selecting save again appears to have no effect.  The only way I found to enable the scales is to turn off reference Y.   Personally, I would rather it cycle through the references and never disable the scales.  Once X is reached, it starts over writing into location A.       

****
It would also be nice to know when looking at reference A-X, which channels they were derived from.   Maybe you could use A1 for channel one for example.  Or maybe just add a new indicator.

I don't know if you have had any feedback on how you allow changing of the scale for example by scrolling the center button.  I like the concept but it doesn't seem to always work.  Sometimes I have the mouse positioned on the scale box and rotating the button will cause the software to scroll through the waveforms rather than changing the scale selection.  Even if I left click and the software highlights the scale text, the center button will instead scroll though the waveforms.   Other times, I am trying to scroll through the waveforms and the mouse rests on one of the scales and it decides to update the scale.   I like the concept but it just needs a little work IMO.   

[SJL-Instruments]

It appears that when selecting the "Automatically determine from holdoff"  the software does not immediately calculate the new values.  Rather is seems to require some event such as changing the trigger level or the holdoff time.   Maybe this is by design but personally, I would rather it calculate the new values based on the current settings instead of having to change values to get it update.   

*******
Also, if I load settings that do not have the CDF settings set to auto, they will take effect immediately.   If I then load settings that use auto for the CDF, the software does not appear to update to the correct values.   Lots of clues.

I tried to save a reference for each sweep.  Doing so, I notice the maximum allowed is letter X.  More than enough.  What is interesting is if you select save reference again, it doesn't round robin.  Rather it saves to Y but it turns off the readable scales.    Selecting save again appears to have no effect.  The only way I found to enable the scales is to turn off reference Y.   Personally, I would rather it cycle through the references and never disable the scales.  Once X is reached, it starts over writing into location A.

Both of these bugs are fixed in the below prerelease:
https://gigawave-releases.s3.us-east-2.amazonaws.com/GigaWave_v2.6.7_PREVIEW20240609_Windows.zip

       

It would also be nice to know when looking at reference A-X, which channels they were derived from.   Maybe you could use A1 for channel one for example.  Or maybe just add a new indicator.

In the pre-release, we have added a user-editable label to all reference channels. By default, the label indicates the source channel (or the filename, for imported waveforms). Should help a lot with bookkeeping.

I don't know if you have had any feedback on how you allow changing of the scale for example by scrolling the center button.  I like the concept but it doesn't seem to always work.  Sometimes I have the mouse positioned on the scale box and rotating the button will cause the software to scroll through the waveforms rather than changing the scale selection.  Even if I left click and the software highlights the scale text, the center button will instead scroll though the waveforms.   Other times, I am trying to scroll through the waveforms and the mouse rests on one of the scales and it decides to update the scale.   I like the concept but it just needs a little work IMO.   

This was accidental and is a consequence of how Windows handles comboboxes by default. In the prerelease, we disabled scrolling the scale setting to avoid all ambiguity. Since you found it useful, we may re-enable it in the official 2.6.7 release, but only after clicking on the scale box.

Your other suggested changes (not listed above) will go into the 2.6.7 release later this week.

[joeqsmith]

Still does not appear correct when loading the setup with CDF set manual vs auto.  Odd is if I expected it to be in auto after loading the setup, I select the  CDF sampling options and it changed to auto and starts sweeping very slowly.  If I select apply (auto is still selected) it appears to change back to manual.   Really odd.

Attached is what I interpreted the OP was asking for.  My software will plot the histogram and min/max but I guess I had not added the transition time vs sweep.   The scope was set for some very fast sweeps just to get some data to plot.   I expect if we set it up for higher resolution, the noise would significantly reduce.  Shown is about an hour or so of data.   Once you have your software working, we can run this test for real.

[joeqsmith]

I added the ability for my software to track the temperature drift, allowing me to collect data for a longer time.   I used a resolution of 5ps and started out with Triggers per Sample and Samples per CDF set to 10.   After 230 sweeps I changed them to 20.   After 540 they were set to 25.  It does help improve the measurement.   Of course, more data, the longer the sweep time.

[joeqsmith]

If we use the LiteVNA for a source and track the measured risetime, we see about a 2.5ps error.  Much tighter than the 12ps we were seeing.    With the faster trigger rate, we can collect a lot more data at a reasonable time, greatly improving the measurement.   

***
It's continued to run over the afternoon.  The combined error is still under 3ps.   

[SJL-Instruments]

Still does not appear correct when loading the setup with CDF set manual vs auto.  Odd is if I expected it to be in auto after loading the setup, I select the  CDF sampling options and it changed to auto and starts sweeping very slowly.  If I select apply (auto is still selected) it appears to change back to manual.   Really odd.

This is fixed in version 2.6.7 (just released). We have added the time drift auto-tracking, liveplotting, and histogram features (alongside various bugfixes). Looking forward to seeing the results of the test.

[joeqsmith]

Thanks!  I will download and try it out over the weekend. 

****

Very nice.  Simple to use.

This is a different LiteVNA (other one is in use) but the values are very close to what I saw using my software with the other unit.  I also tried out some of the problems I brought up.  So far, these all appear corrected.  I think the dialog box popping up when you reach the maximum reference waveforms is fine.  Really, that's a lot of waveforms and I am not expecting this would be a typical case anyway.  I have not yet tried the tracking.   

https://www.eevblog.com/forum/testgear/show-us-your-square-wave/msg5542531/#msg5542531

I'm not liking that when the tracking is enabled, all other functions are disabled.  The problem is when you zoom in and start the tracking, it may already have drifted too far.  By the time you shut off tracking, close the menu, get things aligned again, then re-enable tracking, it may have already drifted too far.   Then you get things finally tracking and want to rearrange all the windows or add something, you start the whole process over again. 

It does track and I can at least make the measurement but it's a pain to use with my old Tektronix pulser having so much drift.   One fix would be let everything warm up for a half hour or so until the drift is less per time unit.  Then enable the tracking.  Personally, I would drop this limitation.  It's not like the PC is being stressed.   

[SJL-Instruments]

As Lasmux has just announced on his thread (linked below), our partnership with Lasmux Devices is now active.
https://www.eevblog.com/forum/testgear/12-ghz-active-probe-project/125/

After purchase of a 2.7 GHz active probe from Lasmux, you can receive a voucher for US$300 off any GigaWave scope. The probe itself costs less than this, so it's no-brainer if you were thinking about buying a GigaWave anyway.

If you already purchased a probe, the offer applies retroactively - just contact Lasmux for the voucher code. We recommend buying the SMA-SMA version of the probe for use with the scope.

In our opinion, the probe is extremely impressive for the price, and complements our scope well. We recommend watching joeqsmith's review of the probe (post #111 in the above linked thread) if you're interested.