New MSOX3104Ts for 1/2 list from Keysight

[rhb]

Keysight has an unspecified quantity of these available on their eBay store.  They are listed as "factory refurbished with all standard accessories, calibration with data and full warranty".  I'm guessing these were used for a trade show booth.  The listing is at 45% off, but I got one at 50% off including the full application bundle.  But they charge sales tax which nearly gave me a heart attack as I wasn't expecting the bill to go up $825 over my offer.  It cost more than any vehicle I've ever bought for myself, including the one new one, a Toyota pickup I bought in 1993.

I don't have a business justification for this.  But I'm about to turn 65 and I have *never* spent this much on any indulgence.  No boat,  No motorcycle. And I never took expensive vacations.  I don't need 1 GHz, but this will fit in with my old HP gear a lot better than a certain Chinese MSO that is so confusing to use I can't repeat a setting 5 minutes later without 3 minutes of stumbling around trying to figure out how I did it the first time.

Given that they are SW upgrade designs, you might be able to negotiate a BW downgrade to get a lower price.  It's an insane amount of money for a hobby, but it *is* a a current top of line HP product even if the name is different.

[nctnico]

I'd cancel it if you can. You can spend such a large chunk of change on a much nicer oscilloscope if you don't mind having less bandwidth.

[TheSteve]

Given that they are SW upgrade designs, you might be able to negotiate a BW downgrade to get a lower price.  It's an insane amount of money for a hobby, but it *is* a a current top of line HP product even if the name is different.

The 1 GHz model is unique and does not support any software bandwidth downgrading etc.

Very nice scope btw, still a painful price.

If you want to see more details on the hardware and bandwidth changes etc some of it is covered here:
https://www.eevblog.com/forum/testgear/keysight-3000t-scope-1-ghz-hardware-upgrade-mod/

[rhb]

Like the pig at breakfast, I'm committed.  That happened when I made the offer.  I've read so many MSO brochures my eyes are on the verge of rolling back in my head,  There are lies, damn lies and datasheets. I *did* get seduced by 1 GHz.  But, except for the sales tax I would have had it for 50% of list and at a better price than eBay used.

I had the impression from the Keysight website that you could do software upgrades from 100 MHz up to 1 GHz.  That not quite being the case would not surprise me.  I've been mauled by RF.

I'm still rather amazed I did this.  But after buying 6 pieces of early 90's HP gear I felt pretty comfortable with it.  Admittedly I did not look at the next level up from the  R&S RTB2K.  But this was a 1/2 off deal on the top end of a product line from one of the best T&M OEMs ever. 

I'm a little sad that it won't say Tektronix on the front just for nostalgic reasons.  I miss the days when "Tektronix" was synonymous with "oscilloscope".  But the engineers lost control to the accountants. 

I worked for several oil companies where that happened.  The best of all was ARCO, Atlantic-Richfield. Amazingly, ARCO was built by R.O. Anderson who was the president of the Dallas Fed before getting involved in the management of an oil company.  Though I learned recently his family had a long history in the financial side of the business.  After Anderson was forced out as CEO in the late 80's he started another oil company.

The only person whom I would rank above Anderson is George Mitchell who developed the shale gas play.  His father died in the office working on a prospect and that was George's ambition.  After selling out of the Barnett, George did a deal with Freeport-McMoran  for Davy Jones.  This was acreage that they had adjacent to Blackbeard.  That was a deep shelf gas play close to the coast.  The partners spent $225 million but had to plug and abandon before reaching the target because of severe over pressure. The lead partner wanted to try again, but the others wouldn't agree.  George hit 400+ ft of pay at Davy Jones.  But at current gas prices, it is not economic to produce because of the technical problems.

The real point of posting this is that it appears to me that Keysight sells off the trade show demos on eBay.  So if you're interested in a piece of current gear, tracking the trade show schedule could be very beneficial.  And you might want to make a lower offer ;-)

[GlowingGhoul]

You made a fine choice. Keysight is refined in ways the others aren't.

It's a pleasure to use. 

[apblog]

Congratulations.  I know you will love it!

They did this deal a few years ago and I jumped on it then, even though it was stretch for me.  Best decision I ever made.

The unit I received was indistinguishable from new, except for faint scratches on the autoprobe interface pad.

It's made a huge difference for me, so I recommend it if anyone is on the fence.

It looks like the full options bundle is lot cheaper too if you buy it from the ebay store when you buy the scope.

There have been some annoyances though, with regards to the software, in particular:

There doesn't seem to be any good way to measure or even just characterize jitter.  The measurement trend math function frequently seems to run out of points halfway across the screen, and there is no histogram display which would be very helpful for understanding jitter. For a scope this expensive, I really expected that it would have some way to perform such a common measurement.   Turning on infinite persistence (as the manual recommends) isn't as helpful as one might expect.

There's also the fact that when you buy the full options package, it doesn't include FUTURE options.  So even though I bought the full options bundle, I didn't get the new options that were released 6 months later (and were included in the new options bundle) -- one of which could have been very useful to me.

That said though, this is a fantastic piece of equipment, and it's saved my bacon several times since I bought it.  It's miles ahead of anything from popular cheap brands, and half price for "essentially new" is a steal. 

[rhb]

I'm saddened to hear that buying the full bundle does not include updates.  I should expect better of Keysight.     The bundle I bought includes jitter analysis.

The only thing I don't like is the memory depth. But I'm an old hand at DSP going back the the days of an 11/780 and an FPS AP120B.  The 11/780 had 4 MB of core and the AP120B had 64 KB.  So it's more nuisance than obstacle.

My main interest is designing  low cost T&M gear that is reasonable for a recent graduate to buy for a personal lab.  A recent grad should be able to make the necessary measurements at home over the weekend without busting the budget.  A couple weeks wages for everything they need to do basic work.

[GerryBags]

I watched a film from the eighties about Tektronix' domination of the oscilloscope market of the time, just as the days of the analog oscilloscope were ending and everyone was moving to DSO's. A few of the Tek employees, including the head of the in-house CRT plant and the guy in charge of the digital side, expressed concern with the risk involved in shutting down the CRT plant.

The whole thrust of the film was that it was having the CRT manufacturing in-house, and having honed it to the point at which they were years ahead of the nearest competitor, that gave them the edge in the market as a whole. Nobody else had anything like the CRT's in the 2400 series, or the 1 GHz 7000 (7104?) series, and even before then many 'scope makers bowed out altogether. It turned out that by shifting to digital processing of the signal they were levelling the playing field and ditching their USP. Perhaps they had no choice, in the face of market pressure, but it seems there is still a market for analog 'scopes and I can't help wonder what advances in CRT tech would have been possible with the advent of materials like graphene, had their state-of-the-art CRT plant still been extant.

I'll see if I can find the link.

[TheSteve]

You will find the scope a true joy to use. The touch screen is much nicer then I thought it would be and is very responsive.

The 3000T series all use the PCB but they have different hardware build lists for the 100/200 MHz, 350/500 MHz and 1 GHz models.
You can upgrade a 100 MHz to a 200 MHz via software license and upgrade a 350 MHz to a 500 MHz via software license but any other upgrades require the unit be sent back to Keysight for a main board swap.

[wpwrak]

My main interest is designing  low cost T&M gear that is reasonable for a recent graduate to buy for a personal lab.

I guess an efficient approach for make very powerful yet inexpensive gear would be to find a manufacturer who a) makes reasonably inexpensive devices, and b) who would be willing to release enough information about their hardware that an Open Source firmware could be made for it. I.e., separate optimizing manufacturing cost from providing flexibility - all the "designed for openness" devices seem to have either abysmal performance or an unpleasant price tag, and a company that excels at manufacturing a top-notch scope for a few hundred dollars may be struggling with software development.

Several years ago, a small German (I think) scope manufacturer who was about to go bankrupt opened their platform. They did this far too late to save them, but people started to develop their own firmware, and there was a brief surge of demand.

Now, imagine a company that is not in trouble doing the same. Nonsense like decoding options would disappear instantly - it would run sigrok or similar natively. You'd have a bunch of user interface styles to choose from, and all sorts of fringe extensions. And overall software quality would improve. Unless the hardware has major flaws, it should sell better than "hackable" Rigols.

One problem is that the companies that do make very affordable scopes are in China. Very few companies in China "grok" Open Source. E.g., I think Siglent would be perfectly positioned to take such a role with their recent models, with well-established chips like Xilinx Zynq, and where generally the main complaints are about software flaws. But would they understand what value such a step would bring to their product ? And who could convince them ?

In the Linux community there is also something called "anti-vendor ports", Linux ports that are based on reverse-engineering, without help from the manufacturer. This is a painful but sometimes viable approach. But, as far as I know, nothing of this sort has happened with scopes so far. There can be several reasons for this, e.g., proprietary chips that are just too difficult to reverse-engineer. But maybe that could also be a path towards affordable "jack of all trades" (and master of some) equipment.

- Werner

[tszaboo]

I had Keysight refurbished  scope at work. Honestly, I couldn't tell if it was used, came with all accessories new. We were joking, that they just slap the refurbished  label on a new scope, because they would still make money on that.

But I wouldn't buy for myself anything faster than 350MHz. You need active probes for that.

[nctnico]

I'm still rather amazed I did this.  But after buying 6 pieces of early 90's HP gear I felt pretty comfortable with it.  Admittedly I did not look at the next level up from the  R&S RTB2K.

When I wrote my previous post I was thinking of the R&S RTM3k  for which you can also buy an option bundle. It has seriously more memory compared to the MSO3k. If you enable all channels the 4Mpts erodes to only 500kpts/channel where the RTM3k has 40Mpts. For that reason alone a Keysight scope wouldn't even make it on my long list. I have owned an Agilent MSO7104A with 8Mpts and that already struggled in some cases.

[rhb]

I am mostly tooled up to develop FOSS FW for Zynq based scopes, Zybo Z7-20, MicroZed, Instek GDS-2072E and a BeagelBoard X15.  I still have a few things to install in the development system.  The Instek is the test platform.  There is a bit more to it than just FOSS as I am going to start with compressive sensing data acquisition routines (*after* doing regular Nyquist sampling).  I'm hoping that might get me some work as I find being retired rather lonely.

Other projects are a diode noise source following Franco Rota's design, but with hydraulically formed enclosure and a low cost thermal RF power sensor. that @Navarro dangled in front of me.

The shallow memory is my sole concern about the 3104.  I really don't understand that.    But the driver in the decision was the discount off list.  I shouldn't take much of a hit if I decide to sell it.

It comes with 500 MHz passive probes, but I shudder at the prospect of trying to get a good signal at that BW with a probe.  Most large BW use will be at 50 ohms with U.FL connectors at the appropriate places which can be enabled by zero ohm jumpers.

But, no, it really wasn't a rational decision.  It was a rational irrational decision.

[nctnico]

The shallow memory is my sole concern about the 3104.  I really don't understand that.    But the driver in the decision was the discount off list.  I shouldn't take much of a hit if I decide to sell it.

Don't count on it. Chances are Keysight is offloading old stock and introducing something new real soon. I don't have any inside information but it is just that their current line-up isn't really competitive.

[HighVoltage]

I few years ago, I bought the same scope MSOX3104AS without the "T" from the Keysight ebay store.
And I have been happy with it ever since.
You will really like the scope, I have no doubt.

Congratulations on a great purchase.

[wpwrak]

I am mostly tooled up to develop FOSS FW for Zynq based scopes, Zybo Z7-20, MicroZed, Instek GDS-2072E and a BeagelBoard X15.

Oh, very interesting ! And I found the thread on your Zynq Open Source project. I'll continue there.

I'm hoping that might get me some work as I find being retired rather lonely.

I think you may get quite a crowd once your project gets rolling. Linux started with a lone Finnish student printing As and Bs on his PC screen, and look where we are today :-)

- Werner

[egonotto]

I am mostly tooled up to develop FOSS FW for Zynq based scopes, Zybo Z7-20, MicroZed, Instek GDS-2072E and a BeagelBoard X15.

Hello,

perhaps Red Pitaya is a good starting point. The hardware is good and there is already ADC and DAC on it.

Only the software is frugal.

So a thankful task.

Best regards
egonotto

[apblog]

I'm saddened to hear that buying the full bundle does not include updates.  I should expect better of Keysight.     The bundle I bought includes jitter analysis.

You do get firmware updates, just not new options.

I'm curious about your jitter analysis package.  Do you have more details on that?  Is it listed as an option on your about screen? 

If it's available, I should be able to purchase it for mine.

As far as I know, the only way get jitter analysis is to buy the Infinium software with the DSA package for $6750.

If you enable all channels the 4Mpts erodes to only 500kpts/channel where the RTM3k has 40Mpts.

That's not quite true.  If I recall correctly, there are two Megazoom chips in this scope, one for channels 1 and 2, and another for channels 3 and 4.

Enabling all 4 channels drops you down to 2Mpts per channel.   The same happens if you are using both 1 and 2, or both 3 and 4.  But if you use 1 and 3 (or 2 and 4) you get 4Mpts on both channels. 

I make a point of always using 1 and 3 if I am just using 2 channels.

There is also another Keysight semi-myth out there that says that all measurements are done "on screen" with screen-pixel resolution.  This isn't entirely true either, I have seen some measurements that seem to keep their accuracy even when zoomed way out -- as long as the sampling rate hasn't dropped, of course.

Don't count on it. Chances are Keysight is offloading old stock and introducing something new real soon.

That's what I thought two years ago when they ran this same deal.  Still nothing new.  Although I think Daniel B. said that they were working on the next generation of Megazoom chip.

As far as Keysight not being competitive goes, I think all of the test equipment companies are very good at pricing their line to reflect the actual value in the marketplace.  And the beat the pants of Tek for most uses.

Now, before I start sounding too much like a Keysight rep, let me add a little balance and say that their FFT updates are pretty slow;  A little faster than once per second, best case.

I am actually thinking about buying a GW Instek GDS-2072E just to get fast FFT updates for audio frequency noise debugging.

[nctnico]

Enabling all 4 channels drops you down to 2Mpts per channel.   The same happens if you are using both 1 and 2, or both 3 and 4.  But if you use 1 and 3 (or 2 and 4) you get 4Mpts on both channels. 

It doesn't work that way. Using 4 channels: divide by two, double buffering: divide by two, enable digital channels: divide by two. That makes divide by 8. Enable reference traces and even more memory goes away. 4Mpts/8=500kpts (without reference traces enabled).

[rhb]

I won' have the Keysight until June :-(

The FFT in the Instek 2000E series is one of the good points.  There is also a very nice spectrum analyzer app, but it is only available on the MDO version.  I discovered I could access it from the demos, but when I mentioned it to Instek they promptly disabled it on the MSO in the 1.35 update.

[2N3055]

AFAIK for 3000 series Jitter analysis option is N8900A Infiniium Offline Oscilloscope Analysis Software..  It runs on PC not on a scope.

[MarkL]

....
There is also another Keysight semi-myth out there that says that all measurements are done "on screen" with screen-pixel resolution.  This isn't entirely true either, I have seen some measurements that seem to keep their accuracy even when zoomed way out -- as long as the sampling rate hasn't dropped, of course.
...

The measurements (including FFT) are made on an intermediate-sized buffer called the "measurement record" which can be up to 65536 points.  It's essentially a decimated version of the main 4Mpts (max) capture buffer.

So, smaller than the capture buffer, but certainly larger than the screen.  Details can be found in the 3000T user manual.

[tszaboo]

There is also another Keysight semi-myth out there that says that all measurements are done "on screen" with screen-pixel resolution.  This isn't entirely true either, I have seen some measurements that seem to keep their accuracy even when zoomed way out -- as long as the sampling rate hasn't dropped, of course.

Keysight doesnt do that. Tektronix TDS2000 will do that, since the sampling memory is the screen memory. How is it even legal to sell that junk in the 21 century is beyond me.