Please recommend oscilloscope for robotics research

[Keysight DanielBogdanoff]

I'll (of course) second the Keysight scopes. Especially the 3000 X-Series or the 4000 X-Series if you have the budget for it.

Of interest to you might be data logging with the BenchVue software, the power analysis applications, and/or the remote interface capabilities that allow you to control the scope from a PC or mobile phone. We also have educational pricing available for universities.

[eeguy]

Which scope would be the next step up from the DS1054Z? How Tektronix scopes compared with the Rigols?

[Wuerstchenhund]

Which scope would be the next step up from the DS1054Z? How Tektronix scopes compared with the Rigols?

Well, Tek just came out with the new TPBS2000 which is its latest entry-level scope - at least feature-wise (price-wise, not so much). There's a thread with more details around in this forum.

As to Tek in general, all I can say is if someone thinks the UI of the Rigol DS1054z in that video is laggy then he hasn't seen a Tek yet. I've a MSO3000 (at least until Monday after which it goes to a colleague), and even with the latest firmware the UI is pretty slow if the scope isn't doing anything special. If it is, then the UI pretty much locks up. Also, the whole UI is pretty cumbersome to use, it's no joy. And that is a scope that sells for around $10k. Tek had great analog scopes but its DSOs were never great but mediocre at best and in many cases just god-awful.

For a very cheap entry-level scope I'd consider the GW-Instek GDS-1000B and GDS-200E, simply because they seem to be a lot more mature than the Rigols. If you can afford to spend a bit more, the Keysight DSOX2k Series or the R&S HMO1202 make good choices. Above that there's the DSOX3000T, but unless you need one of the less common serial decode standards it supports then the LeCroy WaveSurfer 3000 offers a lot more for a lot less money.

If you care about advanced analysis and cheap active probes, try to get a LeCroy. If you want the latest features and raw hardware performance, get a KS.

You'll find that in terms of functionality you usually get more from LeCroy than from Keysight (some features you only get from LeCroy, and for others the equivalents often lack functionality). Same is true re. performance, aside from the low-end and the lower mid-range where KS really offers excessive waveform update rates, but further up the ladder it shows that LeCroy has the stronger architecture compared to other scopes which are mostly based around proprietary ASICs.

On the other side, LeCroy's entry-level (WaveAce, WaveJet, all rebadges) isn't even worth considering.

Seriously, I hate our Tek scopes so much that I bought my own scope. The lagging UI and poor quality LCD just will screw up your day.

I can only concur. I really tried to like that MDO3000 but it's pathetic slowness, the stupid UI and the overall disappointing performance means I can't see it gone fast enough. It's no joke, it really sucks all the joy out of the work. I almost feel sorry for its new owner (but then, he I warned him but he still wanted it  ).

[eeguy]

The Keysight DSOX2k Series seem to be a few years old. How come they don't have new products in this series?

Looks like the DSOX2k Series have 1Mpts as max memory depth and 2GSa/s as the max sample rate. One of the uses is to measure and record the power consumption and current of the motors or the robot. Will such specifications be sufficient? The GW INSTEK GDS-2104A and 2204A have display memory depth of 2Mpts and cheaper than the Keysight. However, Keysight seems to be a better brand. Not sure if 2Mpts is overkill. Why suggestion? Also, shall I go for 100MHz or 200MHz model?

[Faith]

Don't forget that with the Keysight 2000A the 1Mpts of memory is an option. See:

MSOX2000-010: Memory upgrade - 1 M for 2000 X-series.

If that's the route you're going to take however I would recommend buying it from Keysight's Premium Used store on eBay.

From over there the top-of-the-line MSOX2024A (200MHz 4+8 Channel) costs less than the MSOX2002A (70MHz 2+8 Channel) would if you were to buy it "new."

The MSOX2024A which I borrow on occasion was purchased from there and it really is indistinguishable from a used unit. It's thoroughly refurbished and comes with new calibration and new five-year warranty.

[Wuerstchenhund]

The Keysight DSOX2k Series seem to be a few years old. How come they don't have new products in this series?

Because it still sells well enough. Don't forget that, as big brands are concerned, the low end segment pretty much belongs to Keysight (as seen above Tek isn't worth considering, LeCroy's low end sucks, and Hameg/R&S low end stuff isn't as widely known and similar expensive as Keysight).

Looks like the DSOX2k Series have 1Mpts as max memory depth and 2GSa/s as the max sample rate. One of the uses is to measure and record the power consumption and current of the motors or the robot. Will such specifications be sufficient?

Can't say but the low sample memory (which halves in certain situations) means the sample rate (and thus the usable frequency) will drop quickly on longer timebases.

The GW INSTEK GDS-2104A and 2204A have display memory depth of 2Mpts and cheaper than the Keysight.

The GW Instek GDS-2000A is pretty old, and has long been replaced by the GW Instek GDS-2000E which comes with 10Mpts memory. There are various users here and according to their reports it's a very nice scope, and unlike Siglent/Rigol it's also very mature.

[quote[However, Keysight seems to be a better brand. [/quote]

Sure, but that is of little help if your measurements become limited by the tiny sample memory. And it's not that GW Instek is a nobody in the T&M market (they have been around for a very long time).

Personally I'd go for a GDS-2000E over the Keysight DSOX2000A any day.

Also, shall I go for 100MHz or 200MHz model?

That depends on your requirements (keep in mind that 100MHz BW means 100MHz sine, not 100MHz non-sine like square or pulses).

But should you go with the GDS-2000E then the 200MHz variant will still be cheaper than the 100Mhz DSOX2k.

[JPortici]

Which scope would be the next step up from the DS1054Z? How Tektronix scopes compared with the Rigols?

Well, Tek just came out with the new TPS2000 which is its latest entry-level scope - at least feature-wise (price-wise, not so much). There's a thread with more details around in this forum.

As to Tek in general, all I can say is if someone thinks the UI of the Rigol DS1054z in that video is laggy then he hasn't seen a Tek yet.

you might have been referring to the TBS2000
i actually use a TPS2014 daily at work (the battery powered, 4 isolated channel scopes) a 1054z at home and the UI on the tek is a lot smoother on that the rigol. That has become pretty annoying that if i only neet to look and make estimates i just use a tek7704A

For a very cheap entry-level scope I'd consider the GW-Instek GDS-1000B and GDS-200E, simply because they seem to be a lot more mature than the Rigols.

seconded

there's the DSOX3000T, but unless you need one of the less common serial decode standards it supports then the LeCroy WaveSurfer 3000 offers a lot more for a lot less money.

i have the same dilemma (i waaaaaaant a lecroy -ehr, siglent rebadge?- but it's more probable i'll need the exotic serial decoding)

[Wuerstchenhund]

As to Tek in general, all I can say is if someone thinks the UI of the Rigol DS1054z in that video is laggy then he hasn't seen a Tek yet.

you might have been referring to the TBS2000

No, I was referring to the MDO3k/4k, DPO/MOD2k/3/4k/5k/7k/70k, and TDS5k/7k. Which not only are horribly slow scopes per se (weak architecture), they all share that the UI locks up when the scope is doing something processor-intensive. Which is silly, really (and not something I've seen on a Rigol).

I haven't seen a TBS2000 in the wild but what can be seen from the videos it seems to have retained typical Tek annoyances.

i actually use a TPS2014 daily at work (the battery powered, 4 isolated channel scopes) a 1054z at home and the UI on the tek is a lot smoother on that the rigol.

That might well be but then the DS1054z is really the bottom of the barrel regarding scopes (it's only good because of i's very low price), plus the TBS2000 is a very simple scope which can't do anything complex anyways.

That has become pretty annoying that if i only neet to look and make estimates i just use a tek7704A

Well, Tek certainly had a good hand for analog scopes.

there's the DSOX3000T, but unless you need one of the less common serial decode standards it supports then the LeCroy WaveSurfer 3000 offers a lot more for a lot less money.

i have the same dilemma (i waaaaaaant a lecroy -ehr, siglent rebadge?-

The WS3k is not a rebadge. The hardware is a co-development of LeCroy and Siglent (which also acts as the manufacturer), and the software has been developed by LeCroy (Siglent can't touch it). Siglent is also allowed to sell the scope under its own name as SDS3000 in China but not outside.

but it's more probable i'll need the exotic serial decoding)

Fair enough, but you pretty much have to pay for it through the nose.

And considering the DSOX3kT's tiny 4Mpts memory (plus that in many cases it actually halves) I'd consider using some USB device (i.e. a Salae) for decoding exotic standards because it will capture much longer sequences than the DSOX3kT (where you also can't just manually drop the same rate to increase the acquisition time). If a USB device could be a viable way for you then get a WS3k + the Salae and still quite a lot of money.

[Fungus]

One of the uses is to measure and record the power consumption and current of the motors or the robot.

...something that an oscilloscope is totally unsuitable for.

For that you need something like a multimeter with PC connection and some PC software to calculate the numbers. Probably two or three of them in fact.

Even better: Get a power supply with logging ability and power your robot from that. It will tell you the overall consumption directly.

You're in here worrying about 'bandwidth' and stuff but I get the feeling you're asking about screwdrivers when you really need a saw.

[JPortici]

No, I was referring to [...]

yes but earlier you wrote "Tek just came out with the new TPS2000"
i was referring to this with the beginning of the post the TBS is the new one, a small mistake that could arise confusion
it would actually be nice if tek or possibly someone else made a new 4 chan isolated scope (with a decent amount of memory this time)
but let's not derail the topic, the reason why and why not have already been discussed in the "what's missing on the market" topic

you are right on the lecroy!

[Wuerstchenhund]

No, I was referring to [...]

yes but earlier you wrote "Tek just came out with the new TPS2000"

Oh, yes, that should say "TBS2000" (a typo which has now been corrected).

i was referring to this with the beginning of the post the TBS is the new one, a small mistake that could arise confusion

Yes, it does, obviously

it would actually be nice if tek or possibly someone else made a new 4 chan isolated scope (with a decent amount of memory this time)

Agree, most portable scopes are pretty dire in terms of memory (even that expensive R&S ScopeRider comes with 500kpts max, but on the upside it can do 2Mpts data logging and 12Mpts history, plus it's a 10bit scope). There's the Keysight U1620A with 2Mpts but that is 2ch only. Oh well... 

but let's not derail the topic

Agree.

[eeguy]

I am not sure if the Keysight DSOX2k Series can be connected to a computer via USB so that I can store the data real-time. If there is such a built-in function, do I still need to worry about the 1Mpts of memory limitation?

[nfmax]

Yes, the X2000 series can be connected by USB to a PC. The connection library is a free download from Keysight, and I believe the free version of Benchvue lets you download and save data (if not, it's pretty easy to write a short Python program to do so). You are still restricted to the maximum of 1 MPts/channel (in single-shot mode), but these scopes do support segmented memory (option) which lets you partition the 1MPts between multiple trigger events. You cannot stream data in real-time over the USB connection - USB can't support the data rate, even if the scope firmware allowed it.

[Wuerstchenhund]

I am not sure if the Keysight DSOX2k Series can be connected to a computer via USB so that I can store the data real-time.

Yes, it can connect to a computer, either via USB2 (built-in) or Ethernet (unlike with other scopes an optional feature on the DSOX2kA and DSOX3kA scopes, and not exactly a cheap one either). Neither allows the streaming of sampling data in real-time.

[eeguy]

I guess if the scope can send the data to the PC via usb for later data analysis, that might be find. In this case, what memory size do you recommend?

[Wuerstchenhund]

I guess if the scope can send the data to the PC via usb for later data analysis, that might be find. In this case, what memory size do you recommend?

That very much depends on what you want to do, really. If you want to stream data to a PC then a PicoScope might be the better option.

It looks like you don't really know what you need. "Robotic research' is such a wide field that goes from what essentially are tiny remote-controlled vehicles to highly complex autonomous systems that can move even in rugged terrain, so without you giving further details what exactly it is you want to do it's hard to recommend something appropriate.

As others said, if you want to measure power draw of a motor then a scope isn't the right tool. If you however work on control systems then depending on the complexity of your robot you might want a decent mid-range scope with large memory and tools that help you find and isolate glitches.

Post-processing acquisition data on a PC is usually done for stuff like signal analysis but in robotics signals usually aren't complex so I'm not sure why you would want to push them to a PC.

It also helps to narrow down a budget for the scope but be warned that depending on what you want to do that may or may not be enough. In the latter case you should also consider 2nd hand (used) instruments.

[rstofer]

The Keysight DSO2014A (100 MHz, 4 Channels) starts at $2400 and the options can run another $1200 so you wind up with a scope costing $3600 by the time it is useful.  Right around 9 times the cost of the Rigol DS1054Z.

The MSO2014A (100 MHz 4 Channels plus 8 Digital) is nearly $3200 plus options - or about 11 times the 1054Z by the time it does serial decoding.  But the 1054Z doesn't have the MSO option.  I can buy a logic analyzer if I really need one!

I can do an awful lot of robotics stuff (or anything else) with the 1054Z.  Now, true, my projects have never done the DARPA Challenge of autonomously crossing the desert but I would be willing to bet that the 1054Z can handle absolutely everything involved.  And at 11% of the cost of the DSO2014A I wouldn't mind if it got full of sand!

A little realism can be helpful.  What are the real requirements?  What is the budget?  Is this a funded project or is it just for a hobby?  Yes, if somebody else was paying, I might be willing to spend 11x as much of their money.  If it's my money, well, I will get by with something I can actually afford.

If you need blazing bandwidth, maybe a high bandwidth DSO simply costs too much and maybe storage isn't the only way to get the measurement.  For a couple of hundred dollars, you can get a used Tek 485 which is a fairly trivial analog scope with 350 MHz bandwidth.  Considering the lack of advanced features, the 485 probably wouldn't be the go-to scope but it's nice to have the bandwidth should it ever be necessary.

I can see where the Keysight DSO2000X series can be considered 'entry level' when high end scopes are up in the $150k range.  But in my world, I just couldn't justify spending that much money for a scope.  This electronics thing is just one of several hobbies.

[Fungus]

I can do an awful lot of robotics stuff (or anything else) with the 1054Z.  Now, true, my projects have never done the DARPA Challenge of autonomously crossing the desert but I would be willing to bet that the 1054Z can handle absolutely everything involved.

Me too.

And if it's data-logging you want then an oscilloscope seems like the wrong tool for the job. There's plenty of gadgets that will log voltage/current over time with a lot less effort than an oscilloscope. Plus you won't be tying up your most useful tool doing mundane tasks.

[JPortici]

This electronics thing is just one of several hobbies.

To you, and that's fine. Maybe not for the OP.
I'd never use the 1054z for a day job.. infact i don't. The day i needed 4 chan SPI i brought my 1054 (I don't have 4 chan scopes with decoding at the workplace, no need) but it was so frustrating to use at the pace i have at work that i put it away about 10 minutes later and decoded by hand the data from a tek tps. it was faster that way.
At home, for my hobby projects where i don't get paid to finish them in a reasonable amount of time, is okay. Almost fine.

[rstofer]

This electronics thing is just one of several hobbies.

To you, and that's fine. Maybe not for the OP.
I'd never use the 1054z for a day job.. infact i don't. The day i needed 4 chan SPI i brought my 1054 (I don't have 4 chan scopes with decoding at the workplace, no need) but it was so frustrating to use at the pace i have at work that i put it away about 10 minutes later and decoded by hand the data from a tek tps. it was faster that way.
At home, for my hobby projects where i don't get paid to finish them in a reasonable amount of time, is okay. Almost fine.

According to the OP, this was for an educational institution - graduate school.  Maybe they do have funding for something more refined.  OTOH, it could wind up being a designated sacrificial instrument.

All of my comments should be taken in the context of a hobby; I have never spent more than about a month practicing electronic engineering.  My career revolved around electrical engineering - power, lighting, project management, that kind of stuff.  Boring but financially beneficial.

As to measuring power for various motors, there is no sense in using a scope.  It's the wrong tool for data logging.  In fact, I would probably cobble together a high side current sense amplifier to measure current and another op amp to measure voltage and just use a uC with ADC capability.  If a uC wasn't fast enough, I could just use an FPGA with the problem being where to store the data at the rate I would be collecting it.

[Fungus]

If a uC wasn't fast enough, I could just use an FPGA with the problem being where to store the data at the rate I would be collecting it.

A motor is a big inductor. The current can't possibly change at high frequencies.

[rstofer]

If a uC wasn't fast enough, I could just use an FPGA with the problem being where to store the data at the rate I would be collecting it.

A motor is a big inductor. The current can't possibly change at high frequencies.

That's exactly right!  I was thinking that even 1000 samples/sec would be more than adequate but I'm not on the cutting edge of whatever is going on.  Certainly anything in the GS/s is overkill!

Or, oversample the data at, say, 10 kHz and take some kind of average before storing the results.

[ChunkyPastaSauce]

That is great!  What suggestion do you have in the following two cases:

1. Measure and record the power consumption of the entire robot which consists of more than one motor.
2. Measure and record the power consumption of several motors of the robot at the same time.

Maybe consider national instruments hardware and software package over a scope for your app.

[eeguy]

I want to do sensory motor integration and motor control at home. Besides, I need a scope to test some circuits. As for the reason for data logging, I want to publish the data in academic reports. So, being able to store the data to a computer would be desirable. I guess I should not spend more than 3K. Of course, if paying a bit more makes a big different, I would consider it.

[nctnico]

What exactly do you expect to get from logging? If you say reports then I assume graphs based on multiple measurements or screendumps. I usually produce  graphs in reports the old fashioned way: take a sheet of paper, divide in columns, write down the measurement results and type everything into Excel. Ofcourse you can automate this but you have to weigh the amount of effort versus time saved first.