Should I buy a Rigol MSO5000?

[Fungus]

And I'll repeat myself

Yep. 

[dietert1]

By current standards (e.g. USB2, Fast Ethernet...) any scope with less than 1 GHz Bandwidth and with passive probes must be considered a learning toy. In fact, when you look at the MSO5000 sales sheet at the batronix website they show a block diagram including active probes, so once again they sell something unfinished.
Is it really true the MSO5000 does not come with internal 50 Ohm terminators? So Rigol is just creating confusion and i wonder whether they are behind the bandwidth hack. Maybe they want sell by bandwidth because their "features" aren't solid.
I agree completely with ntnico that the best "bang for buck" is a used LeCroy 1 or 2 GHz scope from ebay with a reasonable set of options. The only way to get a serious device for a reasonable price. If you don't trust ebay, you can ask LeCroy themselves for a second hand scope.

Regards, Dieter

[Mortymore]

I just wish that in a near future Rigol or someone else start braking price points in active and differential probes, at the point that we don't have to spend in them as much or more than the cost of the scope.

It's great to have a scope hacked from 50..70 MHz to 200...350MHz, but having the passive probes ruin everything 

[Martin72]

Having any passive probe on high frequency ruins everything.
The derating curves are nothing unusual, you can make it a little bit "better", but you can´t fight against the "nature" of a passive probe:

http://cdn.teledynelecroy.com/files/manuals/pp006a_manual.pdf

the best "bang for buck" is a used LeCroy 1 or 2 GHz scope

If you need the Ghz bandwith, if you have enough space on your bench to place it onto, if you don´t have a problem with old crt-screens, if you generally don´t worry about the age, appx 20yrs, and the risks of something going defective after all these years, if you don´t need usb and network connections because you have 3.5" disks and diskdrive avaible or love to print the screenshot with the internal printer out or having a rs232 port on your pc, finally if you don´t worry about a few kpts of memory, than you should grab your 1000 bucks and go for a lecoy LC/DDA scope.

Is it really true the MSO5000 does not come with internal 50 Ohm terminators?

Yep and it´s not alone, even the RTB series from R&S don´t have it too.

[nctnico]

Having any passive probe on high frequency ruins everything.
The derating curves are nothing unusual, you can make it a little bit "better", but you can´t fight against the "nature" of a passive probe:

http://cdn.teledynelecroy.com/files/manuals/pp006a_manual.pdf

the best "bang for buck" is a used LeCroy 1 or 2 GHz scope

If you need the Ghz bandwith, if you have enough space on your bench to place it onto, if you don´t have a problem with old crt-screens, if you generally don´t worry about the age, appx 20yrs, and the risks of something going defective after all these years, if you don´t need usb and network connections because you have 3.5" disks and diskdrive avaible or love to print the screenshot with the internal printer out or having a rs232 port on your pc, finally if you don´t worry about a few kpts of memory, than you should grab your 1000 bucks and go for a lecoy LC/DDA scope.

Lecroy isn't the only option. For $1000 you can pick up an Agilent 54835A which has a TFT screen, USB and ethernet. Hack it to become a 54845A and have 1GHz / 4Gs/s on 4 channels or 1.5GHz / 8Gs/s on 2 channels on a scope which is designed for high speed signal analysis. I have used this scope to resolve delays in the tens of ps range. The trigger on this oscilloscope is extremely stable.

And there are other options too like the Tektronix TDS700 series. A floppy to USB emulator can solve the problem of needing a disk drive.

Yep and it´s not alone, even the RTB series from R&S don´t have it too.

But the RTB2000 only goes to 200MHz. 500MHz bandwidth without real 50 Ohm inputs is madness.

[2N3055]

It is funny how there must always be this kind of topic..

It is also funny how Rigols and Siglents are held up to a higher standards than A-brand scopes costing 20x more.

So, since everybody else is repeating half-facts ad nauseum, let's repeat some facts:

1. Revision history for very expensive Keysight MSOX-3000T (18000 USD list price) is 27 PAGES LONG... Most of it bugs, some improvements over years.. Does that mean they are shit? No, it means they have great support...
2. R/S RTB2000 came out full of bugs, in some pretty basic stuff like sampling the waveform didn't work properly. It also had only basic math, comparable to Rigol 1000Z..For +5000€ price. After 3 years (that's three years) most of it was debugged, and math was upgraded (under pressure of users).  It is a great scope now, and it has, also, great support. Nobody said " I paid 5000$ for this shit and had to wait 3 years for it to become bug free... Screw the Germans, they are crap.." NO. It was also "We have great support"
....
Just mention exploding PSUs, scopes that loose flash for no reason,  very expensive multimeters that blue screen ....
And so on and so on...

And than, on the other side, for instance, we have a user here which bought a product that was few months old (so called "early adopter") and after few weeks, declared product is crap because it is buggy, that manufacturer is crap because they don't make perfect products first time with no bugs whatsoever....

Funny world, full of double standards.

Rigol MSO5000 was officially released in November 2018. 5 months ago..
And it's basic 70MHz version cost just a bit more than buying RTB2000 MSO option.
If you go and buy 4x 350MHz passive probes that would be 400€.. So you get a scope for 500€...

So let's be realistic on both counts: It is very inexpensive scope, and it is still very new. I personally never was early adopter and couldn't understand people who are.. All brand new platforms are buggy as hell, even from A-brands..
It is all about either having trust in manufacturer they will do it right eventually (and even with A-brands it takes time) or waiting for product to stabilise for a year or so and buy only things that are proven to be stable and dependable.

Full disclosure: Like nctnico, I also believe that if you need equipment for business, and if it makes your process faster and/or better you go and give more money for a product that serves you better.
I personally wouldn't buy MSO5000/7000 for business right now because I don't have time to be part of debug process..
I also don't like their user interface. I like Siglent's UI direction better. But that is a matter of taste too..

If in a year, MSO5000/7000 platform still doesn't have HIRES done right, if it still is not refined in a way that shows it is going towards being mature and stable platform than it will be that MSO5000 was unfulfilled promise and bad product. If they fix problems, it will be GREAT product and a good manufacturer.

We'll see in a year. Until then, jury is still out...
Regards,

[2N3055]

But the RTB2000 only goes to 200MHz. 500MHz bandwidth without real 50 Ohm inputs is madness.

RTB2000 goes up to 300 MHZ and MSO5000 up to 350MHz.
They both need 50 Ohm termination...

[tautech]

Having any passive probe on high frequency ruins everything.
The derating curves are nothing unusual, you can make it a little bit "better", but you can´t fight against the "nature" of a passive probe

Yet some are better than others.

Open this in another tab and compare against the Rigol probe spec above.
https://siglentna-qwavztc8hvq2w.stackpathdns.com/wp-content/uploads/2019/02/SP3025A_3050A_Users_Guide.pdf

V derating and impedance profiles have an edge. 

[petemate]

Im really glad that there is this much debate on the topic. Its a great way for me to reflect on my decision, which at the moment is leaning towards "wait 6-12 months".

One question regarding the missing hi-res acquisition that most people agree is an issue: There are 8 bits at 8GSPS. Thats not going to change with the chipset. As far as I know, hi-res mode is just averaging of a number of sequential samples, providing the possibility to get "increased resolution", since the average of n samples can produce a value that is in-between each sample value.  Why would this be a hardware issue? I get that you can do it in HW, but why can't it be added as a software fix?

[Martin72]

I personally never was early adopter and couldn't understand people who are.. All brand new platforms are buggy as hell, even from A-brands..

But without them, bugs which might appear during daily measuring (and creates countless different measure-situtations) would be hard to find.
I own one since end of november and since december-march I was in contact to the rigol support, telling problems, what could be better and so on.

but why can't it be added as a software fix?

It was implemented in the first firmware update but doesn´t have the effect I expected from:

https://www.eevblog.com/forum/blog/new-rigol-scope/msg2266332/#msg2266332

But they know it, I´ll send them the pics, let´s see what will happen.

They both need 50 Ohm termination...

And it will looks like this:

https://www.eevblog.com/forum/blog/new-rigol-scope/msg2270838/#msg2270838

I personally wouldn't buy MSO5000/7000 for business right now because I don't have time to be part of debug process..

Same here, my colleagues shouldn´t be part of it - we need things that works proper for i.e. acceptance tests.
But if some bugs were eliminated, it could be a cheap tool for daily measurings and if some gets broken, just buying new instead of repairing.

[2N3055]

@ petemate:
That is good question, petemate.
Reason for it to be in hardware is speed. Done right, with 8x oversampling you get 1GS/sec sampling rate and additional 1.5 bit res. and with16x 500MS/sec and 2+ bits.
And all would be real time with no slowdown in any scope operation.
Some manufacturers prefer to do it in software, or better to say postprocessing. Lecroy is such.
It slows down refresh rate but will have an advantage that ORIGINAL samples are preserved. No data is decimated and/or lost. You always get original pure sampled data in a buffer so you can massage it later to your heart's will.

First approach is great for interactive scopes that you use to probe around and general use.
Second approach is great if you tend to analyse data in depth..
Tool for the job. Ideally, you should have both.. In reality, you get one that fits most of your use cases.
Regards,

[nctnico]

Im really glad that there is this much debate on the topic. Its a great way for me to reflect on my decision, which at the moment is leaning towards "wait 6-12 months".

One question regarding the missing hi-res acquisition that most people agree is an issue: There are 8 bits at 8GSPS. Thats not going to change with the chipset. As far as I know, hi-res mode is just averaging of a number of sequential samples, providing the possibility to get "increased resolution", since the average of n samples can produce a value that is in-between each sample value.  Why would this be a hardware issue? I get that you can do it in HW, but why can't it be added as a software fix?

Traditionally hi-res mode is done in hardware (let's not debate when it is done during acquisition or while processing the data to draw the waveform) and it offers a filter which reduces high frequency content and increases the number of bits to get more resolution. The usual oversampling requirements apply: there has to be enough noise and the ADCs need to be reasonably linear. So hi-res mode may not give you the extra resolution you might expect because it is still bound to the laws of physics behind sampling theory. The biggest advantage of hi-res mode is that it cleans up a noisy signal so you get a nice sharp trace.

The downside of high-res mode is that the filter cut-off frequency depends entirely on the samplerate. Therefore IMHO a better feature is input filtering. This is available on some oscilloscopes. Input filtering offers more flexibility because you can set the cut-off frequency independant of the sampling rate, the filter is more agressive and usually you can choose between low-pass, band-pass and high-pass.

Now back to your question: can high-res mode be done in software? Yes it can but it will slow down the waveform update rate considerably. I see in the MSO5000 specs that it should do hi-res mode so why it isn't working is a good question.

[Fungus]

You are not getting the point TK is making. If you leave features out of an ASIC then doing a redesign will take a lot of time and money.

Is that true?

Surely it's all done with CAD systems these days, not teams of people armed with huge sheets of paper and rolls of black masking tape. A minor change to an ASIC could be a few mouse clicks then press "print".

[2N3055]

But without them, bugs which might appear during daily measuring (and creates countless different measure-situtations) would be hard to find.
I own one since end of November and since december-march I was in contact to the rigol support, telling problems, what could be better and so on.

I know and I appreciate effort. It's just I don't have time for it. I wish I had time to play with it, but I don't.

They both need 50 Ohm termination...

And it will looks like this:

https://www.eevblog.com/forum/blog/new-rigol-scope/msg2270838/#msg2270838

I presume proper implementation. But also, it is convenient to just switch it on...

I personally wouldn't buy MSO5000/7000 for business right now because I don't have time to be part of debug process..

Same here, my colleagues shouldn´t be part of it - we need things that works proper for i.e. acceptance tests.
But if some bugs were eliminated, it could be a cheap tool for daily measurings and if some gets broken, just buying new instead of repairing.

I personally have no doubts they will fix bugs. They did for 1000Z, but it might take time.
Once they fix them, it won't be just usable. It will be great.. It is potentially much better than just basic tool. It has many measurements, deep memory, fast refresh rate, decodes....
Regards,

[Fungus]

I personally wouldn't buy MSO5000/7000 for business right now because I don't have time to be part of debug process..

Same here, my colleagues shouldn´t be part of it - we need things that works proper for i.e. acceptance tests.
But if some bugs were eliminated, it could be a cheap tool for daily measurings and if some gets broken, just buying new instead of repairing.

There's room in the world for both types of user, why can't people see that?

Tool for the job. Ideally, you should have both.. In reality, you get one that fits most of your use cases.

Yep, and if you rely on any particular data analysis function then either test it before purchase and/or have another system in place.

eg.

Last year we bought a Rigol DS2202E that came with a free options package.
The scope works well, it's more or less quiet and it is handy for servicing other stuff. I mean with an expensive scope you think twice before taking it from its place.
Some weeks ago we wanted to use the I2C decoder for the first time, but it did not produce any meaningful results. ... I got the work done with our LeCroy Waverunner.

Has the DS2202E given $750 of value? Your comment about "handy for servicing other stuff" suggests it has, if so then it was a good purchase.

In your case the I2C wasn't a showstopper because you had another device handy that could do it. That's the way a sensible company should function, IMHO, not by trying to buy a Nirvanascope(tm) for every single employee because said employee might need a minor function twice a year.

For hobbyists? You can buy a $6 device on eBay to decode I2C. If you only need I2C a couple of times per year then would a 'scope with working I2C for three times the price be a better purchase than a DS2202E+a $6 gadget? (I say "No")

It's fun to complain about bugs on the internet but real life has to be practical.

[2N3055]

You are not getting the point TK is making. If you leave features out of an ASIC then doing a redesign will take a lot of time and money.

Is that true?

Surely it's all done with CAD systems these days, not teams of people armed with huge sheets of paper and rolls of black masking tape. A minor change to an ASIC could be a few mouse clicks then press "print".

ASICs by Rigol are designed differently than Keysight's in 3000T/4000/6000 series.

It has analog front end ASIC with high bandwidth and attenuators for simpler (cheaper) front end design.
Also they have ADC chip ASIC that is pretty much ADC chip that is a bit customized to work with scopes.
Data then goes to FPGA for processing. Waveform engine is in FPGA.
So they can change many things at will, more so than Keysight...

Problems I see is that analog front-end seems noisy, and have low sensitivity (5mV /DIV seems to be most sensitive range, with others doing 1mV/DIV or 500uV/DIV).
That cannot be made better without redesign..
Also, very high bandwidth ADC will be noisy..

We'll see how they handle the situation..

[Mr Nutts]

If you need the Ghz bandwith, if you have enough space on your bench to place it onto, if you don´t have a problem with old crt-screens, if you generally don´t worry about the age, appx 20yrs, and the risks of something going defective after all these years, if you don´t need usb and network connections because you have 3.5" disks and diskdrive avaible or love to print the screenshot with the internal printer out or having a rs232 port on your pc, finally if you don´t worry about a few kpts of memory, than you should grab your 1000 bucks and go for a lecoy LC/DDA scope.

LOL, you don't need to pay a Grand for a LC with CRT there are lots of them which go for a lot less and with a lot more memory than a few kpts 

But Waveruner LT and Wavepro 900 are now so cheap that it doesn't make much sense to go for the older scopes. LT comes with I think 8Mpts and the Wavepro with up to 64Mpts, and instead of USB one can just use CompactFlash cards to transfer data via the rear PCMCIA port 

Or just get one which has the optional network card 

From what I was told both LT and Wavepro 900 were made by IWATSU and should be very reliable, also service manual and schematics are available 

[Martin72]

I see in the MSO5000 specs that it should do hi-res mode so why it isn't working is a good question.

Same was on the 7000 - although it´s implemented in the SPU, see ultravision II platform details.
After the fw upgrade you can select hi-res and something will happen:

https://www.eevblog.com/forum/blog/new-rigol-scope/msg2233980/#msg2233980

But not in the way as expected, as known from siglent or lecroy scopes.

LOL, you don't need to pay a Grand for a LC with CRT there are lots of them which go for a lot less and with a lot more memory than a few kpts

Not here. 

[2N3055]

Last year we bought a Rigol DS2202E that came with a free options package.
The scope works well, it's more or less quiet and it is handy for servicing other stuff. I mean with an expensive scope you think twice before taking it from its place.
Some weeks ago we wanted to use the I2C decoder for the first time, but it did not produce any meaningful results. ... I got the work done with our LeCroy Waverunner.

Has the DS2202E given $750 of value? Your comment about "handy for servicing other stuff" suggests it has, if so then it was a good purchase.

In your case the I2C wasn't a showstopper because you had another device handy that could do it. That's the way a sensible company should function, IMHO, not by trying to buy a Nirvanascope(tm) for every single employee because said employee might need a minor function twice a year.

For hobbyists? You can buy a $6 device on eBay to decode I2C. If you only need I2C a couple of times per year then would a 'scope with working I2C for three times the price be a better purchase than a DS2202E+a $6 gadget? (I say "No")

It's fun to complain about bugs on the internet but real life has to be practical.

In that particular case it could have been user error. I cannot say without more detail. Rigol terminology is, well, weird sometimes, defaults are not logical. Decodes sometimes take time to do it right. Unlike LeCroy or Keysight or R/S that make it easier to setup things.

A bit of trivia: MSOX3000T/4000 doesn't recognize error if you put in an address from 10bit address  range into 7bit address I2C packet. It decodes it as a nice, 7bit ADR packet without any error. Picoscope showed an error, saying packet is malformed 10 bit ADR packet. It KNOWS that was 10bit address and expected different packet.

We are talking about 18000USD list price instrument, and Picoscope 2000 series would do the job right ....
Is it crap for it?? Nope, because it does so many other things right...
Is it perfect? No, that is why I have other scopes..

[Fungus]

ASICs by Rigol are designed differently than Keysight's in 3000T/4000/6000 series.

It has analog front end ASIC with high bandwidth and attenuators for simpler (cheaper) front end design.
Also they have ADC chip ASIC that is pretty much ADC chip that is a bit customized to work with scopes.
Data then goes to FPGA for processing. Waveform engine is in FPGA.
So they can change many things at will, more so than Keysight...

In that case the "hi-res" mode is done in the FPGA so a fix should be just firmware.

I wonder what the problem is, I find it hard to believe they don't know how to do it. The only thing I can think of is that they're low on space on the FPGA and there's multiple paths that all need to access hi-res data so the averaging needs to be implemented all over the place.

[Martin72]

A video from rigol last month published, hadn´t see this before, a hi-res mode comparison between RTB and MSO5000:



LOL, the first seconds…..there are worlds between the two displays.
But that´s not the point.

[petemate]

LOL, the first seconds…..there are worlds between the two displays.

Are you refering to the 1200x800 vs 1024x600?

[Martin72]

No, the brightness….
The resolution ist more than enough.
In everyday use, the dim display makes no problem - but whenever another scope stands byside, the different is mostly huge.

[2N3055]

LOL, talk about marketing..

R/S in HIRES mode drops to 300Ks/sec and shows undersampling. Rigol still samples at 2Gs/sec so shows signal properly.

Only problem is that RTB2000 has same/less noise without HIRES than MSO5000 in HIRES mode....
So if you didn't put RTB2000 in HIRES mode (you don't need it frankly for this example) it would show things properly.

But this is good example that you shouldn't compare things blindly. In this case, if properly used, both could do the job..

[Martin72]

Rigols seems to be convinced of having a functional hi-res mode….makes me nervous.