EEVblog #1146 - New Low Cost Rigol MSO5000 Oscilloscope

[offmar]

Is the probe compensator hitting the endstop? Can you adjust it either side of the "best" position and see visible changes?

No, it's not hitting endstop in any direction, and yes, I can adjust for best position. The screenshots are from that position.

Can you zoom right into a rising edge and look at it, is it 'ringing"?

It's not ringing.

On my previous post, mentioning that I see the same form of the signal at 1x, I mean that I'm switching the probe to 1x.

In principle (if there can be a simple answer  ) is this a serious issue, or I'm just overthinking around it? 

[Fungus]

is this a serious issue, or I'm just overthinking around it? 

99% sure you're just overthinking it. This is about compensating the probes, not looking for artifacts in signals.

I think the idea is to adjust probes so that the horizontal lines are as flat as possible and that you don't go past the top on a rising edge. If the corner of the signal are rounded then that's OK

ie. This is good


Your 'scope (and compensation signal generator) have limited bandwidth, you'll never get it 100% square.

[mindy]

It's not ringing.

Have you tried to measure  AWG (Arbitrary Wave Generator) output ?
Set it to 1-15 Mhz square wave and check signal ringing at rising / falling edge.

[Fungus]

At this stage we're only looking at compensating the probes. We're not looking at rise times, ringing, or anything else.

The compensation signal on the front of the 'scope will be designed for zero overshoot, not for any other characteristics.

[David Hess]

Yes, I tried swapping the probes. Nothing changes, seems it's the scope input. Also, when I switch the probes to 1x, the form is the same, maybe 30% less overshoot. I have no separate generator, but when I use the scopes internal, I get slightly different result, but never perfect.

The internal source is intended for probe compensation and can suffer from considerable leading edge distortion without compromising this.  But any source should display identically on every channel with every properly compensated probe.

I have very similar issue with my MSO7014, but in my case all the channels gives me the same result.
https://www.eevblog.com/forum/testgear/new-rigol-ds7000/msg2190588/#msg2190588

When tried with 50 Ohm input I got clean, perfectly compensated signal.

Ah-oh, that is not good.  See below.

#nctnico suggested that this may be due to a crossover between LF & HF what kind of makes sense.

That could cause it however another source is "hook" from the printed circuit board substrate; see the linked article. (1) It could also be excessive dielectric absorption in a resistor or capacitor in the high impedance circuits.  Shunt protection circuits, typically diodes, can also cause problems like this.

What is odd is that one channel does not display it.  A channel should also not display this sort of difference between using a 50 ohm source and a x10 high impedance probe of suitable bandwidth; both should be identical except perhaps in bandwidth which is not a factor here.

I hope that this can be calibrated and not something messed up on a hardware.

There are no manual front end adjustments on these oscilloscopes, at least as far as I have seen, so it is unlikely and even if there was, this would not correct hook if it existed.  The problem with hook also comes up with cheaply manufactured high voltage differential probes which drift out of calibration over time.

Just wondering if all MSO5k / MSO7k have similar compensation signal issue.

I suspect this represents a poorly specified design or quality control failure which will vary significantly between production runs.

In principle (if there can be a simple answer  ) is this a serious issue, or I'm just overthinking around it? 

How serious this issue is depends on your application.  I would not accept this poor level of performance on any oscilloscope and none of mine have had it except when the high impedance printed circuit board before the impedance buffer become contaminated requiring cleaning and a bake out.

The reason I suggested using a faster source, ideally a reference level pulse generator, is that it might reveal more about what is going on and the difference between the good and bad channels.  The oscilloscope's built in calibration output is only fast enough for compensation adjustment which is *not* the problem being displayed.

At this stage we're only looking at compensating the probes. We're not looking at rise times, ringing, or anything else.

Compensation of the front end high impedance attenuators and low and medium frequency compensation can make use of the same compensation signal used for attenuating probes.  They are all in the same frequency range.

The compensation signal on the front of the 'scope will be designed for zero overshoot, not for any other characteristics.

Often the compensation signal is not even designed for zero overshoot because high frequency content is irrelevant to low and medium frequency compensation.  The compensation signal however is designed to be very flat and have very low levels of tilt which is actually not as trivial to do as it seems because of saturation and recovery characteristics in semiconductors. (2)

These days it is pretty easy to get it right with CMOS designs.  The symmetrical output and what appears to be a first order response in the example you posted indicate a low impedance CMOS switch output driving an RC network to limit bandwidth and transition time which is a great way to generate the needed signal.

(1) The PDF is too large to include as a forum attachment so here is a link: https://drive.google.com/open?id=1PnlWlBESa6v7Uovn2nh1HBtCPShI4y4P

(2) A current switched design using switching diodes should always work to produce the perfect output signal for probe compensation, right?  Wrong!  Some 2 nanosecond switching diodes inexplicably suffer from recovery "tails" which is process dependent so you have to qualify the 1N4148s or whatever to use in this type of circuit.  CMOS outputs made this type of circuit easier.  With some care, it can also be done with bipolar outputs or a linear output stage.

[offmar]

It's not ringing.

Have you tried to measure  AWG (Arbitrary Wave Generator) output ?
Set it to 1-15 Mhz square wave and check signal ringing at rising / falling edge.

I've been mainly using the compensation generator. When testing with AWG at 1Mhz and above, the ringing is present.

[Martin72]

Hm-hm...

I was sure I hadn´t this Problem, but now I´ll grab the rest of the probes and check it tomorrow again…

[offmar]

(1) The PDF is too large to include as a forum attachment so here is a link: https://drive.google.com/open?id=1PnlWlBESa6v7Uovn2nh1HBtCPShI4y4P

Thank you! The entire post and that document are very informative.

[Martin72]

I was sure I hadn´t this Problem

Yepp, test it again, all channels do not have an overshoot in the compensating signal.

Here some pics from channel 4:

With the original probe:



Philips probe, BW 100Mhz:



LeCroy Probe, PP019-2, BW 250Mhz:



Then I took a 1Mhz square, source was a rigol AWG :



Oh, it´s ringing….

Wait..take another scope and let´s see if the ringing is still present:



Yepp, it´s NOT the scope....

Check it with another AWG ( Agilent 33210):



Proved.

My rigol is a MS5074, not "hacked"

Martin

[Martin72]

The other channels:

Your 'scope (and compensation signal generator) have limited bandwidth, you'll never get it 100% square.

Although it seems to me, that the edges are a little bit "sharper" with the LeCroy Probe.

[Bax]

The question is, do the Rigol PVP2350 probes meet their intended specs.

What's the result if the probes are set at 1x? On my 'scope the signal has the same overshoot as with 10x compensated. Both times getting signal from the 1kHz compensation generator.

Same result as you with 1X.

Selecting band limit to 20MHz cleans up the display a bit but the overshoot is still seen.

BTW, my scope has not had the hack applied.

[Bax]

The other channels:

Your traces look a lot cleaner than mine, they are thin. With the default settings, my horizontal lines are wider (noisy).

What settings did you use ? Was band limit selected for the channels?

[Fungus]

The other channels:

looks perfect to me.

Your traces look a lot cleaner than mine, they are thin. With the default settings, my horizontal lines are wider (noisy).

Doesn't mean anything, you might be near a radio station or a noisy light bulb or something.

[Martin72]

What settings did you use ? Was band limit selected for the channels?

Probes were setting to 1:10, no bandwith limit - You can see it in the picture, below the grid, the channel state (white arrow):



It shows V/div. and actual offset, but also a "B" for bandwith limit if selected.

Like here:

[Commodore8888]

Ditto here with an MSO5074(upgraded, Early Jan batch) and an MSO7k.
Tried both with various probes and an agilent arbgen.
Still couldn't get rid of that silly overshoot. Almost looks like it's getting
superimposed after most of the input stage. Using some RG58 and a
50Ohm cleared it up some, but odd to see when the same setup
did nothing on a DS1KZ and an MDO3k.

[Martin72]

Today, I did some measurements on a signalboard which delivers among others a three-phase pwm signal.
And I was a little bit dissapointed.
As known from all the different lecroy models on the work, as known from the SDS1104X-E, the MSO 5074 couldn´t display a "stable" pwm-signal.
I couldn´t describe it better, but the signal "jitters" on the screen, as if you don´t have the trigger proper set.
On the lecroys/siglent, the signal is "stable" displaying.
Sure, with single-shot pressing one after another you can see the signal width going from small to wide and back - But it´s annoying and it would be much more comfortable and less irritating when the rigol does it like lecroy/siglent.
And I try to use the "lowpass" math function to seperate the sinewave out from the pwm.
To set the cutoff frequency of say 1000hz, you must set the timebase of several hundreds of milli-seconds.
It works, but you can´t see much from the sinewave which period is way shorter than the actual timebase set.
Therefore a free selectable cutoff frequency independent from the timebase would be necessary, otherwise the filter functions makes no sense at all.

To the PWM thing, after rigol complaining my "problem" they told me to increase the trigger hold-of time - And this works 
Increasing the hold-off time to (in my case) 67ms (standard 8ns) will give the view on the PWMs described before.

And, Filters:

You must decrease the memory (and increase the timebase) to get lower cut-off frequency, it works I could set the cutoff to 1Khz and get the sinewave out of the signal.
But it seems to me like no top solution, it can´t be triggered proper because of the larger timebase - still I want to set the cut-off frequency independent from timebase/memory.

Still couldn't get rid of that silly overshoot.

Some have the problems, some (most ?) don´t….what could this mean, in worst case a hardware problem - In any case, rigol should be informed and I would give the unit back for warranty in Your case.

[mindy]

Your traces look a lot cleaner than mine, they are thin. With the default settings, my horizontal lines are wider (noisy).

I can confirm that I see more noise on my scope (MSO7014) as well (same issue with a "Hook" on compensation & Overshoot on AWG 1 Mhz and above).

[A Hellene]

Your traces look a lot cleaner than mine, they are thin. With the default settings, my horizontal lines are wider (noisy).

I can confirm that I see more noise on my scope (MSO7014) as well (same issue with a "Hook" on compensation & Overshoot on AWG 1 Mhz and above).

Well, I am afraid that it kind of reminds me of this issue:


Rigol DS1052E nasty surprise!

I hope it is not...

-George

[TK]

I ordered one unit, but I might consider cancelling and holding until this issue is resolved...

[Martin72]

Your traces look a lot cleaner than mine, they are thin. With the default settings, my horizontal lines are wider (noisy).

I can confirm that I see more noise on my scope (MSO7014) as well (same issue with a "Hook" on compensation & Overshoot on AWG 1 Mhz and above).

Reducing the memory depth should decrease the noise, try it out..

[mindy]

Reducing the memory depth should decrease the noise, try it out..

Made some more tests and I can just conclude that my unit is defective and requires some attention from Rigol engineers.
Contacted Rigol support regarding this last week but no reaction yet.

Does anybody had an "opportunity" to try a Warranty service in Europe and knows the procedure & address I should send it to?

[_Wim_]

Made some more tests and I can just conclude that my unit is defective and requires some attention from Rigol engineers.
Contacted Rigol support regarding this last week but no reaction yet.

Does anybody had an "opportunity" to try a Warranty service in Europe and knows the procedure & address I should send it to?

Are you sure your probes are set to 1/10 like Martin72? Do you see the noise also with a 50 ohm terminator on the input (to rule your it is noise picked up from the environment).

[Martin72]

Contacted Rigol support regarding this last week but no reaction yet.

Does anybody had an "opportunity" to try a Warranty service in Europe and knows the procedure & address I should send it to?

https://www.rigol.eu/contact-us/

I´m in touch with them since january, they answered quick and give me great support so far.
I´ll contact them again, with link to this here.

Edit : Done, their answer to this issue will be posted here.

Edit 2 : Both persons are absent until march

[sq6iyn]

Helo
New firmware?  MSO5000_00.01.01.04.04
http://www.rigol.com/Support/SoftDownload/3

[Commodore8888]

Still couldn't get rid of that silly overshoot.

Some have the problems, some (most ?) don´t….what could this mean, in worst case a hardware problem - In any case, rigol should be informed and I would give the unit back for warranty in Your case.

That may sadly ends up what happens. I'll be contacting support to see if they've had other reports. The 7k at work is noticeably worse than the 5k I have here at home I just checked. It seems to be a per channel thing, as 1,2,3 on the 5k show it, but ch.4 is nearly flat.

The 7k is govt owned...which means warranty will probably be done and gone by the time paperwork to ship went through  . That they both do it leads me to think we're dealing with their chipset or firmware decisions.