Rigol DS4014 troubles...

[kinap35]

Hi,

I bought a brand new Rigol DS4014 back in 2016, but bad luck: it had a random failure where it was failing to boot. I sent it back to the reseller, and got it back under warranty a while later. Since the label at the back had a corner slightly bent (where the hidden screw is located), i believe they have change the motherboard.


Probe identification problem:
The boot problem was gone, but i've discovered a new problem: when connecting the probes, there are quite often not correctly identified, displaying sometimes bad ratio (1x or 20x instead of 10x), or sometimes bad type (active, Tek, 50 Ohms). But after repluging them several times, i manage to get it correctly identified as 10x Nor-probe.

I should have immediatly sent it back, but as once connected it stays OK, and because of various personnal and professional circumstances, i've used it more rarely that i was expecting, and i forget to do it, until the warranty expired  Yes, that is really stupid.

Does anybody have an idea of why the probe are randomly identified, and possibly how to fix it?

Fast forward to today: i still have this problem, but noticed some others.


Shifted levels + supposedly too high noise, and self calibration:

First, even with no probe connected, all channels were showing shifted levels, like if there was a DC component, whatever i selected DC or AC coupling.

The noise was also quite too high for me, since at the lowest 10mv/div, i have almost 20mVpp of noise, even when connecting probe's GND to tip.

So i wondered if i should do self calibration, but i'm a little confused about how to do it properly, and i haven't found any video on the Internets.
The scope asks to connect all five inputs (4 signals + trigger) to the "calibration signal".

My first (wrong) try was to connect all probes to the front calibration probe terminals. But it failed with "Selfcal connection error". But after that the levels were no more shifted!

But incomplete calibration was worrying, and reading the manual i understood that i should use the "calibration" BNC at the back. But there's one calibration output for 5 inputs ?!?!

Connecting 1 by 1 is not working. So finally i used a mess of T and I (male to male) BNC adapters + cables to distribute the signal to all inputs. And a mix of 75 and 50 Ohms, since it is all that i have at hand.
Despite being messy, it seems to work since it took about 20mn to complete without error.

But is it reliable? If not, how is it supposed to be done?

After that calibration, the level of noise if about 1.5mV when no probe connected (don't remember if it was like this before or has improved), but still almost 20mV with a probe grounded to itself. But i also notice that the signal is strange at the very center of the screen (see screen captures).
I haven't noticed that before, so could be because of the messy self calibration cabling.


Hot BNC connector:
And last strange thing: i also noticed that the front BNC connector of input 3 becomes quite hot, almost painful to touch. Is it normal with this model?


Conclusion: i feel pretty unlucky with my Rigol experience, with so many defaults/problems when i've bought a brand new unit to be sure to avoid any! And since i've not sent back the unit after its "repair", and now that the warranty is long expired, i'm afraid that sending it back for fixing it would cost quite a lot...
I'm not sure i would buy another Rigol scope if i decide to replace it one day...

But i hope that with the magic of the Internets and the help of some gurus here, i could may be fix (most) of the problems by myself!

Thanks!

[rsjsouza]

I am sorry to hear about your troubles with the DS4014. I have one since 2015 and it still works quite well.

I can't comment on the problem detection, since mine works quite well. I imagine you already checked the scope's detect ring for any excessive accumulation of grease and/or grime. If not, I would proceed to clean it up.

For the second issue, indeed self-calibration corrected many of the similar aspects you are seeing but I only need to do this quite sporadically (perhaps once a year?). The calibration signal is located in the back connector and all the five front BNC inputs are connected to it using a specific board purchased separately from Rigol.

https://www.tequipment.net/Rigol/CK-DS6000/General-Accessories/

Before I had purchased this board I was in a similar predicament as you, suspecting that my oscilloscope had bad channels (2 and 4 were quite different than 1 and 3). Once I went through this calibration with the board above, the four channels were spot on.

You don't necessarily need the board and its cables but instead any way to connect the five connections with identical lengths will do - the board is fully passive.

Regarding the hot BNCs, indeed they might be quite hot to the touch after long hours of work. However, I would open your oscilloscope and see if it has good ventilation and heatsinks inside. Mine had zero heatsinks and the poor FPGAs were operating quite hot.

I describe the heatsink woes on the post below of the ancient DS4014 thread.
https://www.eevblog.com/forum/testgear/rigol-mso4000-and-ds4000-tests-bugs-firmware-questions-etc/msg954693/#msg954693

[kinap35]

I imagine you already checked the scope's detect ring for any excessive accumulation of grease and/or grime. If not, I would proceed to clean it up.

Rings are perfectly clean, as are the probes' pin contacting with them. And the same problem is with all 4 probes, so it is not a failing probe neither.

The calibration signal is located in the back connector and all the five front BNC inputs are connected to it using a specific board purchased separately from Rigol.
https://www.tequipment.net/Rigol/CK-DS6000/General-Accessories/

Ok, thanks for the info. Some more money to spend 

You don't necessarily need the board and its cables but instead any way to connect the five connections with identical lengths will do - the board is fully passive.

My messy cabling was definitely not with identical lenghts, nor impedance, so new calibration is probably not reliable.
But the reference from your link is shown as discountinued, and i can't locate it elsewhere, including on Rigol website. Don't know where i can buy it in France...

indeed self-calibration corrected many of the similar aspects you are seeing but I only need to do this quite sporadically (perhaps once a year?)

I never done it myself since i got it back from repair in 2017. I wonder if a lot of people are doing it?

Regarding the hot BNCs, indeed they might be quite hot to the touch after long hours of work.

The input 3 BNC is already quite hot after only 30/45mn.

I would open your oscilloscope and see if it has good ventilation and heatsinks inside. Mine had zero heatsinks and the poor FPGAs were operating quite hot.
I describe the heatsink woes on the post below of the ancient DS4014 thread.

Interesting, i will probably try to open it and have a look.

[rsjsouza]

You don't necessarily need the board and its cables but instead any way to connect the five connections with identical lengths will do - the board is fully passive.

My messy cabling was definitely not with identical lenghts, nor impedance, so new calibration is probably not reliable.
But the reference from your link is shown as discountinued, and i can't locate it elsewhere, including on Rigol website. Don't know where i can buy it in France...

Unfortunately this board was designed for the DS6000 oscilloscope family, which was discontinued much longer ago than the DS4000. I suspect the easier way to assemble such fixture is to use three high quality BNC-BNC 50 ohm cables cut in half and soldered at a junction: five BNC connectors go to the front inputs and the sixth goes to the calibration output in the back.

indeed self-calibration corrected many of the similar aspects you are seeing but I only need to do this quite sporadically (perhaps once a year?)

I never done it myself since i got it back from repair in 2017. I wonder if a lot of people are doing it?

Probably not, since the extra cost and the very scant information from Rigol would not push people to buy such fixtures.

I would open your oscilloscope and see if it has good ventilation and heatsinks inside. Mine had zero heatsinks and the poor FPGAs were operating quite hot.
I describe the heatsink woes on the post below of the ancient DS4014 thread.

Interesting, i will probably try to open it and have a look.

Now I remember the BNCs reduced their temperatures quite radically when I added the heatsinks to the ADCs. I think it is quite worth taking a look inside, since opening it is an easy task to do.

Overall you might be thinking if the hassle is worth it. In my opinion it is one of the cheapest 500MHz four channel oscilloscopes (when unlocked) that has great memory depth, fully segmented, great protocol decoding (in memory, not only on screen) and it is a breeze to use.
Unfortunately its initial years were plagued by numerous bugs that detracted from its success.

In recent years I was able to also run mine on advanced RF gear (VNA, RF gen) and its bandwidth went up to around 600MHz and its S11 is quite spot on at 50 ohms up to ~550MHz.

I will run mine until something better falls in my dumpster. :-D

[2N3055]

I believe something like this cable arrangement for Keysight MSOX series would do the job..

[kinap35]

Unfortunately this board was designed for the DS6000 oscilloscope family, which was discontinued much longer ago than the DS4000. I suspect the easier way to assemble such fixture is to use three high quality BNC-BNC 50 ohm cables cut in half and soldered at a junction: five BNC connectors go to the front inputs and the sixth goes to the calibration output in the back.

That explains why i can't find it anymore. Oh well... it would have too simple to just buy it!

I believe something like this cable arrangement for Keysight MSOX series would do the job..

Almost, but the DS4014, you have to connect the 4 input + the trigger input, so you need 1 to 5 BNC splitting. But it is more or less what i've done, except i don't have enough 50 Ohms cables, not even of the same length, and mostly 75 Ohms adapters...
And i'm realizing that with only T adapters, it is pretty impossible to go from 1 to 5 with equal lenghts!

[2N3055]

I believe something like this cable arrangement for Keysight MSOX series would do the job..

Almost, but the DS4014, you have to connect the 4 input + the trigger input, so you need 1 to 5 BNC splitting. But it is more or less what i've done, except i don't have enough 50 Ohms cables, not even of the same length, and mostly 75 Ohms adapters...
And i'm realizing that with only T adapters, it is pretty impossible to go from 1 to 5 with equal lenghts!

OMG, you are correct , it is 1 to 5 split...

If you are not in the hurry, It would not be hard to draw a PCB and order one from JLCPCB. In fact I would make a PCB with central vertical BNC and 5 periferal BNC in 5-star layout on the other side of PCB. Make it compact enough and exact impedance would not be that important. Buy 5 same short cables and central one  longer to reach the back.

[rsjsouza]

That or, if you have the time and the energy, create your own in the style of a typical CCTV splitter box with panel-mount BNCs and good quality connections in a star configuration.


Later I can get some dimensions of the board I have here and post in this thread.

[kinap35]

Yes, i think i will have to create a PCB, but i have to find a software and learn how to do that... Since on Rigol's PCB, there are different wriggling traces for each output, having a perfectly equal length for each output seems pretty important, probably because timing adjustements can be in the order of nanonseconds.

In the meantime, it seems my improvised cabling and calibration have completely messed up my scope. It doesn't manage to trigger on a sine 105 kHz signal 

Is there a way to cancel/reset calibration?

[kinap35]

I've just found this document about DS4000 calibration procedure: https://beyondmeasure.rigoltech.com/acton/attachment/1579/f-04b7/1/-/-/-/-/DS4000%20Calibration%20Guide.pdf

First, they show a BNC one-to-five cable instead of a PCB, but with no reference/part number nor indication of where to find it.

Second: they describe a procedure to supposedly trigger a "project" menu with a "ResumeCal" option to reset calibration. The F7/F6/F7/Utility sequence doesn't work, but this menu is accessible deep inside the Utility menu. And "ResumeCal" does effectively reset calibration, and now SelfCal show only zeroes instead of date/time of last calibration.

But the offseted levels are now back (see attached picture, with no probe connected), with offsets from about -5 to -15mV, and now other things are probably also badly calibrated.

I also can see in the document that in addition to the "Selfcal" (Vertical) calibration, there are also "Equivalent cal" and "Delay cal", and the last one requires a signal generator. They ask for a DG5000, but since the requested signal is "square waveform with 100 kHz frequency, 4 Vpp amplitude and 50 Ω impedance", i suppose my DG1022Z should be OK (hoping it is itself well calibrated enough).

What a mess! 

Otherwise, a (maybe) more positive but surprising discovery about the probe identification problem. I had the idea, while connecting the probe, to insert the BNC but without locking it straight away, so basically pushing it in, and wait a second or two before turning it to the lock position. And doing so, it seems that the probe is correctly identified. I tried several times in a row, but will have to confirm tomorrow to be sure.
I had a look in the manual, they don't say anything about this, just "connect the probe". And i don't think i have any super power to able to connect it much quicker than anybody else, so i'm very surprised.
Anybody could check this?

[rsjsouza]

The probe detection works well for me without any need for delay. I wonder if this is indicative of some sort of degradation on the oscilloscope detection circuit...

At any rate, yesterday I put together a Kicad project that mimics the design of my calibration fixture. Please check attached.

[kinap35]

The probe detection works well for me without any need for delay. I wonder if this is indicative of some sort of degradation on the oscilloscope detection circuit...

I wonder that too...

At any rate, yesterday I put together a Kicad project that mimics the design of my calibration fixture. Please check attached.

Great work !
Thanks!

[kinap35]

After some more tests, despite waiting some time when connecting the probe's BNC before turning it to lock position, it stills randomly detect incorrect probe type, but just (a little) more frequently.

So obviously there's something malfunctioning inside that scope...

I've just found somebody who is having the same problem with the same scope (despite i have searched before posting, but if you don't use the right keywords, you can miss interesting posts!):
https://www.eevblog.com/forum/repair/rigol-ds4014-mis-identifies-probes/

Unfortunately the guy had no answer at all...

[rsjsouza]

Just shameless self promotion... If you would like to check how the board was designed on Kicad, check below

[kinap35]

Interesting video with useful informations for a newbie like me. Thanks!