Rigol MSO4000 and DS4000. Tests, bugs, firmware, questions, etc...

[rsjsouza]

I just updated and can confirm the setup save/load works fine. The counter is big but it doesn't bother me. Unfortunately I don't have time at the moment to test it thoroughly.

Regarding the big box in the center, I had a slightly different one (it had two grids with Vrms and Freq). I went to the measure button and cleared all measurements - the box went away.

[trukresom]

Hi,

Pressing the Default button and then OK solved the problem; the mysterious window disappeared.
Indeed it had exactly the size of the History Graph window but was empty.

[rsjsouza]

Could someone on the latest DS4k firmware please try to replicate this bug:

FFT for channel number x is forced to zero when channel number x-1 is switched to GND coupling

1: Connect a probe to channel 2 and observe some trace
2: Turn on FFT for channel 2 and observe some spectrum
3: Switch Coupling for channel 1 to GND
4: Observe FFT trace for channel 2 is now zero.

BTW, this still happens with the latest firmware.

FFT for channel number x is forced to zero when channel number x-1 is switched to GND coupling

Yes, that happens here as well. Latest FW.

Also, did you notice that the trigger line on the display disappears? IIRC this happened with me when I closed the FFT mode. To bring it back I had to go to Trigger menu and fiddle with something there.

This does not happen with the new firmware.

Also, the new firmware fixes the issue with the super slow FAT32 file save.

[Andrew8086]

Fun with 00.02.03.00.03: In the trigger menu falling and rising edge trigger options are exchanged: choosing 'rising edge' triggers on falling edge and vice versa.
Don't want to downgrade just to check - is that a new one or didn't I catch this before? 

[alank2]

It works fine for me.  I connected up CH1 to the square wave output and on rising it triggers on rising and on falling it triggers on falling...

[TurboTom]

Fun with 00.02.03.00.03: In the trigger menu falling and rising edge trigger options are exchanged: choosing 'rising edge' triggers on falling edge and vice versa.
Don't want to downgrade just to check - is that a new one or didn't I catch this before?

Maybe the corresponding input channel is set inverted? On mine, the polarity of the trigger slope is correct after the update (as it has been before as well...).

Cheers,
Thomas

[Andrew8086]

Maybe the corresponding input channel is set inverted? On mine, the polarity of the trigger slope is correct after the update (as it has been before as well...).

Cheers,
Thomas

Yes that's it, my bad.  All 4 channels were set to inverted after the Fw upgrade ...

[rsjsouza]

I just changed the encoder on my DS4014 with the model mentioned in this other thread (ALPS EC12E2424407) and it works really well for push-select - no missed selections at all! The only detail is that its detented force is much higher than the other detented encoders present in the unit (vertical and horizontal encoders). I couldn't see any significant difference in height between the old and the new encoder, as reported in the other thread with the DS1054Z.

If you want to do this you have to remove the back cover and remove only the outer screws from the main chassis - removing the main PCB from the chassis is not needed at all - see the first screenshot. Also, you have to remove the front panel overlay decal as there are screws under it (mine was not even attached - they forgot to peel the adhesive protection).

Also, at least in my unit, the upper buttons (Clear, Auto, Run/Stop, Single) are connected to the main front panel PCB via a very tiny and flimsy connection: the IDC ribbon cable is properly tied to a connector on the main front PCB, but it is just soldered under the button PCB. Both the connector and the solder areas were full of silicone glue; the solder side broke and took me some work to put it back together.

[TurboTom]

Since I replaced the multi-function knob encoder on my DS100Z with a detented one quite a while ago, I could never accomodate myself with the non-detented ones on my recently acquired MSO4000 (multi-function and logic analzyer controls). The non-detented ones are completely acceptable for those functions that emulate analog inputs (like trigger level, trace position and so on) but in my opinion an absolute no-go for for anything "digital".

So I cracked that thing open and swapped in two decent ALPS encoders (the original ones don't appear to be bad quality, it's only Rigol used the wrong type in that place). The usability is so much better now -- I don't get annoyed anymore when I've got to change the settings. Moreover, the detent function camouflages the laggy display update rate quite well. It's amazing how the human brain integrated the different sensoric perceptions. If I need to select an item three lines down, I just turn the know three detents without even having to look at the screen. And most beneficial, pressing the button doesn't affect the setting anymore.

I only hope I'll never have a defect on the device so I'ld have to swap in the original ones to validate warranty again...

Another interesting finding during disassembly: Since my scope is an MSO version and is of fairly recent manufacturing date (february 2016), it shows that Rigol continuously improves their products, see the attached photo. They took care of the rather hot-running ADCs (as reported by several contributors) by replacing the individual heat sinks with a single, big, machined heat sink that covers the whole "high temp area" of the digital signal processing path. Well done! Moreover, the photo finally shows the purpose of the high speed connector on the main board that Connor Wolf was referring to in his teardown video. This connector isn't visible in the photo but it shows that it actually intefaces to the logic analyer subassembly on the MSO models. Altogether, hardware-wise the current production DS / MSO 4000 scopes are very neat and tidy machines and I don't think they lack behind any of the "A brand" designs. The firmware may have its issues but at least Rigol is still providing updates once in a while.

Cheers,
Thomas

[tautech]

Another interesting finding during disassembly: Since my scope is an MSO version and is of fairly recent manufacturing date (february 2016), it shows that Rigol continuously improves their products, see the attached photo. They took care of the rather hot-running ADCs (as reported by several contributors) by replacing the individual heat sinks with a single, big, machined heat sink that covers the whole "high temp area" of the digital signal processing path.

What some may want to know is your DSO's HW revision #.
Can you check it and post it?

[TurboTom]

Sure. H/W of my MSO4014 is 0.1.3.1

Cheers,
Thomas

[hendorog]

Sure. H/W of my MSO4014 is 0.1.3.1

Cheers,
Thomas

Nice work on the encoder, and thanks for thinking to ask that question tautech.

Another data point - My scope is from mid 2014 and the HW version is also 0.1.3.1

[rsjsouza]

My DS4014 had only one heatsink in the main Virtex 5 FPGA - everything else was left to fend for themselves. Also it looks like they used a caulk gun to put the thermal adhesive - I haven't seen so much white gunk on a heatsink before!

I bought some beefy heatsinks and used a very thin layer of Arctic Silver thermal adhesive to tie them to the remaining Virtex 5s. The heatsinks I bought for the ADCs from Thermalloy already had the thermal adhesive pad.

One additional detail is that my board only had the PLL part number scrapped - all others were intact.

I will post the HW rev, pictures and the heatsink models later.

So I cracked that thing open and swapped in two decent ALPS encoders (the original ones don't appear to be bad quality, it's only Rigol used the wrong type in that place).

I agree: the original encoders have good quality but the difference in usability is massive!

Another interesting finding during disassembly: Since my scope is an MSO version and is of fairly recent manufacturing date (february 2016), it shows that Rigol continuously improves their products, see the attached photo. They took care of the rather hot-running ADCs (as reported by several contributors) by replacing the individual heat sinks with a single, big, machined heat sink that covers the whole "high temp area" of the digital signal processing path. Well done!

After the heatsink hackjob on mine, just be sure the large heatsink is actually touching the ADCs, as I recall they were slightly thinner than the Virtex 5s.

Moreover, the photo finally shows the purpose of the high speed connector on the main board that Connor Wolf was referring to in his teardown video. This connector isn't visible in the photo but it shows that it actually intefaces to the logic analyer subassembly on the MSO models.

My non-MSO oscilloscope has the high-speed connector populated. Interesting to see its function.

[tautech]

Sure. H/W of my MSO4014 is 0.1.3.1

Cheers,
Thomas

Nice work on the encoder, and thanks for thinking to ask that question tautech.

Another data point - My scope is from mid 2014 and the HW version is also 0.1.3.1

And with no common heatsink?
Same model?

[hendorog]

Sure. H/W of my MSO4014 is 0.1.3.1

Cheers,
Thomas

Nice work on the encoder, and thanks for thinking to ask that question tautech.



Another data point - My scope is from mid 2014 and the HW version is also 0.1.3.1

And with no common heatsink?
Same model?

Mine is a DS4014 like rsjsouza, not an MS like Tom's. I don't know about the heatsink situation, as I haven't pulled it apart

Anyone know if it can be seen from the outside?

[hendorog]

Sure. H/W of my MSO4014 is 0.1.3.1

Cheers,
Thomas

Nice work on the encoder, and thanks for thinking to ask that question tautech.

Another data point - My scope is from mid 2014 and the HW version is also 0.1.3.1

And with no common heatsink?
Same model?

OK the heatsinks are partially visible from the outside. I can see in through the grill from the right hand end of Tom's picture.

Conclusion is that the heatsinks are not the same configuration as in the picture Tom posted.
Mine definitely has individual heatsinks - so it is like Conner Wolf's teardown: https://youtu.be/_J1pVlqMIHM?t=584

So, yep it would seem that the H/W version number hasn't been changed to match the different assembly.

[rsjsouza]

The HW rev is assigned by a bunch of small resistors soldered in the PCB, close to the coin cell.

I need to check my pictures at home, but I would expect that at least a PCB revision was needed, as I don't recall it having the proper holes to accommodate the screws that fit the heatsink as shown in TurboTom's picture.

[H.O]

I'll put a detented encoder on my next Digikey order for the purpose of replacing the multifunction knob on my DS4k.

Some encoders have a detent per cycle while others have a detent per state change within the quadrature cycle (ie 4 detents per cycle).
Looking at the datasheet for the ALPS encoder mentioned earlier it says it's got 24 detents and a resolution of 24. To me that means this particular encoder has a detent per state change - not per cycle (which seems to be the more common type). Unfortunately there's no "timing diagram" in the datasheet so I'm not 100% sure.

Digikey doesn't stock the specific ALPS encoder or I would've just got that one. Am I correct in selecting an encoder with one detent per state change?

[rsjsouza]

Well, what do you know? My HW version also says 0.1.3.1, but my board was like the one attached.

The heatsinks I used were:
Wakefield 630-60AB for the Virtex 5s
Thermalloy 374424B00032G for the ADCs

To glue the Wakefield heatsinks to the devices I used a thin layer of the Arctic Silver Thermal Adhesive. The Thermalloys already had the adhesive thermal pad.

[TurboTom]

After I replaced the multi-function / logic analyzer incremental encoders with detented ones, I noticed that the user interface of the DS/MSO 4000 is somewhat inconsistent: Some of the (sub-) menues with only few positions like input coupling for inctance need the controls to be turned two detents to change item while others with many positions (like trigger source on the MSO where you can select each digital channel) change line with every detent. Whatsoever, simply the fact that the setting won't change anymore when pressing the encoder button improves usability a lot.

Cheers,
Thomas

[H.O]

I wonder if they've implemented different decoding depending on which menu you're in. A menu with many selections is using 4x while a menu with a less items is using 2x - perhaps. That certainly "works" with a non detented encoder but not so well with a detented one.

[rsjsouza]

Come to think of it, I also noticed a similar thing on some menus, however the alphanumeric keypad has been consistently mapped 1:1 for me - that and the fact the push selection is absolutely stable on all menus.

My suspicion is they employed a different encoding to overcome different delays in the UI, which caused over-sensitivity on a free-running encoder.

[alank2]

Did you guys see the DS4014E and DS4024E they have on the Chinese site.  The DS4014E looks to be about half the price of the DS4014 if you look at the CNY numbers they have.  Oddly, it only has 2GSa/s listed instead of 4GSa/s - one has to wonder why they are still calling it a DS4...

http://rigol.com/Product/Index/88

[Teneyes]

Did you guys see the DS4014E and DS4024E they have on the Chinese site.  The DS4014E looks to be about half the price of the DS4014 if you look at the CNY numbers they have.  Oddly, it only has 2GSa/s listed instead of 4GSa/s -

On Russian site
Is each channel @2GSa/s?
Added flex Ray decoding.
Are there hidden BW / memory options?

http://technica.ru/objects/download_files/article_pdf/Rigol_DS4000E_DataSheet_EN.pdf

[rsjsouza]

Is each channel @2GSa/s?

I don't think it is - similarly to the DS4000, it says "up to 2GSPS for each channel". I would find it hard they would save some money by simply removing the ADC channel interleaving. Perhaps it is a cheaper dual channel 1GSPS ADC and the 2GSPS is only on every two channels. It could also use a different frontend that goes only to 200MHz.

Added flex Ray decoding.

The DS4000 has FlexRay, and at a certain point Rigol offered this for free.

Three additional details: four channel only, smaller memory (14Mpts/channel) and slower update rate (60kwfm/s).