Hacking the Rigol DHO800/900 Scope

[AceyTech]

What you think about passive cooling?
No thermal pad so better transfer of heat.
heatsinks with height of 20 mm less for FPGA and DDR4 that may allow only 19mm and for CPU only 18 mm
Big disadvantage will be weight increase.

Personally, I think about passive cooling all the time.  I think it's sad when engineers don't take the time to do a good job.  IMHO, they chose style over substance in this design. 
I bought a heatsink kit when I got the fans for my project, and will post my findings when I do it.

Weight increase?  Doubtful.  Have you seen the size and weight of the existing heatsink? OMG.  I think it will be less weight to implement discrete sinks.  I wish I had a scale here for a A-B comparison.

@ebastler  I don't think weight on the solder joints will be a concern.  As, solder's secondary trait(behind electrical) is mechanical.  And generally speaking, there is typically significant clamping force to hold a heatsink in place.  -Screws, springs, bands, etc.  Right?  There may be a spec, of course, and thanks for boosting the awareness of that issue.  I just don't think there's enough room internally(<17mm? vertical) to hang enough weight to cause problems.
Also, have you seen some of the heatsinks companies/users/hackers put on little SBCs with these SoCs?? -Wowsa.

[Mechatrommer]

It's yet another poor engineering decision that went into this platform.

probably your and few others' POV only... adding large fan and add the depth is much easier than cutting enclosure into half to get sleeker profile that fit in a laptop backpack needed by some... you cant satisfy everybody... my PC fan makes dho800 noiseless... i need to put my ear to the very back of enclosure to make sure i didnt forgot to plug the fan's connector in during reassembly...

[AceyTech]

Verify this datasheet matches the actual RK chip used in your DHO. My gpio mappings are correct, but if wrong datasheet they could be physically mapped in a different matrix (there are several versions/flavors of RK3399).

Question:  How many "versions/flavors" of the RK3399 chip have you discovered?
 
And do you have any info that Rigol used different versions in the DHO's?
Links please?

I have seen the RK3399 reprsented as "RK3399(K)" "RK3399-K" "RK3399" and "RK3399Pro"
However, only the "Pro" version seems to have a spec sheet that is different than the others.

Links?  I'm well aware of the Pro, but I have not seen any PDFs for the ?K? version.  There is a T version on some SBCs...  Is that what you're thinking?  I have brief info for the T part, and BTW it has the identical pin out that the std RK3399 has.

Again; do you have any info that Rigol used different package parts(as you suggested) on the DHOs?  that would suggest different PCB versions.

[ebastler]

@ebastler  I don't think weight on the solder joints will be a concern.  As, solder's secondary trait(behind electrical) is mechanical.  And generally speaking, there is typically significant clamping force to hold a heatsink in place.  -Screws, springs, bands, etc.  Right? 
There may be a spec, of course, and thanks for boosting the awareness of that issue.  I just don't think there's enough room internally(<17mm? vertical) to hang enough weight to cause problems.
Also, have you seen some of the heatsinks companies/users/hackers put on little SBCs with these SoCs?? -Wowsa.

I was referring to gabiz_ro's suggestion to mount the heatsinks (only) by glueing them to the chips -- no screws, springs, bands. I would certainly advise against that.

And again -- look at a few heatsinks with < 1K/W ratings. Those are a different caliber than anything I have seen mounted on SBCs! And they are probably still not good enough to cool the DHO800, unless you leave its back cover off for unobstructed convection. 

[AceyTech]

@ebastler  I don't think weight on the solder joints will be a concern.  As, solder's secondary trait(behind electrical) is mechanical.  And generally speaking, there is typically significant clamping force to hold a heatsink in place.  -Screws, springs, bands, etc.  Right? 
There may be a spec, of course, and thanks for boosting the awareness of that issue.  I just don't think there's enough room internally(<17mm? vertical) to hang enough weight to cause problems.
Also, have you seen some of the heatsinks companies/users/hackers put on little SBCs with these SoCs?? -Wowsa.

I was referring to gabiz_ro's suggestion to mount the heatsinks (only) by glueing them to the chips -- no screws, springs, bands. I would certainly advise against that.

And again -- look at a few heatsinks with < 1K/W ratings. Those are a different caliber than anything I have seen mounted on SBCs! And they are probably still not good enough to cool the DHO800, unless you leave its back cover off for unobstructed convection.

It's only 6W(TDP) for the RK3399, with few watts for the display, few watts for the storage, a few for misc and efficiency losses. Roughly 20-25W.,  the rest for the FPGA. -total on the 800s is ~35W., with a tiny bit more for 900S because of the AWG board.  How are you coming up with anything close to 1KW?  Edit: Disregard, I misread.

BTW, I have seen at least one industrial SBC maker offering a fanless(touting high reliability) RK3399 board design, that had a relatively small heatsink. -maybe ?35x35mm.  (Has a 2 year warranty as well.)

I think It might be ok using thermal tape/adhesive as @gabiz_ro was suggesting., provided the adhesive keeps them in place after a few thermal cycles.  In my experience with good quality vendors, this hasn't been a concern.  However, we are in the new world of AliExpress/Amazon cheap parts these days, so we'll have to see what happens!

[ebastler]

It's only 6W(TDP) for the RK3399, with few watts for the display, few watts for the storage, a few for misc and efficiency losses. Roughly 20-25W.,  the rest for the FPGA. -total on the 800s is ~35W., with a tiny bit more for 900S because of the AWG board.  How are you getting close to 1KW?

Total power consumption of the DHOs is known to be 4A at 12V, or 3A at 15V if I remember correctly, so about 50W. A heatsink with a 1 K/W rating (that's one Kelvin of temperature increase per Watt of dissipated power -- nothing to do with 1 kW!) seems the lower limit of what you can get away with. It would run at ambient + 50°C, even under the optimistic assumption of free air convection.

Edit: a 35*35 mm² heatsink (with less than 35 mm height, I assume?) will be far from 1 K/W -- more like 10 K/W. You would need a much more hefty heatsink to passively cool the DHO. 

[shapirus]

I don't think dissipating 35 W passively in these dimensions can be viable. 35 W (or what is it actually, 50 W?) is a lot of heat.

Dave tried running the scope with the stock heat sink without the fan in one of his videos (on the second channel IIRC), and it did overheat eventually. Discrete heat sinks may or may not be more effective than that, but either way it gives a decent estimation of what can be achieved with passive cooling. No, I think some forced air flow, even though pretty low, will still be required.

But if someone is willing to experiment with it, there's nothing wrong in that. Quite the opposite: it'll provide useful information to the community, whatever the result will be. I'd only avoid using any kind of permanent thermal glue: the mods should be reversible.

[zrq]

The power consumption is around 35W according to my measurement, of course, this amount of power will make passive cooling very difficult. My surface pro 8 (a consumer product with professionally designed thermal management) will also need to turn on the fan when running at the full load (comparable power and size).

[AceyTech]

Total power consumption of the DHOs is known to be 4A at 12V, or 3A at 15V if I remember correctly, so about 50W. A heatsink with a 1 K/W rating (that's one Kelvin of temperature increase per Watt of dissipated power -- nothing to do with 1 kW!) seems the lower limit of what you can get away with. It would run at ambient + 50°C, even under the optimistic assumption of free air convection.

Edit: a 35*35 mm² heatsink (with less than 35 mm height, I assume?) will be far from 1 K/W -- more like 10 K/W. You would need a much more hefty heatsink to passively cool the DHO.

Doh!  Brain fart. Sorry about the kW vs K/W confusion.  Didn't caffeinate before reading your post..  Agreed.,
--and as others have noted, 35W.  That 12V-4A number was for the power supply that was shipped with a few early models.

[Randy222]

Verify this datasheet matches the actual RK chip used in your DHO. My gpio mappings are correct, but if wrong datasheet they could be physically mapped in a different matrix (there are several versions/flavors of RK3399).

Question:  How many "versions/flavors" of the RK3399 chip have you discovered?
 
And do you have any info that Rigol used different versions in the DHO's?
Links please?

I have seen the RK3399 reprsented as "RK3399(K)" "RK3399-K" "RK3399" and "RK3399Pro"
However, only the "Pro" version seems to have a spec sheet that is different than the others.

Links?  I'm well aware of the Pro, but I have not seen any PDFs for the ?K? version.  There is a T version on some SBCs...  Is that what you're thinking?  I have brief info for the T part, and BTW it has the identical pin out that the std RK3399 has.

Again; do you have any info that Rigol used different package parts(as you suggested) on the DHOs?  that would suggest different PCB versions.

I don't really know what exact RK3399 Rigol used. If you say Rigol used the low end "RK3399" and not K and not Pro, then I guess that's what it is.
I was not suggesting that some DHO's have varying RK3399. What I said in that post was, make sure the datasheet I linked is the right one. I could have perhaps linked to the wrong datasheet.

RK3399
RK3399K
RK3399Pro
RK3399-T

The K version is right in the datasheet. K appears to have faster clock rate. Pro appears to have a "NPU" processor. Since K and non-K are in same datasheet, then I suspect it has same physical pin mapping since it's one datasheet.
So perhaps three-four-five flavors, but maybe Pro and T has a different pin layout?

See pics

 

[Randy222]

EMI shield probably yes to some extend..

That's what I was saying when I said "gnd-plane".
Making those individual glue-on sinks probably takes away from the shielding. If the OEM sink didn't need to be made boxy like that, then they could have saved a ton of aluminum.

[AceyTech]

Edit: a 35*35 mm² heatsink (with less than 35 mm height, I assume?) will be far from 1 K/W -- more like 10 K/W. You would need a much more hefty heatsink to passively cool the DHO.

On closer examination;  While that SBC itself is fanless, their temp spec says "@0.7m/s Air Flow".,   BTW it looks  significantly shorter than 35mm height.  They also don't have a FPGA running balls out, contributing to the heat load.

Incidentally, this is a very warm product, naturally.  Uncomfortably so, IMHO.  When I unplug a USB connector it's HOT., BNC's are as well.  I forgot to measure the temperature on the connectors during my fan testing, but the last mod I did cooled the I/O's considerably.

BTW: @shapirus
I watched that video you mentioned, and have some issues with his methodology.  I'm not saying it was flawed per se', but I just can't put much confidence in some of his findings in general.  Ex: when Rigol did their HALT testing, I doubt they turned it up on end, dumping heat into the power supply section.
  -From what I've observed, these scopes don't really need much of an excuse to reboot.  Sometimes, I'll look over at my DHO because it's clicking and flashing in the middle of a reboot cycle.  For no reason!
(Before anyone asks: ^that's before I opened it)

[AceyTech]

I don't really know what exact RK3399 Rigol used. If you say Rigol used the low end "RK3399" and not K and not Pro, then I guess that's what it is.
I was not suggesting that some DHO's have varying RK3399. What I said in that post was, make sure the datasheet I linked is the right one. I could have perhaps linked to the wrong datasheet.

RK3399
RK3399K
RK3399Pro
RK3399-T

The K version is right in the datasheet. K appears to have faster clock rate. Pro appears to have a "NPU" processor. Since K and non-K are in same datasheet, then I suspect it has same physical pin mapping since it's one datasheet.
So perhaps three-four-five flavors, but maybe Pro and T has a different pin layout?

I didn't say Rigol used the low end RK3399. Nor would I, because the -T is the low end part. 
Rigol used the commonly available RK3399 part that's in almost everything that is RK3399 based.  BTW: I had to diff the PDFs to find info on the -T parts, because I couldn't find anything online, and don't have any factory rep contacts these days.
 
The K part(I.e., commercial) is from "binning" the fast from the slow parts, with extended temperatures range.  -T parts are the slowest of the bunch at 1.5/1.0G clock speeds.

Other than the Pro, the packages, pinouts, etc. are all the same. -You had me going for a minute, I thought you found an undiscovered 3399 variant.

[rifkum]

Hey Andy (or others who know the answer)...
Regarding your post #1507 in this thread (step-by-step mod guide)

First Andy a big thank you for summarizing the process of updating the DHO800!!
It is mind bending to try to follow all the posts in this thread and keep up with where we are in the learning process.

Is it correct that one should update firmware on their new scope BEFORE doing the hack as you have detailed??
My new DHO804 currently has Firmware 00.01.01 but as of this writing I see Rigol has 00.01.02.00.02 available for download.

If the answer to the firmware update question is YES then it might be a good idea for an edit of your post #1507 to clarify this question.
Even better, as someone else suggested, would be to create a new TOPIC in the forum specifically just to hold the guide and have the separate "guide topic" be updated as new relevant steps or options come to light. 
It is a daunting task for a new owner to come to this thread and try to sift through the (currently) 98 pages of discussion.
Or move your guide to the top of the thread would be another option.

But the question I have now is whether I need to update to the latest firmware BEFORE doing the mod??

Thanks again for taking to time to create your step-by-step guide!!

Cheers
Dwight

[AndyBig]

Hey Andy (or others who know the answer)...
Regarding your post #1507 in this thread (step-by-step mod guide)

First Andy a big thank you for summarizing the process of updating the DHO800!!
It is mind bending to try to follow all the posts in this thread and keep up with where we are in the learning process.

Is it correct that one should update firmware on their new scope BEFORE doing the hack as you have detailed??
My new DHO804 currently has Firmware 00.01.01 but as of this writing I see Rigol has 00.01.02.00.02 available for download.

If the answer to the firmware update question is YES then it might be a good idea for an edit of your post #1507 to clarify this question.
Even better, as someone else suggested, would be to create a new TOPIC in the forum specifically just to hold the guide and have the separate "guide topic" be updated as new relevant steps or options come to light. 
It is a daunting task for a new owner to come to this thread and try to sift through the (currently) 98 pages of discussion.
Or move your guide to the top of the thread would be another option.

But the question I have now is whether I need to update to the latest firmware BEFORE doing the mod??

Thanks again for taking to time to create your step-by-step guide!!

Cheers
Dwight

Yes, if you plan to change the model from 8xx to 9xx by hacking, then updating the firmware to version 00.01.02.00.02 is necessary, otherwise you will get unpleasant vertical displacements on the channels.
I still can't get around to changing that description. I'll do this right now.

[Fungus]

Is it correct that one should update firmware on their new scope BEFORE doing the hack as you have detailed??

Yes.

[Randy222]

Is it correct that one should update firmware on their new scope BEFORE doing the hack as you have detailed??

Yes.

Not sure of all the steps in the hack list provided, but I installed the AndyBig APK into my DHO that was running FW 00.01.02.00.00 and a 914 vendor.bin with lics. The scope app worked fine from what I could see. I later did upgrade to 00.01.02.00.02 and then re-installed the AndyBig APK, all seems ok, but I think I have odd offset in some of the channels, some say the channel os at 0.00v but if I locate the trigger in trace it's -7.54mv, even after running selfcal a few times.

[Randy222]

I don't really know what exact RK3399 Rigol used. If you say Rigol used the low end "RK3399" and not K and not Pro, then I guess that's what it is.
I was not suggesting that some DHO's have varying RK3399. What I said in that post was, make sure the datasheet I linked is the right one. I could have perhaps linked to the wrong datasheet.

RK3399
RK3399K
RK3399Pro
RK3399-T

The K version is right in the datasheet. K appears to have faster clock rate. Pro appears to have a "NPU" processor. Since K and non-K are in same datasheet, then I suspect it has same physical pin mapping since it's one datasheet.
So perhaps three-four-five flavors, but maybe Pro and T has a different pin layout?

I didn't say Rigol used the low end RK3399. Nor would I, because the -T is the low end part. 
Rigol used the commonly available RK3399 part that's in almost everything that is RK3399 based.  BTW: I had to diff the PDFs to find info on the -T parts, because I couldn't find anything online, and don't have any factory rep contacts these days.
 
The K part(I.e., commercial) is likely "binning" the fast from the slow not so fast parts, with extended temperatures range.

Other than the Pro, it appears the packages, pinouts, etc. are all the same. -You had me going for a minute, I thought you found an undiscovered 3399 variant.

I don't follow the RK product line, so I myself really don't know what level each "flavor" is at. I only noted that there are several flavors and that we need to check that the datasheet we look at actually matches the RK that's in the DHO.

[AndyBig]

Not sure of all the steps in the hack list provided, but I installed the AndyBig APK into my DHO that was running FW 00.01.02.00.00 and a 914 vendor.bin with lics. The scope app worked fine from what I could see. I later did upgrade to 00.01.02.00.02 and then re-installed the AndyBig APK, all seems ok, but I think I have odd offset in some of the channels, some say the channel os at 0.00v but if I locate the trigger in trace it's -7.54mv, even after running selfcal a few times.

With previous versions of applications there were problems with offsets of up to one major division on the display. For example, with a vertical scale of 100 mV/division there could easily be an offset of 60-80 mV. And it seems that even I saw screenshots with an even greater offsets. This happened when changing the model from DHO8xx to DHO9xx due to different calibration algorithms in these series. In version 00.01.02.00.02, the calibration algorithm was made uniform for the 8xx and 9xx series and the problem of offsets went away.

[AceyTech]

EMI shield probably yes to some extend..

That's what I was saying when I said "gnd-plane".
Making those individual glue-on sinks probably takes away from the shielding. If the OEM sink didn't need to be made boxy like that, then they could have saved a ton of aluminum.

You guys bring up a good point.  Look what I found when browsing through teardown photos of higher-end Rigol scopes.  Individual tin shields with a cutout for the AFE heatsinks to stick through.


I wonder about the temp of these versus the DHO800/900s. 

[souldevelop]

Hello guys!
 I haven't been online for a long time, I want to know the latest DH800 900 firmware v00.01.02.00.02 , is there any other way besides using the modifying apk vendor.bin?

[ebastler]

Wouldn't it be great if the initial post of this thread could be updated with the current hacking instructions? We have AndyBig's up-to-date instructions, but they are hard to find in the long thread.

@sebyon, the OP, is not online daily, but is an active user. @sebyon, if you read this -- could you please update your original post? (I also see that a moderator has once updated the initial post, but the information is outdated again. If sebyon is not available, we could ask the mods for another update.)

EDIT: In the interim, the up-to-date instructions are here:
https://www.eevblog.com/forum/testgear/hacking-the-rigol-dho800900-scope/msg5344076/#msg5344076

[AndyBig]

At the suggestion of AceyTech, I specifically included a link to the description of the hack in my signature so that this description would be easier to find

[AndyBig]

You guys bring up a good point.  Look what I found when browsing through teardown photos of higher-end Rigol scopes.  Individual tin shields with a cutout for the AFE heatsinks to stick through.

I wonder about the temp of these versus the DHO800/900s. 

It seems to me that front-end chips should not consume a lot of energy and get very hot. The main consumers are the processor, FPGA and ADC. Well, the display probably consumes something, but we are not interested in it in terms of cooling

[AceyTech]

You guys bring up a good point.  Look what I found when browsing through teardown photos of higher-end Rigol scopes.  Individual tin shields with a cutout for the AFE heatsinks to stick through.

I wonder about the temp of these versus the DHO800/900s. 

It seems to me that front-end chips should not consume a lot of energy and get very hot. The main consumers are the processor, FPGA and ADC. Well, the display probably consumes something, but we are not interested in it in terms of cooling

Agreed.  It seems an odd design choice that these AFE's are essentially heated by the big thermal generators.