Hacking the Rigol DHO800/900 Scope

[AceyTech]

My Scope came with one of the really noisy fans. Within 0.5M, I couldn't stand the noise/whine.  I decided to find a quiet & hopefully cooler solution.
I know a couple others did mods with 80, 92 and 100mm fans, but I wanted to take the time to illustrate the benefits/tradoffs so people could determine if/how they wanted to take on the challenge. All the pix I remember seeing were 25mm deep, and weren't favored because the fan hits the bench.

First, I attached the stock fan cable to 15mm thick Thermalright fans.(Amazon $6.29USD each)

Fan 1: 80x80x15mm, <23.3 dBA 28.40 CFM 0.07A @12Vdc
Fan 2: 92x92x15mm, <22.4 dBA 42.58 CFM 0.18A @12Vdc
Note: stock fan is spec'd at 0.17A @12Vdc
FYI: All these tests are running on the internal @8Vdc provided by the original fan connector

I removed the stock fan from the heatsink, to decrease the turbulence and allow more laminar flow. I observed >1C drop after removing the fan, so I left it off for the rest.

I started with the 80mm.  The hardware included with the Thermalright fan fits in the DHO hexagon grid hole pattern nicely.
I located the fan so it misses the RIGOL logo.  Less turbulent air noise this way.

Next, using the 92mm fan, I had a bit more difficulty aligning the holes, but managed a location with maximum air flow, but not quite perfect bolt arrangement. i.e., It works, but not as good looking 'cuz I had to turn a couple backwards from the others. (Mechanial crash)

My factory stock fan, starting temps were always very close to 60C(~59.x), but I only found one saved screenshot at 58.3/55.5 @ 25-26C ambient.

After the mod, all channels active, +math, minimum of 1 hour cookin' at 26C ambient:
80mm = 55.5/52.7C 
80mm = 53.8/50.5C  (As a test, I blocked the remaining hex holes around the fan with tape to force all incoming air across heatsink)

92mm = 48.7/46.2C  I didn't tape up the surrounding holes, but I can if someone needs to know what the delta is.

I didn't order a grill for either fan, but definitely found out what happens when you stick a finger in the blades.   It very politely waits for a few seconds, then restarts.  I don't know if that's the fan or the 8V switcher recovering from the stall condition.

Some things to note:
  1.  Oh yeah.  It's Silent now...  At 1/2 meter, you can't hear a thing.  Tiny amount of air noise on the backside, directly at the 92mm fan, even less on the 80mm. (like, not even worth trying to measure)
  2.  It's a well known fact that cooler=better with IC's., but now you can decide if the upgrade is worth it to you.
  3.  If you don't want to clip/solder your stock fan cable, you could get a JST/Molex or other female connector with 2.54mm/0.10" spacing.
  4.  I may try to enhance the heatsink surface area by adding a small heatsink to the area where the stock fan was. I definitely think it would be worthwhile to do.  I have a 40mm² sink, but it was a bit too big for that spot, un-modified.
  5.  If you can find a 80mm at 35 CFM with a reasonable dBA, it might be just about perfect to run at 8V, or maybe run this one at 12V?.
  6.  Thread locker might be important when the nuts are on the inside. Or use plastic?

Hope this helps someone "upgrade" their 'scope to a quiet, cool running one...

Edit:  Update -- Be sure to check out the excellent 120mm fan upgrade by @shapirus that I completely ignored when I did my mods.  --Sorry again for that!

[shapirus]

All the pix I remember seeing were 25mm deep, and weren't favored because the fan hits the bench.

Somebody installed a 120mm x 15mm fan too: https://www.eevblog.com/forum/testgear/hacking-the-rigol-dho800900-scope/msg5338634/#msg5338634

It'd be ideal to find a fan with the same mounting hole spacing as the vesa mount (i.e. 100mm x 100mm), but so far I haven't seen any.

[AceyTech]

All the pix I remember seeing were 25mm deep, and weren't favored because the fan hits the bench.

Somebody installed a 120mm x 15mm fan too: https://www.eevblog.com/forum/testgear/hacking-the-rigol-dho800900-scope/msg5338634/#msg5338634

It'd be ideal to find a fan with the same mounting hole spacing as the vesa mount (i.e. 100mm x 100mm), but so far I haven't seen any.

Yeah, I didn't find any either.  All the ones around the 100mm range that I saw weren't measuring the mounting locations, but the frame of the fan.  Sucks. 

Edit:  Oh, That guy?!!?  What a hack!  he clearly didn't know what he was doing.   

[AceyTech]

Somebody installed a 120mm x 15mm fan too:

^^  I can't believe I forgot about that.  I remember seeing and loving it now.  Sorry buddy. 

BTW: Did you remove the stock fan as well?

[shapirus]

BTW: Did you remove the stock fan as well?

Sure enough. There's no point in it sitting there and shading the surface of the heat sink from the air flow.

[shapirus]

Yeah, I didn't find any either.  All the ones around the 100mm range that I saw weren't measuring the mounting locations, but the frame of the fan.  Sucks. 

The 120 mm fans have 105 mm mounting hole spacing, so for 100 mm the frame size would be someting like 115 mm. Maybe they exist, but it would definitely not be one of common types.

[AceyTech]

BTW: Did you remove the stock fan as well?

Sure enough. There's no point in it sitting there and shading the surface of the heat sink from the air flow.

Good job.  Awesome.  I may try to fit a chunk of heatsink in that area to increase surface area as I mentioned in Note #4.  It is sure to help., given what's almost directly underneath the heatsink there.

[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.

[Randy222]

If anyone is interested, I have attached two files. These are boot logs read from the UART connectors of the processor and FPGA.

RK log is interesting.
1) it confirms what I mentioned a few pages back, this version of droid is using sdcardfs as top layer to the two partions.
2) what's with all the "kernel module not found" messages?

[Randy222]

Regarding the chatter around hacking a channel to be 50ohm. Why bother, just use a 50ohm passthrough connector. That's been a std for ages now (for scopes that don't support internal 50ohm). And if you roast an insternal 50ohm Z network (whatever that may be), then you need to go dig inside the unit to do the fixing.

[gabiz_ro]

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.

[AndyBig]

2) what's with all the "kernel module not found" messages?

I have no idea I only cautiously say hello to Linux across the road, I have no friendly relations with him

[Randy222]

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.

Individual sinks? How do you mount them?

[ebastler]

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.

Getting rid of close to 50W from that small enclosure via convection cooling only seems unrealistic. Look at the size of heatsinks with less than 1K/W -- and those are specified for unobstructed convection, while I assume you intend to keep the scope's back lid in place?

Or, another way to look at it: Look at the heat energy per time that has to leave the enclosure. With just the slow air flow from convection, how hot would that air have to be?

[AndyBig]

We can even simulate the heat transfer process of a specific radiator. In the same SolidWorks, for example.

[ebastler]

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.

Main (only?) difference of the "Pro" version seems to be that it incorporates a dedicated NPU (Neural Process Unit) for neural networks. I don't see how that would be beneficial in the DHO scopes, and have not seen a reference to the RK3399Pro being used by Rigol.

[Randy222]

As for fans, if you can find a thin fan somewhere near 8mm thick, they will fit inside between sink and case. I can find 8mm thick boxer fans, but their XY dimensions are a bit too small, would need too many fans to get the desired airflow. A 40x40x8 would be ideal, you can place 2-3 of those inside the case.

[gabiz_ro]

Individual sinks? How do you mount them?

There are thermally conductive adhesives need to choose good one.

Getting rid of close to 50W from that small enclosure via convection cooling only seems unrealistic. Look at the size of heatsinks with less than 1K/W -- and those are specified for unobstructed convection, while I assume you intend to keep the scope's back lid in place?

Or, another way to look at it: Look at the heat energy per time that has to leave the enclosure. With just the slow air flow from convection, how hot would that air have to be?

Maybe a mixed solution
Passive for what can be done active for other.

[Mechatrommer]

if postage cost is not too much, dont forget to mail me the original heatsink.

[ebastler]

There are thermally conductive adhesives need to choose good one.

Check the weight of adequate heat sinks, and consider whether you want your BGA solder joints to carry that weight. (Plus the extra force when you set the scope onto a table or bump it in transport.)

Maybe a mixed solution
Passive for what can be done active for other.

Requiring the added weight of large heat sinks, an external fan since there will be no room left for an internal one, plus some creative wiring for temperature sensors?  I am not convinced.

[Randy222]

There are thermally conductive adhesives need to choose good one.

Check the weight of adequate heat sinks, and consider whether you want your BGA solder joints to carry that weight. (Plus the extra force when you set the scope onto a table or bump it in transport.)

Maybe a mixed solution
Passive for what can be done active for other.

Requiring the added weight of large heat sinks, an external fan since there will be no room left for an internal one, plus some creative wiring for temperature sensors?  I am not convinced.

For many the little DHO's are not stationary, thus the sinks must be able to stay on while the unit gets knocked around. More cons then pros?

I am also wondering if the OEM sink serves as some form of gnd-plane?

[AndyBig]

I am also wondering if the OEM sink serves as some form of gnd-plane?

Absolutely not.

[Mechatrommer]

EMI shield probably yes to some extend..

[ebastler]

It's quite a neat cooling design in my opinion. Within the constraints of the compact enclosure, it's a pretty good solution -- large surface area, low profile.

Of course, seeing how people eventually add a large external fan as a backpack -- maybe there was no need to aim for such a low-profile enclosure? Adding a centimeter or two to the depth in return for a larger and quiet fan would have been the rational design choice. But probably the sleek form factor selles more scopes... 

[shapirus]

It's quite a neat cooling design in my opinion. Within the constraints of the compact enclosure, it's a pretty good solution -- large surface area, low profile.

Of course, seeing how people eventually add a large external fan as a backpack -- maybe there was no need to aim for such a low-profile enclosure? Adding a centimeter or two to the depth in return for a larger and quiet fan would have been the rational design choice. But probably the sleek form factor selles more scopes...

They could make it deeper at least up to the depth of the carrying handle, which would have allowed to put a larger and quieter fan inside, and that wouldn't have made the bounding box dimensions larger. It's yet another poor engineering decision that went into this platform. And they are also making a signal generator and a bench DMM in the same enclosure! And of course with the same whiny fans (maybe only powered with a lower voltage, but I think that's not likely).