Hacking the Rigol DHO800/900 Scope

[AndyBig]

Lol, I just discovered this nonsense on my oscilloscope board ) A drop of solder shorted one of the capacitors and almost shorted the second capacitor. These are blocking capacitors under the processor. Rigol, where is your quality control? )

[S2084]

Well, I’ll probably do an analog adjustment of the fan speed... I happen to have good NTC thermistors of different ratings.....

[AceyTech]

Well, I’ll probably do an analog adjustment of the fan speed... I happen to have good NTC thermistors of different ratings.....

Old school, I LOVE IT!    I need similar, and think I even have a 80mm fan that has a thermistor on it.  Now where did I put that box?

[shapirus]

Another example of an external fan mod.

Used one of those "slim" form factor 120 mm fans (15 mm thick). Settled on 5.4V / 1030 RPM after a bit of experimenting. Voltage regulation is achieved by two 24 Ohm resistors connected in series with the fan. Added a 560 uf, 10V polymer electrolytic cap to reduce the 4x RPM ripple that I mentioned above -- might be unnecessary, but I don't want any ripple inside the scope, where it can be avoided.

It is really quiet, only becomes audible when you get closer than ~50-60 cm, and that's in a quiet environment. Temps are ~4 deg.C lower than they were with the stock fan (it's now showing 52 CPU / 48 CPU ambient at 22 deg.C in the room). By running the fan at full 8V the temps are reduced by further 4-5 degrees, but the added noise makes it absolutely not worth it.


Overall look:




Mounting: M4x8 bolts. Anything longer won't be possible to get into the gap, and these are just the right size for this fan. Of course I had to drill four new 4.5mm holes for them spaced 100 mm x 100 mm.




Beware of the clearance between the fan's outer lower edge and the surface on which the scope stands! This is one of the reasons to prefer the slim form factor fans. In this case, the required clearance was achieved with the new holes placed at the same distance from the lower edge as the original ones. This placement also provides enough room for the carrying handle to remain usable, and the part of the fan protruding above the scope's housing is just so narrow that the blades can't be reached from inside, unless you have extra thin fingers and/or try really hard.
And yes, the back lid's mounting screw holes remain reachable with the fan installed. No need to remove it to get inside the scope.








Power cable routing:





p.s. Initially I wanted the fan to suck air out of the scope instead of blowing it in, but, alas, this fan (and I would guess that it's a general rule) is only good at pushing, not pulling. It was totally helpless when it tried to pull air through those vent holes: a lot of noise, very little performance. Pushing air into the scope, on the other hand, works really well.

[shapirus]

Well, I’ll probably do an analog adjustment of the fan speed... I happen to have good NTC thermistors of different ratings.....

And there's plenty of room inside even for a THT control board (within reason of course). You mean automatic temperature-controlled fan speed adjustment? This sounds like a viable idea, but for best results, I guess, the temperature sensor has to be screwed onto the heat sink, or otherwise put into tight thermal contact with it, and preferably on the bottom side of the heat sink to prevent air flow from spoiling the measurement.

[AndyBig]

..., the temperature sensor has to be screwed onto the heat sink, or otherwise put into tight thermal contact with it, and preferably on the bottom side of the heat sink to prevent air flow from spoiling the measurement.

Only this way and no other way.

[S2084]

Based on my observations using a thermal imaging camera, the board itself is subject to heating no less than the radiator, which means the NTC thermistor can be mounted on the reverse side, for example, under the FPGA chip, and there will be no protruding wires.  Regarding the method of controlling the rotation speed of the cooler, I meant replacing one of the resistors in the resistive divider with an NTC thermistor, the resistance of which decreases as the temperature rises.  I have already found one whose resistance at room temperature is 8 kOhm, it seems to me that it will be ideal.

[shapirus]

Based on my observations using a thermal imaging camera, the board itself is subject to heating no less than the radiator, which means the NTC thermistor can be mounted on the reverse side, for example, under the FPGA chip, and there will be no protruding wires.  Regarding the method of controlling the rotation speed of the cooler, I meant replacing one of the resistors in the resistive divider with an NTC thermistor, the resistance of which decreases as the temperature rises.  I have already found one whose resistance at room temperature is 8 kOhm, it seems to me that it will be ideal.

That's a clever idea, let us know how it works out. Best done with a bigger fan of course, as the stock one will remain loud af regardless of how you control it (unless you are ok with the target temps of 70-80 degrees or so).

[Randy222]

800 board seems to be missing a lot of SMD parts.

Missing fet to right of fan header, the IC below it is a pfet, one to right is missing on both 800-900, all that in the power area of the board.

Next is the two missing IC's down below the fan header. Are those dc-dc bucks on the 900? Do these items feed power to the RAM chips? Two missing IC's in power section, and two missing RAM chips. Hmmmm.  Oddly though, resistors/caps/inductors (perhaps) are installed all around those missing IC's. Why leave out the IC's but leave in those other small SMD's ?

[Mechatrommer]

Why leave out the IC's but leave in those other small SMD's ?

because the ICs are $3-7/pc and those small SMD's are 0.05cent/pc? so less work to remove fewer parts in pick and place files? attached are spot the difference between 800 and 900 board, and some findings/guesses i made a little while ago.. (thanks to dho900 owners who contributed) fwiw...

[S2084]

Here guys is the result of my modification.  My assumption that reducing the resistance of the resistor in the divider will lead to a decrease in the fan speed turned out to be correct.  The resistance of the pre-soldered resistor I measured was 55.6 kOhm.  In parallel, a 1MOhm variable resistance resistor was connected to it, this gave a very large range of fan speed adjustment.  I adjusted the noise level so that it is not audible at all, even up close.  With this setting, the temperature on the board did not exceed 61°C for about one hour of operation at maximum load.  Photos are attached.

[AndyBig]

Here guys is the result of my modification.  My assumption that reducing the resistance of the resistor in the divider will lead to a decrease in the fan speed turned out to be correct.  The resistance of the pre-soldered resistor I measured was 55.6 kOhm.  In parallel, a 1MOhm variable resistance resistor was connected to it, this gave a very large range of fan speed adjustment.  I adjusted the noise level so that it is not audible at all, even up close.  With this setting, the temperature on the board did not exceed 61°C for about one hour of operation at maximum load.  Photos are attached.

I would do as you wrote earlier - I would replace another resistor with an NTC thermistor with a resistance of 10k and in series with it a resistance selected so that the fan reaches factory speed at a crystal temperature of 65°C. As the temperature rises, the thermistor resistance will drop and the fan speed will increase.

[S2084]

Here guys is the result of my modification.  My assumption that reducing the resistance of the resistor in the divider will lead to a decrease in the fan speed turned out to be correct.  The resistance of the pre-soldered resistor I measured was 55.6 kOhm.  In parallel, a 1MOhm variable resistance resistor was connected to it, this gave a very large range of fan speed adjustment.  I adjusted the noise level so that it is not audible at all, even up close.  With this setting, the temperature on the board did not exceed 61°C for about one hour of operation at maximum load.  Photos are attached.

I would do as you wrote earlier - I would replace another resistor with an NTC thermistor with a resistance of 10k and in series with it a resistance selected so that the fan reaches factory speed at a crystal temperature of 65°C. As the temperature rises, the thermistor resistance will drop and the fan speed will increase.

Yes, of course, but it requires a lot of time to set up, which unfortunately I don’t have much of.  The main thing is that we have jointly dealt with the issue of fan noise.  Thanks to all!

[shapirus]

I adjusted the noise level so that it is not audible at all, even up close.  With this setting, the temperature on the board did not exceed 61°C for about one hour of operation at maximum load.

This is pretty good. Now the idea of trying a centrifugal fan becomes even more promising -- given how inoptimal the stock fan is for the job, a centrifugal one could potentially result in stock or lower than stock temperatures at a very low, if audible at all, noise level.

[Randy222]

If the fan is on a buck dc-dc, should we not see some chopping, 100kHz, 400kHz, something?

As for internal fan control, I gave up looking for PWM control. I won't solder anything to the board, I just gonna put a pot into back cover and inline that with the fan. Then I 3D print a cover to go over the rear which will accomdate fan or fans of some kind. I want to be able to vesa mount it too. I thinking maybe 3D print a wedge cover so that (for now) the lean won't hit any bigger fan sticking out the back. Since I am making a DC filter for the usb-c IN power I will power my fan setup from that.

[shapirus]

If the fan is on a buck dc-dc, should we not see some chopping, 100kHz, 400kHz, something?

I tried to look for it, but found nothing. But the scope I used was a basic handheld one (JDS6052S), which may not have the required vertical sensitivity. Admittedly I couldn't decide whether it would be safe to use the DHO800 to probe its own internals, so I didn't do it. We need someone else to probe that 8V fan power rail with a decent scope to look for the switching noise. If there is any, it's going to be under 10-20 mV -- it seems to be filtered pretty well.

[S2084]

If the fan is on a buck dc-dc, should we not see some chopping, 100kHz, 400kHz, something?

I tried to look for it, but found nothing. But the scope I used was a basic handheld one (JDS6052S), which may not have the required vertical sensitivity. Admittedly I couldn't decide whether it would be safe to use the DHO800 to probe its own internals, so I didn't do it. We need someone else to probe that 8V fan power rail with a decent scope to look for the switching noise. If there is any, it's going to be under 10-20 mV -- it seems to be filtered pretty well.

This is what I was able to see on the fan connector.

[AceyTech]

Admittedly I couldn't decide whether it would be safe to use the DHO800 to probe its own internals, so I didn't do it.

Where is your sense of adventure?!?!?!   
In theory, since it looks like everything is referenced off the system GND, it should be perfectly safe to probe itself. 

[Aleksandr]

Here's a good adventure.Frequency increase up to 400 MHz; replacing inductors expands the zone. https://4pda.to/forum/index.php?showtopic=1080959#entry128030525 Post Но. 21-30.

[ebastler]

But be aware that 400 MHz bandwidth can only be used in single-channel mode, where the scope's sampling rate will satisfy the Nyquist criterion. When using more channels (at reduced sampling rate per channel), it will be the user's responsibility to limit the spectral content of the incoming signal. Otherwise aliasing will occur.

Also, it's not easy to properly feed a 400 MHz signal into a scope without internal 50 Ohm termination. Considering those two limitations, I am not sure it's worthwhile to modify the input filter.

[maxspb69]

The optimal bandwidth for this scope is 125 MHz maximum  in 1-2 channel mode, and 70 MHz in 3-4 channel mode. In extreme cases, 250 MHz in single channel.
There is no point in expanding it.

[Fungus]

Note: I can get aliasing on the 1kHz probe compensation signal if I set it up wrong.

The trick to aliasing is awareness.

[ebastler]

The trick to aliasing is awareness.

But the trick to opening up the scope's physical lowpass filters on the input is that awareness alone will no longer help you. 

[Mechatrommer]

The trick to aliasing is awareness.

But the trick to opening up the scope's physical lowpass filters on the input is that awareness alone will no longer help you. 

there are people doing it for a specific purpose to achieve something, and then there are people who follow the step because its cool without knowing the purpose or consequences. either way, i'm not one among any of them. God gave me strength to work side income to afford used GHz dso and some probes, the consequences? i currently dont have space for them to operate in comfortable way please note getting a properly used GHz scope will still requires $5K or above, let alone the matching active probes. cheers.

[S2084]

If you read carefully what RX3AM wrote on that forum, you can see that he wrote about the pass resistance of 50 Ohm, and even recommended not to modify all four channels, and even pointed out the approximate error in the measurements.  If you don't need it, then you don't have to do it.  In my opinion, the purpose for which he did this is clearly described there.  I consider this man extremely competent in what he did, unlike some of the “great theorists” who often speak out here.