EEVblog Electronics Community Forum

Products => Test Equipment => Topic started by: extronic on July 15, 2018, 11:16:23 am

Title: [research] Do Rigol oscilloscopes really need a fan?
Post by: extronic on July 15, 2018, 11:16:23 am
Hot summer came to place where I live. It's good time to perform some experiments with Rigol scopes, which are as noisy as vacuum cleaner. I'd like to share with you results of my last research on power consumption, cooling, fans, etc. I have Rigol DS1052E and 1074Z-S - first consumes only 18-20W of power, whereas the second eats 25W of mains power. So the question is - is it possible to remove the fan? How much can I reduce the noise level? Let't try

Rigol DS1052E

At the beginning I wanted to find out which parts heat up the most. I don't have thermal imaging camera, so I used simple thermocouple meter. I attached thermocouples to several components with a thermocondictive sticky tape and measured hottest components:

(http://extronic.pl/foty/rigol/DS1052_temp_table.jpg)

So the hottest component is LM317 in power supply. Firstly I wanted to replace it with cheap Chinese switching converter but this may be not the best idea ;) I noticed that all ICs have no heat sinks. Then I ordered some heat sinks and have another try

(http://extronic.pl/foty/rigol/heat_sinks.jpg)

IC with heat sink and thermocouple

(http://extronic.pl/foty/rigol/ic_heat_sink.jpg)

And here comes the first fail during this experiment - measurement with no neat sinks I made day before, which was colder. The next day was much hotter (ambient temp in my room was 30'C!) and temperatures measured were higher than with no heat sinks! So, I don't know how much heat sinks improve cooling.

Anyway, the LM317 with no fan heats up to 90'C. Datasheet says the maximum temperature is 150'C. Any other components don't heat over 70'C. So this means, that Rigol DS1052E doesn't need fan at all! To prove it, I powered up the scope, set highest sampling rate on two channels, turned on FFT, alternate trigger, measurements, ref channel, auto cursors, etc to make the scope as busy as possible, and then I left the scope for 8 hours. What happened after? Nothing. Scope was still working.

Although this test succeeded, I was worrying that having sauna inside the scope may reduce its lifetime. Then I check the impact of fan voltage on temperature of LM317 and FPGA in the scope. I measured noise level with my phone. No idea how accurate this measurement is, but it shows noise level approximately.

(http://extronic.pl/foty/rigol/ds1052e.jpg)

Here are the results of temperature measurement. I started with no fan and after half an hour I started to increase fan voltage by 2V on every 20 minutes.

(http://extronic.pl/foty/rigol/1052_temp.jpg)

Another chart shows temperature of LM317 and FPGA when themperature stabilized and fan noise level:

(http://extronic.pl/foty/rigol/1052_noise.jpg)

You see, that fan powered with 4V is practically noiseless but makes significant difference on temperature. Many people replaced Rigol's original fan with quieter one, but I had better idea. Let's just add a serial resistor to reduce fan voltage. So I added 100R resistor on fan cable as shown in the photo:

(http://extronic.pl/foty/rigol/fan_resistor.jpg)

Final conclusion is that Rigol DS1052E doesn't need a fan, but it's better to add a 100R 1W resistor to lower it's voltage. This way I reduced noise level from 32dB to 5dB!

Remember to perform calibration after this mod, because ADC offset depends on the temperature. Power up the scope and left if for at least half an hour to stabilize temperature and then run self calibration.

To verify if scope is calibrated correctly you can run simple test - switch mode of operation from Y-T to roll mode, set vertical scale to 2mV/div and horizontal to 50s/div. Then power your scope as you should see something like this. When scope is cold there are some offsets on both channels but after it heats up, the offsets should reduce and remain stable then.

(http://extronic.pl/foty/rigol/NewFile0.bmp) (http://extronic.pl/foty/rigol/NewFile1.bmp)

Rigol DS1074Z-S

Next scope to mod is Rigol DS1074Z-S. I figured out that hottest components are Spartan 6 FPGA whis already has a huge heat sink and Freescale CPU. Power supply doesn't heat over 50'C even with no fan. This time I advice not to run the scope with no fan for long time. Freescale CPU heats up over 90'C so I glued a heat sink to it and to other large ICs.

I performed the same test with fan voltage, temperature and noise. You see that FPGA and CPU heats a lot and they could reach 100'C if I perform the test longer. That's why the fan is a must.

(http://extronic.pl/foty/rigol/1074zs_temp.jpg)

(http://extronic.pl/foty/rigol/ds1074zs.jpg)

When assembling the scope I accidentally hit the power source shield with a screwdriver and it sounds like a bell. Noticed there's a fan next to the shield. Hmmmm? Maybe I should remove the shield? Lets try.

(http://extronic.pl/foty/rigol/1074zs_noise.jpg)

Red dots show noise level with power supply shield attached, green dots with no shield. Interesting phenomenon happens when voltage is between 6 and 8 volts - an air flow causes a resonance on the shield?

Again, as in DS1052, when voltage is low the fan is noiseless. So I added 100R 1W resistor on fan wire to have about 5V on the fan. Then I performed calibration and here's the result:

(http://extronic.pl/foty/rigol/DS1Z_QuickPrint59.png)

Conclusion - DS1074Z-S heats more than DS1052E and can't run without a fan. But adding a simple 100R resistor to lover fan voltage reduces noise level from 31dB to 7dB.

!!!

This modification is easy to perform but if you are not experienced you'd better not do it. During this tests I have broken supporting leg and the scope, but cyanoacrylate adhesive works excellent :D

(http://extronic.pl/foty/rigol/leg.jpg)
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: EEVblog on July 15, 2018, 11:35:57 am
Nice work.
I think it's best to use a better fan or the dropper resistor as you say. A small amount of airflow is always vastly better than none.
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: Fungus on July 15, 2018, 11:48:08 am
As noted in the fan swapping threads: The ambient temperature specification of a DS1054Z is 50 degrees Celcius. If your room is at 25 degrees then there's a bit of headroom.

It's always good to see real data though.

I measured noise level with my phone. No idea how accurate this measurement is, but it shows noise level approximately.
...

... I reduced noise level from 32dB to 5dB!

I wouldn't put much faith in the noise measurement. Many acoustics researchers would sell their grandmother's soul to have a workroom at 5dB. The quietest place ever built isn't much quieter than that.

The difference is notable though. From 32 to 5 is huge difference on a logarithmic scale.
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: Fungus on July 15, 2018, 11:51:14 am
Nice work.
I think it's best to use a better fan or the dropper resistor as you say. A small amount of airflow is always vastly better than none.

Swapping the fan for a silent model works, too.

I still do double-takes when I switch my DS1054Z on (and often put my hand round the back to make sure air is still coming out).
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: Candid on July 15, 2018, 12:59:36 pm
I replaced the original SUNON fan in my 1054Z with a SUNON HA50151V4-10000-A99 that has about 60% of the airflow of the original one.
And I added some heatsinks to all chips that felt warm or hot with my fingerchecker. Now I barely can hear the fan of my 1054Z and I am very happy with this. I checked every single chips heatsink after several hours of runtime and no problems with temperature.

As Dave says a small airflow is way better than no airflow. I recognized this when I replaced the SMPSU in my FeelTech 6600 function generator by a self made linear type with 7805, 7812 and 7912 regulators with heatsinks. I added a small fan with low airflow and added some holes in the housing so that a controlled airflow is possible. The inside temperatures and so the heatsink temperatures went down dramatically. A heatsink with no airflow is not helping much especially for long time running or you need very big size heatsinks so that the mass factor is catching the air flow factor.
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: IDEngineer on July 15, 2018, 05:58:34 pm
As a wild thought on this topic (I have a Rigol 4000 series and HATE the fan noise)...

I wonder if some of the parts are intended to run significantly warmer than ambient. I first started wondering this with respect to the U800 (?) custom IC in the old Tektronix 2465 series scopes. That part is notorious for running hot (and failing, which might be related!) but it's not like Tek didn't know how to design good equipment. If they wanted that part to run cooler, they could have made it happen.

Fast forward to some even more integrated DSO's and their run-hot ASIC's, ADC's, and FPGA's. Could it be that the device presumes certain components will run warmer than ambient and the design is based on their behavior at these elevated temperatures? I recall that years ago IBM designed a CPU to run at some inhumanly warm temperature - they couldn't figure out how to run it cool, so they intentionally ran everything warm and the design relied on that. You had to let it come up to temp when first powering it up, sort of like a tube based amplifier!

It's extremely tempting to put good heat sinks on all the hot devices in my Rigol 4000 series and then swap in a much quieter fan, but I've held off because I don't know what presumptions were made based on those device's expected operating temperatures....
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: Fungus on July 15, 2018, 09:00:34 pm
Digital logic chips aren't designed to run hot, and Rigols are all digital once you get past the front end.

Fast forward to some even more integrated DSO's and their run-hot ASIC's, ADC's, and FPGA's. Could it be that the device presumes certain components will run warmer than ambient and the design is based on their behavior at these elevated temperatures?

All those ASICs have datasheets...  :popcorn:
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: james_s on July 15, 2018, 09:30:31 pm
I think the quieter fan and/or reducing the voltage is the way to go, some of those numbers with no fan are a bit higher than I'd be comfortable with.

As for the 50C ambient rating, if it's 50C in the room then I'm probably going to be laying on the floor in shorts with a cold beverage with using a scope being one of the last things on my mind.
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: lordvader88 on July 15, 2018, 11:29:17 pm
Personally I would keep the heatsinks and then use the fan aswell for lower temps just for piece of mind.

My Tek2430A hybrid chips sure get hot. It's noisy and I still wish it was 2x cooler. IDK what they run at but I wish it never went over 50C
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: james_s on July 16, 2018, 01:03:22 am
I wouldn't worry *too* much about the temperature, silicon doesn't have the same temperature needs as a human. Also the Rigol scopes are inexpensive mass produced tools, I doubt there are going to be a lot of people coveting them 20-30 years from now the way it is with classic Tek scopes. Those old Tek scopes were state of the art when they came out and very expensive, and still quite capable. The Rigols are good scopes but I don't think they're going to really stand out from the crowd by the time they start failing and parts are unobtainium.
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: Keysight DanielBogdanoff on July 16, 2018, 05:24:02 pm
I'd throw in a lower noise fan instead of getting rid of it all together. I know what the chip data sheets say, but we also have an oven full of oscilloscopes in my building and I know first hand what extra heat does to these things over time.
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: extronic on July 16, 2018, 06:05:40 pm
I'd throw in a lower noise fan instead of getting rid of it all together. I know what the chip data sheets say, but we also have an oven full of oscilloscopes in my building and I know first hand what extra heat does to these things over time.
Do you bake scopes like cookies during production tests? ;) I'm interested in details of this test and the results.
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: Ice-Tea on July 16, 2018, 06:11:39 pm
(http://extronic.pl/foty/rigol/ic_heat_sink.jpg)

That *really* isn't the way to measure temps of ICs...
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: ttelectronic on July 16, 2018, 07:09:09 pm
I'd throw in a lower noise fan instead of getting rid of it all together. I know what the chip data sheets say, but we also have an oven full of oscilloscopes in my building and I know first hand what extra heat does to these things over time.
I feel the same way, heat over time even if it's within tolerance will shorten the life of electronics.
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: Keysight DanielBogdanoff on July 16, 2018, 07:22:32 pm
I'd throw in a lower noise fan instead of getting rid of it all together. I know what the chip data sheets say, but we also have an oven full of oscilloscopes in my building and I know first hand what extra heat does to these things over time.
Do you bake scopes like cookies during production tests? ;) I'm interested in details of this test and the results.

Basically, but in a big walk-in thing. I don't know the specifics, but you can cycle temperatures to artificially age electronics to get an idea of common failure modes and product lifespans. Here's a picture of our chamber. I can't show the racks of equipment inside because they're prototypes.
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: Ice-Tea on July 16, 2018, 07:32:13 pm
... but you can cycle temperatures to artificially age electronics to get an idea of common failure modes and product lifespans. Here's a picture of our chamber. I can't show the racks of equipment inside because they're prototypes.

HALT. Highly Accelerated Life Test.
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: TheSteve on July 16, 2018, 08:13:21 pm
I'd throw in a lower noise fan instead of getting rid of it all together. I know what the chip data sheets say, but we also have an oven full of oscilloscopes in my building and I know first hand what extra heat does to these things over time.
Do you bake scopes like cookies during production tests? ;) I'm interested in details of this test and the results.

Basically, but in a big walk-in thing. I don't know the specifics, but you can cycle temperatures to artificially age electronics to get an idea of common failure modes and product lifespans. Here's a picture of our chamber. I can't show the racks of equipment inside because they're prototypes.

Hmmm, prototypes...  >:D
Title: Re: [research] Do Rigol oscilloscopes really need a fan?
Post by: Keysight DanielBogdanoff on July 17, 2018, 05:36:32 am
I'd throw in a lower noise fan instead of getting rid of it all together. I know what the chip data sheets say, but we also have an oven full of oscilloscopes in my building and I know first hand what extra heat does to these things over time.
Do you bake scopes like cookies during production tests? ;) I'm interested in details of this test and the results.

Basically, but in a big walk-in thing. I don't know the specifics, but you can cycle temperatures to artificially age electronics to get an idea of common failure modes and product lifespans. Here's a picture of our chamber. I can't show the racks of equipment inside because they're prototypes.

Hmmm, prototypes...  >:D

No spoilers!