GW Instek GPP-3323 PSU fan noise on idle

[kakci]

Hello!

I'm planning to buy a PSU and i could find lots of infos about the Rigol and Siglent current models.
Also I checked the GPP-3323 but sadly not much review or teardown about that model exists on the net.

I'm especially interested in the noise factor (cooling fan), how loud is it on idle state?
If any1 could check it (or even document it) on the Instek GPP series I would be greatful.

Thanks.

[er12]

Hi,
I bought a GPP-4323 last month. The fan was a nice surprise !

Without load the fan noise is very low, much quieter than my oscilloscope for example (GDS-2204E) !

The built quality is really good, an ADUC7060 with an ADR03 reference for each channel






The UI is very nice too.

[kakci]

Thank you for your help and the teardown pictures. So would you recommend it if you would need to run it under small load (< 500mA) overnight in your room?

[er12]

Hi,
I did the the following test : channel 1 : 15v, channel 2 in load mode 250ma connected with channel 1, during more than 8 hours, the fan stay allways at the lowest speed.
For the noise, it's the same level than a good quality PC power supply (Seasonic) with a low load.
At a 3m distance, I can't hear the fan noise.

[kakci]

Perfect thank you, Im gonna order the GPP 3323.

[embedded_tom]

electrobob just did a nice review on the GPP-4323:

https://www.electrobob.com/gpp-4323-power-supply-review/

[KedasProbe]

My GPP-3323 arrived today, (Europe Version with LAN)
Delivered 10 days after order. (Eleshop 643€ all-in)

To start: a few pictures (I will turn it on this evening after work, and warm up)

[Hydron]

Am interested in what your offset voltage/currents are. That's the most annoying thing for me, my worst channel is a little over 1mA out (though I have also noticed the aliasing that electrobob did when I was driving an oscillating load).

edit: to be clear, I like mine, just nitpicking really!

As for fan noise, I don't find it objectionable unless I'm drawing a lot of power from it and it spins the fan up, in which case I'm getting exactly what I should expect from a nice compact _linear_ supply - heat has to get out somehow!

[KedasProbe]

The offset is about 1mA but when you just set it to 32V it shows 1.8mA read back, less when you reduce the voltage (in reality there is only a few µA current)
Ch2 seems to be a little better than CH1.

CH3 voltages are: 5.15V, 3.34V, 2.51V, 1.77V  (stays the same with 10 Ohm load)

A few Tests: (all measured with 10 Ohm as load, excluding DMM6500 in series)
Set: 500mA   DMM6500: 499.4mA   CH1Meas: 500.1mA
Set: 2mA   DMM6500: 0.85mA   CH1Meas: 1.9mA
Set: 2.1mA  DMM6500: 1.07mA   CH1Meas: 1.9mA (trying to get as close to 1mA as possible)

For lower than set 1.2mA the output is nothing, Off means -0.72µA on CH1 and CH2, on CH3 -0.076µA
Off Open is -7.4mV
Off Load 10 Ohm -0.15mV

Firmware was version v1.06, March 2019, I upgraded to v1.12, March 2020

Taking a screenshot fails, it hangs until I remove the USB stick. (writing an empty .BMP file)

And yes it is silent with slow fan speed, it didn't run faster yet, I can't hear it when the fan of the DMM6500 is on.

[Hydron]

I think you're getting similar results to me - using it to source or measure small currents requires care and is not just set-and-forget. I guess this is why it isn't >1k EUR/GBP though, as the specs/features would otherwise suggest!

I just wish there was a public calibration procedure, as it seems like some tweaking would help a lot with this. Sure it's in spec, but I bet they didn't have the right tongue angle during factory cal.

The firmware does not seem to be encrypted btw, I'm just unfortunately clueless at reverse engineering software so it isn't a lot of help

[KedasProbe]

I think the first mA are a bit unstable so I'm not sure if they even calibrate that point.

But on the other hand if you want a small current source you may be better of putting it to 10V and set a 10kohm resistor in series.

During the weekend I will do some more tests, maybe a video.

Are you able to make screenshots? (not that I really need it, it's also a strange way of making screenshots)
Did you ask them for a Calibration Procedure?
(I will ask after they explained my issues, don't want to mix subjects too much)
Edit:
FYI I got this 'error' short on my screen during upgrade:

[Hydron]

I did ask eleshop for the cal routine, they were like "what do you need that for?" and pressing further didn't help. Would appreciate if others have a go at getting it as well.

As for screenshots, it _seems_ like it has crashed, but page 93 of the manual says the following:

Code: [Select]

In the target interface, press System button >
F1 (Setting) > F5 (Hardcopy) button and rotate
the scroll wheel (Encode) to locate the target
interface followed by pressing Enter key to
save screenshot. A confirm voice will sound
when successful.And sure enough, the encoder and enter key do what they say, and it comes back to life with a .BMP on the USB drive root.

[KedasProbe]

Maybe I need to wait longer, not sure, or try another smaller USB stick. (if I can find that)

I also cannot find build date or calibration date of this unit, nothing has a date except the release date of the firmware and that was 2019

edit, no I only get an 0 Bytes .BMP file.
saving a SETSTV.STA file on it of 448bytes worked also .CSV works, so I don't think the problem is the USB stick

[Hydron]

Mine GPP-4323 came with a newer firmware installed - V1.14, Oct 9 2020. Can only see V1.12 on their website however (always possible though that the newer version relates to production changes rather than anything the end-user sees).

No obvious build date either, though it must at least have been powered up for an update after that Oct-9 date. While I ordered in October, I had to wait for a new shipment to arrive by air in November as they were out of stock, so a build date not long before I got it makes sense.

[KedasProbe]

Yeah looks like I have one of the old first units since the 3 channel with LAN isn't bought much.

I can still return it no questions asked I'm still in my 7 days.

Did your screen had a screen protector? (mine didn't)

[Hydron]

Did your screen had a screen protector? (mine didn't)

Not 100% sure, but I don't think so.

[KedasProbe]

Some On and Off graphs:
Risetime (10%-90%) is about 20ms unless the CC kicks in sooner, can be 14ms.
Falltime is max 700ms without load and decreases depending on the load, 10 Ohm is 14ms
(see scope screenshots)

Power consumption is about 16W and with the screen off it's about 1.5W less (on mains, not very accurate)
Using 2.5W DC adds about 10W on mains (230V 50Hz)
Using 10W DC adds about 30W on mains (doesn't increase fan speed, at least not immediately)

Network commands in the browser works but is a bit slow.

[KedasProbe]

Some extra test Output (CH1 and CH2)
There is a load resistor present of 500 Ohm  (Measured in power off 494 ohm, CH3: 938 ohm)
To confirm I placed a 500 ohm load CH1, the results should be a falltime of half of the no load 700ms -> 350ms confirmed.

Output capacitor size?
Calculated based on measured falltime CH1: 0.700s/2.2/494ohm = 644µF   (2.2 = conversion from falltime to Tau)
Measured at 100Hz during power off:
CH1 621µF
CH2 649µF
CH3 92.7µF (different for GPP-4323)

So about 640µF parallel with 500 Ohm on CH1 and CH2

Speed of OVP and OCP:
Testing method: set it so you know OCP or OVP has to kick in when you turn the channel on.
I tried only OCP, only OVP and both at the same time they resulted in the same result.
Min 100ms Max 200ms before the output is turned off.

That is longer than I thought it would be and I assumed with OVP it wouldn't even turn the channel on since it already knows what is going to happen, but it did turn it on like it had no idea that Set voltage was higher than OVP.

Speculation:
I don't think there is any hardware involved, checks the readback and then acts, if it just missed the last update it has to wait 100ms for the next one and then it takes 100ms to act.
They do claim it is hardware OVP/OCP protection, I would like to see prove of that, 150ms is very slow for hardware reaction time, I would also expect the reaction time to be the same every time you do the same test.

Utilizing Hardware to Realize Over Voltage Protection / Over Current Protection / Over Temperature Protection

(Technically this quote is always correct though)

Also setting 0V CH1 gives about 1.8mV, with 10 Ohm 1.7mV
(Channel off -7mV, with 10 Ohm -0.15mV)

[KedasProbe]

The firmware version 1.16 (15 July 2020) replacing 1.12 (17 March 2020) is now available to download from their website.

I'm still waiting for a response from GW-instek about the error message during firmware update to 1.12, the failure to take screenshots, some reaction about the OVP/OCP in there documents is less than 50ms, not 100ms+ and calibration/build date info.

[samipetri]

The firmware version 1.16 (15 July 2020) replacing 1.12 (17 March 2020) is now available to download from their website.

I'm still waiting for a response from GW-instek about the error message during firmware update to 1.12, the failure to take screenshots, some reaction about the OVP/OCP in there documents is less than 50ms, not 100ms+ and calibration/build date info.

I got change log also when i asked GW why there is no new FW at their website.

GPP_X323
Version V1.14->1.15 Modify the problem that the secondary CPU cannot be upgraded by the main CPU
V1.15->1.16  1. Modify the continuous boot exception caused by the removal of the special treatment of CH3 and the problem that the USB disk cannot be recognized when the boot is inserted
2. Modify the issue of backlight hardware for the software compatibility. Note: After upgrading the software, the version before 1.16 cannot be upgraded

[KedasProbe]

I did the update to 1.16 but it looks like it didn't do it correctly: shuts down too early I think, the update of 1.12 took much longer and had an end message. (the version shows 1.16 now though)

Because I didn't want to miss an error on the screen this time I recorded it. (to send to GW instek)
Looks more and more that there is something bad with my unit although it seems to work. (obviously I didn't try everything)
https://youtu.be/iCOGoF51ncU

Press the F4 (Recall) button and start to upgrade
The procedure is done when the message, upgrading successful, shows up.

P.S. You can hear the fan spin up and down, to stay on topic

[KedasProbe]

Bonus Video while I was recording the firmware update:
https://youtu.be/jTXdN6LOPTs

Edit: Updated measured output of CH1 after warmup, output measured with DMM6500

Code: [Select]

import socket
from time import sleep


ip_address_DMM = "192.168.0.211"
s_DMM = socket.socket()
s_DMM.connect((ip_address_DMM, 5025))

ip_address_GPP3323 = "192.168.0.183"
s_PS = socket.socket()
s_PS.connect((ip_address_GPP3323, 1026))

sleep(1)
Channel = "1"
Voltage = 0.0
s_PS.send(("VSET" + Channel + ":" + str(Voltage) + "\r\n").encode()) # set voltage
s_PS.send((":OUTPut"+ Channel + ":STATe ON\r\n").encode()) # turn channel on
sleep(5) # discharge without load (via 500ohm internal load)

RBValues=[]
Stepsize = 10 # 'mV'

for mVoltage in range(0, 32000, Stepsize):
    s_PS.send(("VSET" + Channel + ":" + str(mVoltage/1000) + "\r\n").encode()) # set voltage
    sleep(0.5)
    s_PS.send((":MEASure" + Channel + ":VOLTage?\r\n").encode())  #VOUT1? includes 'V'
    RBVoltage = float(s_PS.recv(30).decode())
    RBValues.append([mVoltage/1000, RBVoltage])
    print(mVoltage/1000, RBVoltage)
    s_DMM.send("dmm.measure.read()\r\n".encode())
    sleep(0.4)
   
with open('GPP-3323Readings.txt', 'w') as f:
    for item in RBValues:
        f.write("%s\n" % item)

[KedasProbe]

I also measured the second channel it's clear that the two voltage measurements are almost identical, and are always a little too high.
(this will even increase with a load and voltage drop on the wires)

The voltage output is again not very smooth with 10mV steps but this time with CH2 the readback and the setting are more equally wrong.
(meaning set and readback show little difference on the display)

I'm still waiting for support from GW-Instek, they wait for the engineers to answer....It are slow default messages:

"We are checking your inquiry to our engineers.
We'll keep you updated if we receive the engineers' feedback."

The engineer did say that the firmware update errors weren't supposed to be there but I guess that was a given...
And it takes time before they realize that the USB stick is not the problem of the sheenshot failure. (I tried a few I even bought an old type)

I made the settings & readback in reference to the output to make it less confusing. So positive error means they display more than what is on the output terminals.

[KedasProbe]

Using the power supply as extra voltmeter?

The resistance is about 100kOhm in load mode with channel off.  (max. resistance is 1KOhm with channel on)
It will show -.----- below 1.0V (since the load does not work lower than that)

You do have a capacitance of about 400µF on it.
In source mode the display will also be grayed out but it will always display 0.0000V

[Hydron]

Should have my own DMM6500 in a few weeks, will run your script to get some more datapoints (with channels 3 and 4 in addition).

Really hope they release the calibration procedure; while realistically the supply easily does what I need it to (especially given the addition of a Keithley 236 SMU for accurate low current sourcing/sinking, at least once I get some triaxial cables/connectors for it), it is super annoying to see the offsets at zero. Would be good to know what they mean by "Note: After upgrading the software, the version before 1.16 cannot be upgraded" too.

Edit: Oh and by the way, while it won't display voltages below 1V you can still read them back via SCPI. I tried CC mode to discharge a NiMH to ~0.8V and that worked too.