GW Instek PSU Calibration

[mawyatt]

Has anyone been able to perform a self calibration on a GW Instek PSU? More specifically the GGP4323?

Also, how accurate is the current limit, and voltage at lower limits, like ~20ma and ~5 volts?

Any responses from those that actually have these GW Instek PSUs is greatly appreciated.

Best,

[ZeroCubic]

I would also like to know about this for my unit

[coromonadalix]

Did you ask at Instek ??   they where helpful in the past for an variable smps i had ?

[Momchilo]

I have a GW Instek GPP-4323. I can do some measurements in the evening. What do you want to know exactly?
My unit has an offset of -500µA for CH1 and +1mA for CH2. It is more or less always the same offset, but also depends on the output voltage. I couldn't figure out how to change this or find a calibration routine for the unit. The current readback on the unit itself has also an offset of 1mA for CH2. CH1 is quite good with 200µA offset.
Because of this constant offset, you don't have a precise current source for low currents or you take into account the offset at low currents. But each unit has a different offset, so you can be lucky or not.
If you knew how to do a calibration, you would have a pretty accurate PSU. Even at low currents. At least I think you could improve its performance a little bit. But at some point the measured current is in the noise of the ADC. 

There are already some good test results and reviews. You can find them here:
https://www.eevblog.com/forum/testgear/gw-instek-gpp-3323-psu-fan-noise-on-idle/
https://www.electrobob.com/gpp-4323-power-supply-review/

Overall, I would say that the device is more than okay. Especially since it can also be used as a load from 1.5V and has 4 useful channels. I bought this for 600€ with the LAN option, it looks like it's quite expensive at the moment like everything else.

For a precise power source below 50mA I think you would need a SMU. Especially if you want to switch between CC and CV in operation without any significant spikes.

[mawyatt]

Thanks for the information and pointers (had already read both these).

Your results are very good regarding the current offset, so as you say "lucky" since this is well below the error spec for the GPP-4323. How accurate are the low voltages & current limit (1 thru 5 volts & 5 thru 20ma) relative to the setting, and are they stable over time?

With the extra digits on the display, this begs the question from GW Instek, why not have a user cal since the basic accuracy error specs make the last digits worthless from an accuracy perspective. Basically 0.01% +3mv Line and 0.01% +3mv Load Voltage and 0.2% + 3ma Line and 0.2% +3ma Load Current Limit completely swamp the nice display capability!

One would think GW would recognize this as an opportunity to allow self calibration and then the user could have a somewhat limited but useful "Poor Man's SMU" rather than just a nice PSU

Agree a SMU is in order for our usage, however these are quite expensive. Wonder if a DIY SMU that augments a basic good linear supply is in order

Anyway, thanks for the help!

Best

[Hydron]

What Momchilo said pretty much sums the situation up for mine too - the small set/readback offsets are mostly annoying because they let you set/read to such a resolution that you want to use it for a more SMU-like job as mentioned. I actually do have a SMU (an old brown era Keithley 236) to use for the low current stuff, but sometimes I want more than one channel!

I can probably do some logging if it would help (already contributed some measurements to the linked thread).

Finally I had a poke around in the firmware file - there are some strings in there regarding calibration but I'm not any good at reverse engineering software so I wouldn't know where to start to figure out how to do it. If it helps anyone that _is_ an expert at it, the main processor is a LPC1788 I think, likely with internal and also QSPI flash, there are also the isolated processors/ADC chips on there that do the regulation/measurement - there is a intel-hex section in the FW for flashing them from the main processor (part of the FW update process) it seems.

Would be great if someone manages to get an answer from GW Instek (or can RE the firmware) - I did try with the company I bought mine off but no luck sadly.

Oh and btw the load does work below 1.5V, at least in CC mode (the SCPI setting and read-back range is more generous than the on-instrument controls).

[nctnico]

Agree a SMU is in order for our usage, however these are quite expensive. Wonder if a DIY SMU that augments a basic good linear supply is in order

For your purpose a used Agilent 66311 or 66312 series PSU is likely a better choice. These offer near SMU performance with adjustable loop compensation as well.

[mawyatt]

Those Agilents are nice supplies indeed, but would really like something that can provide multiple outputs (3 or 4) at precision levels that can be switched ON-OFF at same time. The SPD3303X-E has an ALL ON/OFF switch as does the GPP-4323.

We've already had a mishap due to not sequencing the multiple supplies properly, and added some cross coupling diodes to the power supply distribution PCB to prevent this from reoccuring. This PCB has multiple regulation, filtering, over-voltage, over-current and reverse-voltage protection from the initial design, but didn't consider the possible issue of cross coupling from the multiple supplies thru the various chip substrate diodes 

BTW evidently the GPP-4323 uses a ADR03 voltage reference, which is a good stable reference same as used on SPD3303X-E which uses the ADR7792 ADC, but don't know what DAC this uses. Anyone know what ADC and DAC the GPP-4323 uses?

Probably should just order the GPP-4323 and do some evaluation tests and see how it behaves, although I don't like to return equipment to suppliers because we didn't do our homework and something behaves differently than we expected. This places a burden on the suppliers and OEM since the returned item now must be depreciated as it can't be resold as new.

Best

[nctnico]

Just ordering the GPP-4323 and test it is likely the best path. Another option would be a Keysight E3631x series PSU but that is much more expensive. I agree with the burden on the suppliers and it is not my favorite way as well but sometimes there is no alternative than to evaluate a device yourself.

[mawyatt]

Agree, just went ahead and ordered the GPP-4323 from Saelig.

Would have liked to stay with the Siglent, since not having to learn the nuances of another piece of test equipment, but the experience we've had with one of the three SPD3303X-E we have, and the other poster's issue about the calibration procedure doesn't instill a lot of confidence. Also like the ability to set voltage and current on all 4 channels on the GPP-4323, rather than just 2 on the Siglent.

In Dave's review he points out some important issues with the SPD3303X-E that we overlooked in favor of the accurate, stable, clean output (major requirement). One issue of course is the terminal spacing, the other is why the display has so little information conveyed?

Best,

[Hydron]

BTW evidently the GPP-4323 uses a ADR03 voltage reference, which is a good stable reference same as used on SPD3303X-E which uses the ADR7792 ADC, but don't know what DAC this uses. Anyone know what ADC and DAC the GPP-4323 uses?

See the post here: https://www.eevblog.com/forum/testgear/gw-instek-gpp-3323-psu-fan-noise-on-idle/msg2406369/#msg2406369

The ADUC7060 in use has a number of channels from 2 24-bit ADCs plus a 14-bit DAC (though I guess you also have the option of PWM as a DAC too).

[mawyatt]

See the post here: https://www.eevblog.com/forum/testgear/gw-instek-gpp-3323-psu-fan-noise-on-idle/msg2406369/#msg2406369

The ADUC7060 in use has a number of channels from 2 24-bit ADCs plus a 14-bit DAC (though I guess you also have the option of PWM as a DAC too).

Yes, took a couple looks at that post, very interesting but couldn't find the DAC is use. 14 bit DAC would give ~2mv per step for 32V max output which is good resolution without dithering.

Your measured results for all 4 channels are very encouraging, is the setting voltage and measured voltage on the PSU stable over a few hours when read with your DMM6500, btw which is a superb DMM

Best,

[Hydron]

I've got my DMM6500 (and SMU, to give the PSU a little bit of a load) warming up now.
Once they've had at least half an hour I'll set the DMM logging the output of the GPP-4323 (set at say 10V, 50mA load unless you suggest otherwise) and just grab the graph/info from the DMM6500 (I'll try and separate it into warm-up and steady state, should be able to do that on the DMM once I have captured the data). Should be a pretty easy test with the 6500 without writing any scripts even.

[mawyatt]

If you have the time and don't mind could you also look into a current limit setting, anything from 5 to 20ma and see how stable this remains over an hour or so?

We've ordered a GPP-4323 from Newark (Saelig was no stock), due Friday

Best,

[Hydron]

See zipped csv and screenshot of data recorded by an in-cal DMM6500 of the PSU outputting 10V/10mA (1A limit = CV mode).
Log started with the PSU at <5min on-time (probably 2-3 min), so it shows the warm-up ramp. DMM and SMU (10mA load) were fully warmed up when logging started. DMM in auto-Z input mode, 10PLC.

Will log the current next, though I won't be capturing the warm-up curve.

Note that I think that there is a FW update available that will allegedly make the DMM6500 Y axis scale a bit more useful!

[mawyatt]

That looks very good indeed

Thanks.

Best,

[Hydron]

Yeah not complaining about the voltage accuracy!

See attached current measurements - PSU in CC into 10V (SMU as a load again), DMM @ 10PLC, 10mA CC setpoint.

Obviously it has a fair offset on this channel (CH 1, same as the previous measurement) and the noise is higher, though 10mA is a much smaller proportion of full-scale than 10V.

[mawyatt]

Also looks very good, your GPP-4323 is certainly a keeper

Thanks so much for the help and measurements, it's folks like you that are knowledgable and standout above the "noise" that make this site so valuable and worthwhile

Happy Thanksgiving from this side of the pond

Best,

[Hydron]

No problem - happy to help. It is indeed a keeper - not perfect, but good enough that I made work buy one too, and it's always the one I use by default. The electrobob review is pretty fair and useful if anyone else is looking at one, yes there are some small issues but overall very nice unit with 4 real channels in a small form factor.

As for thanksgiving, we don't get the cultural and family aspects of it here and it's just a (relatively recent) commercial thing sadly - tomorrow all the shops are going to be a madhouse. On the other hand we recently had our traditional anti-catholic november celebration (to be fair these days it's less about burning effigies and more about going to a park to see some fireworks! see https://en.wikipedia.org/wiki/Guy_Fawkes_Night for those not familiar with it.)

[djsb]

Please be aware that the latest firmware update for the GPP-4323 is V1.16 (attached). The version on the GW-INSTEK website is (or was if they have since corrected this omission) is V1.17 which may not work with your GPP-4323 depending on the hardware revision. I had a support call with GW-INSTEK in the Netherlands about why I could not update my firmware. They were VERY helpful and supportive.

[Hydron]

Did they say how to tell which firmware to use? And whether both newer and older hardware would continue to get fixes? Does anything break if you use the 117 fw on an older unit?
This kind of thing can be annoying, though needing to spin new HW revisions are certainly understandable in the current climate.

[djsb]

When I tried to install V1.17 my GPP-4323 kept showing a "file error" message in red letters and the firmware would not install. I tried the V1.16 and it updated properly. I don't know what if any issues the new firmware fixes as there is no change log included with the update. I didn't ask them about continuing support and assumed that V1.16 was the last update. Maybe you can contact them to clarify this. They are quite helpful.

[mawyatt]

Just received the GPP-4323, it has the V1.17 firmware.

Preliminary results look very good, all 4 channels are reasonably accurate in Voltage and Current settings/readings. Noise level (only checking one channel at this time) is ~60uV SD, similar to the SPD3303X (~55uv), at 10V output unloaded over a 10 minute span.

Initial first impressions:

Like:

A) 4 independent isolated channels with settable voltage and current ++++
B) Voltage and Current accurate on all channels ++++
C) Low output noise level ++++
D) Programmable Load Function on Ch1 and 2 ++++
E) Display Screen reconfigurable with useful display parameters ++++
F) Keypad data entry ++++
G) Small Form Factor considering 4 Linear channels ++++
H) USB on Front ++++
I) Quiet Multi-Level Temperature Fan Control ++++

Dislike:

A) Channel arrangement, Ch4-Ch1-Ch2-Ch3 ----
B) Terminal + and - identification (white terminals, not Red & Black) ----
C) Doesn't remember channel settings at power down, then use these during power up (maybe this is configurable)
D) Doesn't remember system settings (Beep Off for example) at power down, then use these during power up (maybe this is configurable)

Anyway, need much more time with this supply before putting into service, but first impressions are quite good indeed

Best,

[Hydron]

System menu -> "Power On" soft key will solve complaints C and D. The terminals are colour coded on mine but not very obviously - just the inner bit of the safety banana socket. Does yours have sockets or binding posts?

I would be interested as to what sort of set/readback offsets you get in the low current/voltage ranges - this is the #1 reason that everyone wants access to the cal procedure.
When set to output 1V I see <1mV set/readback error on all channels (albeit one is very close to 1mV), which is pretty fantastic given that 1 mV is the minimum increment for set!
Not quite so good with current - when set to 1mA my set offsets are all <1mA though some only just, readback is a little worse (this is without much warmup though). This is where the cal routine could really help, though you'd need to be careful not to make it worse!