Power supply for home lab - do I really need a R&S?

[2N3055]

You can read some older threads

Old indeed   Most of the information in those threads is from 2017 or 2020 and refers to the older version, which has been discontinued.  The current HW is the NGE100B

Old:  https://www.rohde-schwarz.com/us/products/test-and-measurement/dc-power-supplies/rs-nge100-power-supply-series_63493-387267.html
New: https://www.rohde-schwarz.com/us/products/test-and-measurement/dc-power-supplies/rs-nge100b-power-supply-series_63493-576967.html

But the NGE100B series also has its flaws. One of them is the CC mode.

In constant current mode, the control loop is way too slow. For instance: if you set the voltage to 20V and the current to 10 mA to test a LED and enable the output with a led connected, the voltage will overshoot to 20V for almost 20 milliseconds, before settling on the Vforward of the LED. This will kill the led as the current will be far beyond the maximum limits of the LED.

A test with a 50 Ohm load and a current set to 50 mA let the voltage spike to 18V for 20 ms (I = 360 mA!), after that is decays exponentially over a period of 60 ms to 2.5V (50 mA * 50 Ohm). This is such a serious design flaw that I feel this design should have never made it to market.

This problem has never been fixed in all the firmware updates that have been released over the last couple of years.

Control loop is probably not in software but in analog hardware. So no FW can fix it...

[nctnico]

I think the control loop is in software though. I have an Agilent 66311B which has the same 'problem' (delayed current limit). If the current limit is done in an analog circuit, then I don't see how you can screw up a power supply circuit that bad. OTOH, it could be done to deal with circuits which have a rush-in current which otherwise wouldn't start up properly. Some DC-DC converter modules are prone to not starting when the current limit is set too low. But I can't see this as a good point on a general purpose lab PSU though.

[mhsprang]

I think it's a software loop. The flat top and decay time (see image) are always exactly the same length, no matter the set voltage or current.

[mawyatt]

This sort of issue with the R&S PS brings up an interesting point. This can happen even to the best of the best, which R&S is a member.

What really defines the company's character is not that they have a defect/problem/issue, every company will eventually have issues with some product(s), but how they deal with such and how they treat the little folks out there. This exposes the true character of the company hidden within the cloak of famous names/brands/history and such, of course the smaller "Class B" companies have no cloak to hide behind!!

This is one area we use to evaluate companies (or attempt to), even more so now that we are retired (well semi) and paying for all our TE out of OUR pocket, not some big company budget that it completely divorced from what goes on in the labs with folks doing the real work!!

Best

[mawyatt]

I think it's a software loop. The flat top and decay time (see image) are always exactly the same length, no matter the set voltage or current.

That's totally unacceptable behavior from any supply, it's a disaster in the waiting for just about any sensitive electronics use we can think of.

Here's an inexpensive SPD3303X with settings of 20V and 10mA with a Blue LED similar setup and PS settings. Note the DSO Vertical and Horizontal scales, 500mV/Div -2V Offset at 50ms/div.

Best,

[moerm]

... I overlooked this important specification from it

Not only that one...

[quote author=FranciscogGomez link=topic=427153.msg5494642#msg5494642 ey pdenisowski,
For the price of a NGE103b, I can get a mid-quality but decent power supply (that will do its job) for less money; which will leave my wallet with  extra cash to get more equipment.
Regards
[/quote]

How to provide advice when your requirements a bit all over the place, meager, and "floating"?

Based on what I seem to have understood as your requirements I'd pick the Siglent X (not X-E) out of those you listed. Simple reason: Not known for having significant flaws, specs seeming to meet your requirements and at about half the cost of the R&S one.
Explanatory note: I consider Siglent to be a chinese "halfway A class" brand that is, I wouldn't (yet) put them next to western A brands like R&S. BUT: I guess we've all seen quite a few cases of A brands selling mediocre or somewhat flawed products.
All in all I look at Siglent as a brand with, for China, good engineering and build quality, certainly good enough for a hobby or (not high-end) lab with a, for China, matching price that is, usually not among the cheapest chinese instruments but still significantly cheaper than comparable western brands.

Side note: I (but that may be a personal thing) would absolutely not buy a multi-channel PSU without properly and fully independent/separate channels. I didn't check that yet for the Siglent PSU but definitely advise you to check that.

Finally, yes, there are quite a few excellent western (good) brand PSUs which are within $ reach for you. Their main advantage IMO is that they are "battle-proved" and most of them are very well known here, have been torn down and analyzed, etc. and also often have schematics available. Their disadvantage is that buying second hand, e.g. from ebay, always carries some risk unless you buy from a very well reputed seller.
All in all I personally would go that route (old, second hand) but one thing I'd certainly *not* consider: to buy a R&S PSU with a known, and known since years yet still not fixed (AFAIK) significant flaw. A Siglent PSU that meets my requirements I would, however, consider.

[thm_w]

Here is rigol dp932:


edit: older R&S PSU I have also has a similar issue, but not as bad.

[mawyatt]

I think the control loop is in software though. I have an Agilent 66311B which has the same 'problem' (delayed current limit). If the current limit is done in an analog circuit, then I don't see how you can screw up a power supply circuit that bad. OTOH, it could be done to deal with circuits which have a rush-in current which otherwise wouldn't start up properly. Some DC-DC converter modules are prone to not starting when the current limit is set too low. But I can't see this as a good point on a general purpose lab PSU though.

There are "certain" applications like motor startup, SMPS startup, and so on that require a significant startup current but this high current at Turn On shouldn't be "built-in" to the basic PS as a default setup parameter. This should be a user settable parameter either by a panel switch (old school) or modern UI Input.

Remember long ago when a Power Designs (think this was the type) PS had a Turn Off glitch that cost us  a couple $50K custom chip sets. When discovered we immediately took a hammer to the PS and returned it the "Calibration Dept"

In Power Designs defense (or whoever it was), this wasn't a design defect, but a component failure, as we had many of these supplies in our lab and checked every one after this incident!!

Best

[mhsprang]

The behaviour of the CC mode is even worse than I imagined. I did the following test: set the voltage to 8V, current to 10 mA, load is 50 Ohm. Image Scope_1 shows the transient when I hit the Output button on the NGE100. The first peak is almost 7.5V, so the current is appr. 150 mA. Then it remains stable at 120 mA for over 30 ms and then it decays to 20 mA (1V over 50 Ohm).

Measured over a longer period, see image Scope_0, the current decays almost linear to 10 mA. This takes a whopping 12 seconds!!

And finally, I attached a LED to the power supply. To no surprise, the LED did not survive.

I will send all of this to R&S and see what they have to say. This behaviour is unacceptable for a reputable brand as R&S.
@Anyone from R&S on this forum: feel free to comment...

[mhsprang]

Just for "fun", I set the Ramp time of the output to 0.1 s. That didn't help either.

[pdenisowski]

I will send all of this to R&S and see what they have to say. This behaviour is unacceptable for a reputable brand as R&S.
@Anyone from R&S on this forum: feel free to comment...

Thanks everyone for the posts and information.  I've passed this on to our product management and R&D folks for power supplies and we're investigating.  If you haven't already, please also submit this to your local technical support team and R&S account manager / distributor.  I'll post updates as I have them.

[pdenisowski]

I'm also posting the relevant specifications

https://scdn.rohde-schwarz.com/ur/pws/dl_downloads/dl_common_library/dl_brochures_and_datasheets/pdf_1/service_support_30/NGE100_bro_en_5214-8748-12_v0101_96dpi.pdf

[rsjsouza]

My Keysight E36312A has settings to delay the overcurrent protection "OCP Start" and "OCP Delay Start" - perhaps there is a similar configuration in the R&S supply that might be overloading the output?

[mhsprang]

My Keysight E36312A has settings to delay the overcurrent protection "OCP Start" and "OCP Delay Start" - perhaps there is a similar configuration in the R&S supply that might be overloading the output?

There is no such setting on the NGE100 series. Only a settable ramp up, which does not solve it. See reply #38 (four upwards).

[rsjsouza]

Similar response on the E36312A with various settings

(Edit to avoid confusion in the future) In this particular case, a 47 ohm resistor was placed in the output of channel 3 (set to 20V and 10mA). The curve was generated when the "output enable" button was pressed.

I think it's a software loop. The flat top and decay time (see image) are always exactly the same length, no matter the set voltage or current.

That's totally unacceptable behavior from any supply, it's a disaster in the waiting for just about any sensitive electronics use we can think of.

Here's an inexpensive SPD3303X with settings of 20V and 10mA with a Blue LED similar setup and PS settings. Note the DSO Vertical and Horizontal scales, 500mV/Div -2V Offset at 50ms/div.

Best,

[blackdog]

Hi,

Later this evening i will show some pictures of GW Instek, Rigol and Siglent power supply's

mhsprang
Can yo ushow us the puls responce @ 35mA please.

Kind regards,
Bram

[mhsprang]

@blackdog: I can't before Monday when I'm back at the office. But I can tell you the response is the same up to about 140 mA. Above that, the first "plateau" is still there at around 130-140 mA but the curve rises afterwards.

I can make some curves tonight from the PSU's I have at home: a small and cheap Tenma (< €100) and a somewhat larger TTi QL355P (€800-ish).

[mawyatt]

Here's our GPP-4323 with same setup.

Best

[mhsprang]

showly charging the output capacitor 

[mawyatt]

Exactly, at the selected CC level.

BTW this is an effective way to measure the Power Supply Effective Output Capacitance using:

C = I/(dV/dT) in the linear range

Best

[rf-messkopf]

That's totally unacceptable behavior from any supply, it's a disaster in the waiting for just about any sensitive electronics use we can think of.

I don't disagree with that statement but this behavior is quite common for a number of supplies from reputable manufacturers.

I just tested the supplies I currently have on the bench. All set to 8 V, 10 mA, loaded with 50 ohms. The attached pictures show the output voltage when the load is connected to the supply.

HMP4040-E3646A.png: green trace: R&S HMP4040, white trace: Agilent E3646A.
6632B.png: Agilent 6632B, with rear panel "Fast/Normal Operation" switch set to "normal" (notice the time scaling).

The Agilent 6632B hangs at 8 V for about 150 ms.

[ArdWar]

Keithley 2281


Keithley 2281 with 100V/s slew limit


R&S NGP800  (Note the different vertical scale)


Keithley 2461 (I mean, I'll burn their HQ if it didn't work properly)

[mahi]

Rohde & Schwarz HMP4040 (pre-facelift).

Tested at 8V and 20V with 10 mA current limit in a 50 ohm load. Even at 32V the output looks the same as the 20V test.

[2N3055]

I don't disagree with that statement but this behavior is quite common for a number of supplies from reputable manufacturers.

Maybe it is time we start reconsidering difference between good reputation and legends and lore..
And assign brand value according to actual results and not "reputable brand" urban legends...

[rf-messkopf]

I don't disagree with that statement but this behavior is quite common for a number of supplies from reputable manufacturers.

Maybe it is time we start reconsidering difference between good reputation and legends and lore..
And assign brand value according to actual results and not "reputable brand" urban legends...

True. Attached: same test with a 20 year old homemade supply, completely analog. 

Note: My tests show the load being connected to the supply, with the output already on.