EEVblog #1173 - Rohde & Schwarz Power Supply Bonanza

[EEVblog]

Rohde & Schwarz were offended at the quality of the power supplies in the EEVblog lab, so sent in some "real" lab power supplies!
Unboxing and first look at the Rohde & Schwarz HMP4040, HMP2020, and NGE103 programmable lab power supplies.

Comparisons with other brand PSU's: There are some comparisons with other brand PSUs:
https://www.rohde-schwarz.com/brochure-datasheet/hmp/

https://www.rohde-schwarz.com/au/product/hmp4000-productstartpage_63493-47360.html

https://www.rohde-schwarz.com/au/product/nge100-productstartpage_63493-387267.html

[Mr. Scram]

Hey Rohde & Schwarz, the power supply situation in my lab is outright outrageous. Luckily for you, I don't have a problem with that being corrected at all. I'd even throw in a review!

Getting sent kit like that is certainly one of the perks of being a Youtube personality. Thought leader? Influencer? Whatever. I don't know what it is about R&S equipment, but the slightly understated and fairly German design always appeals to me in ways I can't quite explain. Nothing flashy or attention grabbing, just a measured "Want to get work done? Let's get going".

[nctnico]

IMHO the binding posts and keypad are bit of a hit & miss. I'd like binding posts AND a keypad (to enter voltage and current) but none of the power supplies shown in the video seem to have both. Using a keypad is much quicker and precise than turning a knob.

[EEVblog]

I don't know what it is about R&S equipment, but the slightly understated and fairly German design always appeals to me in ways I can't quite explain. Nothing flashy or attention grabbing, just a measured "Want to get work done? Let's get going".

Agreed. Even their new stuff which is trying to be "trendy" still manages to pull that off.

[Fungus]

Accepting freebies from big companies...? 

[Fungus]

Clever use of the foam in the packaging, saves foam and no need to glue anything!


At 23 minutes I think you mean "convection", not "radiation". If heat is radiating out of the holes in the case then you've got a big problem. 

[johnlsenchak]

But   Dave  we  didn't  get  any tear downs , that's what people want to see !!! Bloody  hell !   

[Mr. Scram]

Agreed. Even their new stuff which is trying to be "trendy" still manages to pull that off.

Their recentish oscilloscope is a device I lust after. I don't even care how good it is, even though it is. It's a perfect example of how to make what is still a tool attractive without falling in to the boy racer trap of flashy LEDs or over designed shapes. Of course, the attention to detail and fit and finish are another part of the appeal. Shahriar calling the internals "beautiful" is some of the highest praise I can imagine.

[NikkiC]

But   Dave  we  didn't  get  any tear downs , that's what people want to see !!! Bloody  hell !

Give the bloke a chance to make em mate.   

[MT]

Soooo, the Rigol is now just a very distant memory from the past!

[mariush]

I don't like the direction buttons with weird shape around the rotary knob, they just seem hard to press and use.  There is room to add the buttons to the right of the knob like you have them on keyboards.

There's an issue with the smaller psu ... the usb connector in the front is not centered properly and when you pressed on the power button, the connector also moved inside... almost like maybe the circuit board is locked with only one screw, or maybe locked in place with friction (pcb slid between some plastic channels)

I don't like the way they soldered those wires to the outputs. With a bit of routing on the circuit board, they could have used bus bars between the main circuit board and the board with the connectors.
Same for that main wire and the wire connecting the usb shielding ... they could have used a bus bar bent a specific way to keep the clearance instead of a wire on the other side.  They could even dip the whole bar in some enamel or use some heatshrink, if they wanted insulation.

Also not a fan of how they screwed the fans to the metal case, they could have used some rubber things to reduce vibrations and maybe glue some rubber spacers on the corners of the fan so that if you press down on the case, the metal won't hit the fan.

[Brumby]

But   Dave  we  didn't  get  any tear downs , that's what people want to see !!! Bloody  hell !   

Back it up there....  Teardowns HAVE been done and are available: EEVblog #1174 - Rohde & Schwarz PSU Teardowns

[Fungus]

Back it up there....  Teardowns HAVE been done and are available: EEVblog #1174 - Rohde & Schwarz PSU Teardowns

I think Dave needs to put one of his couch feet on the end of that bolt through the transformer.


(maybe that's the reason he has so many - they're really toroidal transformer condoms!)

[Romain]

Anybody knows where are the hi-res pictures of the teardown?
I'm looking for the digital isolator used for each channel.

[maginnovision]

I don't like the direction buttons with weird shape around the rotary knob, they just seem hard to press and use.  There is room to add the buttons to the right of the knob like you have them on keyboards.

There's an issue with the smaller psu ... the usb connector in the front is not centered properly and when you pressed on the power button, the connector also moved inside... almost like maybe the circuit board is locked with only one screw, or maybe locked in place with friction (pcb slid between some plastic channels)

I don't like the way they soldered those wires to the outputs. With a bit of routing on the circuit board, they could have used bus bars between the main circuit board and the board with the connectors.
Same for that main wire and the wire connecting the usb shielding ... they could have used a bus bar bent a specific way to keep the clearance instead of a wire on the other side.  They could even dip the whole bar in some enamel or use some heatshrink, if they wanted insulation.

Also not a fan of how they screwed the fans to the metal case, they could have used some rubber things to reduce vibrations and maybe glue some rubber spacers on the corners of the fan so that if you press down on the case, the metal won't hit the fan.

I don't think vibration from the fan is a problem at all, everything is very rigid. Nothing rattles(not sure about nge) They have pretty much always used wires, not bus bars, so asking them to change that is 100% futile.

[electronic_eel]

I'm a bit puzzled about their architectural choice of the big-ass, custom-wound mains transformer and a buck preregulator afterwards.

Why didn't they use a common flyback or forward regulator directly from mains as preregulator for each channel? That way they wouldn't need the big and probably expensive transformer but still had individual isolated channels with preregulation. The transformers for a flyback or forward can be much smaller and thus cheaper because of the higher frequency.

[Cnoob]

No scoping the outputs? to see how the voltage rises when switched on.
Standard test I thought.

While I'm envious it's a piece of lab equipment I like to be simple no frills and no spills.

I have a Tenma Duel psu. apart from the memories (which I don't like.) It has everything I need.

[sorenkir]

The big one would perfectly stack on my 1996 vintage Hameg HM8142 !

[IanMacdonald]

I'm a bit puzzled about their architectural choice of the big-ass, custom-wound mains transformer and a buck preregulator afterwards.

Why didn't they use a common flyback or forward regulator directly from mains as preregulator for each channel? That way they wouldn't need the big and probably expensive transformer but still had individual isolated channels with preregulation. The transformers for a flyback or forward can be much smaller and thus cheaper because of the higher frequency.

There's always a fair amount of capacitive coupling through a SMPS transformer. Not a problem if the secondary is earthed either directly or capacitively but would be in a floating output application as it might induce tens of volts of common-mode noise. A 50Hz transformer would be less problematic in this respect.

[luzotug]

Every manufacturer this days do "overpriced" software options, the NGE102/103 options are (in Europe) :

• NGE-K101 ethernet : 120 EUR
• NGE-K102 wireless lan : 180EUR
• NGE-K103 four digital I/O : 90EUR

If you take all the options the unit is more expensive than the older brother HMC8042/HMC8043 :

• HMC8042 without GPIB : 980 EUR / NGE102 with full option : 650+390 EUR
• HMC8043 without GPIB : 1100 EUR / NGE103 with full option : 790+390 EUR

Sure you get the new feature wireless and four digital I/O but the HMC8042/8043 have better spec, sense line on the back, 1 analog modulation input on the back, data logging (NGE only use USB for screenshot and firmware update) and the key panel can also be used as a keypad.

[ergya]

Young player Beginner fanboy here who likes any fancy puppies like these. I think I understand what (offical) calibration means just could not find any information on the net how often do you need to calibrate things like this? You just walk in to any calibration firms (if I am correct Dave was doing this for living, sorry if not ) and tell them "hey fella here is my R&S little puppy please make it great again" or the have to be R&S certified or something or they can just calibrate it? If you don't calibrate a fancy toy like this, how off they can go in 1/5/10 years (in average, I know ot varies from device to device)?

[electronic_eel]

There's always a fair amount of capacitive coupling through a SMPS transformer. Not a problem if the secondary is earthed either directly or capacitively but would be in a floating output application as it might induce tens of volts of common-mode noise.

Isn't that mostly an issue of the design of the SMPS transformer? I guess you can get rid of most of the common mode noise with a separate shield winding around the primary winding. That reduces efficency a bit, but that wouldn't be an issue here.

[Fungus]

No scoping the outputs? to see how the voltage rises when switched on.
Standard test I thought.

Yeah, but, "Hameg". You really think...?

[floobydust]

Can anyone confirm if these have a massive output capacitor? This is frequently used to cover for the lag in the digital (MCU) control loop. When a PSU switches from CV to CC there is a large current overshoot until the O/P cap discharges or the MCU backs off.

You can can set current limit to say 0.01A and output voltage to 25V, repeatedly short the PSU output and do spot welding! Some PSU's have huge i.e. >680uF output caps that make CC mode kind of a joke.

I used a popular manufacturer's fancy digital PSU and it was so bad I had to laugh and go back to using a 30 year old analog PSU. I've never seen specs on CV->CC response time.

[Cnoob]

Yeah, but, "Hameg". You really think...?

It would be nice to see how  smooth and clean the output rises including the switch mode one or is there something wrong.

I would of expected to see a rubber cap on the toroidal even before Dave rasied the issue, it was that obvious.

[Fungus]

Can anyone confirm if these have a massive output capacitor

You could try watching the teardown video. 

Short version: No, there's hardly any capacitance on the output at all.

[Fungus]

I would of expected to see a rubber cap on the toroidal even before Dave rasied the issue, it was that obvious.

Definitely a fail, something that R&S should be looking into after Dave's "QC inspection".

Fortunately Dave did a whole segment on it in the teardown video, I'm sure R&S will see it.

Maybe Dave will get another one in the post in the next couple of weeks, at which point the only reasonable thing to do is to send me the defective one (along with an old couch foot)

[maginnovision]

I put this on youtube but... The case there is powder coated. Unless someone deliberately removes the coating and puts a heavy object there it's a non issue. Since someone would have to remove the coating it would also not be their fault, that's a VERY durable coating.

[IanMacdonald]

You can can set current limit to say 0.01A and output voltage to 25V, repeatedly short the PSU output and do spot welding! Some PSU's have huge i.e. >680uF output caps that make CC mode kind of a joke.

That will be the main reason for the postregulator. Without it you'd need a pi filter with a couple of honkin' great caps to get the ripple low enough for a lab spec. That and the postreg can cut the current quickly in a s/c condition, which a buck converter cannot because of the stored energy in the inductor.

BTW if if output reverse polarity protection diodes are used, I'd like to see a polyswitch or fuse as well.  Otherwise an accidental reverse current into the output terminals could damage the wiring or PCB. People do sometimes use bench PSUs to charge batteries... and sometimes they get them the wrong way round.

[Wolfgang]

You can can set current limit to say 0.01A and output voltage to 25V, repeatedly short the PSU output and do spot welding! Some PSU's have huge i.e. >680uF output caps that make CC mode kind of a joke.

That will be the main reason for the postregulator. Without it you'd need a pi filter with a couple of honkin' great caps to get the ripple low enough for a lab spec. That and the postreg can cut the current quickly in a s/c condition, which a buck converter cannot because of the stored energy in the inductor.

BTW if if output reverse polarity protection diodes are used, I'd like to see a polyswitch or fuse as well.  Otherwise an accidental reverse current into the output terminals could damage the wiring or PCB. People do sometimes use bench PSUs to charge batteries... and sometimes they get them the wrong way round.

... but the price for a switching postregulator is broadband PARD at the output. In this case R&S specifies 1.5mV RMS from 20Hz to 20MHz,
a RIGOL DP832 has 350uV RMS over the same frequency range. For sensitive RF or analog stuff the R&S noise level could be a bit high.

[splin]

At 23 minutes I think you mean "convection", not "radiation". If heat is radiating out of the holes in the case then you've got a big problem. 

Nah, just means it's a rebranded Weller design.

[LapTop006]

I'm a bit puzzled about their architectural choice of the big-ass, custom-wound mains transformer and a buck preregulator afterwards.

Why didn't they use a common flyback or forward regulator directly from mains as preregulator for each channel? That way they wouldn't need the big and probably expensive transformer but still had individual isolated channels with preregulation. The transformers for a flyback or forward can be much smaller and thus cheaper because of the higher frequency.

Possibly simply to give it the feel of a linear, comfort for those of us who still go looking for linear bench supplies.

I put this on youtube but... The case there is powder coated. Unless someone deliberately removes the coating and puts a heavy object there it's a non issue. Since someone would have to remove the coating it would also not be their fault, that's a VERY durable coating.

If you do have something heavy enough to rub on it (which is probably harder than Dave's demo since the case wasn't fully together), plus some vibration, it wouldn't last all that long.

[Fungus]

I put this on youtube but... The case there is powder coated. Unless someone deliberately removes the coating and puts a heavy object there it's a non issue. Since someone would have to remove the coating it would also not be their fault, that's a VERY durable coating.

I smell a challenge! 

That bolt looks quite sharp to me, I wouldn't like to rely on the powder coating for long-term safety.

[Dundarave]

I put this on youtube but... The case there is powder coated. Unless someone deliberately removes the coating and puts a heavy object there it's a non issue. Since someone would have to remove the coating it would also not be their fault, that's a VERY durable coating.

When I viewed the video, I was hoping Dave would have had a quick peek inside the case to see if R&S had perhaps stuck a bit of insulating material to the inside of the case above the bolt end...  A square inch of adhesive insulator would do the trick.

[RoGeorge]

For a linear power supply, I was expecting to see 50/60Hz rectifiers and huge filtering capacitors near the big toroidal mains transformer.

If the ones in the video are linear power supplies, how come each channel have a ferrite donut transformer?  Wouldn't that turn them into switching power supply?  And, if we use high frequency anyway for the smaller ferrite transformers, why bother with the huge toroidal mains transformer anyway, what am I missing?

[stranger]

As said earlier I prefer binding posts on PSU, sometimes there has to be more than one wire to a terminal and stacked plugs are a poor solution. The bit that is surprising is the lack of grounding provision on the front panel. If developing some PCA that has been designed for star earthing back at the PSU it helps a lot if there is provision on a lab test psu that the appropriate terminals can be earthed at the psu to mimic the ground distribution of the final product.

At the moment I'm using Agilent PSUs and really miss not having a keypad to set demands but at least I can ground appropriate terminals at the PSU using the ground binding post.

[Mr. Scram]

For a linear power supply, I was expecting to see 50/60Hz rectifiers and huge filtering capacitors near the big toroidal mains transformer.

If the ones in the video are linear power supplies, how come each channel have a ferrite donut transformer?  Wouldn't that turn them into switching power supply?  And, if we use high frequency anyway for the smaller ferrite transformers, why bother with the huge toroidal mains transformer anyway, what am I missing?

I think they're hybrids with switching preregulation, but I stand to be corrected.

[0xdeadbeef]

The HMP device etc. are linear supplies, the HMC and NGE devices are hybrids. The teardown video shows three different devices, first an HMP2020, than an NGE100, than a HMP4040. At least I recall this order, maybe I'm mistaken.

[Wolfgang]

... what would point to some sort of switching regulator is the relatively high broadband noise level. Not sure either, there is no schematics available, I suppose.

[splin]

... what would point to some sort of switching regulator is the relatively high broadband noise level. Not sure either, there is no schematics available, I suppose.

Or the bits in the teardown video where he points out the switching pre-regulator ICs and shows the datasheets?

For those interested, the HMP2020 & HMP4040 (at least - the NGE100 may be completely different but Dave didn't show any close-ups of the control logic) have two MCP3421 18 bit ADCs (6 pin devices marked CAxx and CGxx @ 5:37), presumably for the voltage and current readback. At around $.50 a piece they're sparing no expense on these moderately priced PSUs.  They are good enough to meet the specs but they could have done much better given the high price point. I guess it leaves room in the line-up for the high precision/speed, uber expensive, PSU variants should they decide there is enough demand.

A REF5020 can also be seen which looks to be the 3ppm 'I' version. The ADCs don't have a ref i/p so I guess it's used for the set voltage and current. Anybody spotted the DACs?

[EDIT] corrected MCP342 typo

[onlyrgu]

There is an Wifi Card installed (ie. You paid for the hardware) , but you need to buy a license  to activate it.
Their BOM cost goes lower if they sell the Supply without the (off the shelf)card?
Or is easier just make a single DFM for the supply?
Or can the make up the cost if the sell wifi sw / activation code separately to 50% of their consumers.
These questions are going to make me up tonight!!

[Mr. Scram]

There is an Wifi Card installed (ie. You paid for the hardware) , but you need to buy a license  to activate it.
Their BOM cost goes lower if they sell the Supply without the (off the shelf)card?
Or is easier just make a single DFM for the supply?
Or can the make up the cost if the sell wifi sw / activation code separately to 50% of their consumers.
These questions are going to make me up tonight!!

Some would argue that everyone gets the upgraded hardware for a lower price, because a few people pay extra. By having some people pay for the add-ons, the final price for the cheap model can be lower. Its a somewhat complicated affair which isn't as black and white as it initially seems.

[ali_asadzadeh]

Dave That was great

Please make some video on how the sense lines work?

[onlyrgu]

There is an Wifi Card installed (ie. You paid for the hardware) , but you need to buy a license  to activate it.
Their BOM cost goes lower if they sell the Supply without the (off the shelf)card?
Or is easier just make a single DFM for the supply?
Or can the make up the cost if the sell wifi sw / activation code separately to 50% of their consumers.
These questions are going to make me up tonight!!

Some would argue that everyone gets the upgraded hardware for a lower price, because a few people pay extra. By having some people pay for the add-ons, the final price for the cheap model can be lower. Its a somewhat complicated affair which isn't as black and white as it initially seems.

Sending a technician or engineer to upgrade the hardware is significantly more than the BOM cost. So sending a software upgrade code via email or post is way cheaper. Only logic explanation I can think of!

[Radiosonde]

I don't know what it is about R&S equipment, but the slightly understated and fairly German design always appeals to me in ways I can't quite explain. Nothing flashy or attention grabbing, just a measured "Want to get work done? Let's get going".

Yeah..thats really fascinating, but it isnt true on all there instruments today, some things are just to futuristic.
The golden age of Rohde & Schwarz design was about 1990-2004 or so ..back then there design was absolutely top.
Furthermore most of the used equipment comes from this era, they build alot of extremely nice stuff back then:
SMIQ,CMU,EB200,EK890,SMT,SME,FSEx,ZVx.....
you see I am crazy.


Gesendet von meinem SM-J730F mit Tapatalk

[pa3weg]

Right! So after some time owning these PSUs (10 pcs), I have one with a fault now. Real life teardown and repair happening.
One of the channels has no output, and the input fuse on the regulator module has blown, the rectifier measures short in-circuit. Next step is taking the rectifier out and checking the pre-regulator

Service wise the fixed wiring is a bit annoying, whole front panel needs to come off and  the outputs desoldered to remove the module PCB...

[KoF]

Did you manage to fix it? - I also have a nearly totally destroyed HMP2030 on the table. 

The first channel controls the soft start of the transformer and the fan. But if the first channel is broken, what do you think will happen?
Right!   The relay does not switch and the resistor to be bypassed burns. At the same time the other channels die a heat death because the fan is not running.

At two channels, one out of two large 2200μF capacitors were exhausted and had almost no residual capacity. - But these are not brand parts ...
In one channel, the transistors of the pre-regulator had burned out and in another, the linear post-regulator transistor was destroyed.

So far, I've spent nearly 10 hours finding all the problems  - two of them just went for disassembly and desoldering.
Without a service manual, this is really a tough job. Does anyone have a service manual for the HMP series?

By the Way... the DAC on the Board is an DAC8552 (MSOP8 Marking D82) from TI. Dual Channel 16Bit with SPI

[tramax73]

Hi Guys, anyone has the adjustment procedure of HMP4040 ?

Regards,
Mirko

[mhsprang]

Can anyone confirm if these have a massive output capacitor

You could try watching the teardown video. 

Short version: No, there's hardly any capacitance on the output at all.

But the control loop spoils it all. Set to 20V, 0.050A and loaded with 50 Ohm, the power supply overshoots to 18V steady during 13 ms, after that the voltage drops exponentially to the expected 2.5V.