EEVblog #1293 - New NGP800 800W Power Supply

[EEVblog]

A look at the new Rohde and Schwarz NGP800 800W system power supply. Teardown and a brief play around.

Subscribe on library: https://lbry.tv/$/invite/@eevblog:7

[sibeen]

*drool*

[johnlsenchak]

Dave,   let  me  get the circuit topology   right , the unit  has a common   primary  to all  four  isolated   secondary  stages . This  would mean their  is   one  switching
transformer   with  four independent  secondary's   for  each modules?

Or is their   is a common  high  voltage   primary  side     with   four  isolated  switching  transformers  on each module ?

Maybe    you could  have clarified   this point on the white board  drawing a block diagram

That touch  display is  "sex  on the  stick" along  with those  magnetic    coils .

How  come  you didn't   test the "sense  pins"  on  each independent   output

[Phoenix]

The general topology appears to be a single PFC input stage (can't see what style) producing a common HVDC rail (probably around 400V). Each output then has a separate DC-DC isolation and regulation stage.

I can't quite make out the exact DCDC stage topology - I doubt it's a resonant type as there are no resonant capacitors and the power level is quite low. The two pairs of gate drive transformer and CT suggest that the transformers operate in parallel. They could be flyback or forward converters? I originally thought the big output inductor would be for a buck converter, but I can't see the associated freewheel diode.

[johnlsenchak]

The general topology appears to be a single PFC input stage producing a common HVDC rail (probably around 400V). Each output then has a separate DC-DC isolation and regulation stage.

That's   what I thought  their is something   like  a  450 DC  volt   highly  regulated   high voltage  primary  rail   for  all four  independent   secondary module   so you  can   series and parallel   the outputs.   

Dio-Dee's  !

[KaneTW]

Man, I love R&S' stuff. I have a HMC8043 and would upgrade to one of these in a heartbeat.

[prasimix]

Thanks Dave for a great video. To me it's a sort of "reality check" for something that I started few years ago on your forum with EEZ H24005 and continue to improve with EEZ BB3. So, it seems that someone in R&H resonate with my ideas when so many things that we added a while ago now become publicly available on their latest device. Perhaps they just followed my path, backed up with multi-million R&D budget . Still, they have a lot to do, and some stuff they are not going to catch up soon, like our ~3 sec boot time (at least it seems that they are running Linux instead of embedded Win). Perhaps they could much easily add MicroPython support ...



... or MQTT:

.

Not to mention that design will not going to be open source any time soon (that include also software on the PC side for GUI editing and SCPI remote control like FOSS EEZ Studio).
Anyway, I can say that both H24005 and BB3 are still the only power supply (and more in case of BB3!) that comes with touchscreen well under $1000. Let's see how soon Rigol et al. will catch up .

[EEVblog]

The general topology appears to be a single PFC input stage (can't see what style) producing a common HVDC rail (probably around 400V). Each output then has a separate DC-DC isolation and regulation stage.

I can't quite make out the exact DCDC stage topology - I doubt it's a resonant type as there are no resonant capacitors and the power level is quite low.

There are three large film caps behind the heatsink.

[Phoenix]

The general topology appears to be a single PFC input stage (can't see what style) producing a common HVDC rail (probably around 400V). Each output then has a separate DC-DC isolation and regulation stage.

I can't quite make out the exact DCDC stage topology - I doubt it's a resonant type as there are no resonant capacitors and the power level is quite low.

There are three large film caps behind the heatsink.

I guess 1 could be the HVDC input decoupling and the other 2 as resonant capacitors for 2 LLCs.


Now I'm still looking for 2 resonant inductors - unless the transformers leakages are sufficient - probably quite resonable.

[ali_asadzadeh]

The thing of beauty and joy forever!

[dr.diesel]

HMC8043

My most used power supply as well, love it.

[cgroen]

I bought a R&S HMP4040 after seeing Dave's review video on it. I'm trying hard to resist this one 

[H.O]

Very nice indeed, wouldn't mind being sent one of those out of the blue :-)

One UI thing that would irritate the crap out of me is how the menu always seems to return to the top when you "go back", ie leaving you to scroll down in order to continue where you left off.

[nctnico]

From a previous discussion on an HP 6024A I remember that flyback is the only topology which allows a very wide voltage range. Likely this PSU has a 2 phase flyback converter in order to reduce the ripple. The two current transformers are likely there to sense the current during each switching cycle and shutdown in case of an overcurrent situation.

[Paul Moir]

Regarding the two pretty symmetrical stages on each channel, perhaps configuring for series/parallel operation?  The MOSFET doing the switching?  I've seen some very wide voltage range PSUs with that topology.

[Phoenix]

From a previous discussion on an HP 6024A I remember that flyback is the only topology which allows a very wide voltage range. Likely this PSU has a 2 phase flyback converter in order to reduce the ripple. The two current transformers are likely there to sense the current during each switching cycle and shutdown in case of an overcurrent situation.

I don't think flyback is correct either as I can't see anything that could be a flyback clamp circuit. The power supply has operation down to 0V (as you'd expect from a bench supply). You can't easily do this with an LLC or flyback unless you use some sort of burst mode - which gets quite messy. I'm still wondering if the large FET and inductor form a buck output - but where's the diode?

Regarding the two pretty symmetrical stages on each channel, perhaps configuring for series/parallel operation?  The MOSFET doing the switching?  I've seen some very wide voltage range PSUs with that topology.

They could be doing series/parallel between the 32V and 64V models with configuration jumpers or smaller PCB changes.

[nctnico]

From a previous discussion on an HP 6024A I remember that flyback is the only topology which allows a very wide voltage range. Likely this PSU has a 2 phase flyback converter in order to reduce the ripple. The two current transformers are likely there to sense the current during each switching cycle and shutdown in case of an overcurrent situation.

I don't think flyback is correct either as I can't see anything that could be a flyback clamp circuit.

In the HP6024A they use two switching MOSFETs so the energy (from the leakage inductance) flows back into the supply rails. Look up the service manual for the HP6024A; it has a large section on how it operates.

Looking at the video and HP6012A / HP6024A schematics again I'm convinced it is a dual phase flyback converter. The 4 MOSFETs and two identical transformers (identical part numbers!) point towards that.

It is interesting that R&S choose to use that expensive current sense resistor. A 3W surface mounted current sense resistor would have done the job but the thermal coefficient is likely to throw the accuracy off.

[Phoenix]

Dual switch flyback still requires 2 diodes on the primary as the aforementioned flyback clamp circuit. I can see one 1000v smd on each circuit that appears to go to a ceramic capacitor, if there is another one might be hidden.

There are also 2 output diodes for each transformer. They could be in parallel for a flyback.

[johnlsenchak]

The above  diagram  is more  of a  "push pull "  switching  design  using MOSFETS.   I  always thought  that  the fly-back   topology  was used mainly in  CRT  television
in the horizontal  output stage, to create  high voltage   for the anode 

[EEVblog]

I guess 1 could be the HVDC input decoupling and the other 2 as resonant capacitors for 2 LLCs.


Now I'm still looking for 2 resonant inductors - unless the transformers leakages are sufficient - probably quite resonable.

I figured they reused the isolation transformer as the inductor. Neat if you can get away with it.

FYI:

[johnlsenchak]

It  looks  like  the output  of a  class  AB (push  pull) audio  amplifier, with a  crossover circuit and  T1 is the speaker  voice  coil, something   you would see  in a high price  car  audio  amplifier

You  would  also  see a  circuit like  above   as a step up  switching  power  supply    in  car amplifiers to take   the   13.8   battery DC voltage   and increase it  over  60  volts DC, rail to rail   for the  audio   output  stage I would imagine  in  the power  supply  shown  this video,    it does the complete  opposite

[Maxlor]

Really nice power supply, but it's 5000 bucks, yowza.

[Paul Moir]

Regarding the two pretty symmetrical stages on each channel, perhaps configuring for series/parallel operation?  The MOSFET doing the switching?  I've seen some very wide voltage range PSUs with that topology.

They could be doing series/parallel between the 32V and 64V models with configuration jumpers or smaller PCB changes.

Yes possibly, but the case I was thinking of they did it on the fly.  Sort of like using a relay to hook up different taps on a linear supply.  Looking at the specs, I think it fits "under the curve".

[nctnico]

I'm not buying the resonant mode story. For starters resonant mode power supplies are so efficient that at 200W you don't need a heatsink for the MOSFETs (see the demo board from Infineon). Also using exactly the same transformers for both tank circuit and output transformers is highly unlikely. AFAIK often the leakage inductance from the transformer is used for the resonant inductance so you don't need two transformers / seperate inductor (the Infineon demo board only has one transformer and no seperate inductor). Besides that the transformers are way too big for a 200W resonant supply. I've seen >300W resonant power supplies using integrated transformers (E-core ferrite glued to a board and using PCB traces as the windings) over 2 decades ago. Also the current transformers don't look fast enough to deal with cycle-by-cycle current limiting at the typical frequencies of resonant power supplies (and why would it need two?).The capacitors on the board are more likely power supply decoupling capacitors. Again: why would you need more than one for a resonant supply?

Look at this picture from the HP6024A (200W switching flyback PSU from the early 80's) and spot the similarities. You can easely identify the switching MOSFETs (top left), transformer (top middle), current sensing transformer (middle, next to the MOSFETs) and rectifier + output filter (top right).


It would be interesting to do a further analysis of what R&S actually implemented in this power supply. Also I'm not sure I like the way the NGP800 is built. To me it looks like a nightmare to repair. How are you going to replace any of the parts on the heatsinks? Access is horrible and to me it looks like once a part is unstuck from the thermal tape, a new part will be hard to glue back.

[Phoenix]

Having mosfet, diodes transformers and cts is normal for every topology. For a dual switch flyback - where are the input flyback clamp diodes? Yes it can recirculate the energy, but you need the diodes! If you can find them it's a good argument.

Also with an llc or flyback, how do you get down to zero volts output, with current control? Flyback maybe operate discontinuous, but what fun that would be from a control stand point.

I still can't shake the feeling of that extra big inductor doing more than a little filtering. And that big mosfet... Or is that just output enable?

The small relay looks to be for selecting sense location, I can't see any series/parallel self configurability.