R&S 4-Output Bench Power Supply, Prog (HMP4040.04) - Review

Table of contents

RoadTest: R&S 4-Output Bench Power Supply, Prog (HMP4040.04)

Author: qhirmer

Creation date:

Evaluation Type: Power Supplies

Did you receive all parts the manufacturer stated would be included in the package?: True

What other parts do you consider comparable to this product?:

What were the biggest problems encountered?:

Detailed Review:

1. Introduction

Hello everyone! Today I have the pleasure of presenting to you the results of my road test of the HMP4040, which was my main supply for the last one and a half months. I had great fun doing all the testing, so let’s jump right in.


1.1 Overview

The power supplies in the HMP family are available in two form factors. The smaller ones come with either two channels (HMP2020) with one up to 32V and 10A and one up to 32V and 5A, or with three channels (HMP2030) where all three are able to provide 32V and 5A, both with a total power of up to 188W. The two bigger ones are equipped with three (HMP4030) or four (HMP4040) equal channels, all able to provide up to 32V and 10A with a total output power of up to 384W.


Some shops offer the power supplies with a grey front, which, as a Rohde & Schwarz sales person told me at electronica 2018, is equipped with a slightly older hardware revision.


1.2 Shipping Box and Contents

After receiving a battered-up package that bulged at its sides I feared that the device might not have survived the shipping in one piece. The power supply dropped down and the Styrofoam pieces which originally held it in place were lying partly on top of it. (Please excuse the bad pictures, I only had my smartphone at hand and couldn’t wait to open the package image)


The power supply itself thankfully only received minor cosmetic damage at the rear where both sides got bent outwards and the bottom was dented.


An additional piece of cushioning material below and above the device might help in such situations as shipping damage is to be expected for such a heavy device, especially when it is sent to another continent as in my case. However having a look at the other two reviewers I was apparently still very lucky, and the damage was easy to fix during the teardown.


Included in the package was, next to the power supply itself with two 230V fuses preinstalled, the following:

  • Important Product Records, with
    • Getting Started Manual
    • Basic Safety Instructions
    • Reference China ROHS
    • Calibration Certificate
  • Power Cord Set with type B, F, G and I plugs
  • Fuses for 115V operation


1.3 Main Specifications and Features

The specifications of the tested power supply already sound very promising: four equal channels with a maximum output power of 160W each (and a total maximum of 384W), wide current and voltage ranges of 0-32V and 0-10A and the possibility to fully remote control via a variety of connections (although their availability depend on the installed connection options). The channels can also be operated in parallel or series, allowing either up to 40A or up to 128V. The output has a (stated) maximum ripple voltage of < 1.5mV(RMS) and ripple current of < 1mA(RMS).


Arbitrary waveforms can either be programmed on a computer with the official HMExplorer software suite, or directly on the device, with up to 128 voltage- and current-steps.

Protection for connected circuits is offered by built-in overvoltage protection based on the voltage measured at the terminals, or on the programmed value. A feature similar to over-current protection is given by programmable electronic fuses, and FuseLink allows, as its name suggests, to turn off other channels as well, either immediately or after a selectable delay. This enables basic protection of multirail circuits when something happens on one of its rails.


And let’s not forget one of the best features: the "real", clunky main power switch, that completely turns off the power supply instead of turning it into a power-hungry device in idle mode.


1.4 Look and Feel


When compared to alternative power supplies, which nowadays often feature colored screens and a "modern" design, the interface of the HMP series might look a bit dated, but in my opinion, this is a very welcome thing. The screen is clearly readable, bright and has a good contrast.


One thing some people might miss when using these power supplies in a lab environment are the banana sockets that are used instead of the more traditional binding posts, which means that, at least on the front, you cannot directly hook up loose wires but need either banana plugs, or an adapter.


The buttons on the right side of the screen have a clear layout and allow the operator to "feel" his way around, as I have caught doing so myself already after only a short time of using it. The lack of advanced features, like datalogging and -plotting, greatly reduce the complexity of the device and make it actually very fun and intuitive to use on a daily basis, but also from the very first minute. Very rarely the buttons record two presses when pressed lightly and at an angle it seems, but I was not able to reliably reproduce this behavior.


Turning on and off channels is done by selecting it using the buttons labeled CH1 to CH4 (which turns the button green) and subsequently pressing Output.

Setting up current and voltage values can be done individually for each channel or for multiple channels at once using the numeric keyboard, the knob, or the arrow buttons located around the knob. Also, by pressing Tracking to enable tracking mode, selected values can be changed for the same amount.


During operation, voltage and current (selected value when turned off, and actual output when turned on) are displayed on the screen, as well as the output power for each channel.


After pressing the Menu button, as you might have guessed, the menu is presented, with the following structure:


*: these entries may depend on the installed option


Some example pictures of the screen and the menu structure are shown in the following gallery. Please note that the real screen has uniform brightness and no ghosting, which only appears on these photos:

{gallery:width=400,height=400,autoplay=false} Example menu screens


Main menu screen


Fuse linking setup


Fuse delay setup


Over voltage protection setup


Arbitrary waveform editor


Let’s also have a quick look at the other sides.

Not much going on on the left side apart from the two fans, which are quite quiet and even under full load do not seem to run at full speed. Also notice the small crack in the side of the back plate that happened during shipping.


On the back side there is the IEC socket which was covered by a sticker telling the user to check the voltage settings underneath. On the left the terminals for all channels including the sense wires are located, as well as the module slot above that.



A small video of the start up procedure and sound. Please note that the sound was recorded directly next to the device, so consider this as a sort of "maximum" loudness. During operation, the fans are quite quiet (much more silent than my Siglent oscilloscope and the dynamic load), and only a slight humming sound can be heard from the device in idle.


Look Inside

After removing the two phillips screws of the connectivity module and the six torx screws (one hidden under the warranty void if removed sticker, and one under the connectivity module) the back can be taken off and the frame slides out of the hull.


We can clearly see where the majority of the weight comes from: the big toroidal transformer in the back with a fan to its right. Below some filtering is happening and there is also the heatsink with the second fan.

One thing I didn’t like very much was the close proximity of the ribbon cable that connects the front board to the connection module to the heatsink. I fear that under constant high load this part might get quite hot and the ribbon cable damaged, although I am sure that R&S did their homework and thought about that.


A closeup of the channels (there is another one to the right) and the sockets at the bottom, with some great soldering. Somebody there with a bit too much free time who wants to draw a quick schematic?


Some filtering and a relay (possibly for inrush current limiting?) on its own PCB in the back.


And finally, an overview of the bottom. You can see that each channel sits on its own PCB, so I guess that the 3 channel versions simply leave one of them out. Again some good soldering for the cables at the bottom.


2. Measurements

Ok, let’s get to a more technical point, the electrical performance of the power supply. The measurement hardware used for these tests include a Siglent SDS1104X-E oscilloscope, an NHR S320 DC Power Supply Mini Tester with a 4110 load module serving as dynamic load and an HP 3456A as well as a Gossen Metrawatt METRAHIT X-TRA for recording voltage and current, respectively. Please note that, to my knowledge, none of this equipment has been calibrated since it was bought.


2.1 Turn-On and Turn-Off Behavior

Let us first have a look at the ramping behavior. Designers of power supplies need to make tradeoffs between the ramping speed and the prevention of producing overshoots, as they tend to be mutually exclusive.



     5V ramp up, no load                                                                                                                                 12V ramp up, no load



32V ramp up, no load


Without load the power supply shows excellent behavior with an average ramp up time of about 2.14 ms.



Voltage before ramp up



Sometimes, especially for low voltages it seems, there is a small peak before the ramp up occurs. I can not tell where this comes from but it is insignificant and its amplitude is not constant.




5V ramp up, 1A                                                                                                                                                   5V ramp up, 10A


12V ramp up, 1A                                                                                                                                                 12V ramp up, 10A


32V ramp up, 1A                                                                                                                                                 32V ramp up, 5A



For maximum output current the power supply enters constant current mode for a brief period (as indicated by a red channel button) and, once the desired output voltage is reached, switches back to constant voltage mode.




12V ramp up, 250mA source current



The same behavior occurs when a small current is set up and no load is attached. That’s when the output capacitor acts as a comparatively big load that first needs to be charged to the desired voltage. This behavior is perfectly normal, and the shown performance is great.




5V ramp down, no load                                                                                                                               12V ramp down, no load


32V ramp down, no load



When turning off the channel without load the output capacitor obviously gets discharged via an internal resistor, which leads to a mean ramp down time of about 16.9ms.


Ramping into constant current mode

When we set up a smaller current limit on the source than on the load, we can observe the behavior when ramping into CC mode.




5V ramp up to CC, 2A load                                                                                                                         12V ramp up to CC, 2A load


32V ramp up to CC, 2A load



The power supply produces overshoot for these cases. It seems like the power supply needs some time to activate its CC circuitry, so it ramps up the voltage close to the selected output voltage before coming down to the CC voltage.




   5V ramp up to CC, 10A load                                                                                                                       12V ramp up to CC, 10A load


32V ramp up to CC, 10A load



When a higher current limit is set up, the overshoot decreases and the voltage drops faster.


2.2 Ripple

Next, we have a look at voltage ripple. The manual states a voltage ripple of <1.5 mV(RMS) for a 20MHz bandwidth limit, as well as a voltage ripple of <150uV(RMS) and a current ripple of <1mA(RMS) for a 100kHz bandwidth limit. The old Hameg website of the power supply showed scope screens of the voltage ripple for a full frequency range and with a 20MHz bandwidth limitation, which can still be visited at https://web.archive.org/web/20121016192547/http://www.hameg.com:80/595.0.html.


To measure directly at the output port of the power supply I used a DIY BNC to scope probe tip adapter, and a Siglent PP510 probe in the 1:1 setting. For all measurements a bandwidth limit of 20MHz and AC coupling was enabled.




     5V ripple, channel off                                                                                                                          5V ripple, channel on, no load



Interestingly there is still some ripple when the channel is turned off. While the ripple voltage increases somewhat with voltage when the channel is turned on (as seen below), it is constant for a turned-off channel.




5V ripple, 1A                                                                                                                                                 5V ripple, 10A


12V ripple, 1A                                                                                                                                               12V ripple, 10A


32V ripple, 1A                                                                                                                                                 32V ripple, 5A



The  "ringing" contents of the ripple voltage increase in amplitude and length with increasing load.




FFT of 5V ripple



A quick look with the FFT function of my scope reveals major frequency contents in the range 2MHz to 8MHz, and additional components above 10MHz. I cannot completely exclude the possibility of some of my equipment interfering in this measurement, so additional measurements will be done once I get access to our RF measurement chamber at Uni image.


2.3 Transient Response

To have a look at the transient response of the scope, I connected a pulsed load that alternates between two constant current settings, each with a dwell time of 0.5s which turned out to be sufficient for the power supply to regulate. For all images except the last one the source was set to maximum current, and to 1A for the last case.




5V, 0A <-> 1A                                                                                                                                                5V, 0A <-> 10A


        32V, 0A <-> 5A                                                                                                                                      32V, 0A <-> 2A, 1A source



The power supply is able to recover fast without producing oscillations, but a small voltage spike seems to occur when a step happens. For the last case, where the power supply was put into CC mode periodically by setting the load spikes to 2A and the source only to 1A, a big voltage spike of about 30V can be observed, as described above.


2.4 Power Consumption

For recording the power consumption of the power supply I only had a consumer grade power meter available, so take these values with a grain of salt.

The main power switch completely switches off the supply, as compared to the more modern approach to utilize a "soft" power switch, where even when turned off some power will always be drawn. When sitting in idle on the main screen and all channels switched off the power supply needs 30.1W, which is a considerable amount, so it should be turned off whenever it is not needed for an extended period. Turning on one channel without a load connected increases the power draw to about 30.5W for 5V and to about 31.7W for 32V output. Activating all channels again increases the power draw to 37W.


2.5 Power Supply Turn on and Turn off

Turning on the power supply did not produce any considerable voltage at its output terminals, owed to the relay that disconnects the socket to the power circuitry. However, turning off the supply can create a voltage spike at the output.




PSU turned off at 1V output                                                                                                                          PSU turned off at 3V output


PSU turned off at 1V output



A hard switch off using the main power switch creates a voltage spike that increases with decreasing output voltage when a channel was left turned on. This spike can easily damage attached loads, so I can only recommend to first turn off the channels, or even better disconnect the load, before completely turning off the supply.


3. Software and Remote Control

The power supply unit provided for the road test came equipped with option HO732, enabling remote control via USB and Ethernet. Alternatives for this module provide either an RS-232 and USB port (option HO730) or a single GPIB connector (option HO740).


3.1 Web Interface

When connecting to the device with a browser, some information about the instrument (such as device type, serial number and firmware version) as well as connection info of the ethernet interface are shown. All webpages are available in English and German language.


Under the menu point Screenshot, a picture of the current display content is presented and reloaded when the button below the image is pressed. After some digging in the source code of the page I noticed that the image is located at /crt_print.bmp and reloading it also refreshes its content. So for a custom remote control application that connects via ethernet, it should be sufficient to simply redownload the image regularly.


Under SCPI Device Control the power supply can be controlled remotely with SCPI commands and the last error message can be queried. Although there are two buttons, one for all and one for only the last error message, both produced the same output in the textbox in my tests.


When querying the content of the display or communicating with SCPI commands, the power supply is automatically put (back) into remote control mode, which defaults to the SCPI command SYSTem:REMote, locking out front panel control. I think that for most applications, at least in a research environment, the SYSTem:MIX mode would be a better default since it enables both remote and local control of the device and so, when an application remote controlling the device crashes for example, the operator can intervene and turn off the channels or adjust the settings locally, instead of having to first restart the application or the controlling computer.


Please note that some examples in the manual (as of 11/2018) contain errors, mostly consisting of missing spaces (both INST:OUT1 and FUSE:LINK?2 produce error code -100, "Command error"). Also choosing nonexistent options does note produce an error (for example OUTP:SEL 2 is accepted without error, where only 0, 1, ON and OFF make any sense).


3.2 HMExplorer

On my computer an error message is shown every time the application is started or Add device is pressed, reporting a missing FTDXX.dll. Reinstalling the ftdi driver does not change this, however the application works without any noticable problem so I didn’t investigate this further.


Only three components of the application are available for the HMP series: SCPI Terminal, Screenshot and EasyArb.


SCPI Terminal

This terminal is a handy little tool for sending SCPI commands to the device manually or in a semi-automatic way. After selecting a supported device under Device in the menu bar, possible SCPI commands and a small description of their syntax are listed on the right side.


Under Script, a new script can be created or an existing one can be loaded. These scripts support repeated send and read operations as well as timed waits. While this might be useful for short and quick experiments, usually a more sophisticated environment is used for automation purposes. Maybe a scripting language like Lua might have been more helpful, especially considering how easy it is to implement it using one of the numerous existing libraries.




This tool can be used to create an arbitrary waveform (or, for the HMP family, a stepped voltage/current sequence) or load one from a file, and send it to the device.


Both voltage-time and current-time diagrams can be zoomed using the mouse wheel. The green area in the diagrams correspond to the selected point from the list on the left side.

Under Curve->Import an existing waveform can be imported. This part of the tool felt quite unfinished, partly because of the missing translation on the left side and the comma as decimal separator in the boxes on the right, even for English language. Also a chosen setting under Columns can not be set to "nothing", so if the user makes a mistake he has to restart the import process.

When an error occurs (for example when the chosen file is still open in another program) a generic error message is shown without any information, sending the user to HMExplorer->?->Exception browser for further details.


After an import all points are inserted into the table on the left side, and for the HMP family steps with negative voltage or current are shown in red color for which the power supply will output 0V or 0A, respectively.



The power supply showed consistently great behavior in all electric tests. While an overshoot can occur when switching to constant current mode as well as for load spikes, this behavior is to be expected and the supply still handled these situations well.


The user experience is great and intuitive, I had much fun using the device as my main supply. Reaction to button presses happen swiftly and the LCD is clearly readable even in bad viewing angles (although the color might invert).


Despite its simplicity, it leaves nothing to be desired. I can definitely recommend the power supply.


The only downsides that need to be considered are the ripple in the 2MHz to 8MHz band which is also present when the channels are turned off, the voltage spike in the event of a sudden power loss on active channels (e.g. by turning the power supply using the main switch) and the somewhat unfinished look and feel of some parts of the HMExplorer software suite.



Thank you very much for giving me the opportunity to do this road test, as well as to Rohde & Schwarz for giving away three of these units. Also a big thank you to Mr. Scasny and element14 for organizing the road test program.