Author Topic: I want to buy a (electrically) noisy power supply.  (Read 2088 times)

0 Members and 1 Guest are viewing this topic.

Offline forrestcTopic starter

  • Supporter
  • ****
  • Posts: 646
  • Country: us
I want to buy a (electrically) noisy power supply.
« on: March 03, 2020, 09:24:47 am »
Ok, forgive a the clickbait title, although it is at least close to accurate...

I'm finding that I need to simulate various "bad dc power" situations on a fairly regular basis here.   I.E. power dropouts, AC ripple, switchmode noise, etc.

Over the years, I've done various things such as coupling the noise into a DC line with a transformer or hooking up an AC power supply in series with a DC supply to get what I'm looking for.   It's always a pain to set up, and get right, and it often doesn't result in what I want waveform-wise.

For reference:  The gear I'm powering is typically 24V or 48VDC nominal, which means up to around 60V (although I could live with just over 48V if I had to).   I'll need at least a couple amps at the high end, but for this box more will definitely be better, since there are a few cases where 2A@60V wouldn't be enough.

My mental picture of the perfect box would be a DC supply where I could select various 'imperfections' to apply to the output.    I realize this likely either doesn't exist, or is so expensive that I might have to scrape myself off the floor once I see the price.   

I have seen some power supplies with a modulation input, but so far I haven't found any that do up to the amount of power I feel I need.   I did find one box which effectively amplifies your signal generator output to increase the voltage/current, but I haven't been able to get a price yet.  Both of these have the disadvantage (Compared to the mystical box I'm describing) that you then have to get your arb to output a signal which will correctly replicate the noise, which has been part of the challenge.  This also assumes that the frequency of the noise isn't higher than the maximum modulation bandwidth of the supply or amplifier.

Oh, and I'd rather avoid building something for this, although I'm not beyond that.  I just have enough projects to get done without having to put development time into a box like this.

So, Ideas?  Suggestions?
« Last Edit: March 03, 2020, 09:26:46 am by forrestc »
 

Offline dzseki

  • Frequent Contributor
  • **
  • Posts: 508
  • Country: hu
Re: I want to buy a (electrically) noisy power supply.
« Reply #1 on: March 03, 2020, 09:58:25 am »
I believe an older high(ish) power PA amplifier could be modified so it would be DC coupled, so you could feed your arbitrary waveform for cheap at 60V @ many Amps at least.
You haven't stated any bandwidth requirement!
Also part of power supply noise are not just noise on the rail, but emitted interference, you'd need a different source for that.
HP 1720A scope with HP 1120A probe, EMG 12563 pulse generator, EMG 1257 function generator, EMG 1172B signal generator, MEV TR-1660C bench multimeter
 

Offline forrestcTopic starter

  • Supporter
  • ****
  • Posts: 646
  • Country: us
Re: I want to buy a (electrically) noisy power supply.
« Reply #2 on: March 03, 2020, 10:45:12 am »
You haven't stated any bandwidth requirement!

I haven't really stated one because it's hard to nail down.   I've seen some crap chinese supplies which have some horrible noise in the hundreds of khz range and up, probably harmonics from the switchmode supplies.   I'm not sure if that's useful to simulate, but it might be.

 

Offline tom66

  • Super Contributor
  • ***
  • Posts: 6680
  • Country: gb
  • Electronics Hobbyist & FPGA/Embedded Systems EE
Re: I want to buy a (electrically) noisy power supply.
« Reply #3 on: March 03, 2020, 11:11:38 am »
Something along the lines of EEZ Bench Box 3?

It's a programmable power supply, open source hardware/software, so it should be possible to do what you want provided the hardware is capable of reasonably rapid slew rates.
 

Offline dzseki

  • Frequent Contributor
  • **
  • Posts: 508
  • Country: hu
Re: I want to buy a (electrically) noisy power supply.
« Reply #4 on: March 03, 2020, 11:44:43 am »
You haven't stated any bandwidth requirement!

I haven't really stated one because it's hard to nail down.   I've seen some crap chinese supplies which have some horrible noise in the hundreds of khz range and up, probably harmonics from the switchmode supplies.   I'm not sure if that's useful to simulate, but it might be.

The noise you may encounter may be high frequency or high energy, but probably not both at the same time, otherwise you'd indeed need a deep pocket.
HP 1720A scope with HP 1120A probe, EMG 12563 pulse generator, EMG 1257 function generator, EMG 1172B signal generator, MEV TR-1660C bench multimeter
 

Offline Berni

  • Super Contributor
  • ***
  • Posts: 4923
  • Country: si
Re: I want to buy a (electrically) noisy power supply.
« Reply #5 on: March 03, 2020, 12:03:46 pm »
I built one as a test rig.

Its just a MCU with a DAC and a really high power opamp running the output. It was being used to play back various engine cranking waveforms on a 12V rail. Worked pretty well.

Tho for simulating high frequency noise of switchmode supplies (>10KHz) its hard to beat a signal injection transformer. At the high frequencies the transformer is pretty small lets you have very high currents without the actual signal amplifier having to handle the DC load.

So if you want both id join together a big powerful AB amplifier for low frequency and adding a injection transformer on its output for high frequency. To keep it simple you could just give it a BNC connector on the front for feeding it from a waveform generator while having an internal crossover that splits the low and high frequency part and feeds it to the corresponding amplifier.
 

Offline forrestcTopic starter

  • Supporter
  • ****
  • Posts: 646
  • Country: us
Re: I want to buy a (electrically) noisy power supply.
« Reply #6 on: March 03, 2020, 12:25:15 pm »
How deep is your pocket? You can buy a high speed SMU to emulate a noisy power, and it is all programmable (waveform of the output voltage).
Or an AWG coupled with a high current OPAMP buffer (or just a DC coupled non-class-D audio amp).

Probably low to mid 4 figures.   If an off the shelf solution was $1K it would just get ordered.  At $5K I'd have to think long and hard about it.

If the instrument turns out to be more useful than for this application, it of course ends up being easier to justify.  A SMU kinda falls into that category, depending on the feature set, especially if I get something which can do 4 quadrant work and be useful for impulse load tests and the like.

One thought I did have was to look at a much lower voltage solution but with the current range I need and use a DC supply to 'offset' it.  I.E. use my lab supply to provide the 48V base, then have another instrument which can do at least a volt or two @ a few amps add the ripple/noise, so a high speed SMU might fit into this category, with most likely a lot more voltage than I would need. 

If I built something myself, I'd most likely go the power/high current opamp direction and find something which could do a few volts @ the amperage I would need, and plan on using it as above (on top of a bulk bench supply).  Just a quick digikey search shows lots of options.


 

Offline grifftech

  • Frequent Contributor
  • **
  • Posts: 369
  • Country: us
    • youtube channel
Re: I want to buy a (electrically) noisy power supply.
« Reply #7 on: March 03, 2020, 04:43:49 pm »
Search eBay for a PSU with the required voltage/amperage and get the cheapest one
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 26757
  • Country: nl
    • NCT Developments
Re: I want to buy a (electrically) noisy power supply.
« Reply #8 on: March 03, 2020, 05:01:53 pm »
Ok, forgive a the clickbait title, although it is at least close to accurate...

I'm finding that I need to simulate various "bad dc power" situations on a fairly regular basis here.   I.E. power dropouts, AC ripple, switchmode noise, etc.

Over the years, I've done various things such as coupling the noise into a DC line with a transformer or hooking up an AC power supply in series with a DC supply to get what I'm looking for.   It's always a pain to set up, and get right, and it often doesn't result in what I want waveform-wise.

For reference:  The gear I'm powering is typically 24V or 48VDC nominal, which means up to around 60V (although I could live with just over 48V if I had to).   I'll need at least a couple amps at the high end, but for this box more will definitely be better, since there are a few cases where 2A@60V wouldn't be enough.

So, Ideas?  Suggestions?
An old HP6012A PSU could do. This is a 1000W (minimum) PSU which can go up to 60V. It has a 0-5V input which can be used to set the output voltage. You can use a function generator to create voltage waveforms. Full swing up response time is 120ms according to the specs. The swing down time depends on the load (but the PSU also has an internal current sink).

Anyway, I recommend against using tests with a randomly noisy power supply. Always make sure a test is repeatable so you can re-test using the exact same conditions. That is the only way to make sure you fixed a problem.
« Last Edit: March 03, 2020, 06:05:13 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16549
  • Country: us
  • DavidH
Re: I want to buy a (electrically) noisy power supply.
« Reply #9 on: March 03, 2020, 11:29:35 pm »
Tho for simulating high frequency noise of switchmode supplies (>10KHz) its hard to beat a signal injection transformer. At the high frequencies the transformer is pretty small lets you have very high currents without the actual signal amplifier having to handle the DC load.

Using a signal injection transformer also avoids the conflict of providing a low impedance high bandwidth output with one stage.  Power supplies usually rely on 50 to 100 microfarads of output capacitance per amp for stability but this limits bandwidth.  Higher performance designs can reduce the output capacitance, and closely resemble audio power amplifier output stages, but using a signal injection transformer will always provide higher performance.

With a single stage, I think 10 MHz is feasible but 20 MHz would be pushing it with commonly available bipolar transistors so this is one of the places were MOSFETs would be better, but it would take a lot of design work.  The transformer injection method is much easier and can use a common power supply; it just requires a suitable lower power higher frequency amplifier and the transformer of course.
 

Offline forrestcTopic starter

  • Supporter
  • ****
  • Posts: 646
  • Country: us
Re: I want to buy a (electrically) noisy power supply.
« Reply #10 on: March 03, 2020, 11:47:38 pm »
I was going to recommend Keysight N6705C and some modules, but that would certainly blow your budget.

I actually saw that on the keysight giveaway (without the right modules) after I posted this question so I went looking at it, seems like a sweet box.    There is also some of the 'A' rev's up on ebay which I'll have to research more, as it looks like a used autoranging module and a used chassis might be in my price range.

In relation to your other suggestions, I've added them to the list of what to consider here.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21609
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: I want to buy a (electrically) noisy power supply.
« Reply #11 on: March 04, 2020, 01:13:33 am »
The "enterprise grade" approach is modulated sweep, into RF PA, into ISN, into EUT.  EUT is grounded (or not) as applicable, ISN is CM, diff, normal mode, whatever you're testing, and other EUT connections are terminated appropriately (more ISNs, CDNs, etc.).  Or replace generator chain with transients for applicable tests (usually EFT and surge).  This should just fit within your budget, but I will admit setup time and verification/calibration and actually running the test isn't the most expedient process.

A source simply being "dirty", likely won't be a high enough level to be offensive -- 100s of mV perhaps, not 3 or 10V (or as many V/m in the case of radiated tests) -- and will only be dirty at certain offending peaks, not the whole band.  If your EUT happens to be relatively immune at those peaks, and vulnerable in the gaps between, well, you haven't learned anything with such a test.

The "dirty" source may also be predominantly CM or DM, or both but in different frequency ranges.

If the EUT is such that, you can make assumptions about bandwidth and linearity -- that is, that it's not particularly likely to have peaks and valleys in its response, and that its response will be roughly proportional to the stimulus -- then a more generic source can be used.  I suppose audio amplifiers might be such an example, where the supply noise may be audible in the output even at fairly low levels, and effects like input stage RF rectification aren't very frequency-selective.

Cases where this won't be true, include for example, a lot of things with wide thresholds or common mode ranges (digital logic won't screw up until a volt or two appears across relevant inputs; diff stages won't be screwed up until CM range is violated; etc.), and a lot of things with selective response (radios most obviously, but also synchronous (clocked) digital logic that is only sensitive during a sampling period, etc.).

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline Berni

  • Super Contributor
  • ***
  • Posts: 4923
  • Country: si
Re: I want to buy a (electrically) noisy power supply.
« Reply #12 on: March 04, 2020, 06:48:30 am »
Using a signal injection transformer also avoids the conflict of providing a low impedance high bandwidth output with one stage.  Power supplies usually rely on 50 to 100 microfarads of output capacitance per amp for stability but this limits bandwidth.  Higher performance designs can reduce the output capacitance, and closely resemble audio power amplifier output stages, but using a signal injection transformer will always provide higher performance.

With a single stage, I think 10 MHz is feasible but 20 MHz would be pushing it with commonly available bipolar transistors so this is one of the places were MOSFETs would be better, but it would take a lot of design work.  The transformer injection method is much easier and can use a common power supply; it just requires a suitable lower power higher frequency amplifier and the transformer of course.

From playing around with amplifier designs id say large BJT output stages for these sort of powers are only really usable up to audio frequencies. These large power BJTs are pretty reluctant to turn off in a hurry once a large current is flowing trough it. So MOSFETs are pretty much a must, to make it worse they need to be rated for linear operation. Even then getting past 1MHz at high voltages and currents takes careful design. And even once you do get such a thing working after a lot of experimentation, then you might find that giving it any sort of weird lowish impedance load will make it unstable (Like large value capacitors found on input of the 'load'). You are basically building a RF amplifier at this point.

Tho i am surprised i haven't yet come across a piece of test equipment that implements this low frequency amp + injection transformer as some sort of "Power immunity testset"

I do have a PSU that can sort of do it up to a point, its the HP 66332A (20V 5A Dynamic Measurement DC Source). Basically it is a fast two quadrant PSU (or four quadrant if you have the option installed, just a relay that flips polarity) where there is an internal waveform buffer that can be loaded trough GPIB and then played back on the output at up to about 64KSPS. The actual bandwidth stops at about audio range in the fast mode (Also has a slow mode if the load is annoying enough to make it oscillate in fast mode). You can do things like transients, brownouts, audio frequency noise. But none of this functionality is available trough the front panel, so i mostly use it as a small electronic load or battery simulator because of it being two quadrant.
 

Offline duak

  • Super Contributor
  • ***
  • Posts: 1038
  • Country: ca
Re: I want to buy a (electrically) noisy power supply.
« Reply #13 on: March 05, 2020, 05:16:19 am »
Some ideas on how I'd attack this:

- I'd start with the frequencies or bands of interest, the amplitudes required and the impedances at those frequencies.  I'm sort of expecting that the HF EMI will be on the order of a few watts that translates into a few volts into a few ohms.  Mini-circuits (and others) make broadband amplifiers that could work here.

- the higher frequency ranges could have EMI injected by a series transformer or a series inductor.  An inductor with an air gap to prevent saturation from the DC load current is probably easier to get than a transformer.  The EMI could also be injected by a capacitor following a series inductor.  The mode of the EMI is also important; ie., common-mode or differential?

BTW, I went to Germany to work with a company that was integrating some of our stuff into their system.  While I was there, they were performing an EMI susceptibility test, so they brought over a sweeper, a big RF amplifier and an antenna.  After clearing the lab of people, which was pretty much the first storey of the building, they ran the tests on the system by blasting it with RF to see what would happen.

- an off the wall idea is to use one or more power FETs across the DC bus after passing through one or more series inductors to isolate the FETS from the source supply at high frequencies.  Signals applied to the FET(s) gate(s) will vary the drain current thus modulating the voltage applied to the unit under test.  In essence it's a broadband amplifier integrated right into the test jig.  Bandwidth can be extended by arranging the FETs in cascode, ie. common source driving a common gate.  I can envisage using different valued series inductors to also broaden the frequency range.

- the line frequency related ripple could probably be simulated by injecting a signal into the sense inputs of the source power supply.  I understand that power supplies usually have a bandwidth on the order of 500 Hz so 100 to 120 Hz plus a few harmonics should be possible.  Vicor DC-DC converters use HF ZVS designs and have +10/-50% or so trim range and tend to have small output capacitors.

 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16549
  • Country: us
  • DavidH
Re: I want to buy a (electrically) noisy power supply.
« Reply #14 on: March 06, 2020, 01:32:33 am »
Using a signal injection transformer also avoids the conflict of providing a low impedance high bandwidth output with one stage.  Power supplies usually rely on 50 to 100 microfarads of output capacitance per amp for stability but this limits bandwidth.  Higher performance designs can reduce the output capacitance, and closely resemble audio power amplifier output stages, but using a signal injection transformer will always provide higher performance.

With a single stage, I think 10 MHz is feasible but 20 MHz would be pushing it with commonly available bipolar transistors so this is one of the places were MOSFETs would be better, but it would take a lot of design work.  The transformer injection method is much easier and can use a common power supply; it just requires a suitable lower power higher frequency amplifier and the transformer of course.

From playing around with amplifier designs id say large BJT output stages for these sort of powers are only really usable up to audio frequencies. These large power BJTs are pretty reluctant to turn off in a hurry once a large current is flowing trough it. So MOSFETs are pretty much a must, to make it worse they need to be rated for linear operation. Even then getting past 1MHz at high voltages and currents takes careful design. And even once you do get such a thing working after a lot of experimentation, then you might find that giving it any sort of weird lowish impedance load will make it unstable (Like large value capacitors found on input of the 'load'). You are basically building a RF amplifier at this point.

I base what I said off of my own experiments designing audio power amplifiers many years ago where I easily achieved full power bandwidths of 1/4 the bipolar output transistor Ft and higher using current feedback topologies.  Getting the transistors to turn off was the major problem and it is not uncommon in the 100+ MHz class-AB amplifiers found in old function generators to find active drive of the output transistors to achieve this although I did not know it at the time.

As you point out, layout is critical; the circuits tend to turn into HF RF transmitters.  Construction at high power levels is more like plumbing.

The advantage of such an implementation is being able to simulate events which require large signal transients at least within the bandwidth and slew rate of the power supply output.  Usually though a custom transient generator is a better solution.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf