PassMark Inline PSU Tester

[OneMoar]

way to much money you can buy a pro lab scope for this kind of money

how to field test a psu
short black to green
does the psu turn on ? : yes
do you have 12v ? :YES
goto good psu
else
bad psu replace psu
also this exists

Amazon.com: Thermaltake Dr. Power II Automated Power Supply Tester Oversized LCD for All Power Supplies - AC0015: Industrial & Scientific

Buy Thermaltake Dr. Power II Automated Power Supply Tester Oversized LCD for All Power Supplies - AC0015: Voltage Testers - Amazon.com ✓ FREE DELIVERY possible on eligible purchases

www.amazon.com

a bad psu is the easiest thing in the world to field check
its on or its not
if a machine is hard powering off = bad psu
duh

 

[Ferrum Master]

It is a pointless product. The design idea is botched in the core. I am with Earthdog and OneMoar with this. It has zero real usage with such poor load test. You cannot heat up the tested thing. It is kinda not for this forum. Too specific. It is too costly for an average user, and useles as there is no need for it really. Pro users already will have professional solutions and don't tinker with such funny product actually.

This product is not targeted to the average Joe, in any other way, slang usage in a tech article is a bad style, as it is meant to be consumed by a tech specialist, it is not a leisure gossip article.

If it is really is targeted for cheap(and it is not) ripple/current meter, it must have be designed as an adapter connected in serial with a real load, not like this as a standalone unit. You need a set of the them. Attached to to GPU/ATX24/12V CPU lines. So you can get rid of the need of some like Sunmoontec electronic load machine, like most reviewers use, using a real life load, thus saving a lot of money. On the cheap you get some R9 290 or mining gear and skyrocket the load with real life test for as long you wish. Log the ripple/voltages/current/response during real gaming/power on/torture test loops and response in graphs using log features or close the PSU in a box to heat it up and test.

Basically it is 80$ BOM cost dollar device sold for 600$ and with a recommended stamp on it. Nope... never... it doesn't work.

 

[OneMoar]

what he said ^
if you feel safe pulling 40 or 60A though this device then you probly deserve the 3d degree electrical burns you are going to get

a good scope and a ATX breakout board is infinitely more useful then whatever this is

 

[crmaris]

This tool is supposed to be used along with a system, which will provide the load. They just added a standalone mode in case you have the PSU out of the system and you want to run a quick test. I have also criticized the short standalone testing time in the review but the fact is that this is a tool for the field, where the load will be provided by the system.

As for field testing a PSU like this, just shorting green to black and check voltages, yes for the most basic stuff it could work. But it the PSU has high ripple and causes instability under load or when the load regulation is going down because the feedback is broken or when the PSU's timings cause problems in the system's startup, there you need something more advanced.

As I mentioned before, this is not for everyone, but mostly for technicians who want an all in one solution, for field usage.

 

[OneMoar]

This tool is supposed to be used along with a system, which will provide the load. They just added a standalone mode in case you have the PSU out of the system and you want to run a quick test. I have also criticized the short standalone testing time in the review but the fact is that this is a tool for the field, where the load will be provided by the system.

As for field testing a PSU like this, just shorting green to black and check voltages, yes for the most basic stuff it could work. But it the PSU has high ripple and causes instability under load or when the load regulation is going down because the feedback is broken or when the PSU's timings cause problems in the system's startup, there you need something more advanced.

As I mentioned before, this is not for everyone, but mostly for technicians who want an all in one solution, for field usage.

moot point passmark ... missed the mark
dead psu is dead no technician needs a glorified multi-meter to determine if the psu is at fault

my knock off portable oscilloscope does all of that

ripple does not cause stability full stop, if there is enough ripple on the line to cause a issue its going to show on your run of the mill meter or even on the motherboards sensors and even if it doesn't


I am never going to get that far(or care) because the second I see the machine hard power cycling or exhibiting other tell-tale signs of a power issue I am replacing the psu -stop

this gadget needs to be like 60 or 120 bucks
or re engineer it put it in a case that sits on your desk with more i/o connectors and market it to wanna-be youtubers with no real experience

and using the machine experiencing some kind of fault to load test a questionable psu

THAT IS A TERRIBLE IDEA if it blows kablewy goes everything downstream tester and machine included

 

[PassMark]

We came across this post regarding our PSU Tester and thought it might be worth responding to some of the points raised.

Cost
The cost of any product is more than the physical parts. In this case there is a large intangible firmware and software component (R&D). Not to mention marketing, packaging, storage, and other one off costs like getting plastic moulds and custom cables done.
At the moment it is super low volume. Just 200 units were manufactured for the world. If you factor in the intangibles we are selling them at a loss at the moment.
If we can get the volumes up a bit (e.g. 5000 units) then the cost will come down.

Portable oscilloscope does the same thing
They aren't even close (nor is a multi-meter)

• Typical portable oscilloscope has 2 channels. The PSU tester has 7 channels for voltage + 9 for current (16 channels). So you would need 8 separate oscilloscopes and a ton or wiring to match the tester.
• With the tester you don't need to cut wires
• Oscilloscope doesn't generally measurement currents (only voltage). At least not without cutting wires and using special probes. So an oscilloscope also doesn't measure power (watts). So it can tell you nothing about the efficiency of the unit. Or the power consumption per rail, which is critical if you some one like Dell or HP and trying to determine the minimum PSU requirement for a machine.
• Oscilloscope doesn't measure all the timings without getting involved in a very complex time consuming procedure. And you would still need to be a PSU expert to remember all the valid timing values. PSU Tester has the valid values built and measurements take under a second to perform.
• Oscilloscope doesn't check Power sequencing of voltages against the spec. For example, 12V & 5V must always be higher than 3.3V at startup, and how can you measure 3 voltages when you only have 2 channels?
• Oscilloscope doesn't measure minimum slew rates. At startup there are requirements for voltages to ramp up by certain amount per millisecond. Are you going to manually check each millisecond and compare it with the last one with an oscilloscope?
• Oscilloscope doesn't allow for testing of the PSU under load (this is assuming you use the trick of shorting the PS-ON wire to ground). Oscilloscope places zero load on the PSU. Voltage might look OK with zero load, but drop under load.
• Cheap oscilloscope don't do data logging
• Oscilloscope doesn't compute ripple measurement
• You can't take power measurements with an Oscilloscope when the machine running. On the other hand you could leave the PSU Tester connected for days to pick up intermittent problems (e.g. spikes and brown outs).

PSU faults are obvious - no diagnostics are required
Sometimes they are obvious (e.g. you see the smoke). But many times they aren't obvious. Device driver bugs, motherboard faults, bad RAM and many other issues can cause a PC to reset. So having a PC restart isn't a sure sign of a PSU failure. Sometimes faults are intermittent and sometimes there are compatibility issues where a PSU will work with one motherboard but not another. Do you RMA the PSU or the motherboard?

If there is a fault it can be useful to know what the fault is. If you know for sure the 3.3V rail is bad, then there is a better chance of a quick repair and better chance the vendor will accept the unit back as faulty.

We totally get that this isn't a product for everyone. If you only deal with one bad PSU a year, then this isn't a product for you. But if you are looking at a few bad units a month we think it makes sense. (e.g. Doing IT support in a large organisation, computer repair centre, data center, server farm, etc..). It's a specialised niche and as far as we are aware there is no other comparable product available at the moment.

 

[_UV_]

Typical portable oscilloscope has 2 channels.

sure, rigol1054z one of the most popular cheap oscilloscope

Oscilloscope doesn't measure minimum slew rates.

really?

Oscilloscope doesn't allow for testing of the PSU under load

i need to rethink what i'am doing with oscilloscope

Cheap oscilloscope don't do data logging

are you talking about something poor made handheld trash?

Oscilloscope doesn't compute ripple measurement

not directly, but if you know what you looking for that's a matter of second to tune it

At the moment it is super low volume. Just 200 units were manufactured for the world. If you factor in the intangibles we are selling them at a loss at the moment.
If we can get the volumes up a bit (e.g. 5000 units) then the cost will come down.

sell it at 100 or below and even 25000 units will be too low to fulfill initial demand for the first month)

 

[OneMoar]

sure, rigol1054z one of the most popular cheap oscilloscope

really?

i need to rethink what i'am doing with oscilloscope

are you talking about something poor made handheld trash?

not directly, but if you know what you looking for that's a matter of second to tune it


sell it at 100 or below and even 25000 units will be too low to fulfill initial demand for the first month)

ouch
I was gonna post something similar but you got this
literally everything he said is incorrect
again moot point no onsite technician need this I get called out to a computer thats turning its self off randomly I am bringing a spare psu with me and replacing it no questions
if its a expensive server psu or something I might take it home or put it on the bench and have a look at fixing it but usually it just goes in the bin for liability reasons

and if I am repairing said unit I don't give a flying fawks about the slew rate or mutli channel testing even tho my scope does that (and as uv pointed the the DS1054Z does that and more if you can do a little bit of math)

if I need load I will hook a couple resistors up in series Jesus electrical repair 101

because usually you blow a transistor or a mosfet or even a cap

I am going to replace those components turn it back on check I have voltage and then put it back in the spare/repaired pile

I am sorry nobody needs a portable psu load tester and I use the term nobody loosely no experienced technician is going to drop 600 bucks on this when they can buy a higher end Rigol for that

this thing was designed by somebody with a lot of technical understanding but very little common/business sense XOR somebody thats never been in the field

 

[jonnyGURU]

Late to the party, but I wanted to chime in.

So, I bought one of these out of morbid curiosity. I hooked it up to a Chroma 8000 to see how accurate it was. I have to admit, I was pretty amazed that this thing is pretty accurate for current and voltage measurement.

The only thing that isn't accurate.... at all.. is the ripple. My best guess is there's a good deal of noise being measured as well. But this is so easily bypassed. You can use a cheap scope as long as you can limit bandwidth to 20 MHz. Use a probe tip spring adapter (picture attached) and put it into the back of the + and - of the connector of whatever rail you're measuring and even without the differential caps. We have the differential caps built into our fixture board and when we compared the ripple through the fixture versus just using the probe tip spring adapter, we only saw an extra 3 to 5mV of noise, which was still WAY LOWER than what the Passmark tester was measuring.

So in a nutshell, I would say that if you wanted to review PSUs, you could easily use this device in conjunction with an actual oscilloscope and get some pretty good results.

 

[OneMoar]

jonny but

 

[jonnyGURU]

jonny but

You can still buy it and it's still relevant. Aris is the only one that did a proper review of it. I just thought I'd reinforce it.

 

[OneMoar]

You can still buy it and it's still relevant. Aris is the only one that did a proper review of it. I just thought I'd reinforce it.

I legit did not know I figured after the scathing reception it would have been DOA

 

[jonnyGURU]

I legit did not know I figured after the scathing reception it would have been DOA

Since it's introduction, a number of reviewers have latched onto this device for PSU reviews. It was brought to my attention when those reviewers were feeding back that the ripple reported by the device was unusually high. Naturally, since ripple is something that is something somewhat easy to capture outside of using this device is easy to do, I wanted to check it out.

 

[OneMoar]

the device being 'portable' still makes zero sense

 

[PassMark]

I have to admit, I was pretty amazed that this thing is pretty accurate for current and voltage measurement.

Thanks for the positive feedback. Very much appreciated.
Honestly, I used your PSU review site often enough in the past, that we would have sent you a free unit.

The only thing that isn't accurate.... at all.. is the ripple

We've been asked about the ripple measure a few times. As far as we known there is no standard way of measuring ripple for PSUs i.e. the standards are vague on the exact measurement method to be used.

Some of the points that aren't covered in the standard, and are left up to the technician to decide on are,

• Should any attempt be made to differentiate between infrequent (random) spikes and periodic ripple in the ripple algorithm. If random spikes are taken into account then the sampling frequency and sampling time starts to play a big role if the spikes are fast transients.
• Over what time period should the the max / min voltage values be taken? e.g. 1ms, 100ms, 1sec, 10sec, hours?. This can make a huge difference in the result.
• Should Max / Min be reported or some type of average max and average min be reported
• Should differential scopes be used or measurements taken against common ground.
• What load level should be used when measuring ripple and how constant/stable should the load be?

Our current method is as follows:
We sample the voltage on each rail at 10K samples / second. We then take 1 sec of data and search for the absolute min and max of each rail, every second. The min/max difference is reported as ripple. These 10K and 1sec values are somewhat arbitrary, as there is no spec. As far as I know most people either aren't reporting on what their test method is or don't know what what they are measuring. (JonnyGURU being an exception here, as I am pretty sure he knows what he is measuring).

Note that this is under firmware control. So if some standards body did publish a spec on the measurement procedure, we can update the ripple computation in firmware to match the spec.

The voltage measurement precision in our tester is +-4 mV in the A/DC. If you suspect numbers are off more than this, then there is a calibration process that can be followed (if you have the high end gear to compare to). Page 20 of User's Guide
https://www.passmark.com/downloads/PSUTesterUsersGuide.pdf

 

[jonnyGURU]

the device being 'portable' still makes zero sense

I certainly wouldn't put "portable" as a marketing bullet point, sure. But if you look at the tear down, there's not much to it. It doesn't NEED to be larger. The only reason to make it larger is to allow it to support a larger load in "stand alone mode", but then you're increasing the price exponentially and you might as well be buying a TechRed or SunMoon.

As far as we known there is no standard way of measuring ripple for PSUs i.e. the standards are vague on the exact measurement method to be used.

Like I said, you would normally limit your bandwidth to 20MHz to eliminate high frequency noise. My guess that's where the tester's high test results come from. Bandwidth may not be limited by youtr device and you're reporting some high frequency noise as ripple. Obviously, labs using ATE's have direct BNC connections from the ATE to the scope so you don't have to take this into consideration. But as we used a probe tip spring to confirm the discrepancy between the BNC on the load fixture and what we were seeing with the Passmark tester. Probing the connector through the back of the connectors showed us that we could still reliably measure ripple using a run of the mill oscilloscope. Like I said, we only saw about 3mV of additional ripple due to the lack of the 0.1uF and 10uF caps that Intel suggests using, which is built into the ATE.

Maybe a next generation version can provide test probe ports so it's easier for someone to use their own oscilloscope, or better yet.. a BNC connector that they can attach directly to their scope.

Honestly, I used your PSU review site often enough in the past, that we would have sent you a free unit.

You still can.

Honestly, I can never have too many ways to test power. I could put the second one in our Taiwan office.

 

[PassMark]

...you would normally limit your bandwidth to 20MHz to eliminate high frequency noise

This is done in the device with a RC filter on each rail. So this isn't the cause.

Obviously, labs using ATE's have direct BNC connections

Yes, we agree this might part of the cause for the ripple measurement difference. BNC connectors with coax cable would be a lot more immune to external EMI than the typical PSU cables used in a PC.
Using a probe with a spring tip that minimises the loop distance would also help. What I have done in our device is that the signal travels to micro controller via one of the internal layers which is between two ground planes to minimise interference from other signals and environment but there is still a large loop (about half the size of the tester's PCB) that can not be eliminated easily in this type of device. EMI inside the case of a PC, right next to a PSU with inductors is always going to be worse than a lab environment.

It is also a cost benefit trade off as well. We could design a unit that measures ripple more accurately, but not at this price point.

I'd still be much happier if there was a clear specification for ripple measurement. Will never be consistent results otherwise.

...additional ripple due to the lack of the 0.1uF and 10uF caps that Intel suggests using

These two caps are in the device. Photo attached.

You still can

EMail me your address details to david at passmark dot com.

 

[jonnyGURU]

These two caps are in the device. Photo attached.

No, I'm saying WE only saw +3mV of ripple/noise without the caps in the test fixture.

Have you reached out to Intel for ripple test protocol?

 

[PassMark]

Have you reached out to Intel for ripple test protocol

We haven't.
But surely if there was a protocol it should be in the ATX/SPX specification documents, not kept in secret.
And it partially is already included, but the text in the spec is somewhat vague, like this.

"3.2.6. Output Ripple/Noise.

Ripple and noise are defined as periodic or random signals over a frequency band of 10 Hz to 20 MHz. Measurements shall be made with an oscilloscope with 20 MHz bandwidth. Outputs should be bypassed at the connector with a 0.1 µF ceramic disk capacitor and a 10 µF electrolytic capacitor to simulate system loading".

Clearly they intended (some) noise to be included in the measurement as "random signals" should be included. But how much effort should be made to find these random signals is unclear. To be fair however they do suggest using the (unavailable?) P6046 differential probe + amplifier and power supply for the amplifier, which obviously we can't do.