EEVblog #468 - Uni-T UT513 Insulation Resistance Meter Teardown

[EEVblog]

Teardown Tuesday.
Inside the Uni-T UT513 5KV Insulation Resistance meter.
And what happens when you apply 5KV across the range switch of a cheap arse multimeter?

Relay datasheet: http://www.okita.co.jp/home/okita-up/english/products/reed/relay-pdf/lrl-10x-100pcv_eng.pdf

[c4757p]

They used a special power adapter and plug for isolation? God I hate having to keep track of those! Why couldn't they just use a little inverter to provide their own isolation to whatever adapter you use? There can't be much cost difference between the external transformer and an internal one - throw in a couple $0.03 transistors and you have yourself an inverter!

[Psi]

I'm more concerned about why it keeps crashing. 

If it keeps locking up it's not all that useful, even if the pcb is well designed with good attention to detail.

[c4757p]

I bet it has to do with that shielding over the MCU - not enough. With the rise time of those arcs across the range switch, there must be a ton of EMI in that thing.

[EEVblog]

I bet it has to do with that shielding over the MCU - not enough. With the rise time of those arcs across the range switch, there must be a ton of EMI in that thing.

Yep, it's the likely cause.
These things usually aren't used to generate sparks though 

[Rufus]

I doubt the relay switches the output. The biggest danger of insulation testers is leaving capacitance in what you are testing charged up.

The relay probably shorts the multiplier and output through the chain of power resistors above it.

The other relay probably switches the transformer secondary windings to give two output ranges.

The cut outs around the 'earth return' line are not for high voltage. Measuring the current in the 'earth return' is what the whole instrument is about and 1T ohm at 1kV is only 1nA. The cut outs and lack of solder resist will be to reduce board leakage. It would have been interesting to see what that return current fed into.

Is Villard Australian or something? I thought Cockcroft-Walton was more common and Greinacher invented it earlier anyway.

Anyhow thanks for the tear down. I nearly bought one of these for a job where I should have been testing at more than 1kV (it would have identified earlier a clearance screw up which broke down at 1.5kV - lol). I haven't needed more than 1kV since but nice to see what I would get if I do in the future.

edit: Or maybe the relay just shorts the output through the big resistors, the chain with capacitors across them is strange, maybe it is feedback for the switcher and the 200M resistor is just for measurement.

[c4757p]

These things usually aren't used to generate sparks though 

Oh, come on! That's the first thing you do when you have something that puts out 5 kV!

[nitro2k01]

With regard to the crashes, I'm thinking that maybe the lead placements actually matter. As you placed the leads, you've created a nice antenna loop around LV part of the meter. Maybe the crashes would go away if you actually placed the leads and device under test strictly behind the meter.

[NiHaoMike]

Maybe try using it to measure the resistance of a piece of ESD mat as well as common insulators?

I was expecting a TV type flyback with integrated rectifier. I would expect such a thing to be cheaper than a lower voltage transformer and voltage multiplier.

[Involute]

[c4757p]

If you use the subject line like that, people are going to miss the question.

Among other things, it prevents debris from compromising isolation distances.

[oPossum]

The USB chip is made by WinChipHead. The stylized WCH is their logo. They make the horrible CH341 USB to serial chip - possibly the worst chip of it's kind.

I wouldn't be surprised if the USB on that thing has some problems.

[sleemanj]

For more sparkly goodness, forum user Lightages posted this a few days ago:
https://www.eevblog.com/forum/testgear/how-to-let-out-the-magic-smoke-in-a-ut61e/

[Lightages]

Yeah! This insulation tester is lots of fun! The UT71E buzzes really good too, but mine survived 2500V, for how long it would do it I don't know. Would it take 5000V? I don't want to ruin it.

[nitro2k01]

And what happens when you apply 5KV across the range switch of a cheap arse multimeter?

What would happen if you pulled the same stunt on a Fluke?  I went back and reviewed Ep#64 and at around 5:30 you (Dave) were raving about the design of the input section. Would that prevent the arcing you showed the "cheap arse" multimeter suffering? Because to my, admittedly inexpert, eyes the range switch seemed pretty similar as a set of curved tracks on the PCB.

If the input protection works as it should, the MOVs should kick in and be the path of least resistance for the current. The moves are connected through a beefy 1 kOhm power resistor, which will dissipate power while the current flows through it, while clamping the voltage so that there's never potential between any two points that are so close that you would get potential (no pun intended) for arcing.

[baoshi]

The C8051F352 is made by SilLabs. Datasheet here http://www.silabs.com/Support%20Documents/TechnicalDocs/C8051F352-Short.pdf
These are sold as "Mixed Signal MCU"s as they usually come with better Analog circuit. For F352 is has built in 16bit ADC.

[rolycat]

Is Villard Australian or something? I thought Cockcroft-Walton was more common and Greinacher invented it earlier anyway.

Many references on the net identify the Swiss physicist Heinrich Greinacher as the original inventor of the voltage doubler in 1913, and of extending it into a cascade in 1920. The British John Cockroft and the Irish Ernest Walton rediscovered the cascade while working together in 1932 (and won a Nobel prize with it).

Villard apparently invented a more primitive diode clamp circuit. However, none of these references identifies who Villard is or was, or why his (or her) name is so frequently misapplied to the Greinacher cascade.  Can anyone here shed light on this mystery?

[EEVblog]

Villard apparently invented a more primitive diode clamp circuit. However, none of these references identifies who Villard is or was, or why his (or her) name is so frequently misapplied to the Greinacher cascade.  Can anyone here shed light on this mystery?

I was taught that the basic doubler without the filer is the Villard doubler, and once you add the diode/cap filter it becomes a Greinacher doubler / Cockcroft-Walton doubler. Adding more becomes the cascade/multipler. Cockcroft-Walton is probably the most popular, but I have often used the name Villard.
I have no reference for that though. Maybe buried away somewhere...

[M0BSW]

 Dave, does this mean I can send my lame arsed vc99 to you for total destruction

[HKJ]

When I got the UT513 some time ago I wondered why it hat a E-STOP, you could just push the TEST button to turn it off, it is larger and it is illuminated.
First time it arced over, I found out why. The E-STOP is the only button that always works.
It might have a nice hardware design, but I do not like that the microprocessor locks up when it arcs on 5000 volt (I have not seen it on 2500 volt).

[Phantomix]

The C8051F352 is made by SilLabs. Datasheet here http://www.silabs.com/Support%20Documents/TechnicalDocs/C8051F352-Short.pdf
These are sold as "Mixed Signal MCU"s as they usually come with better Analog circuit. For F352 is has built in 16bit ADC.

Thanks, just wanted to post that here
I'm developing software for these Silabs MCUs (that means, I "have to"), they have an crusty old 8051 core tweaked with more peripherals and a higher clock (usually up to about 25 MHz and single cycle instructions)...

Edit: The f352 used in the uni-t can run at 50 MHz

[tom66]

Nothing wrong with Golden Power et.al.
Cheap alkalines often perform equally well (or nearly as well) as branded batteries
http://www.batteryshowdown.com/

[Entropy]

Interesting teardown, they put a lot more stuff into this than in the UT511, which I own. The UT511 only goes up to 1000V and is approx. 80 EUR. Attached is a photo of the main pcb for comparison:

[tom66]

Interesting they still use the mu-metal on the processor, so perhaps they have problems with it crashing at just 1kV too.

[EEVblog]

Interesting teardown, they put a lot more stuff into this than in the UT511, which I own. The UT511 only goes up to 1000V and is approx. 80 EUR. Attached is a photo of the main pcb for comparison:

HUGE difference!, thanks.

[Spawn]

Interesting teardown, they put a lot more stuff into this than in the UT511, which I own. The UT511 only goes up to 1000V and is approx. 80 EUR. Attached is a photo of the main pcb for comparison:

HUGE difference!, thanks.

Yeap it is totally different, the 511 is a lot cheaper doesn’t have USB connection and missing a lot more, here some more pictures and a video from my preview:
https://www.eevblog.com/forum/testgear/uni-t-ut511-insulation-tester-unboxing-and-preview/

Thanks for the nice teardown Dave

[retiredcaps]

@Dave,

Someone is already stealing your video and uploading it as his own.

https://www.youtube.com/channel/UC9gt4SbTnoiXsQ3NkeObKnQ/videos

[EEVblog]

Someone is already stealing your video and uploading it as his own.
https://www.youtube.com/channel/UC9gt4SbTnoiXsQ3NkeObKnQ/videos

Thanks.
Complaint submitted.

[MBY]

I have a UT 511 and decided to follow Daves request and do a short teardown. I did it yesterday but now I'm too late since Entropys post!

Anyhow, I can confirm that my unit does look the same as Entropy's.

The UT 511 is a insulation tester with 100, 250, 500 and 1000 V test voltages. It ALSO has the problem of just dying or locking up at random, often just on the 1000 V range.

People in the thread has suggested EMI and thats probably what happens. But my idea before was peak current and undervoltage cut out. My unit got better with new alkaline batteries instead of the included crappy "heavy duty" zinc-carbon ones.

Has anybody tested with the AC adapter? Do the unit shows random lockup or shutdowns at higher voltage (and, presumably, arcing) even when powered from AC?

Another question: What does the USB interface do? Its one-way from the unit to the PC, so I presume its just a serial link with data? The UT 511 doesn't have a USB interface, but it has the same PCB (I show you in the pics), but the USB part it non populated. Is there any reason to try to populate it? I can perhaps test by mounting a visible light LED on the transmitter.

Edit: The forum went unreachable during posting. The images disappeared.

Edit: I was quite disappointed. I was hoping to touch the interesting potting compound!

Edit: To clarify. The attached images shows a closeup of the HV part and the DC-plug/USB-interface card.

[gotnoise]

I would like to understand the purpose of the shielding tape. Shouldn't an EMI shield be connected to ground? Or does a "floating" shield help as well?

[MBY]

Is the forum okay? This is my third attempt to post this message! 

I have a UT 511 and decided to follow Daves request and do a short teardown. I did it yesterday but now I'm too late since Entropys post!

Anyhow, I can confirm that my unit does look the same as Entropy's.

The UT 511 is a insulation tester with 100, 250, 500 and 1000 V test voltages. It ALSO has the problem of just dying or locking up at random, often just on the 1000 V range.

People in the thread has suggested EMI and thats probably what happens. But my idea before was peak current and undervoltage cut out. My unit got better with new alkaline batteries instead of the included crappy "heavy duty" zinc-carbon ones.

Has anybody tested with the AC adapter? Do the unit shows random lockup or shutdowns at higher voltage (and, presumably, arcing) even when powered from AC?

Another question: What does the USB interface do? Its one-way from the unit to the PC, so I presume its just a serial link with data? The UT 511 doesn't have a USB interface, but it has the same PCB (I show you in the pics), but the USB part it non populated. Is there any reason to try to populate it? I can perhaps test by mounting a visible light LED on the transmitter.

Edit: The forum went unreachable during posting. The images disappeared.

Edit: I was quite disappointed. I was hoping to touch the interesting potting compound!

Attached is a close up of the HV parts and the DC-jack/USB-interface PCB.

[jnissen]

Dave, great tear-down but please use the proper terms. You mentioned  couple times in the video and also in other video's that this or that was done to knock down the "EMC". You can't change the compatibility levels on a whim! You can however change the interference or "EMI" that comes from the product. I've made the same mistake in the past and had the long version of the discussion delivered to me by a crusty old EMC testing facility engineer. Those guys take it personally when you don't use the proper terms!

[MBY]

Now I soldered a LED on the unpopulated PCB and connected it to a photodiode in a heat-shrunk tube. Total bummer. The LED lits on key press (along with beeps) and it is steady on during a key press. So no serial data and only a level change. Clearly the SW has disabled serial output on this model.

Attached is a picture of the LED/photodiode "tube" and a scope screen shot showing a key press.

[nitro2k01]

Hmm, have you tried flipping the polarity of the LED? It's a bit of a stretch, but maybe the unit keeps the diode reverse biased when idle, and then when it's transmitting it goes between trying to sink current through the LED and high impedance.  This would explain why it's lit when idle and off while transmitting, if you have the LED in reverse. just a thought. (Oh, it's on when you press the button. Nevermind.)

[lapm]

These things usually aren't used to generate sparks though 

Oh, come on! That's the first thing you do when you have something that puts out 5 kV!

 , first thing i would do with high voltage dc source. Sparks...

BTW we need some sort of sparcky as smiley in here

[M0BSW]

 VC99, As some of you know Ive been measuring the voltage on my VC99, which has been dropping a volt per month even with fresh batteries, well it died tonight, and I thought I'd take a look, instead of the small ceramic fuse and a large ceramic fuse, they fit two stupid thin lame arsed glass fuses in them, glad I never went near the mains with it, anyway it's dispatched in the bin now ready for land fill tomorrow , see even a VC99 has it uses, I suppose in won't rot in the land fill will it, the bugger will still be there when I'm dead and gone  Grrrrrrrr

[karan279]

The usb IC on there appear to come from this chinese manufacturer http://wch-ic.com/ , though this particular part no isn't listed , but one of their usb to serial IC , CH340 is common and about half a $  

[jesuscf]

The usb IC on there appear to come from this chinese manufacturer http://wch-ic.com/ , though this particular part no isn't listed , but one of their usb to serial IC , CH340 is common and about half a $

Beware of the CH340 and friends!  It is poorly (as non-existent) documented and there is only a complete driver available for Windows.  The driver for Linux is semi-functional and zero drivers for Macs.

[jesuscf]

I'm developing software for these Silabs MCUs (that means, I "have to"), they have an crusty old 8051 core tweaked with more peripherals and a higher clock (usually up to about 25 MHz and single cycle instructions)...

Edit: The f352 used in the unit can run at 50 MHz

What MCU would you use instead?  I ask because I used a C8051F352 a few years back and it worked very well.

[karan279]

Beware of the CH340 and friends!  It is poorly (as non-existent) documented and there is only a complete driver available for Windows.  The driver for Linux is semi-functional and zero drivers for Macs.

Thanks for that info , most of these cheap ones don't have anything other than a windows driver and that too not reliable at all

[amyk]

The UT511 HV section looks similar to that of a CCFL inverter, not surprising given the max of 1kV.

[Neilm]

Dave - If you really want some sparky fun get a 10kV tester and put it across an old CD. You get LOTS of sparks as the oxide burns off the plastic  Probably won't work too well with only 5kV.

Its also a good way too see how immune your tester actually is. These testers are often used to burn out small manufacturing defects in equipment when it is under test so a tester that dies during this sort of test will cause all sorts of problems.

For those interested, the 10kV instrument I was testing when I did the above with the CD quite happily carried on working and burnt out the oxide.

Neil

[KD0CAC John]

I've got couple of older Meger units , one is just AC - 4,000 volts and the other is AC / DC 6,000 volts .
Got them both needing miner repairs .
I got them because I was getting more & more used vacuum variable caps that I wanted to know if they were good .

[T4P]

No wonder there's such a huge disparity in price between UT511 and 513!
The 511 IMO is not even worth the money it is going for but the 512 and 513 is 
I did consider them before i left for the PC world because of it's MASSIVE screen and how it looks so i guess, until i earn some money and get my computer upgraded ... sadly i'm not getting my hands on one of these, i love it! But ... No application or use for  Maybe blowing up old hardware  So yes my U1273AX is sitting there with nothing to do 

[T4P]

Dave - If you really want some sparky fun get a 10kV tester and put it across an old CD. You get LOTS of sparks as the oxide burns off the plastic  Probably won't work too well with only 5kV.

Its also a good way too see how immune your tester actually is. These testers are often used to burn out small manufacturing defects in equipment when it is under test so a tester that dies during this sort of test will cause all sorts of problems.

For those interested, the 10kV instrument I was testing when I did the above with the CD quite happily carried on working and burnt out the oxide.

Neil

What sort of defects?    I think if you position the meter further away it might not crash. Could be because of low current supply from those weak batteries 

[Fezder]

nice teardown, bigger one relay

[SeanB]

I normally erase CD's and DVD's in the microwave. 20 seconds and the pile is unusable. If I had access to that method of Photons though............

[larry42]

When I got the UT513 some time ago I wondered why it hat a E-STOP, you could just push the TEST button to turn it off, it is larger and it is illuminated.
First time it arced over, I found out why. The E-STOP is the only button that always works.
It might have a nice hardware design, but I do not like that the microprocessor locks up when it arcs on 5000 volt (I have not seen it on 2500 volt).

My UNI-T UT502 always has voltage on the output connection - regardless of the "TEST" button. Which is not good on a 2500V tester (Have to switch to DCV to prevent getting shocked). Anyone got a schematic for me to try and fix this UNI-T P.O.S.?

[Phantomix]

I'm developing software for these Silabs MCUs (that means, I "have to"), they have an crusty old 8051 core tweaked with more peripherals and a higher clock (usually up to about 25 MHz and single cycle instructions)...

Edit: The f352 used in the unit can run at 50 MHz

What MCU would you use instead?  I ask because I used a C8051F352 a few years back and it worked very well.

Hi! Don't get me wrong, sure these chips work, and if you prefer them that's fine :-) I was just a little frustrated by the limitations of the old 8051 core...

So, which MCU should be used in a commercial project? the one that fits best. This also includes development time - sometimes, for example if it is commonly used in your company, it's better to stick with an old 8051 architecture instead of moving to a more modern one.

I personally prefer the STM32 ARM Cortex for connectivity and calculation tasks, or the STM8 for "as-cheap-as-possible" tasks. I also worked with MSP430 which seems to be a nice architecture, and AVR. The thing with AVR is that there's no good debugging standard for all chips (I'm talking On-Chip-Debugging here)... It also lacks the ability to run Code from RAM (which is possible with STM8, for example)

The 8051, and especially the ones delivered by Silabs are, in my opinion, the worst architecture:

Limitations of the 8051 core:
- Not a linear memory model (e.g. you have multiple memory areas starting at Address 0: DATA, PDATA, XDATA, ...)
- Only one Accumulator register that has to process all data
- Small Stack pointer - limited to the lowest 128 Bytes of Data RAM
- Limited amount of 128 SFR (Special Function registers) for all peripherals
- Inefficient access to all RAM above 256Bytes (XDATA)

Limitations of the Silabs 8051:
- The Silabs Crossbar is a nice idea for flexible pin assignment, but most times it makes the project more complex without real benefits.
- They stuck with the old 8051 UART, which uses TIMER1 as time base
- They use Paged SFR access (you have to write to the SFR_PAGE register before you can access the desired SFR) this makes coding harder
- Very limited capabilities of PWM output (pwm frequency). Silabs statement from their datasheets: For a PWM, it doesn't matter how fast it is, as long as the on/off proportion is right.

Limitations of the Keil C51 compiler:
- Data overlays are a good idea to make the best effort of the limited amount of Data RAM
- Function pointers work, as long as the target functions don't use local variables (Overlay screwed up)
- Sometimes, changing variables that are function parameters will result in unpredictable results.
- Reentrant functions are problematic, that means if a function is called from normal code, and then from an interrupt, it also can screw up your variables.

[iXod]

Looks like all listings of this meter model on U.S. e-Bay do not include the AC adapter (extra USD 35).

And it looks like the adapter is 220v only!

[larry42]

My UNI-T UT502 always has voltage on the output connection - regardless of the "TEST" button. Which is not good on a 2500V tester (Have to switch to DCV to prevent getting shocked). Anyone got a schematic for me to try and fix this UNI-T P.O.S.?

Followup from myself: I had hoped that my faulty UT502 might have a similar relay that was either incorrectly wired, or the wrong "form". Unfortunately there's not safety relay in the UT502, from what I can tell.

[larry42]

My UNI-T UT502 always has voltage on the output connection - regardless of the "TEST" button. Which is not good on a 2500V tester (Have to switch to DCV to prevent getting shocked). Anyone got a schematic for me to try and fix this UNI-T P.O.S.?

Followup from myself: I had hoped that my faulty UT502 might have a similar relay that was either incorrectly wired, or the wrong "form". Unfortunately there's not safety relay in the UT502, from what I can tell.

Just for completeness I'll mention that I fixed the UT502 - the was a solder bridge on Q4 right out the box, causing the "test" button to be ignored, and hence the HV output was always present.

The UT513 looks like has has better construction.

[JuanPC]

1500 Volt Bug Zapper Racket Power Mod,.

[ilanko]

Hi Dave,
Way not blasting Fluke 87-V Digital Multimeter http://goo.gl/FKZYrL with 5k volt ?

[mboensel]

They used a special power adapter and plug for isolation? God I hate having to keep track of those! Why couldn't they just use a little inverter to provide their own isolation to whatever adapter you use? There can't be much cost difference between the external transformer and an internal one - throw in a couple $0.03 transistors and you have yourself an inverter!

Old video and old comments, but I'm optimistic.

Could you explain what you mean by isolation? Isolation from Mains Ground?

Also Dave at 8:11 says "Linear adapter higher isolation voltage" when referring to the input. Could someone explain or point me in a direction? Thanks