Marco Reps - 2022-03-02
This board on PCBWay projects hub https://www.pcbway.com/project/shareproject/ADRmu_10V_DC_reference_standard_5fd4746f.html NITECORE 165AH Power Station https://www.banggood.com/NITECORE-NPS600-165000mAh-165AH-220V-300W-LCD-Display-Portable-Outdoor-Power-Station-21700-Battery-Power-Generator-Power-Source-For-Hunting-Camping-Beach-Tour-p-1718850.html?p=KR28032004379201507P coupon code: BG670c89 Raspberry Pis & JBC tools in stock at https://www.welectron.com/ More from me https://www.patreon.com/marcoreps The list https://docs.google.com/spreadsheets/d/139WT1bAHAmnD6NX4IW7SPVKkw-OCP7QnEriZ-MRHLrw/ Split bobbin transformer https://www.digikey.com/en/products/detail/signal-transformer/DP-241-4-12/952944 Triple junction concentrator solar cells http://www.azurspace.com/index.php/en/products/products-cpv/cpv-solar-cells Fluke patent https://patents.google.com/patent/US8253427B2/ Keithley patent https://patents.google.com/patent/US9478351B2/ Wavetek paper https://eprintspublications.npl.co.uk/1888/1/bemc99-7.pdf 00:00 Intro 00:43 Normal transformers 02:43 Fluke 732A 04:11 Keysight 3458A 04:55 Cheap switcher 05:19 Good switcher 06:02 CERN HPM7177 06:52 UPS transformer 07:33 PCB transformer 07:48 Fluke 5700A 08:18 Qi Charger 09:18 Split bobbin transformer 10:34 Keithley 182 11:09 Keithley 617 11:21 eXtReMe iSoLaTiOn 12:57 Silly? ideas 13:43 Solar cell 15:10 Battery power 15:48 Fluke patent 16:27 Keithley patent 17:34 Wavetek patent 22:09 What I did 24:06 Results #metrology #voltnut #isolation
You are absolutely one of my favorite and most talented content creators. Your videos sit well among the ranks of CuriousMarc, Applied Science, and Stuff Made Here.
Great video, as always. I'm excited to watch your following grow.
In the past I was designing mains filters for rf test chambers. One limiting factor is the Barkhausen effect (magnetic core switching delay). There was a Russian paper we looked into that used mechanical filters (effectively a band pass filter) very interesting results.
At my first company back in the early 70's they produced the Vidar 510 DVM. It had a special power transformer. The mains primary was on the outside of the EMI chassis and the secondary on the inside. The transformer was fully encased in a die case shield, such that only the magnetic field could be conducted to the inner chassis. IIRC it was developed for an Aluminum smelter to monitor current shunt voltages with huge multi thousand amp of common mode currents all over the place with shunts located a long distances away.
Huh, that sounds demanding. Did the core of the transformer actually pass through the shield?
Was there anything to prevent the shield from acting like another winding on the transformer (slits, gaps, etc...)?
It looks like a couple of the switch mode converters you tested have Y-capacitors across the isolation barrier. These can sometimes cause currents large enough to feel. It seems like their contribution might outweigh the transformer inter winding capacitance, but it would be interesting to do the measurement with them snipped out to see.
Same thing I wanted to point out.
I have lit up LEDs and even neons from such things! And felt it too haha
Cutting out the Y capacitors will remove the return path for high frequency electromagnetic interference (EMI). Which means taking them out may cause other types of interference with the sensitive measurements.
There are other ways to design power supplies to achieve low EMI. It is much simpler to avoid the designs that require the Y capacitor.
Sorry to disturb you, but I am desperate. :-)
I think about everything he is doing in this video, has taken a space rocket over my head. May I ask what he is measuring and what it influences?
@Fried Mule common mode rejection. The voltage on output of the charger relative to earth ground. Due to capacitive coupling, there will be a small current that can flow, and a voltage of up to nearly 100V that can develop. It can cause problems with precise measurements and audible noise in audio circuits.
Interestingly enough motor-generator pairs are actually used in very high voltage applications such as particle accelerators where you have to get a few hundred watts, or a few kW of power into an area (deck) that has a voltage of say 100kV relative to ground. It's a very fascinating.
Apparently back in the day the military used these to do frequency conversions as well, usually between 400Hz and 60Hz. Since maybe the 70's solid state electronics got good enough to replace those.
They used to be used in a lot of machine tools (lathes/mills/grinders/etc) to allow for more optimal spindle speeds via frequency conversion.
Can't beat a meter of insulating shaft for isolation.
@Power Max if by "back in the day" you mean "today", you are correct.
@Auxiliary Panther I work for the US army and we do not use motor generator pairs for frequency conversion. There are a number of different MIL COTS converters we use. They all use solid state switch mode converters and inverter stages. AFAIK they work by rectifying the 115VAC 400Hz to DC, with power factor correction, then feeding an inverter stage to convert that back into AC, which is filtered with some output LC circuits. Usually these are fed with 12, 10, or 8 AWG MS22759/43 or /87 cables (200°C and 260°C, respectively.)
Great job! I guess you could increase isolation while not compromising efficiency by building a resonant converter. It should reduce the uncoupled inductance and thus deliver more power.
Delighted to see one of my favorite YouTubers in the comments of my newest addictive channel. Marco reminds me of a combination of FesZ and GreatScott. Love both of your work
I’m an electronics hobbyist at best, and I only have a concrete understanding of let’s say 20-35% of your topics and observations. But I love each and every video, your approach is fascinating and it fuels context for so many ideas or projects (ahem, problems) I have. ❤️
One thing you should probably still try are the transformers that are commonly used in physics classes with those interchangeable coils and everything. Here, the cores are a few centimeters apart, however, I assume that coupling is less than ideal.
LED solution: use a high power blue LED facing directly into another blue LED of slightly higher wavelength. Best case energy recovery is only about 10% but if you are only looking for ~ 1W that's not a huge problem
is that a diy optoisolator?
Great video! I've been working on low leakage currents using off-the-shelf DC-DC converters, mostly at switching frequencies. Love the AC leakage comparison.
I just figured out a neat trick to reduce leakage on a split-core AC transformer even lower than the typical 1uA, by about 10x. Tried it on a Hammond 162G12, 12.6VCT, 12W. It reduces the AC leakage current by about 10x.
The trick only works with 120VAC US power and off-the-shelf dual-primary transformers. You connect one primary of a dual primary AC transformer to the 120V line and neutral. Then connect the other primary reversed, with only the neutral connected, so it generates the opposite phase120VAC. The other pin of the second winding is not connected, it just floats. This out-of-phase winding cancels out the electric fields from the primary, and so most of the leakage current also cancels. I get almost a 10x reduction, a drop from 0.6uA to 0.06uA!. Of course the transformer won't perform as well with only one of its 2 primary windings connected: Less total power, more V drop. But damn is the leakage current low! Check the bottom of site below for more details.
Dave
http://www.djerickson.com/rpi-power
I love your content! Thank you for all your hard work making these awesome, technical, funny videos!
Good video, but Marco, you missed explaining the most important instruction on 732a: You never take measurements from 732a while it's plugged in, that is the whole point of the 732a having a 24V sealed lead acid battery pack built in. The main point in the 732 manual: Never take measurements while the 732a running from Mains power. The Mains power is only kept on to keep the 732a warm and charged between measures. Typically you're not going to be using that device for long term tests, a 732 is mainly used to transfer a known calibrated Volt from your local cal lab to your other in-house 732's, and from there you can periodically align your 3458a's to the 732a's.
We've found that a battery / supercap cycled power system gives best low noise results, by far. There is a reason a lot of those systems are patented.
I think he was purposefully leaving it plugged in since batteries are perfect isolation and he just wanted to compare the transformer isolation.
I remember talking about that subject on eevblog forum years ago because I had the project of transforming relatively cheap but accurate dmms into table dmms. The goal was also to add logs, network connection etc...
Unfortunately I never had time to go very far.
For the solar part, I went the opposite way using cheap panels and lighting them with monochromatic LEDs.
I don't remember the power I could transfert. The biggest issue with that solution was its size.
9:00 that qi reciever may be pulsing because it's not detecting a load on the output. Once it has no draw it signals back to the send coil "fully charged" by shutting itself off (with rf magic to the spec).
I think "Medical grade" means there is no capacitor to the ground in the filter, as in medical installations there are no RCDs (not to switch off any life supporting equipment) just alarms for signaling naughty electrons flowing where they aren't supposed to.
Medical grade power supplies can have capacitance to earth as long as the earth leakage current requirement is met. Up to 5mA is allowed, although rarely is the measured value more than 500uA.
Medical power supplies have a lot of requirements, but the most relevant here relate to the #1.) Construction of the transformer/PCBA (to provide sufficient isolation) and #2.) Output (AKA "patient" or "touch") leakage current to earth, which mostly depends on the value of Y-class capacitors across the primary-to-secondary isolation boundary.
In a non-medical power supply, up to 500uA "touch" current is allowed.
In a Type-B/BF medical power supply, up to 100uA "patient" current is allowed
In a Type-CF medical power supply, up to 10uA "patient" current is allowed
(All these currents referenced to a standardized 1K or 2Kohm measurement resistor.)
Basically, low leakage medical power supplies need to use small value Y-capacitors (or none at all), and usually need to play games with the transformer to keep the EMI low without using large value of Y-capacitors.
For this application, Type-CF has the lowest primary-to-secondary capacitance, and would probably be a good pick.
@Solid State Workshop Hey! I recognised your profile pic instantly! Found your videos great when doing my introdoctory course in electronics :) Hope you are doing well
@txcpnae It’s always so nice to hear that. Thank you. Once my life settles down a little, hope to put out some more 😉
No, "medical grade" means you can eat it and you won't get sick.
@Solid State Workshop In medical ultrasonic cleaners I have seen them get dangerously close legally to the legal line at ~490uA.
That coaxial inductor sure sounds useful in high speed switching converters (SiC/GaN based). Self capacitance is a real pain at 10s of V per ns.
Whoa this is so good and inspiring - not everybody can make a long list of transformers exciting!
Woah, I am currently developing an isolated high-impedance ADC module array designed to measure high-impedance voltage sources (~100k of internal resistance) with 5 digit resolution and 2.5V dynamic range.
I've learned about FR4 leakage, guard traces, solder mask leakage, chopper amps etc already, and now you've covered probably the most subtle topic - isolation transformer leakage and how to reduce it. This was of a great help, thank you!
Look up "austin ring transformer". These are used to feed single phase AC power to the signal lamps on AM broadcast towers, while not passing the RF to the power grid. They are dual, interlocked toroids, with one AC winding on each one.
"...as a DC person, I hardly know what a dB is..." this was gold! :D
As a digital person, I tend to have the same viewpoint.
How about this idea: Charge up super capacitor for a given time period from mains (ex. 10ms) and then switch off the charging from mains power and run on the super cap during the measurement?
As the power draw of the circuitry is likely much lower than the power you can supply to the super cap to charge it, I think the duty cycle might be appropriately quite low, thus making it possible to do measurements.
That's essentially the approach discussed at 15:48 with the Fluke thermometers.
@AJMansfield Damnit, I was just hunting down that timecode, and you beat me to it.
That was my first thought, flying capacitor power supply -- to be safe, think you would want to use relays to fly the cap, but that's a new can of worms.
Or an array of super capacitors that you charge and then disconnect to discharge on the DC side
Really you only need two super capacitors and then just have a timer or gasp a micro controller on each side controller a realy or two with some sort of opto coupler interrupt pin so you're never actually connected to mains. Like charging a bunch of batteries and switching them out but faster
I kept thinking about this capacitive isolation requirement that you where trying to achieve; you mentioned around the end the use of an "amorphous metal transformer core" - this looks like a very electrically conductive material. I am bringing this up since if you would be using a ferrite core, that should not be electrically conductive, and also work better at the high frequency of a switching converter. So even if you are sacrificing a bit of magnetic permeability, the core its self would not be a path of capacitive coupling. In the final designs, it seems (to me at least) that the smallest physical electrical distance between primary and secondary is not over the PCB, but rather from wire insulation to the core, and core to insulation on second side. What do you think?
Interesting point indeed! I was swooning over amorphous cores so much that that didn't even occur to me. I'll try and find a ferrite core to test some time. (Or if you got around to it first, I'd love to hear your result!)
@Marco Reps I guess the other important thing to consider is that the leakage current between primary and secondary is frequency dependent - for a fixed voltage, the same capacitance will yield much lower current at say 50Hz (mains frequency) than 50KHz (a switching converter). In the end I guess its a compromise - get higher capacitance but work at low frequency or get low capacitance and try high frequency... However, high frequency means easier common mode filtration. This is quite an interesting subject :D
What?! FesZ and Marco Reps in conversation!? Not sure if my heart can take the excitement!
"Light emiting resistor"
My god that made me laugh out loud HAHAHHA
Every electronic component act as a heater and emits light if you operate it wrong enough
That and the diagrammatic goatse.
one of the oldest jokes, but always appreciated 😀
Caught me off guard Hahaha
Diode = voltage dependent resistor
sort of
If you want a low cost low stray capacitance DC/DC converter, you can use a common mode inductors as a transformer in a half bridge or full bridge converter.
Perhaps the idea of one BLDC motor turning another as a generator using a plastic rod is worth revisiting. Some good quality bearings can last plenty long enough.
Mains transformers with an intermediate shield winding (connected to secondary-side ground) are part of the "secret sauce" of traditional hi-fi equipment. Beyond about 100 VA (i.e. sizes needed for typical integrated and power amplifiers or receivers) this is rather necessary if you insist on unbalanced connections and an IEC Class II device. Otherwise open inputs may be rather prone to hum. It probably still isn't as good as the "windings on opposite sides of a grounded core" approach suggested, but I reckon magnetic coupling and as such efficiency is a whole lot better.
(I have to wonder whether the wondrous "medical grade" device may have had a shield winding but it wasn't properly hooked up or something. Mind you, it's a fairly big transformer. The only mains transformers with a shield winding that I've been able to find at parts distributors were in fact medical grade.)
This issue has been further exacerbated by the increased ubiquity of switch-mode power supplies, which often are in dire need of safety capacitors in the nF between primary and secondary ground to meet EMC or even have full-fledged mains filters whose leakage currents have nowhere to go in a Class II device. (I have some JBL 104s in the office, the level of hum/buzz when using an unbalanced audio cable that is unplugged is scary. It entirely vanishes once plugged in.)
Speaking of SMPS, shield windings can also help their transformers out as well. Mind you, going by disassembly channels, getting basic electrical safety right seems to be enough of a challenge as-is...
I am glad you mentioned the approach of nuking remaining leakage by redirecting it to ground (PE). This is also made use of in a wiring mod that I have previously suggested for IEC Class II equipment with mains filters:
Replace
L o---||---> SGnd
N o---||---> SGnd
by
L o---||---> PE
N o--||---> PE
PE o--||---> SGnd
(capacitors -||- being a Y2 rated 2.2-4.7 nF safety type)
Works wonders on old Yamaha A-760, 960 and 1060 series amplifiers, for example (they are using some thyristor business for primary-side regulation and accordingly need mains filtering badly but are IEC Class II; at least on the 220 V version the mod actually is quite straightforward). Stock, they like to hum on open inputs rather noticeably, as you may be able to guess. I have termed the modified wiring "IEC Class II with benefits". It follows the tradition of PAL in combining multiple cost-increasing elements from existing systems for the best result. ;)
This only works if your PE is actually present and properly hooked up, of course, which is not a given everywhere in this world (while I would be unconcerned about Germany, the UK and newer installations in the US, parts of Asia may be a different story).
This is my first vid I have seen by this guy. Quite amazing -- most went over my head and I'm going to have to rewatch this a few times. Wow wow wow!
i just don't know why youtube has not recommended before. Such a good channel, but seems to be well hidden in the deep caves of youtube.
The Topaz ultra-isolation transformer is old tech, I worked there in the 70's.
The primary and secondary are wrapped with copper, insulated with tape to avoid a shorted turn. The one we did had aluminum cases around both winding (not just the plate you showed), the copper and aluminum were grounded. Any noise on the primary saw:
Capacitance to the primary copper shield
Inductance of the shield to ground
Capacitance from copper shield to aluminum shield
Inductance of the aluminum to ground
Aluminum shield to the secondary copper
Inductance of the copper shield to ground.
Isolation was very good, as long as all the shields were grounded properly (primary and aluminum to earth, secondary to the chassis ground of the powered product. They did side by side, and inner and outer when leakage inductance was critical.
Interesting, thanks. I have a 500VA (probably 1980s) Topaz unit that I occasional use (mostly for the safety benefits if I am working on powered equipment where the mains wiring may be exposed. But it's good to have further understanding.
The heater requirement is a challenge for power, but once the reference gets up to the regulated temperature the power needed to maintain this temperature in a thermal isolated chamber should drop considerably. This gives two power modes to your reference, warm up and steady state. So, you do not need low noise during warm up, because the reference will not be used until up to temp and stable.
The answer: a non-isolated warm up supply (mode 1, high current) and a charge exchange isolated supply (mode 2, low current).
For isolation you could use two supercapacitors (5F 12V) and exchange charge between them and the AC mains transformer using two relays (DPDT, isolate both the capacitor feed and the return, much like some chopper charge exchange controllers).
So, a relay isolates the first supercap (let's call it the source supercap) from the AC (mains) derived line charging circuit and another relay isolates the second supercap (let's call it the load supercap) from the source supercap. If both relays are on the supercaps are cross connected and powered by the AC mains derived circuit, quickly charging both and providing warm up power for the heater. After warmup the relays are sequenced to exchange charge between the mains AC supply, first to the source supercap, and subsequently to the load supercap, in such a way that there is never a direct connection back to the AC line transformer and its nasty inter-capacitance leakage and coupling noise.
The load supercap is maintained at a voltage higher than the linear regulator voltage used to supply the reference circuit and the heater regulating current, and can be regulated by the switching circuit. The reference supply after warm up is essentially a varying charge/discharge voltage across the load supercap, much like how a bridge rectifier periodically charges the smoothing capacitor. But, in this case the switching speed is very slow (like once a second) in order to extend the life of the mechanical relays (high reliability 10M cycles?). If you have a switch on the front panel for standby you could just feed power across the caps saving some cycles on the relays until you switch to ultra quite isolation mode. Some PCB mount relays are very quite when they switch. But, the tiny relay switch click(s) may give some useful audible feedback
The above has been done with batteries for ages, but now we have super or ultra caps.
I like the motor generator idea. Chop the blades off a couple of PC fans and glue them back to back. Otherwise you could rip the cones out of a couple of loudspeakers and mechanically couple the coils together.
"The main winding was of the normal lotus-o-delta type placed in panendermic semi-boloid slots of the stator, every seventh conductor being connected by a non-reversible tremie pipe to the differential girdle spring on the 'up' end of the grammeters." - what I hear as a mechanical engineer
Now if he'd used spurving bearings be would have dropped at least a decimal on those ppms
LOL me too, but your humor made my day! :-)
This is just silly. You can't use tremie pipe on odd-numbered conductors mate
@Callan Fair Not even if he tightens the flingnut on the reversible trom-joint?
@Fried Mule we're trying to have a serious conversation here
I love the transformer headed transformer animation slice. Very funny.
Fantastic content as always. Thanks 👍
hahaha I won't judge if you take your test gear on holidays! That was brilliantly done mate, I've been looking for a battery scope but now I'm convinced to get one.
Incredibly cool. Thanks for sharing a few bits of forbidden arcane knowledge with us.
Great video, both technical content and entertainment value!
Would piezoelectric transformers be good for low leakage?
Most of the ones I see have high voltage output but there must be some low voltage ones.
You might not want a lot of ultrasonic vibration inside a sensitive instrument.
Love it!
Your sens of «electronics humor» is the best🙂🙌
How much improvement could you get by preceding the isolating stage by an ordinary transformer with an earthed center-tap on the secondary?
Thanks for the video. Is it possible to use coaxial wire for the inductor in HF filter? Will it reduce parasitic capacitance (if one side of the shield will be grounded)?
CORRECT, This is extreamly important, for many things, I have been playing with low capacity transformators for a long time, I need this mostly on mains powered things, i really like the ideas with toroide types, and 3D printed holders / wires seperators, i think this idea can be scaled up to mains size on toroids too, i am clearly going to make a few experiments with this. maybe even another rotation of the cores ? to make even less capacity ? my comment got lost ? so now i try again hope it works this time
As I was taught in school, the cleanest voltage supply is a battery!
Just to mention, you explained why the audio chip in my computer blew up. Amplifier and computer - polarity was backwards. Same power from the wall but... All it took was a leaky filter cap.
Consider looking for a Type-CF medical AC/DC adapter. These have very low leakage at mains frequency.
19:15 I believe this is how fast Marco actually draws his schematics 😂
Always enjoy watching your deep knowledge and dad jokes. Sometimes i know what you are talking about, other times i just hope i will absorb something through osmosis x
Can you test or make one of those horribly expenisve military grade nuclear explosion detectors? (e.g. Maxwell Technologies HSN-1000 Nuclear Event Detector) It might be worth investigating to protect your precious meters from an EMP considering the current situation.
I am constantly baffled as to why I'm subscribed here lol. I watch most of the videos and often come out understanding less than I went in. This stuff just goes way over my head lol. Keep it up man with magic words.
Welcome to that rabbit hole called emc :) Common-Mode problems are 99% of that work. To be honest, it's mostly solved by altering dimmensions after failing to "just" add filters.
Having that said, have a look at ethernet magnetics and gate transformers and those old ISDN/POTS transformers.... at least the latter 2 should give you 1W+
The higher switching clocks needed may sound counter intuitive, but remember the impedance of an CM choke :) so there's a geometric sweet spot in the hunt for the lowest CM noise...
BTW, I have a propper EMC measurement setup... so if you feel your converters should get some characterization, I may be able to get you numbers :)
Regards
Wonderful video. This one took a lot of preparation, too. Thanks!
Bert Holtappels - 2022-03-02
This is ridiculously good. High grade technical content, but even the puns would make it worth it 😁
Spookie Ham - 2022-10-25
I never really noticed them before but after watching a few videos they are riddled with them and really makes them enjoyable.