My high power test bench objectives

Having been frustrated and appalled by the specious illiterate rubbish posted about home dc power grids and their components I am resolved to build a credible high current bench 24/48V (300/150A) sourced from an LFP bank (which I will also use for night rate energy storage).
I will have a DSO 50g/s and 6.5DVM, Picolog 16b 8 channel data logger plus hall effect ammeters to carry out stress tests, breaking high currents.
Components to examine

dc MCBs
Switches
Fuse types
Fuse links
Wire ampacity
bolted joints
connectors
effect of speciality grease?
sustained arc flash vs gap vs voltage (with methods of extinguishing etc)

What have I forgotten?

Feel free to comment on this pursuit of knowledge by test evidence (rather than plausible conjecture BS).
Hope to add to the fount of knowledge on Solar power systems fundamentals started by our Board master @Will Prowse
You will doubtless find that most text books are centred on AC power/theory which is of course invaluable but when it comes to dc Power >15A say then you are in the wild west (as an amateur and professional Power Engineers would scoff). But who has exposure to dc traction, electroplating, telephone exchanges, large server UPS's and so on?

Yet we have discussions on 48V 300A solar panel grids and of most concern is higher voltages of ca 400Vdc

My source is the 1954 Busbar Design Manual issued by the Copper Development Association --Highly recommended to source a copy on Ebay as I did.
You would never realise the complexity of a grid of flat copper bars/fittings/switchgear/panels

I also draw inspiration from the boaty crowd ABYC. https://abycinc.org/standards/ They have many decades of experience drawn from LABs and dc systems when afloat together with a raft of official regulatory standards required by Insurers - think about it. They have a recent standard on the use of LFBs aboard. Systems must be certified by an official EE inspector of you wont get insured and may not be allowed to moor in exclusive marinas and be a danger to other craft. Electrical fires are quite common.
Heavy current in a hot marine environment - could you get a more severe test of integrity. Cowboys cant swim!

below 600Vdc 5000A busbar and switchgear room for a London underground substation

This is the newer version of the design guide - would be interested if you checked it against your paper version and see if it is just and updated doc or whole different publication

I think 50 giga-sample/second DSO is somewhat overkill for monitoring solar production.
That's more what I would use to characterize HV pulse generators used for nanosecond frame rate imaging, or drivers for streak cameras.
But if you've got it, flaunt it.

I've dabbled in testing breakers and surge arrestors.
I use a 1 GHz 4 G-sample scope for that. Not that my high-voltage or current probes are that fast.

robbob2112 said:

This is the newer version of the design guide - would be interested if you checked it against your paper version and see if it is just and updated doc or whole different publication

Design guide for copper bus bars

Very details guide on copper bus bars

diysolarforum.com

Click to expand...

Thanx rob, that is a very valuable edition to our KB, glad to see this posted under category "

How-to PDF's or Ebook's​

as a sticky reference. My HC1954 ed (£5) was 179pp inc many vintage pix or dc gear (I couldnt possibly justify a HC of the latest ed, the 2022ed is 108pp and doubtless focusses on more efficient and cost effective use of Copper + alloys etc but same principles and basic approach

Depends what you are trying to test and measure I suppose, and to what degree of accuracy.

You don't need a time source with an accuracy of microseconds per year to time the boiling of an egg.

Hedges said:

I think 50 giga-sample/second DSO is somewhat overkill for monitoring solar production.
That's more what I would use to characterize HV pulse generators used for nanosecond frame rate imaging, or drivers for streak cameras.
But if you've got it, flaunt it.

I've dabbled in testing breakers and surge arrestors.
I use a 1 GHz 4 G-sample scope for that. Not that my high-voltage or current probes are that fast.

Click to expand...

Well its amazing what you can get for a couple 100$ today on sale offer.

I want to see what practical versions of fusible links reveal. In 1930's autos this was the method of protection - a thin insulated wire ca 3" long, for ca 100A x 6V. I want to test the modern equivalent for 12 and 24V systems using thin copper wire inside a Si rubber sleeve (say 50mm long with 6mm ring terminals). This type is also known as a strip fuse ca 1" long and these are readily available and use a Zinc strip suitably sized. I dont care for the Cheap Charlie plastic housing that holds the fuse strip. Fuses in use heat up considerably within their prescribed rating - easily enough to melt this type of plastic. Si Rubber can stand 400C and degrades to a powder at 600C. The function of the sleeve must contain a very hot say 250C under operation and also the product of explosive evaporation under sc conditions - so as to mitigate spreading incandescent mtl. IMHO an easy job for a thin wire inside a Si sleeve (its how glass fuses work)

There has been much ill informed debate about the need for Class T or NF type cartridge fuse protection. IMHO this is a grossly over engineered solution as we are dealing with low volts <48V. Much is made of the I2t criteria - look closely and ask what does this really mean in engineering terms? I believe it was a neat marketing trick for pushing high cost cartridge fuses for HV apps into area for high current low dc volt apps.

There have been many posts on various forums to show the limitation of Maxi type strip fuses etc and how they fail and can initiate an arc flash fire. This is IMHO simply because the design doesnt allow a sufficient separation gap so that the arc self extinguishes quick enough (all in the pursuit of $$$ profit and worse still Cheap Charlie knock offs). So my pursuit is back to basics and demonstrate that the original 1930's fuse link in a Silicone rubber sheath is a perfectly adequate safety mechanism for ca 1$ - heresy and big Corp will go after me with their heavies.

For any boaties out there would you prefer to trust your Maxi type fuse or my Si DIY fusible link with your on board Lifepo energy storage

Below is a truck fusible link for alternator protection 100A 24v (GM IIRR)

Fuse protection depends on what you are attempting to protect and how much you are prepared to spend.
Trying to prevent fixed wiring from bursting into flames is one thing.
Trying to protect some fragile semiconductor switching device with a fuse is an entirely different matter.

Critical applications always require significant engineering input and testing, and its not always either practical or cost effective to use fuse protection where the fuse costs more than what it is supposed to be protecting.

Generally with fuses, you get what you pay for. High performance fuses from reputable companies are vastly different to cheap Chinese fuses advertised on e-bay or Ali.

Same with mil spec or aerospace qualified anything.
Super expensive, but at least you get a full ISO documentation trail to prove what it is, and where it came from.

Not many of us are prepared to spend several hundred dollars on a fuse.
But if you were on your way to Mars, and it was part of your life support system, it might be a real bargain if something goes wrong.

Warpspeed said:

Depends what you are trying to test and measure I suppose, and to what degree of accuracy.

You don't need a time source with an accuracy of microseconds per year to time the boiling of an egg.

Click to expand...

Missing my point, what I got on sale for $200 vs a pedestrian performance at $100 - the difference is outstanding.

Golly.
I cut my firewood with a axe.
But I bought a high precision optical firewood length measuring system that can measure the length of logs of wood to sub micron accuracy.
It only cost me a hundred bux, a real bargain.

Now I can measure the length of every single log to 0.00005 millimetres, and I am both warm and happy.

For serious members pls ignore comments that have no added value to our combined KB
I came across this YT by chance whilst checking Beryl and the weather hazard for a pal living in Jamaica. As it happens I was blown away by this extraordinary piece from curators of a museum of old relay type Telephone Exchange equipment and the 48Vdc 5000A power grid needed to run. IMHO its a fascinating account of the principles of running such a grid and a "must learn" for all budding Solar system builders to see the precaution taken with handling such current. It makes you realise what risks we may be taking and how trivially we appear to be treating such a technical area (doubtless carrying over our experiences with conventional domestic mains ac supply.)
Some may view this as a vastly over-engineered system (look at the cct breaker gear for a 500A 48Vdc live load compared with what you may be doing from your panels). This gear had to operate 100% 24/7 with multiple operations. See the size of that 500A cartridge fuse?

An interesting point is made around 28:30 of the vid which I have never seen discussed on this forum (or any other solar site). I cars we conventionally expect it to be a negatively earthed (to chassis) system as the electrolytic corrosion occurs at the positive pole (hence all the green stuff that collects around the +ve battery post). It was assumed that this reduced the amount of chassis corrosion in a car. This may well be an Urban Myth as many UK/Japanese cars were positively earthed to chassis. I quote here a knowledgeable discussion on +ve vs -ve vehicle earthing systems

However back to solar grids - surely as earthing is mandatory by authorities (no more floating dc systems) - then it makes sense to have the -ve side grounded. Has this been discussed?

PS for the masochist

Boron said:

Missing my point, what I got on sale for $200 vs a pedestrian performance at $100 - the difference is outstanding.

Click to expand...

Good deal. The high sample rate may have come at a trade-off for other performance.

I paid $2000 for my first $25,000 msrp Tek scope (on eBay, seller would have bought from a liquidation auction.)
Paid $400 for my most recent at a local liquidation auction.
These are 8-bit native, 13 bit effective in high-res mode.

What software functions do you have available?
Mine has FFT. Does not have Track and Trend. Does not have Integral, which would be useful for BH curves (also want high resolution, because it plots one input (integrated) vs. the other, not vs. time.

For fuse and breaker testing, mechanical shorting should be fast enough stimulus.
Do you just use a battery bank and wire resistance?
How high a fault current have you created?

For surge arrestors, like MOV and those (useless) Delta "Oil Field Lighting Arrestors" I've charged a capacitor and closed a knife switch. Or simply used a HyPot, if capacitance was low enough that current flow doesn't shut off tester before it fires.

I should have a faster switch method to see how well these eat fast spikes vs. what gets through.
And I don't have anything yet which will cause the Delta unit to fire. 8kV peak didn't. Ignition coil, maybe?

@Hedges Im still building my bench so no results yet be glad to share these via pm (send me a pm to set up link. My DSO is Rigol DSO 1054 with s/w mod to 100Mhz https://www.rigol-uk.co.uk/product/rigol-ds1054z-50mhz-digital-oscilloscope/

Oh, 50 MHz 1 Gs/second. At 20 samples per cycle, easy to reproduce the waveform.
When frequency is pushed higher it tends to be 4 samples per cycle.

Boron said:

I will have a DSO 50g/s and 6.5DVM,

Click to expand...

I took that to mean 50 giga-sample per second.
That would be in another realm.

Yep that was a typo its 1g/s and 50Mhz with s/w ug to 100 Mhz It was ca £210 - what a bargain.,
I took that to mean 50 giga-sample per second.
That would be in another realm.
Joke is that a set of differential probes cost same as the DSO

Coming from a test equipment calibration background 30 years ago and a lot of statistical data analysis in the last 10. Telecom jobs with large data sets like 20k calla per second. You can tell a lot from call duration and DSP used. I calculate the 6 sigma values and spew out tickets and auto disable bad DSP until the card is reset then watch again and have the card replaced if a certain number of DSP are marked bad.

I can say that high sample rates can sometimes cause weird oddities when looking at things to granular. For high sample rates I will a lot of times calculate a moving mean and median window for the data to even out the blips from the sample rates. Sometimes you have to play with the window size to get meaningful graphs and output

Using the signal's inherent noise and oversampling to interpolate and get higher number of bits?

I've seen how dramatically effective that is, with high sample rate and signal slowly trending down between two adjacent levels of oscilloscope ADC resolution. The raw data capture was PWM modulated between the two levels, and moving average appeared to recover the signal.

"When oversampling a signal, there should be noise present to satisfy this demand of small variations in the signal. The quantization error of the ADC is at least 0.5LSB. Therefore, the noise amplitude has to exceed 0.5 LSB to toggle the LSB. Noise amplitude of 1-2 LSB is even better because this will ensure that several samples do not end up getting the same value."