500A power source

Early high power low voltage (+5 V or -5.2 V) computers power supplies used sometimes a center tapped transformers and two diodes instead of a 4 diode bridge, thus suffering only a single diode voltage drop. If you can find such a transformer, run the full secondary into 4 diode bridge, it might be suitable for 10 Vdc,

To find suitable three phase transformers, look also for 277/480 V primary and run them at 120 V or 208 V and you might find a suitable transformer.

The Bioenno I originally recommended. You have to either:

1. Increase your MOSFET heat sink.
2. Use a larger MOSFET.
3. Use more MOSFETs.
4. Add a dropping resistor.
5. A combination of 1-4. Add a series resistor is the easiest.

I looked at your PCB layout - what kind of heat sink are you using now? What is your average power dissipation for normal, continuous operation? Improving that heat sink is probably the easiest solution.

The PCB can either take a 10W heatsink with the MOSFET mounted vertically, or a 100W heatsink with the MOSFET mounted underneath. The power limitation comes from the TO-247 MOSFET, a IXFH400N075T2, and I don't think a bigger heatsink will help much, as its 8ms Transient Thermal Impedance is right at the limit. But simply expanding the PCB to spread the work across two FETs would not be difficult, and perhaps much easier than futzing around with making lower supply voltages (a bigger heatsink, to handle an added bolt-on power resistor, is another possibility).

That would be yet another new RIS-976 version, maybe RIS-796B, but it's probably inevitable anyway. Diode half-sine testing wants a microcontroller interface.

--
 Thanks, 
    - Win

Those are awesome parts, yes that would work well.

--
 Thanks, 
    - Win

Ok, then you need to chose door #2 and get a bigger hammer:

...and here is a smaller, and much cheaper, hammer:

ND/2354478

Yep. They would do the job.

--
 Thanks, 
    - Win

My knowlege is not current, I used to do a lot of marine work. Large working vessels like salmon seiners and towboats used 32VDC systems, obtained by series connecting four large 8V lead-acid batteries. I'm sure the starting load on those big diesel engines could approach 500A. I believe GM made the batteries, could be worth a look.

Yes, I agree they would work.

--
 Thanks, 
    - Win

found this-

You didn't mention there would be reps. What PRF?

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Design 
Website: https://www.seventransistorlabs.com/

Half-sine test would be at 60Hz or 50Hz. When measuring the peal operating capability of a diode, peak current must be lowered with more cycles. Manufacturer's plots typically show about 1/16 peak current for 100 cycles as for 1.

--
 Thanks, 
    - Win

That 1000A supply was used for a Z80B based computer system, with a lot of slow, power hungry RAM. Enough to store two pages of live video, and chr oma. Things ave really changed in 35 years. A used, high speed 12 bit A/D c onverter hybrid was $1400, and the RAM was already obsolete.

Whoa, a 5kW Z80 system! Most of ours were under 100W.

--
 Thanks, 
    - Win

It was a full relay rack of circuit boards and fans. 768KB of 12 bit wide 4 Kx1 TI DRAM per video page. It was so slow that 16 chips per bit were addre ssed at once, then read in sequence to get the required data rate. There wa s a huge sticker inside the front door showing where each memory board appe ared in a video page. That used most of the 1000A. Other voltages for the a nalog circuits came from another set of power supplies. The thing was on 24 /7/365, along with the video switcher for master control. That filled anoth er full rack. It was a PITA. Someone had 'recapped' it, butt hey replaced t he sealed liquid tantalums with cheap imported 85C electrolytics. I would r eplace about $500 of them with the original style per month. It was obvious that the component costs made up a large part of the %250,00 from the Sque ezeZoom, and about $50,000 for the switching system.

One interesting thing about the SqueezeZoom. Even though the video circuits were very low noise, you could zoom into infinity and generate a great mov ing star field like you would expect to see on the view screen of a faster than light star ship. It was much better than the paper with holes punched in it, that was common in movies and TV shows. We played a prank on the Vid eo engineer one day. I set it up to zoom to infinity, while another enginee r set the Chyron to flash 'GAME OVER' in huge letters. He freaked when he g ot back from lunch! :)

Did it really go from 3 phase rectifiers directly to 5 Vdc ? The standard practice with big computers needing several kilowatts was to generate low ripple 8 Vdc from 3ph rectifiers and using local 7805 or

309K regulators on each PCB. This helped avoiding the DC power harness voltage drops.

Even when using 2102 memory chips (1024 x 1 bit 50 mA static RAMs) that 1000 A current would have been able to easily power a full megabyte of static RAM.

That amount of memory would have store multiple NTSC frames, especially if the composite video was sampled at 3 x Fsc (3 x 3.57 MHz) as was the practice in he 1970's.

Still wondering why they did not use core memory as large computers did ?

Sounds like it after all stored component video (3 x 4 bit) or 6 bits for luma and 6 bit chroma, not composite.

With 768 KW x 12 bits that makes just more than 2200 chips, consuming less than 100 W, since all chips weren't accessed all the time.

More likely 12 or 24 chips/sample.

It had a pair of very heavy copper buss bars that ran most of the height of the rack. Each board would have needed multiple regulators, and the ext ra space for heatsinks would have split the system into two full racks. It would have required even more cooling, as well. A single lare regulated sys tem eliminated the extra space, heat and load balancing.

Static ram was too slow, even using that paging system. This was designed i n the early '80s by a Professor of EE at the University in Gainesville Flor ida.

It stored two frames.

Core was even slower that the available SRAM.

ot of slow, power hungry RAM. Enough to store two pages of live video, and chroma. Things ave really changed in 35 years. A used, high speed 12 bit A/ D converter hybrid was $1400, and the RAM was already obsolete.

It stored 12 bits of video, and six more of Chroma.