recommend a decent/proper dump load

Tim,
Do not trust that controller circuit very long.
The fets are operating in their linear region, and they Are going to fail. Maybe not today, or this week, but they will.
G-

Rhubarb dammit and so on, but once again this forum proves its worth. Any ideas?
Luckily I haven't yet ordered any of the bits I don't already have.

Edited to add: Ghurd - tried to send you a PM but couldn't get that to work. Text as follows:
Ghurd
You'll have guessed I'm not a microsparks, what I do think I know I gleaned from my father, who is. I did put the diag under his nose, and he thought it was OK (he's mid-70s though so may have missed something).
Where did I go wrong? I didn't want to drive those MOSFET gates to their max V, but rather to somewhere in the middle of their stated range - had some nebulous idea that would be better than whacking them with max V, even with the sharp corners taken off by that 100pF, and they'd still get into saturation - was that where I cocked up?
Tim

Also edited to add: Ooo -er and yikes. Have just noticed (I think) a need for a bleed resistor between gate and source, otherwise the bloody thing will never turn off. Hmm. Think I'll discuss circuit from first principles with Dad.

Remember. You asked.
I only looked at it briefly.

MIGHT be OK if the 2.7Vz is changed to a 5.1V.
I would change it to 12V myself.
I would hang a 500K or 1M in parallel with it.
Don't need 4xNPN. One is enough.

I strongly suggest using 4 fets, one for each glow plug. Very Very strongly.
Really should have a gate resistor, maybe ~50 ohm, on each individual gate, and they should be on separate heat sinks.

The fets will still be operating in the linear region, mostly. That's not good.

The LED will probably never light. If it does, the dump load is not big enough.
Could move it over, feeding it from the NPN, with a much larger LED resistor (5K).

It would probably be OK as it is if the fets were changed to some big old transistors, which are kind of rare anymore (and expensive!). (I figure most of the heat will be in the transistors instead of the water)
That is probably how your father is thinking about it. Can not think about fets the same way.

And after I saw the price break down (BTW- Yikes!), wondering why you don't get a ghurd kit instead.
Looks like somewhere around 12 GBP would do it.
G-

Thanks Ghurd. Not sure how I'd get the cash to you in the States (don't have paypal) but am looking for around 80-100A switching capacity at 24V, so presume I'd need to ask for two of those IRFZ44Ns.

You're probably right about the transistors - it's that era and Dad still says he misses the simplicity of valves.

[Gotwind Ben]

Most, well all here would advise the Ghurd kit.
For the money, I suspect around £15 delivered to the u.k, It can't be beaten.

I don't see any problems with a Paypal account, just sell something cheap on eBay and have the funds there before buying if you want to be super safe. I've used Paypal for over 10 years without problems.

"am looking for around 80-100A switching capacity"
Again- Yikes!

For something with what is considered anywhere near decent control, it takes at least 2 top-of-the-line interconnected controllers.
If the battery bank is large (pronounced 'expensive') enough to handle the charging system, you want top-of-the-line stuff. 3 stage PWM control with temp comp, etc.
I really have to wonder if you are no being a bit overly optimistic.
That is somewhere around Four 10' good air core dual rotors running about their peak continuous rating.

The ghurd kit comes with 2 IRFZ44N fets. And I tell people not to push them past 6A without a decent heatsink, and never ever push them past 10A.
The IRFZ44N is 0.0175 ohms. A 7A load makes it 0.8575W 'hot'.
The SPW47N60C is 0.07 ohms. A 7A load makes it 3.43W 'hot', or 400% more Watts in the fet to try to deal with.

Anyway,
"ghurd" kit, 2 extra fets (4 total, 1 for each glow plug), 4 heat sinks, '24V adapter kit parts', S&H (anywhere), total US$21.50. (GBP 13.14 today)

Kind of freaks me out about GBP 19 for 2 fets,
and needing GBP 38 worth of them for a circuit that is questionable.

Ask your friends or people at work or something if they have paypal. Could always give the the cash to send. Probably not good for ebay sellers, but I don't do ebay. LOL
G-

As I am CERTAIN ghurd knows FETs are voltage controlled devices while "transistors" are current controlled devices. FETs have a much faster Ton/Toff than traditional bipolar semiconductors. A $5 uController would be a godsend to utilizing the FETs to their full capacity. Operating in full conduction for 50uS~1mS is much better than operating in linear mode constantly. The Average current/power that a FET can dissipate is MUCH higher in "switch mode". As long as the "switch times" occur multiple times during any given "dump cycle" the effect is the same. That is, if your three phase, rectified source has a frequency of 200hz, and your switching frequency is 1kHz or more, PWM to your "dump load" will be "transparent" to your source, and your FET can dump a great deal more power than it would be able to in linear mode. If you need to use linear mode dumping, you should consider IGBTs or Thyristors. If your purpose is to "Break" the turbine, using thyristors is an economical solution. An additional half-wave rectifier, in parallel with your "normal rectifier", constructed of SCRs and leading to a "dump load" would achieve the result at minimal cost. When an over-voltage or over-current situation is encountered, the SCRs are enabled to dump the load. The SCRs will stay in conduction mode until the current and the trigger drop below the SCR's threshold. Even after the trigger is removed, SCRs will frequently stay in conduction until the current drops (That pesky phase shift between V & I in highly inductive circuits ;-) ) SCRs are rugged, cheap components, perfectly suited for the task. If anyone is interested in either a uController & source code for use with dumping via FETs/IGBTs, or a schematic for using SCRs, let me know and I will be happy to help. I cannot promise I will prototype anything, just /write the code and/or design.

Fish

Thanks both, I quite realise that at the moment, having advertised two 6ft diameter 400W jokes-on-sticks coming in and one bell-end glowplug radiator of less than 1 kW going out, my stated reqts are clearly gross overkill for RE purposes. However there's a different non-RE medium-term reason I need LVDC switching capability that heavy.
Will consider a bit more.

Tim,
"two 6ft diameter 400W jokes-on-sticks"?
That's both more and bigger than I can install or use. :oops:
400W, or 100W, is only a joke when someone expects to go off-grid... And heat their house with the surplus power.

Fish,
uController looks neat. (and if you are affiliated with them, or very intimately familiar with their use, shoot me an email- I may know someone looking for that type of thing)

But it seems like overkill unless it has temp comp or 3 stages or something?
Have you seen the O-scope pics of my controller?
The HZ were sped up for a specific commercial application (which removes the battery from a small motor vehicle and substitutes a cap- the charging system is a reactance current limited dynamo-type charger), but the switching slopes should be about the same.
Instead of a fixed frequency PWM, I think of it as a PWM with modulating frequency.
With some tweaks (involving un-doing what I did to intentionally slow it down) the standard ghurd controller can go past 1 or 2 MHz, though I do not believe it is a good idea in a battery based system.
The O-scope battery voltage (which was actually a cap) will be tighter in a more normal battery based system, if that's how it is assembled.
http://photobucket.com/GhurdControllerScope

Not sure I agree with inexpensive (raw) IGBTs being a good idea in the hands of 99.99% of the DIY people. I just think there are too many complicated 'gotcha' scenarios, like latched, not firing on, etc.
And their Vf in a 12V system has its own issues.
Just my 2 cents, which I am willing to refund on request. ;)
G-

ghurd,

Your scope shots demonstrate you are avoiding the linear range very nicely. I find uControllers "Cheap and Easy logic". I would define my skill set as "experienced hobbyist", certainly not "professional". Accordingly, I would not pursue professional venues, but I would be more than happy to assist in hobby oriented pursuits. My experience is primarily with the AVR family of uControllers with an emphasis on PC based interfaces for Voltage/Current control in external circuits. Things like stepper motor controllers, SMPS prototyping and some darker arts. I have a firm grasp on AVR-ASM, never played with C much, though there are numerous C compilers for the AVR family (I find the code generated by these compilers bulky and awkward). For $5 you can get an AVR with multiple ADCs, multiple PWMs, numerous USART options, ram, EEPROM and flash memories. So, yes, you could temp-sense several components, test voltage and current and still have plenty of I/O's to control active components. You also get counters and a very efficient processor with the ability to track data trends. If you can find fast enough switches, you can directly switch with ~50nS precision (20MHz), certainly not requisite for dumping power, but handy for some other purposes.

I have not seen the schematic for your "ghurd kit"; I must admit I am curious. Sounds cool :-) Your scope shots seem to indicate ~12A for ~2mS to ~7mS pulse widths with a wide range of pulse frequencies. You could certainly mimic this with a uController; however, it would likely be better to use a fixed switching frequency with simple pulse width modulation. In fact, this capability is built into most of the AVR family (virtually no code required). The only "advantages" I see to the uController route over your approach would be if you wanted some type of interface with a PC or other controller, or if you wanted to "load manage" your dump load; that is, if you wanted to drive multiple devices independantly based on temp/Volt/Current feedback, it would be much easier from a hardware stand point. Keeping track of the power "dumped" might provide some insights into better power management/utilization, the draw back of the uController approach to a DIYer would obviously be learning enough about the uController to make the firmware/hardware "bullet proof"; prolly beyond the scope of the average DIYer unless provided with firmware and PCB, and likely assembled rather than a "kit".

Fish