Home-brew 13.8v, 30A PSU regulation prob

Thread Starter

Dave_UYZ

Joined Jan 16, 2014
29
If you're using a MOSFET as a series pass source-follower, don't forget you need at least 6V Vgs to turn it fully on.

Some people use a charge-pump voltage doubling rectifier alongside the power one to develop a higher voltage to drive the gate.
No, the MOSFET is ONLY in the Solid State Switch.
It has only two states; on or off, both of which are lines to its own PCB.
 

crutschow

Joined Mar 14, 2008
34,847
A "crowbar" circuit by definition drops a short across the output in case of a fault condition, which takes care of an over-voltage failure no matter what the cause. You could use a circuit-breaker instead of a fuse to disconnect the power when the crowbar trips.

A typical crowbar circuit consists of a zener or other voltage sensing circuit to determine the over-voltage trip point which then triggers an SCR connected across the output. The SCR has the advantage of staying tripped until the power is removed.
 

Thread Starter

Dave_UYZ

Joined Jan 16, 2014
29
A "crowbar" circuit by definition drops a short across the output in case of a fault condition, which takes care of an over-voltage failure no matter what the cause. You could use a circuit-breaker instead of a fuse to disconnect the power when the crowbar trips.

A typical crowbar circuit consists of a zener or other voltage sensing circuit to determine the over-voltage trip point which then triggers an SCR connected across the output. The SCR has the advantage of staying tripped until the power is removed.
Thank you.
Frankly, I'd rather not "put a nail across the output" and pop the fuse.
Following the detection of an over-voltage state (such as by use of a 3423 crowbar chip), I had in mind, originally, to switch the MOSFET Solid State Switch to OFF (open circuit). But it is too high an on resistance and drops too many volts at full current output.

I realise I may have to simply bite the bullet and go back to first principles here, but I'd like to get it working proper!
 

bountyhunter

Joined Sep 7, 2009
2,512
Thank you.
Frankly, I'd rather not "put a nail across the output" and pop the fuse.
Following the detection of an over-voltage state (such as by use of a 3423 crowbar chip), I had in mind, originally, to switch the MOSFET Solid State Switch to OFF (open circuit). But it is too high an on resistance and drops too many volts at full current output.

I realise I may have to simply bite the bullet and go back to first principles here, but I'd like to get it working proper!
If the voltage regulator is working "proper" you will never need a crowbar. The voltage will be where you set it.
 

Thread Starter

Dave_UYZ

Joined Jan 16, 2014
29
If the voltage regulator is working "proper" you will never need a crowbar. The voltage will be where you set it.
Point taken, but this is a "belt & braces" gadget for a bunch of cadets.
The operative word is "IF".
Can any one recommend an SCR type to trip a 30A fuse (the blade type seen in a modern car, for example) ?
 

bountyhunter

Joined Sep 7, 2009
2,512
A crowbar circuit is the emergency stop.

Series pass transistors can and do fail S/C.
But the pass transistor and regulator circuits are no more likley to fail than the crowbar circuit if both are properly designed. Most commercial power supplies do not have crowbars for that reason.
 

ian field

Joined Oct 27, 2012
6,536
But the pass transistor and regulator circuits are no more likley to fail than the crowbar circuit if both are properly designed. Most commercial power supplies do not have crowbars for that reason.
Crowbar protection only carries current when tripped - series pass transistors carry a lot of current all the time, sometimes so much they discolour the PCB around them.
 

bountyhunter

Joined Sep 7, 2009
2,512
Crowbar protection only carries current when tripped - series pass transistors carry a lot of current all the time, sometimes so much they discolour the PCB around them.
My statement was:

But the pass transistor and regulator circuits are no more likley to fail than the crowbar circuit if both are properly designed.
Assuming the heatsink and power distribution is done right, the pass transistor(s) will not operate at an unsafe temperature. hence my point, one circuit is no more likely to fail than the other if both are properly designed.

In reality, the electrolytic cap's life expectancy is many orders of magnitude less than the semis unless the latter are just being abused to death.

The bench supply I built in 1980 is still working with all original components.
 

Thread Starter

Dave_UYZ

Joined Jan 16, 2014
29
Thanks chaps.
Obviously, more work is indicated than I previously thought.

I think I'd feel happier if there IS a sort-of crowbar in it though; if only for the purposes of placating the Hon Treasurer if there's a problem ('oh yes, protection against over-voltage was definitely incorporated')

Anyone know how long it takes to blow a blade fuse ?
 

ian field

Joined Oct 27, 2012
6,536
Thanks chaps.
Obviously, more work is indicated than I previously thought.

I think I'd feel happier if there IS a sort-of crowbar in it though; if only for the purposes of placating the Hon Treasurer if there's a problem ('oh yes, protection against over-voltage was definitely incorporated')

Anyone know how long it takes to blow a blade fuse ?
As long as its *FIXED* 13.8V - tracking crowbar circuits can be tricky devils!

The most important part of any crowbar circuit is the fuse - followed closely by the crowbar thyristor, don't forget it has to handle the peak instantaneous current of dumping the charge in the reservoir caps.
 

Thread Starter

Dave_UYZ

Joined Jan 16, 2014
29
1] Laying hands on a beefy SCR seems a tad tricky [and I thought I had one in my junk box]. Any practical suggestions welcome.

2] I notice that the 723 has two Voltage inputs [one for the internal supply, V+].
Would the stability of the thing be improved if I have a stable 12v into the V+ ? (from, say, a 7812 and a separate supply for it)

Thanks for the help.
 

#12

Joined Nov 30, 2010
18,224
The LM723 datasheet says .01% regulation for changes in its power supply voltage. Providing the chip with a pre-regulator would make this number even smaller, but .01% is pretty small in the first place. 100 uv/v

The 2 power inputs are for the last stage current delivery transistor and "everything else". It isn't always an input. It can be used to connect the collector to a pull-down transistor instead of using that transistor as a voltage follower.

Looking for a crowbar, you're interested in the single pulse maximum current, similar to ratings for rectifier diodes. I don't even know if that spec in on the datasheets for SCR's.

Incomplete because my real life is pressuring me right now. Gotta go.
 
Last edited:

ian field

Joined Oct 27, 2012
6,536
1] Laying hands on a beefy SCR seems a tad tricky [and I thought I had one in my junk box]. Any practical suggestions welcome.


Thanks for the help.
A crowbar thyristor doesn't have to be that beefy - as long as it fails S/C.

Replacing it is still likely to be cheaper than wrecking something expensive connected to the PSU.
 
Top