Internal Circuit Breaker?

[exclamationmark]

I'm in sort of an unusual situation with regards to safety in one of my 'creations'. It's basically a rackmount pre-amp powered from a 48v SMPS laptop supply. However, this is converted to 100v and -100v rails internally. The main thing I am concered about is the possibility of the chassis becoming live with either of these voltages. While probably not innately lethal, I don't really appreciate little love shocks from my equipment.

Normally the chassis in amps with power transformers is earthed, so that if it becomes live, the mains breaker (and whatever poor fuses in the amp) trips. Grounding the chassis and sticking a fuse in series with the power supply probably wouldn't work in case, as I'm assuming the SMPS's thermal shutdown or internal fuse would trip before my fuse ever did (I'm using something like 0.8A of the 1.3A the SMPS is rated for), probably destroying it in the process. Also, my -100v rail only supplies roughly 50mA maximum from a rather expensive DC-DC converter, which I would very much like to keep alive.

My first question is, would grounding the chassis turn it into essentially a hum receiving device? The entire pre-amp is star grounded at the power input (there's literally like 20 wires attached here - hum free).

The second rather broad question is, how exactly do I solve this problem? I know that no current is supposed to flow in the ground to chassis connection normally, so could I somehow develop a circuit breaker that disconnects power when it detects a current is flowing in the chassis?

[kg]

i assume the smps is grounded via a third prong? and that smps is isolated from chassis and signal ground?

one thing you can do is put a couple of high current diodes in parallel, facing opposite directions, between chassis and signal ground.

voltages less than ~0.7v will be open circuit, voltages higher than that will be clamped.

experiment with bypassing with a cap.

[trobbins]

The main thing I am concered about is the possibility of the chassis becoming live with either of these voltages.

Are you saying the DC +/-100V rails have a 0V reference, and the 0V reference is taken to the star chassis point? If so, then grounding isn't an issue - the DC supply is correctly grounded (touching the + or - 100V rails will always be dangerous).

[J M Fahey]

the chassis becoming live with either of these voltages.

Live relative to *what*?
The danger in line powered equipment is that (sad heritance of yesteryear) one pole is grounded, in fact very well grounded, *you* are grounded by definition (unless you are a mosquito or hummingbird) so touching in any way the other, "live" pole places full line voltage across two points your body, usually from one arm to your feet or from arm to arm, *very* dangerous.
And the live pole may be the big, present everywhere, easy to touch chassis, to which you will usually be connected through guitar cable ground.
But if you ground this particular preamp's chassis, only dangerous points are the +/-100V rails, which are internal, hopefully not easily accessible, much harder to touch accidentally.
You will also be grounded to it, meaning your body has 0 voltage differential with it.
In a nutshell, although the physiological effects of 100V DC and 120V AC must be roughly equivalent, your situation is definitely not as dangerous as having a 120V live chassis.

[exclamationmark]

The SMPS is one of those laptop bricks (three pronged, however) and is connected to the pre-amp via one of those 2.5mm lugs. At this point it only has the positive terminal and negative terminal (of which I've taken as the reference for ground). It's then converted to the respective voltage rails.. So in a worst case scenario, if for some reason the B+ or bias rail (100v or -100v) comes loose (no doubt due to my abysmal soldering skills) and comes into contact with the case, I'm assuming the case would be at a +/-100v potential to ground (ie, the point where everything is star grounded). Wouldn't that make the now-live case +/- 100v in reference to me (assuming the SMPS star-ground is the same as earth potential)?

I get a little bit confused by the whole notion of references. Could I market a battery that is thousands of volts above earth ground, yet has only 9 volts between terminals as a 9 volt battery?

[trobbins]

Probably best to completely describe, or show a schematic of what happens after the laptop smps DC output. What voltage is it, and how do you then get to + and - 100V rails, and what is the 100V rails powering that could make it to the outside of the box (ie. to a user). Do you bring the mains protective earth in to the equipment, and connect it to the chassis (assuming you are using a metal chassis?).

Ciao, Tim

[J M Fahey]

Yes, I guess you are somewhat confused, but it can be explained.
By "referred to" we mean what you see happening with your multimeter: if you touch some point with one probe, say the red one, by itself it means nothing.
You need to touch the black (common) probe to something else, and the multimeter will show the voltage *difference* between those points.
In the same way, if you touch one point with some part of your body, some current will pass through you if and only if another part of your body is touching some point which is at a certain voltage different from the first one.
In a way, you *are* a multimeter, which instead of a needle or screen goes "ouch" or even "pfffftt" given enough voltage (and current).
What most laymen don't realize is that you are almost always touching ground (that's one voltage) so touching any other voltage will put it across you.
What's important is the voltage *difference* between ground (usually you) and what you are touching.

So in a worst case scenario, if for some reason the B+ or bias rail (100v or -100v) comes loose (no doubt due to my abysmal soldering skills) and comes into contact with the case, I'm assuming the case would be at a +/-100v potential to ground (ie, the point where everything is star grounded). Wouldn't that make the now-live case +/- 100v in reference to me (assuming the SMPS star-ground is the same as earth potential)?

No, if the case is grounded and +100V (or any other voltage) touches it, that voltage will be shorted and become "0".
That's why cases are grounded for safety!!

I get a little bit confused by the whole notion of references. Could I market a battery that is thousands of volts above earth ground, yet has only 9 volts between terminals as a 9 volt battery?

Yes, and consider two possibilities:
1) you are standing on the floor, you touch *any* terminal in that battery, you are zapped by thousands of volts.
2) You are walking on that, say, 9000V carrying wire, circus style, you pull your multimeter and measure that 9V battery, or touch each pole with one hand: you will measure 9V , you won't be zapped by it at all.

[exclamationmark]

This is exactly the problem I have. I know the voltage will be shorted to 0 (the situation I was talking about was a chassis WITHOUT a ground connection), but I have no safety mechanism from stopping my negative rail from being destroyed from drawing max current.

I recently tested my power supply in a short-circuit situation (most laptop supplies have short circuit protection, presumably to stop them bursting into flames) and it made a nice little beeping noise until the short circuit was removed. So, the positive rail being shorted seems to no longer be a problem.

My DC-DC converter for the negative rail, however, only has a max dissipation of 3 watts, so can supply around 30mA before it enters thermal shutdown (ie, melting). While not cheap, I guess it's easily replacable if anything does happen... I don't actually know how much current it will draw at peak in a short circuit situation, so a fast-blo fuse in series *may* be an option.

[trobbins]

! - I find your descriptions and details ambiguous and difficult to appreciate exactly what you are proposing. A picture (ie. circuit schematic) is worth a thousand words.

[J M Fahey]

I don't think you'll find a 30mA fuse (although the NASA might have a few ..... at U$500 each) but it's relatively easy to provide some active protection for your SMPS converter which does not allow more than 30mA to pass .
We'd need to see some schematic though.

[Enzo]

My HP bench DMM has a 32ma fuse in the input leg. Seems to me they cost about $1.50-$2.00 each.

[J M Fahey]

WOW !!!
Never saw one.
I guess the wire must be *real* thin.
Do you have a link to some supplier?
I sure could use a few.
TIA.
EDIT: I mean "real" glass fuses.
I can get some "Polyswitches" in that range, but their time constant is too long and they are not good over, say, 32V or some small value like that.

[Enzo]

These are standard size 1-1/4" glass fuses. SLow blow, in fact. I think the box I have is Littelfuse, but they were not some rare brand.

Go to Mouser and enter "32ma fuse" in the search. They have 1-1/4" as well as 20mm sizes, both fast and slow.

[exclamationmark]

I actually just replaced the DC/DC converter for the negative rail with a high-current inverting boost controller (basically the same as what the positive rail has - however, it requires a proper PCB and handful of external components). Stuck a 0.2A fuse before the converter and shorted the negative rail to the case, and it blew in the blink of an eye. So it seems everything is working as intended now. Kind of a lazy solution though, haha.

[J M Fahey]

Go to Mouser and enter "32ma fuse" in the search.

Thanks a lot.
Always thought that less than 100mA was special order.