very low capacitance high voltage and current SSR

Folks,

I'd like to pick your collective minds for a challenging circuit. (at least for me)

I have an application where I need to switch on, 1mS later deliver five AC bursts of 300Vp at 50amps peak. then in less than 1mS later open the relay.

When the contacts are closed, the forward drop should be low to minimize self heating but it is not critical to the signal. tens of volts can be dropped without affecting the outcome (besides heating)

When open the capacitance across the contacts needs to be less than 10pF (!!!) with

Reply to
Mook Johnson
Loading thread data ...

Sounds like a nuclear test. How many times does it have to work? ;)

Sounds like a job for a big thyratron or possibly a mercury wetted relay. There are things you can do with low-capacitance diode bridges at lower current, but 50A is a lot.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

It's very hard with just a FET in series with the load

But, you could put a rectifier bridge towards the load and then a FET on the + and - of the bridge. Then the load only sees the diode capacitance. Put more diode bridges in series to reduce the capacitance if needed

Cheers

Klaus

Reply to
Klaus Kragelund

te.

anti parallel pin diodes?

-Lasse

Reply to
Lasse Langwadt Christensen

Mook - first question: what's the circuit that the burst circuit switches out of (and into) look like voltage wise when the burst circuit is disconnected? If it's less that the +- 300 volt needed for the burst circuit and the ground/common connections are manerly, put 2 diodes in the burster's output and when the need to switch the burster from the other, switch the bursters power rails from positve to negative & negative to positive. A

50 amp diode w/ 10 pf reverse capacitance with several hundred volts may not be hard to find. Easy to be optimistic when some other fellow is doing the work.

Hul

Mook Johns> Folks,

Reply to
Hul Tytus

Mercury-wetted reed relay with an over-driven coil? The 1msec timing might just be feasible, and mercury-wetted relays don't bounce - the mercury film damps the impact.

Contact resistance is low, stable and predictable, and they are good for 10^8 operations.

They do need to be mounted within 15 degrees of vertical. Orientation-insensitive mercury-wetted reeds have been advertised from time to time, but don't seem to be reliably available.

--
Bill Sloman, Sydney
Reply to
bill.sloman

Isn't 50A a bit high for a reed relay?

Reply to
John S

formatting link

shows the contact resistance to less than 70 milli Ohm, which is typical of all the mercury-wetted reeds I've run into. 50A through that is 175W, which is a lot, but the application is specified as handling five short burst of AC.

Sadly, Mook Johnson hasn't defined what he means by "short". He can make a ball-park estimate of the thermal mass of the closed reed contact, and work out how hot it might get during each burst, and how hot it might end up after five such bursts.

Until we know what he means by "short" we can't really know if 50A is too high.

--
Bill Sloman, Sydney
Reply to
bill.sloman

Which reminds me of another mercury relay that is (was?) popular for industrial applications: it's a vertical cylinder, filled with mercury (not just wetted!), with a steel slug floating in it, and a contact on top. A surrounding coil tugs on the steel slug, displacing mercury into the contact. Typical ratings 240/480V 25A and thereabouts. But I doubt they're anywhere near fast enough, let alone the bounce!

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
 Click to see the full signature
Reply to
Tim Williams

Again, if you could tell us what the AC aand load are, we might be of more help...

Offhand, consider IGBTs for high peak loads. Lower voltage drop than MOSFETs (more current density per die area -- cheaper, too!). You do have to be careful to avoid the combination of too low Vge and too high Ic, which leads to desat and huge power dissipation (linear -- short circuit -- conditions!). (MOSFETs do this too, but 'desat' (linear range operation) occurs at higher voltages and lower currents, so it's destructive after milliseconds, not microseconds.)

For a bidirectional switch, you need two, back to back (emitter to emitter), with co-pack diode, and G-E supplied from an isolated gate driver circuit.

You may be able to get away with less, such as if the load doesn't need DC -- in ye olde CRT circuit, the "S correction" capacitors were switched with single MOSFETs to GND (not back-to-back), to maintain a linear sweep over a wide range of horizontal frequencies (31-106kHz). The signal was cap-coupled, so when 'off', the body diode simply charged the capacitor to peak voltage and that was that.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
 Click to see the full signature
Reply to
Tim Williams

Reply to
whit3rd

They system as many tranducers that act as both transmitters and receivers. The relays selectively connect one of these to the 300Vp-p burst driver to fire a shot on only one of them. The others listen to the ring down after the firing.

if the relay contact have enough capacitance the other channels will couple to each other and muddy the signal. The signal coming back is in the 1mV range so anti-parellel blocking diodes are in the running. Finding some that can source 50 amps peek repetitively with low duty cycle and still have low capacitance at near 0V is a challenge.

That why I was thinking interms of something that acts like a switch but has very high blocking impedance at 100ish Khz range that can be put in series with traditional highly capacities SSRs to keep insulation of the small received signals.

I was looking at GDTs. Anyone used these. how quickly do they turn on and off? Can they be used over and over like a spark plug?

Reply to
Mook Johnson

That is another requirement that it has to work in any orientation. :/ Like the idea though.

Reply to
Mook Johnson

The bursts are 5 cycles of a 100Khz signwave AC (50uS). There will be eight of these burst with 2mS spacing (50uS on / 2.5mS = 2.5% duty for

16mS) followed by about 200mS of rest time (open) before doing it again.
Reply to
Mook Johnson

Not that explosive. :) Its and acoustic pulse-echo thingee. it will make your ears ring but not bleed. LOL!

Reply to
Mook Johnson

Why not use more switches:

Connect two switches in series with each transducer, and at the midpoint between the two series switches, connect another switch going to ground. (a "T" configuration)

When the two series switches are OFF, the shunt switch to ground needs to be ON, to improve the isolation.

Even if there is a large capacitance in each series switch, the capacitive leakage current across the switch won't be able to cause a big signal across the low resistance of the shunt switch, so it will be blocked effectively.

Each of the three switches probably has to be composed of 2 series-conected MOSFETs in opposite directions, if you need to be able to block voltages of both polarities. Driving the gates will be a pain. You might be able to buy off-the-shelf SSRs based on MOSFETs, but they might be too slow.

Reply to
Chris Jones

Relays are not going to cut it, you don't have the control you are looking for..

I assume you want to start and stop each sinewave at a marked position, like the zero axes for example, of course it may depend on the load you are driving. Driving inductive loads require you to shunt the return, dump the return so that it does not burn things out.

You need a full bridge circuit that can be biased from your signal source of interest with a close loop front end.

You'll want 100 amp capable bridge with current limiting.

I did a magnetic sweep amp using a quasi output config via rows of

20 amp 500 volt fets for each side. The design requires that you don't exceed the watts of each fet, not amps, and lots of heak sinking..

our heat sinks are submerged in glycol that are heat exchanged from a chiller.

Jamie

Reply to
M Philbrook

Jamie hasn't been paying attention. The relays are to be closed about 1msec before the pulse sequence starts, and opened about 1msec after it ends.

The relays aren't there to provide control, but rather to provide isolation between bursts.

In fact the mercury reeds would see a long term power dissipation of about 27mW, in bursts of about 0.4W every 216msec. Doesn't sound too bad.

--
Bill Sloman, Sydney
Reply to
bill.sloman

Before reading the responses, my knee-jerk reaction ("answer") is: why not try the diode sampling bridge circuit?

Reply to
Robert Baer

The bridge would need over 50A through and voltage compliance of over

300V to switch the required signals. You need a forward DC current exceeding the signal current in 'on' state, and a reverse voltage exceeding the peak voltage of the signal voltage in 'off' state to switch with diodes.
--

-TV
Reply to
Tauno Voipio

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.