High Voltage Analog Switch ASIC

Who could make this? I'm looking for +/-100V switching with logic level control and maybe 32x32 crosspoint array.

Any ideas on cost?

Cheers

--
Syd
Reply to
Syd Rumpo
Loading thread data ...

l

The closest I know about would be some Supertex analog switches & multiplexers. No crosspoint arrays that I know of. Someone else (Allegro???) has a smaller competing line of products.

Reply to
cassiope

Thanks, Supertex parts are good, but I'd need quite a few which is why I'm considering an ASIC or some other custom solution. I'll have a look at Allegro.

--
Syd
Reply to
Syd Rumpo

On a sunny day (Thu, 16 Feb 2012 17:40:24 +0000) it happened Syd Rumpo wrote in :

Switching speed? Bandwidth? Numnber of swicthes per second?

Use relays if slow.

Reply to
Jan Panteltje

1024 relays? No thanks, Jan. Size is important or I'd use lots of COTS parts. Bandwidth, say 20MHz. Switching speed, say
Reply to
Syd Rumpo

Besides Supertex take a look at X-Fab. Doing a chip right now that'll run your required bandwidths with ease but it's only +100V. Not sure if they have a process extension that lets you go negative as well, we just didn't need that.

Using an external design team? From white sheet to tested and bumped/bonded samples you'll be looking at $250k to $500k, depending on your volume commitment and whether you let the design house handle procurement.

Start writing the spec, that's going to be your living document and will become part of the contract.

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

Thanks, Joerg, that's a good starting point.

Cheers

--
Syd
Reply to
Syd Rumpo

On a sunny day (Thu, 16 Feb 2012 21:52:02 +0000) it happened Syd Rumpo wrote in :

Let us know the part count and power consumption once you have an all semiconductor solution. Pulsing bistable relays could have a very low part count, also easy to replace (sockets). Almost like driving ringcore memory ;-)

Reply to
Jan Panteltje

semiconductor solution.

(sockets).

Yes, a place I worked at a while ago used them in thousands, they're great in the right place, and the quiescent current is good :-)

But not here - speed, vibration, size, cost.

Cheers

--
Syd
Reply to
Syd Rumpo

We are using their XH035 process. But Supertex might be a better fit for you, especially if your yearly quantities are not in the high five-digits and maybe you are even willing to share the IP with them.

Another piece of advice: While X-Fab and Supertex are very responsive don't be too dismayed if others in the high voltage arena are not. Some companies won't really talk to you unless your are from General Motors or so.

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

On a sunny day (Thu, 16 Feb 2012 21:52:02 +0000) it happened Syd Rumpo wrote in :

Been thinking about this. It depends a lot on the application. One solution for *some* cases would be to attenuate the input swing to say +4 and -4 V, use common 74HC4051 switches, and then amplify back to +100 and -100V. This reduces your signal to noise, but has an number of advantages, one of those that HV amplifiers are easier to make than HV switches. Only needs 32 HV amps... Thinking about this more would require me to charge you :-) It is a fascinating requirement though.

Reply to
Jan Panteltje

and -4 V,

It's an option which has been considered. The signal one way is volts at most but +/-100V the other way, so we could use low voltage switches and multiple drivers, but we'd need an impractical number of high speed, high voltage drivers. 32x32 is suggested as a useful size for a single chip crosspoint array, but we'd actually need many of these.

Cheers

--
Syd
Reply to
Syd Rumpo

On a sunny day (Sun, 19 Feb 2012 09:22:22 +0000) it happened Syd Rumpo wrote in :

There is one option that is cute and bidirectional, but does not meet your timing requirement I think: Use CdS light dependent resistors 'LDR's at the cross points, lighted by some bulb each (or a white LED?). Those work fine at a few hundred volts, have a good on / of resistance ratio, but are slow. All depends on the application!

Reply to
Jan Panteltje

There is one option that is cute and bidirectional, but does not meet your timing requirement I think: Use CdS light dependent resistors 'LDR's at the cross points, lighted by some bulb each (or a white LED?). Those work fine at a few hundred volts, have a good on / of resistance ratio, but are slow. All depends on the application!

And to expand on that LDR idea, you do not need 32x32 light bulbs or LEDs, you could use a good LCD monitor with back light, one that is advertised as

10,000 : 1 contrast ratio, and draw a checkerboard of 32 x 32 on it, a LDR on each square. Use some honey comb grid (old floor racks?) to prevent cross-lights. Now it is a software problem to light each square when needed :-) Can do that with a few lines of C code... LOL Now how is that for taring up the thinking box? Gonna give myself an award for this. :-)
Reply to
Jan Panteltje

bulb

10,000 : 1 contrast ratio,

Jan, you need help :-)

But just to show that you're not the only nut-case round here, how about a grid of 32 wires crossed by a perpendicular grid of 32 wires separated by a small distance and enclosed with a low pressure noble gas. Strike a low impedance gas-discharge path where needed, like a Neon or similar

- I only need AC signal connectivity so the DC could be blocked.

Anyway, this needs to fit in a smallish space and run much faster that an LDR could do.

Cheers

--
Syd
Reply to
Syd Rumpo

On a sunny day (Sun, 19 Feb 2012 17:37:51 +0000) it happened Syd Rumpo wrote in :

bulb

10,000 : 1 contrast ratio,

Sounds good, but would it not be noisy? Gas-discharge paths are not very smooth things perhaps?

Reply to
Jan Panteltje

bulb

Use green LEDs, CdS response peaks around 520nm

--
?? 100% natural

--- Posted via news://freenews.netfront.net/ - Complaints to news@netfront.net
Reply to
Jasen Betts

On a sunny day (Sun, 19 Feb 2012 17:37:51 +0000) it happened Syd Rumpo wrote in :

Sounds good, but would it not be noisy? Gas-discharge paths are not very smooth things perhaps?

The other thing is, that for that to work, you need an ignition voltage of > 200V in case of +-100V signal. This then requires +/-300V drive signals. For a lower ignition voltage (ionisation voltage I should say) the signal itself will switch the crosspoints on. As to the LDR LCD thing, maybe OLED would be better, as OLEDs have a much higher contrast ration (OLED can be 100% off) . Using small LDRs (5 mm diameter) 32 makes 16 cm square OLED module... Of course LEDs on a small PCB would do too, wonder if you could multiplex those with a high frequency so as to make use of the slowness of the LDRs :-)

Reply to
Jan Panteltje

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.