Do any octal 3 to 8 demultiplexers exist?

...

...

So, do you want eight independent one-of-eights, or a one-of-64? You could do that with 4X 74HC154, 4-to-16 decoder. That at least reduces your chip count by half, albiet it's still 96 pins. And, of course, you still have to select only one of them with a 2-to-4, like a 74HC139.

Good Luck! Rich

That be a lot of pins (91 I/O, at least)! An FPGA would be a possibility, though perhaps an expensive one.

--
  Keith

I'm not sure exactly what you're trying to do, but if you're making eight byte-wide addressable output ports you should know there are easier and denser ways of doing it! Microcontrollers with 64 bits of digital output will be the least-pins approach. But simply 8 octal registers will be easier to lay out on a PCB than a bunch of multiplexers.

Tim.

You could get a quad in a cheapish cpld.

Ah, yes. I see what you want.

High pin-count devices aren't generally so specific. I highly doubt you'll find such a thing. ...even a quad.

An SRAM based FPGA would need a PROM programmer, but I suspect that's your only choice, other than SSI. If this is for hobbyists, then package type is going to be a major problem too. High pin-count packages tend to be hobbyist unfriendly.

--
  Keith

There ya go - use a pic!

There's a couple of them with more than 64 I/Os.

Cheers! Rich

Rich Grise wrote in news: snipped-for-privacy@example.net:

I'd rather eight 1 to 8 demuxes that had their control bits tied together - so essentially that'd be like a 8 channel 1 to 8 demux. If I'm not being clear - I want to switch an 8 bit parallel signal to 8 different 8b buses.

-Michael

Ok, what about tri-state line drivers, with the inputs tied together and OE controlled by a 3-8 decoder? I believe you should be able to get at least dual-octal drivers. ...perhaps quads. It's more pins (and more input capacitance), but the parts might be more readily available.

--
  Keith

Then you don't want demuxes at all - except the one to select your port. Splitting up an 8-bit port into 8 ea. 8-bit ports using demuxes would be an absolute nightmare to wire up/lay out.

If you need them latched, use 8X HC573; if they don't need to be latched, then 8X HC244. That's the way they've been doing output ports since time immemorial. If you need them bidirecitonal, use HC245's.

Or, use one of those mongo PICs:

Good Luck! Rich

--
No.

Keith Williams wrote in news: snipped-for-privacy@news.individual.net:

Yes it sure would be, and even more pins if I have to use 8 individual chips! I should mention though, that I only need a chip with common control lines - so really the ideal chip would only have 8 + 64 + 3 + 2 = 77 pins (input, output, control, power). Honestly though I realize there's little to no hope in finding such a thing, so I'm hoping to just find something like a dual or even a quad to make things a bit simpler. I'd rather not use an FPGA, as, as you said - it's a more expensive option, and also the project I'm working on is intended to eventually be released to the public (open source and all that nonsense) - and I don't think the average joe has access to an FPGA programmer.

-Michael

Yep. Even assuming no shared pins, it should fit in a hand solderable TQFP package.

Marc

"Tim Shoppa" wrote in news: snipped-for-privacy@g14g2000cwa.googlegroups.com:

Well OK - so I've been working on a microcontroller based servo controller. Right now I just use an 8 bit port and software pwm, giving me 8 motors that can be controlled. But I got to thinking about it - the PWM signal you feed a servo has less than a 10% duty cycle, and a constant frequency. Thus, if you were to put a 1:8 demux on a single pin from the microcontroller, if you were very careful with timings, you could control 8 servos with just ONE pin (technically you could control 10, but the timings would be getting close enough there that I wouldn't push it, and also that'd be another control bit, so 8 it is). Thus, if you were to use a full port - you could control sixty four servos! The software that I've written will require just a couple very minor modifications to handle the additional pins and the multiplexer, so mostly I just want to try it out for s**ts and giggles, but I also think that this could be developed into a very small, very effective servo controller if I could find the right parts.

-Michael Noone

You may want to check out the cheap and available 74HCT259 3 to 8 demultiplexer chips (even many Radio Shack stores still carry them). Sounds like you would like to send pulses to a servo to get it positioned, then stop the pulses so it stays positioned, send pulses to the next servo to position it, etc. If you have 8 bit control (like the data pins on the parallel port), you could use 3 of the 8 bits for input to fifteen of the 259 chips and use two more of the chips with another 4 bit input to two more of the

259 chips for fifteen output lines to turn on/off the other fifteen chips. This would give you 120 control lines for the servos. I haven't tried this with servos, so YMMV. Below are some pages where I've used the 259 chips and servos for a webcam.

Sounds familiar. I thought about demultiplexers, but favored a PIC only approach. At a dollar and a half per piece, it's a great price. And if you take the 16F628, you can use 14 pins out of 18 to control servos. The other four pins are power, RxD and one input-only pin that I use to configure the baud rate. Then you can chain as many as you like via a single serial communication wire.

What about building a bus with 8 octal tri-state drivers?

Thomas