mosfet inverter.

[brad]

I'm kinda partial to cmos inverter circuits, and ever since seeing the discrete opamp, I can't help but wonder if the same can be done with an inverter.  I'm an electronics luddite, but isn't this basically what's going on inside a chip like a 4049?



Is there much more that would be needed before it would be possible to use it in place of a single inverter from a 4049?

[puretube]

[Peter Snowberg]

Please do remember that when making ICs you have the option to almost perfectly match complementry devices. You can't do that with discretes unless you buy 1000 and spend all day matching a pair. It won't perform perfectly, but that may be a benefit.

Try looking at the output stage of small to mid sized MOSFET audio amps.

[puretube]

:?:

[Joe Davisson]

You can do that using complemetary parts (BS170/BS250). Works with other transistor types also, such as bipolar/JFET. Mixing types should also work, but the output might not be centered.

[gez]

Stay tuned, schematic coming...

[gez]

I came up with this for a fellow forum member a year or so ago but never got round to trying it (and I don’t think he ever did):



The 22u cap decouples the input/output and keeps crap from the rails reaching the input.  The trimpot gets round the problem of unmatched MOSFETs, though the transfer function will probably be asymmetrical (not necessarily a bad thing) even if you bias to half supply.

The 1M pot is in parallel with the ‘infinite’ impedance of the input and so sets zin.  The 100k is in parallel with the output impedance of the set-up and therefore loads it to a lesser or greater extent.  Unfortunately, if you make this resistor too big the thing will take ages to bias up.

The (minimum of) 1k source resistors help keep current consumption down and prevent you exceeding max power ratings (might be able to do without them with power MOSFETs and mains supply).  

Never tried the above, but am interested to hear from anyone who does.  Should imagine that gain will be higher than an IC inverter.  Who knows how it will clip though?!

[R.G.]

I've done this for years. No particular advantages other than getting your hands on the internal notes to muck about, and you can do that with the 4007, which is cheaper.

The BS170/BS250 work fine for that. That's how I always simulate a 4000 series inverter when I need the analog response, not the digital response.

The transfer function is almost the same, with some minor exceptions. The discrete one has bigger thresholds (typicallly 2v instead of half volt) and will do a LOT more current before it burns up.

[gez]

BP, if you're reading this I replied to you but my email bounced.  :?

[brad]

Thanks for the replies!  

I was hoping to be able to make a small board with several discrete inverters on it so I could use it to substitute the IC in circuits such as the Tube Sound Fuzz, Calavera, Three Legged Dog ect.  That is, connect the discrete inverter board directly to the IC socket.  This is just purely to see how (or if) different transistors affect the sound, or maybe diode compression could be added somehow.

To make a discrete inverter board to replace an IC, is it really as easy as setting up the two mosfets and connecting the in, out, +, and gnd to the corresponding pins in the IC socket?  ...because that's what I'm basically seeing in hex inverter datasheets.  For example, this is from a Fairchild 4049 datasheet, and it just looks like the only other components are diodes to protect the transistors.  Or is that zener and trimmer (??) setting the input impedance?

[puretube]

private...