Why not use an exponential based resistor ladder instead of linear so that one can get very precise and a large range of resistance from a single digital resistor?
Instead of n*R/(n-k) for the resistance at the kth "step" we have R*2^k.
so the ladder would be
R, 2R, 4R, 8R, 16R, etc..
Then add some logic so that when switching resistance values it occurs smoothly.
This seems much more useful and cost effective than just using a linear set of resistors.
I think this has to do with the design and manufacturing process of the digital pot. It must be a lot easier to fabricate N pcs of same-valued silicon resistors on the die, and then switch them in and out by FET switches. This way, at least you can surely provide pot resistance value that is monotonically changing.
If they have to do it as you suggested, they would have to very tightly maintain the resistance of each portion over wide range of resistance value. I'm not an IC guy, so someone cal tell the real reason.
Atsunori
The advantage of a set of equal-valued resistors is that the resistance versus code thing is inherently monotonic. That's hard to do with 2^n steps.
MDACs are often used in place of digital resistors. They are often monotonic. DACs can be designed to compensate for switch resistances pretty well.
We just announced this:
It uses MDACs and opamps to simulate resistances. The competition mostly uses real resistors and relays, which has monoticity and "smooth switching" problems, not to mention lifetime problems.
I think there are some "audio taper" digital resistors, but they just skew the resistor values some.
John
I've just designed a 10-bit DAC, full-range monotonic, constructed with 3% ratio resistors (monolithic/ASIC).
As for weighted "ladders", I do it all the time to create dB-linear gain controls.
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
Stormy on the East Coast today... due to Bush\'s failed policies.
Why would that be? Linear DAC R/2r resistance ladders are far too useful (and so very easy to build, in the IC or externally) to quit using. Then just feed that to a log or exponential function to get whatever you want.
Got a quick circuit for resistance "simulation"? I've been trying to simulate something similar but I'm not having much luck. I ended up with switched resistors and caps.
One basic idea is...
ftp://66.117.156.8/Fake_Resistor.JPG
The effective resistance is Rs (the only resistor) divided by the gain of the MDAC, which is usually in the range 0..1. This is unipolar only. Other circuits can simulate bipolar resistance.
This is the simplified circuit. It needs more parts for loop stability and may need a voltage divider ahead of the first opamp, depending on the input voltage range. Then of course the resistance of the divider has to be factored in. The thing we wound up with on our VME module (floating, full bipolar, multiple ranges, protections) got to be quite complex.
John
That's similar to the one I was playing with but I couldn't get the simulator to converge. I'm pretty sure their FET models are junk (the whole simulator is junk, but it's what I have).
I think I could get around bipolar operation and my voltage needs are quite modest. I'd just use a programmable pot (the previous design used them) but they're too high resistance and won't take any power. A "fake cap" would also be interesting. I'll play some more when I get a little time.
LT Spice is good and free.
You can do a programmable fake cap, or an inductor, with an MDAC.
John
Its UI is pretty bad. I may have to go that way though.
Yes, looking at the "fake resistor" it's clear these are also possible, though the inductor would be a bit more difficult, it seems.
I hate learning new user interfaces, but I got used to this one pretty quick.
The cap can be done by applying +IN to one end of the cap, and driving the other end with a scaled version of +IN, namely amplified through an MDAC path. If you drive the low side of the cap in phase with the input (ie, the mdac path doesn't invert) the gain reduces the effective C. If you drive it out of phase, the effective C is increased, sort of a Miller effect.
John
Would these digitally simulated components function in both polarities, non-polarized, and would these same strategies be easily adapted to be non-polarized?
occurs
I see no problem. A little trickier design for 4 quadrants at power. They should handle AC components very much like the implementing circuits handle AC components. Single battery radios work, so do old style split supply analog circuits. Or for that matter variable frequency motor drives.
I tried learning LT Spice but had to get some work done so went back to the crap. The LT Spice schematic editor, in particular, is horrid.
Yes, I see. That's even "simpler" that I was envisioning, given the "fake resistors" I've found. Perhaps I'll have time to play soon. Maybe I'll even give LT Spice another try.
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.