Multiplying Resistance

[Yoa01]

Hey guys.

So I am working on building a filter (yeah, I know, I'm experimenting), and it calls for a 10M resistor. I know these exist, but I sure as hell don't have something with such a resistance. However, I do have a 100k resistor. Is there any way I can multiply the number of ohms? I know they add, but I don't have enough resistors or room for that.

I've heard of people using FETs to multiply resistance, but no one actually says how that would work, and knowing how FETs work I can't see much of a way to do it.

Or, should I just buy a 10M resistor? That would be easier, of course, but I figure knowing how to multiply (and divide) resistance is a good skill to know, as most skills are.

 

[(*steve*)]

I'd just go out and get the resistor.

However a 10M resistor sounds awfully huge. Did you get the design from somewhere, or did this come from your own calculation.

The problem is that you are likely to find that other impedances are way lower than 10M and this may cause problems if the 10M is in series or parallel with them.

 

[MrEE]

Just a guess here, not knowing the details of your circuit, but if you are dealing with an RC filter, then you may want to increase the value of C (say x 10) then you can use R=1M instead for the same time constant.

 

[Yoa01]

Well, I'm using this:

Now, looking at it, it looks like a spinoff of a basic RR active High Pass filter, just with the lead capacitor being part of the resonance section (I guess). Looking at it, I really can't tell how it would work because it certainly doesn't look much like any filter I've seen.

To be totally honest, I'm trying circuits like this because I can't get any low pass filter to work other than a simple 3dB RC one, which really acts more like a volume control than a filter (and yes, I did run the signal through a spectral analyser to figure that out).


You'd think a low pass filter would be easy, but NOPE. Either that or I suck with electronics, which is generally also true.

 

[(*steve*)]

What do you mean that the low pass filter acted like a volume control?

What is the impedance of the spectrum analyser (it must be large in comparison with R)?

You should see a halving of amplitude for each doubling of frequency beyond the 3db point of the filter.

 

[Yoa01]

It acted just as a normal pot with a signal going through it: more resistance, lower overall volume. Only difference was that the capacitor made it overall quieter.

I honestly can't tell you that. I used a virtual one on my computer (Virtual Analyser 2011), whose input signal (line in signal) comes from a mixer, which is fed with cables of whatever impedance (also don't know). There's a lot of variables and unknowns to tell you, but basically it's filter output>6' cheap guitar cable>mixer (track volume at 50%, master at 80%/-5dB), output to two coupled 1/8" stereo cables>computer input.

Yeah, if that occurs with my build it's not audible or picked up by my rather shoddy analyser... :/

 

[(*steve*)]

OK, if we assume the input impedance of your PC is around 50k, then the low pass filter should not use a resistor larger than about 5k. Also the signal dricing it needs to have a low impedance.

What is the signal? If it's not a pure sine wave, you'll see other artefacts too.

 

[duke37]

The 10M resistor is to provide negative DC feedback to set the op amp in the center of the supply. You could try a 1M if you have one but this would reduce the Q of the resonance.

Some simulation would be advantageous.

 

[Yoa01]

Yeah, a 5k wouldn't do anything in this situation (tried it, no avail). I may have forgotten to mention I'm building a synthesizer. All I need to work right now is a square wave oscillator (works) and a 12dB variable low pass filter (doesn't even need resonance). Hell, a working Sallen-Key or Butterworth filter would work by this point, but I can't even get that to work.

Ok, how about this: if someone can find a working filter design that can successfully, variably, filter the output of a 555 running on 9v, I would be bloody happy.

And Duke, I actually did try that as it is the highest resistor value I have, yet for whatever reason I got the exact output of the 555 through the filter's output. I may have wired something wrong, sure, but I'm pretty much stuck as that is the case with any low pass filter I build.

PS: sorry, I'm not having the best day today. I don't mean to be rude, you guys are some of the nicest and helpful people I've met.

 

[duke37]

I have had a look at 'Electronic Filter Design' by Arther B. Williams
A simple low pass filter is shown in the top circuit running at 1rad/sec turn over frequency.
For a Butterworth filter, C!=1.414; C2=0.7071

The components have to be scaled to convenient values.
Taking the frequency scaling factor = 1000 (160Hz)
Taking C2 = 100n (1e-7)
Z=0.7071/(1e-7*1e3) = 0.7071e4 = 7k

The gain is one at low frequencies and you do not need to go to the exotic heights of an op-amp, an emitter follower will do as shown in the lower diagram.

To get a variable turn over frequency you will need a twin ganged potentiometer. Try 10k?

 

[Yoa01]

And what's funny is that that's the first time I've seen a filter designed like that. From what I've seen, there's about 10 different ways to make a basic low pass filter. Huh.

So, uh, is it bad that I have no idea what most of that means, nor how to implement it physically? Let's try something simpler:

Ok, standard Sallen-Key lowpass, at least as far as I can tell for 3 sources. Make pin 3 of a NE555P the input, and assume the output has no load. What values can be used where to create a simple 12 dB lowpass? Best design I've seen so far is to have 10nF (.01 uF) caps and 10k resistors (or a 10k dual ganged pot). This design works, but having a 100k pot being my lowest I can't say for sure.

I've never had an opportunity to take any classes on electronics, so I'm pretty much using Google as my classroom.

Also, can we get a Mod to change the thread title? This literally has absolutely nothing to do with multiplying resistors anymore!

 

[duke37]

The circiut you show is the same as the one I showed. Mine was a Butterworth which needed one capacitor twice the other and the two resistors the same.
You do not say what frequency you want for the turnover. If you divide the capacitor values by 10 then the turnover frequency will go up by 10.

If you have a 100k pot, use 1n capacitors.

I could not get your circuit to expand (too big a file?) so cannot read the values.

To make it use your op amp circuit or my emitter follower circuit. The transistor can be any small silicon npn with good gain or a Darlington.

Have you tried a simulation?

 

[Yoa01]

Oh, yeah, you're right, sorry! I was looking at it wrong, terribly sorry. Still though, I've only seen Ladder filters using transistors. Interesting.

Well, as I say, I need it to be variable. The resistors will vary as a ganged pot, so I need a kind of catch-all capacitor that can handle everything from zero to max resistance. It doesn't need to be an exact 12dB cutoff (turnover) response, just generally close.

So then, a 1M pot could use 10n caps? I have that setup, I don't have 1n caps.

The resistors are 10k and the caps are 1nF. It's from Wikipedia, just search for Sallen-Key Filter. It is a large image, sorry. Tip: you can open the image in a new tab or window to see it, no matter the size.

I'm sorry, I don't understand the statements about the transistor, do you mean to use the transistor variation instead of the OpAmp variation? I have npn's, and given that I don't know how they really work even after hours of research, this could be a helpful project.

After learning a bit on how to do it, yes. Here is my 555 oscillator (graph in blue) going into the Sallen-Key Filter (graph in green). You can see both the waveforms and schematic:

It's my first time using LTSpice, sorry if it's not what you were looking for :/

 

[duke37]

I have done a simulation in 5spice using a 2N2222. The response is not ideal and deviates a bit when the resistor is high. There is also a strange response in the high audio frequency range.

The top diagram shows the response with resistors from 100k on the left to 10k on the right. You can see that the 100k is a bit high for the transistor circuit as the curve should be at 0dB at low frequencies. This is the Butterworth circuit with one capacitor twice the other.

The lower diagram shows the difference between the Butterwoth circuit and the S&K circuit (which has identical capacitors). You can see that the Butterworth circuit has a better response, holding up longer and then falling quicker.

I also, using another program, looked at an ideal op amp and a 741. The741 showed the a similar high frequency effect as the 2N2222.

 

[Yoa01]

That response is normal, I think. I've seen a few analog filters (namely older ones) have a curve like that, but they were exponential combination filters (like, say, a band reject and a low pass). Thank you for doing that, it's certainly helpful! Wish I had something I could do that with.

So it seems that a transistor-based Butterworth filter is the best based namely on response.

Though testing it in a program made my computer spaz out, I built the actual circuit and tested it with the lowest pot I have (50k). I had to change a cap because, well, I don't have a lot of what few components I do have, but I can say that, for maybe a 3k range, the filter does in fact work! It seems that low resistances create a high frequency shifting sound (over a time that's simply not worth showing with an oscilloscope), whereas high resistances create silence. I'll try to get the correct parts and try it again, but so far it seems to work.

Now, if I could get an OpAmp filter to work, that'd be super. Though I do like the tone of this one.

Thanks!

 

[duke37]

All capacitors should be the same value. The ones you used are very large. All three at 100nF would be a good try.

Using an op amp would hardly be likely to change the response greatly.

 

[Yoa01]

I figured they should be the same, but I only have so much on hand. I will try that, I need to get more 100nF's anyway.

I'm not worried about response with opamps, I would just like a filter using one to work! haha

 

[duke37]

I have done an analysis of your big caps circuit. It mostly attenuates anything in the audio range.

 

[Yoa01]

Certainly explains why I had to turn my amp up.

I've actually found, through my own tinkering and some research, a new filter design. It's based on the KORG MS20's filters, which is generally based on a S-K.

It generally works, but works better when the bottom-right pot is 100k. It has a brighter tone than the transistor filter, but I like it.

 

[Yoa01]

Current flow problems

I must totally suck at electronics. I ask this forum something almost daily.

Ok, so I seem to have current flow problems with my latest audio filter design. Firstly, here is the circuit I have. The red shows what I think is going on, just going by how things are reacting.

I think it acts like that because the waveforms that come out of this look purely like a high pass filter, and turning knobs only seems to change volume (attenuation).

I suppose my question would be 'how can I get this circuit to act like a low pass filter?' I know this circuit works because I have one right next to me in my KORG monotron (I've checked the schematics and board itself, it is most definitely this schematic), yet our circuits are nearly the same (well, aside from the fact that KORG's cutoff control circuit is far longer with many more parts).

(Edited slightly, original includes auxiliary input. Input is from an oscillator, VCF Gate is from a ribbon controller)

Thanks in advance! I swear, I should start paying you people after how nice and helpful you've been.