Author |
Message |
widdly
Joined: Jun 25, 2007 Posts: 268 Location: singapore
G2 patch files: 2
|
Posted: Sat Nov 08, 2008 1:05 pm Post subject:
|
|
|
The opamps are inverting...so two in row will be non-inverting. Last edited by widdly on Thu Aug 05, 2010 10:04 pm; edited 1 time in total |
|
Back to top
|
|
|
gogmagog
Joined: Nov 08, 2008 Posts: 19 Location: Nashville, TN
|
Posted: Sat Nov 08, 2008 1:16 pm Post subject:
|
|
|
Thanks for the speedy reply, and sorry for digging up an old thread (didn't notice the date until it was too late).
Anyway, I'm still wondering what the practical purpose of using two inverting opamps in this scenario is versus using one non-inverting opamp. I'm thinking of implementing a mixer like this, but on a much larger scale as a patchbay... |
|
Back to top
|
|
|
blue hell
Site Admin
Joined: Apr 03, 2004 Posts: 24079 Location: The Netherlands, Enschede
Audio files: 278
G2 patch files: 320
|
Posted: Sat Nov 08, 2008 1:39 pm Post subject:
|
|
|
gogmagog wrote: | sorry for digging up an old thread (didn't notice the date until it was too late). |
That's ok, we do it all the time
aboard. _________________ Jan
also .. could someone please turn down the thermostat a bit.
|
|
Back to top
|
|
|
blue hell
Site Admin
Joined: Apr 03, 2004 Posts: 24079 Location: The Netherlands, Enschede
Audio files: 278
G2 patch files: 320
|
Posted: Sat Nov 08, 2008 1:41 pm Post subject:
|
|
|
This is not even an old thread it's from June, this year even _________________ Jan
also .. could someone please turn down the thermostat a bit.
|
|
Back to top
|
|
|
blue hell
Site Admin
Joined: Apr 03, 2004 Posts: 24079 Location: The Netherlands, Enschede
Audio files: 278
G2 patch files: 320
|
Posted: Sat Nov 08, 2008 1:52 pm Post subject:
|
|
|
gogmagog wrote: | Anyway, I'm still wondering what the practical purpose of using two inverting opamps in this scenario is versus using one non-inverting opamp. |
The first opamp is a summing node, the non inverting buffer would not be suitable for that. Penalty is that an extra opamp is needed to reverse the phase. I just saw that the Ken Stone circuit has an interesting feature, the bipolar/unipolar switch ... but maybe Moog has that as well. _________________ Jan
also .. could someone please turn down the thermostat a bit.
|
|
Back to top
|
|
|
gogmagog
Joined: Nov 08, 2008 Posts: 19 Location: Nashville, TN
|
Posted: Sat Nov 08, 2008 2:20 pm Post subject:
|
|
|
Ok, thanks for the explanation. Looks like I need to hit the books some more |
|
Back to top
|
|
|
blue hell
Site Admin
Joined: Apr 03, 2004 Posts: 24079 Location: The Netherlands, Enschede
Audio files: 278
G2 patch files: 320
|
Posted: Sat Nov 08, 2008 2:34 pm Post subject:
|
|
|
gogmagog wrote: | Ok, thanks for the explanation. Looks like I need to hit the books some more :-( |
You want that 2nd opamp out?
Guess you could make the ingoing buffers inverting, that way the outgoing inversion would be a feature really. _________________ Jan
also .. could someone please turn down the thermostat a bit.
|
|
Back to top
|
|
|
gogmagog
Joined: Nov 08, 2008 Posts: 19 Location: Nashville, TN
|
Posted: Sat Nov 08, 2008 3:40 pm Post subject:
|
|
|
Not really so concerned about having a second opamp per output, so much as I am about the fact I don't know why it needs to be inverting in order to function as a mixer. |
|
Back to top
|
|
|
blue hell
Site Admin
Joined: Apr 03, 2004 Posts: 24079 Location: The Netherlands, Enschede
Audio files: 278
G2 patch files: 320
|
Posted: Sat Nov 08, 2008 4:04 pm Post subject:
|
|
|
Ok ... well let me try to explain ... the inverting summer has a special property in the minus (-) input of the opamp being at virtual ground (due to the feedback regulating the - input to the same level as the + input), this results in the inputs being virtually short circuited with no possibility of the signal from one input going back into another. When you'd try to sum on the + input you'd not have that feature and signals would influence each other back through the matrix in this case. _________________ Jan
also .. could someone please turn down the thermostat a bit.
|
|
Back to top
|
|
|
blue hell
Site Admin
Joined: Apr 03, 2004 Posts: 24079 Location: The Netherlands, Enschede
Audio files: 278
G2 patch files: 320
|
|
Back to top
|
|
|
blue hell
Site Admin
Joined: Apr 03, 2004 Posts: 24079 Location: The Netherlands, Enschede
Audio files: 278
G2 patch files: 320
|
|
Back to top
|
|
|
gogmagog
Joined: Nov 08, 2008 Posts: 19 Location: Nashville, TN
|
Posted: Sat Nov 08, 2008 4:28 pm Post subject:
|
|
|
Aaaaah! Now I get it. The negative feedback essentially cancels out the signal behind the opamp, yes? I had actually wondered about input signals bleeding into each other, but was ready to ignore the consequences, since I had seen some patchbay designs (like this one: http://monopole.ph.qmw.ac.uk/~thomas/synthdiy/patchpanels.htm) ignore it too.
Thanks a bunch for clearing that up for me. |
|
Back to top
|
|
|
doctorvague
Joined: Mar 14, 2007 Posts: 281 Location: new mexico
Audio files: 2
|
Posted: Sun Nov 09, 2008 6:55 am Post subject:
|
|
|
Thanks Jan. That answers some questions for me too. Appreciate your input on things like this - I've learned a lot from you - in bits in pieces
Cheers
Phil _________________ doctorvague's youtube channel |
|
Back to top
|
|
|
neandrewthal
Joined: May 11, 2007 Posts: 672 Location: Canada
|
Posted: Sun Nov 09, 2008 10:50 am Post subject:
|
|
|
So, now that everyone is happy, can I ask a question?
Can I build a matrix mixer using these CGS DC mixers (with the optional attenuverting circuitry shown in red)? The official CGS matrix mixer has input buffers to "avoid loading your output" I'm not too sure what this means, but I'm wondering if I can get rid of them for simplicity's sake since I will be need 12 of these circuits for my intended project. Getting rid of the buffers would save me 32 op-amps
_________________ " I went through quite a few trannies til I found one I liked" - Wild Zebra |
|
Back to top
|
|
|
urbanscallywag
Joined: Nov 30, 2007 Posts: 317 Location: sometimes
|
Posted: Sun Nov 09, 2008 12:21 pm Post subject:
|
|
|
For each input you connect a single output to, the input impedance goes down because the input resistors are in parallel. This happens any time you use a multiple module or connect 1 banana output to several inputs.
With a matrix mixer each input channel is always connected to N inverting summers, so the input impedance goes down quite a bit. Assuming 100kohm input resistors the effective input impedance is:
2 channels: 50kohm
3 channels: 33kohm
4 channels: 25kohm
5 channels: 20kohm
6 channels: 16kohm
7 channels: 14kohm
8 channels: 13kohm
What this means is that the signal that is input into the matrix mixer will start to droop. Assuming 100kohm input and 1kohm output resistors:
1 channel: 1% error
2 channels: 2% error
3 channels: 3% error
4 channels: 4% error
5 channels: 5% error
6 channels: 6% error
7 channels: 7% error
8 channels: 7% error
For audio the error is not too significant, less than 1dB attenuation. But for CV, especially accurate 1 V/octave signals you will have tuning issues.
|
|
Back to top
|
|
|
neandrewthal
Joined: May 11, 2007 Posts: 672 Location: Canada
|
Posted: Sun Nov 09, 2008 12:42 pm Post subject:
|
|
|
urbanscallywag wrote: | For each input you connect a single output to, the input impedance goes down because the input resistors are in parallel. This happens any time you use a multiple module or connect 1 banana output to several inputs.
With a matrix mixer each input channel is always connected to N inverting summers, so the input impedance goes down quite a bit. Assuming 100kohm input resistors the effective input impedance is:
2 channels: 50kohm
3 channels: 33kohm
4 channels: 25kohm
5 channels: 20kohm
6 channels: 16kohm
7 channels: 14kohm
8 channels: 13kohm
What this means is that the signal that is input into the matrix mixer will start to droop. Assuming 100kohm input and 1kohm output resistors:
1 channel: 1% error
2 channels: 2% error
3 channels: 3% error
4 channels: 4% error
5 channels: 5% error
6 channels: 6% error
7 channels: 7% error
8 channels: 7% error
For audio the error is not too significant, less than 1dB attenuation. But for CV, especially accurate 1 V/octave signals you will have tuning issues.
|
Would a trimpot in series on one side of the pot help to "center" it? I don't care if the scaling is 4% off, as long as the "off" position is at 12 o'clock on all the pots. Maybe center detent pots would get rid of this issue? _________________ " I went through quite a few trannies til I found one I liked" - Wild Zebra |
|
Back to top
|
|
|
urbanscallywag
Joined: Nov 30, 2007 Posts: 317 Location: sometimes
|
Posted: Sun Nov 09, 2008 3:01 pm Post subject:
|
|
|
Well I think you could up the gain a little bit by increasing the 100k* resistor between nodes A and B. What is the output resistance of your other modules? If its 1kohm like I assumed above and you have a 4 channel mixer you could add a 4kohm resistor in series with 100k*. If the output resistance of each of your modules isn't 1kohm this trick won't work. That's where the input buffers come in. |
|
Back to top
|
|
|
neandrewthal
Joined: May 11, 2007 Posts: 672 Location: Canada
|
Posted: Sun Nov 09, 2008 3:47 pm Post subject:
|
|
|
Yeah, my output impedance is 1k for just about everything. So, all I do is use 101k for channel 1, 102k for channel 2, 103k etc...?
Sounds easy. I can build 8 separate adding/subtracting mixers that cascade into a big matrix with only 4 TL074's
But, what if I happen to plug in a module with a 10k impedance?
None of this would change the center point of my pots, would it?
If I have a 4% drop on channel 4 and it just means that -5-+5 signals are now -4.8-+4.8 and 0-10 signals are now 0-9.6 that wouldn't bother me the slightest bit, as I'm building a separate mixer specifically for v/oct voltages _________________ " I went through quite a few trannies til I found one I liked" - Wild Zebra |
|
Back to top
|
|
|
urbanscallywag
Joined: Nov 30, 2007 Posts: 317 Location: sometimes
|
Posted: Sun Nov 09, 2008 5:13 pm Post subject:
|
|
|
OK looks like I have caused some confusion above, let me try and straighten things out. The error will be constant across all channels of the matrix mixer, but the error will depend on how many channels your mixer is. If you make a 4 channel mixer only look at that row in each of my "tables".
So hopefully now you've got the idea that the "error correcting" resistor should be constant across all channels of the matrix mixer.
I don't think you'll have any modules with a 10kohm output impedance, but maybe a designer could choose 100ohm instead of 1kohm. In this case the errors will all a magnitude smaller, 1% -> 0.1%, 2% -> 0.2%, and so on. So if you added a 4kohm resistor in series with the 100kohm resistor you'll have created a new error by adding 4% more gain for sources with 100ohm output resistors.
Your droop voltage numbers look correct. I'm not sure a separate matrix mixer for 1V/octave signals will help any unless you have buffered inputs. It will only really be a problem if you have a signal from a keyboard, sequencer, or MIDI -> CV and you have the mixing pot all the way clockwise (for no attenuation) in which case you wouldn't have correct scaling. |
|
Back to top
|
|
|
neandrewthal
Joined: May 11, 2007 Posts: 672 Location: Canada
|
Posted: Sun Nov 09, 2008 5:40 pm Post subject:
|
|
|
Ok, thanks a bunch. Makes sense now.
I meant I'd build a regular mixer for mixing V/oct, just 2 in 1 out, not a matrix. Also, I already have a CGS cascade mixer, which is a matrix mixer for v/oct CV's _________________ " I went through quite a few trannies til I found one I liked" - Wild Zebra |
|
Back to top
|
|
|
urbanscallywag
Joined: Nov 30, 2007 Posts: 317 Location: sometimes
|
Posted: Sun Nov 09, 2008 5:42 pm Post subject:
|
|
|
|
|
Back to top
|
|
|
Pehr
Joined: Aug 14, 2005 Posts: 1307 Location: Björkvik, Sweden
Audio files: 2
|
|
Back to top
|
|
|
noisedeputy
Joined: Nov 15, 2008 Posts: 17 Location: Boston, MA
|
Posted: Sun Nov 16, 2008 8:49 am Post subject:
|
|
|
so is this just 8 ins and 8 outs?
I want to make something like the aux pander, with 8 fx sends. |
|
Back to top
|
|
|
vtl5c3
Joined: Sep 08, 2006 Posts: 425 Location: PDX
Audio files: 13
|
Posted: Sun Nov 16, 2008 6:24 pm Post subject:
|
|
|
That's how I built it, but you can really do any number of inputs and outputs. Not sure what the aux pander is, but I built it thinking of it as an 8 way effect send, where any of the 8 inputs could be sent to any of the 8 outputs in varying amounts.
noisedeputy wrote: | so is this just 8 ins and 8 outs?
I want to make something like the aux pander, with 8 fx sends. |
|
|
Back to top
|
|
|
fonik
Joined: Jun 07, 2006 Posts: 3950 Location: Germany
Audio files: 23
|
Posted: Wed Jan 21, 2009 8:16 am Post subject:
|
|
|
urbanscallywag wrote: | OK looks like I have caused some confusion above, let me try and straighten things out. The error will be constant across all channels of the matrix mixer, but the error will depend on how many channels your mixer is. If you make a 4 channel mixer only look at that row in each of my "tables".
[...]
I don't think you'll have any modules with a 10kohm output impedance, but maybe a designer could choose 100ohm instead of 1kohm. In this case the errors will all a magnitude smaller, 1% -> 0.1%, 2% -> 0.2%, and so on. So if you added a 4kohm resistor in series with the 100kohm resistor you'll have created a new error by adding 4% more gain for sources with 100ohm output resistors.
[...]
I'm not sure a separate matrix mixer for 1V/octave signals will help any unless you have buffered inputs. |
i read thru the thread. and i was thinking of the following: an inverting opamp buffer at each input, and an inverting opamp mixer at each output row - this way i would have buffered inputs (no error), a non-inverting matrix mixer and lower part count. with selected resistors it should be suitable for keyboard voltages too (but what different keyboard voltages should one want to mix?).
an inverting input buffer would actually be the same thing as an inverting mixer with just one input _________________
cheers,
matthias
____________
Big Boss at fonitronik
Tech Buddy at Random*Source |
|
Back to top
|
|
|
|