Buffered mults

[airfrankenstein]

When I saw the thread title, I thought it said: "Buttered Muffs"...

Looking forward to THAT module!

[sinemod]

Ok so what is the good schematics
Does it need 2 op amp ?

and what about a simple inverting

[airfrankenstein]

why would it need two op amps?

[Nantonos]

Here you go:



Place a temporary load, the 100K resistor with the dotted line, on one of the outputs. Put a steady voltage signal into the input. Adjust the potentiometer so that output voltage measured at the top of the load resistor equals the input voltage.

Then place three more temporary loads on the other three outputs, what does the voltage read now.

In other words, that looks like a fine input circuit but shouldn't the outputs on a buffered mult be, well, buffered to isolate each them from the load on that output and on the other outputs.

[daverj]

Here you go:



Place a temporary load, the 100K resistor with the dotted line, on one of the outputs. Put a steady voltage signal into the input. Adjust the potentiometer so that output voltage measured at the top of the load resistor equals the input voltage.

Then place three more temporary loads on the other three outputs, what does the voltage read now.

In other words, that looks like a fine input circuit but shouldn't the outputs on a buffered mult be, well, buffered to isolate each them from the load on that output and on the other outputs.

Adding temporary loads to the other outputs won't change the voltage on the first output. The common point between outputs is the feedback point of the amplifier, which is constant.

However this circuit has a couple of other issues that make it a problem.

1 - If the load of the input you connect to is not the same as the temporary load that you tested with, the voltage will be wrong. And if two inputs connected to two of the outputs have different loads, they will get different voltages.

For example, if you calibrate it using a 100K resistor, but the module you connect it to has a 50k input load, the voltage on the input will be off by about 50mv at full scale (+5v or -5v). Or 100mv if feeding it +10v. That means 59 cents difference for a 5v signal (or 118 cents for a +10v signal)

2 - with 1% resistors the difference between two outputs could be about 1 cent different.

3 - With a gain of x2 and a TL07x amplifier, the worst case input offset voltage could make the outputs be off by 20mv, or about 24 cents, relative to what the input signal was. This will be the same amount for all outputs, and assuming that you are sending the voltages to a 1v/oct input on a VCO or filter, the frequency control of that module can be used to compensate. So this isn't that big of a deal, but something to be aware of.

The best way to do a buffered mult is to have one amplifier as the input stage to buffer the other stages, and then each output have it's own amplifier. Those amps should each have their own gain adjustment and should have the output resistor inside the feedback loop so that the load they are feeding becomes irrelevant.

Ideally they would also each have their own offset trim, though that isn't as important as long as everything they feed has a bias control (such as frequency on a VCO) to correct for offset differences.

It is possible to make one with a single output amp, if you put the protection resistor in the feedback loop and basically make a passive mult on the output. The problem with that is that as you plug additional modules in, if the cable momentarily touches ground it will short out all of the signals. By having separate amps for each output you eliminate that problem.

[roglok]

dave, would it be possible for you sketch this up? i think a schematic for your solution would be greatly appreciated by many...

[Ginko]

Sounds like it would be quite a good DIY kit, if you need a couple of TL074s (to make 1:3) and trimpots, I reckon you could squeeze about 4 buffered mults into a 4hp panel - a stacked pair of PCBs maybe?

[horstronic]

Sounds like it would be quite a good DIY kit, if you need a couple of TL074s (to make 1:3) and trimpots, I reckon you could squeeze about 4 buffered mults into a 4hp panel - a stacked pair of PCBs maybe?

That would be nice! I'm just a bit unsure if its worth the effort as you can do this on stripboard pretty easily.

[Ginko]

true, it would have to be convenient to build and priced conveniently

[daverj]

dave, would it be possible for you sketch this up? i think a schematic for your solution would be greatly appreciated by many...

There's many ways to design a buffered mult. If I were designing one for my own use, this is probably what I would do:



The first stage has 7 percent gain. The trimpots ideally give +/-7% adjustment. With 1% resistors, the gains could be off by up to about +/-4.5%. With a 10% trimpot the adjustment range could be off by another 1.5%-2%. So the adjustment range of the trim will still cover all possibilities.

The feedback cap values are based on a printed circuit board. With point to point hand wiring or stripboard I might double those.

Since all amps are on the same die, if their offset voltages are similar then using two inverting amps in series may provide some cancellation of that voltage.

I didn't put in bias controls because 1) to make it temperature and supply voltage stable requires a bunch of extra parts, and 2) I am assuming the places these outputs go already have a bias control in the form of the frequency pot on a VCO or filter.

I did this using a Euro power connector and voltages. It would be identical except for the connector for a +/-15 system.

The two Schottky diodes provide power cable reversal protection and their small voltage drop won't effect the operation of the circuit.

[diablojoy]

hmm i did recently do a board for myself but certainly not as good as what dave is suggesting , there is no trimming provided for and the output resistors are not in the feedback loop so it isnt going to be as accurate.
it works but i am still testing it to see how well it works.
I still might revise it to improve on it, though there is no room left for anything extra component wise without going to SMT
board is 8HP , 8 inputs split into 2 groups of 4 to 2 groups of 4 buffered outputs using 6 dual opamps. (I used 6 x LF442)
output 4 normals to input 5
Its all on a single double sided board measuring 100 x 40 mm with 16 x erthenvar jacks but it also includes headers on the ins and outs for different formats so it is quite tight . mixed euro / motm power header and filter caps are silk screened on the back so power connection for either format is easy
I perhaps may have some spares going if anyone is absolutely desperate

[Ginko]

Nice!

[cloudscapes]

Individually buffered outputs is probably a better idea, like the schematic on electro-music. The one with the single opamp input buffer will work, but it won't be rock-solid when you start mixing the output "destinations". If you leave the outputs unbuffered and start passively mixing/stacking multiple outputs further down the line, you might have bleed interfering with your mixes. Buffered outputs ensure that you can do whatever you want down an output without it bleeding back and up another output. *

*speaking as a modular noob so I reserve the right of being totally wrong

[roglok]

dave, would it be possible for you sketch this up? i think a schematic for your solution would be greatly appreciated by many...

There's many ways to design a buffered mult. If I were designing one for my own use, this is probably what I would do:

Thank you very much for the circuit and especially the great explanation, dave!

[qfactor]

dave, would it be possible for you sketch this up? i think a schematic for your solution would be greatly appreciated by many...

There's many ways to design a buffered mult. If I were designing one for my own use, this is probably what I would do:

Thank you very much for the circuit and especially the great explanation, dave!

Indeed, thank you, daverj, for your insight and thoughts about having buffers in both inputs and outputs!

[horstronic]

I made a PCB design for the schematics daverj posted. Haven't tried it yet but I will soon. I'm pretty shure it's gonna work as it is a pretty simple design.

Here it is:
Bottom
Top silk
Top

[roglok]

I made a PCB design for the schematics daverj posted. Haven't tried it yet but I will soon. I'm pretty shure it's gonna work as it is a pretty simple design.

Here it is:
Bottom
Top silk
Top

nice! are you planning to do a run?

[horstronic]

I made a PCB design for the schematics daverj posted. Haven't tried it yet but I will soon. I'm pretty shure it's gonna work as it is a pretty simple design.

Here it is:
Bottom
Top silk
Top

nice! are you planning to do a run?

Not right now. Maybe if there is enough interest I can order a few boards but I have to test it first and I don't have the time right now.

[horstronic]

I finally had the time to build it. Here's the result:




It works pretty nice
The only thing you need to keep in mind when adjusting the outputs is that this circuit doesn't have offset trimmers, so you have to measure the offset first. It's not very high (+- 5mV on my build) but it can affect the results if you don't do it.
So here's what I did:
Leave the input unconnected and measure its voltage. Then measure the voltage on the output you want to adjust. Subtract the input voltage from the output voltage. That's your offset.
When you trim the output now, always add that to the voltage you measure.

Here's an example in case my explanation was too confusing (sorry, I'm not a native speaker):

First measurement with nothing connected to the input:
Input Voltage: +1mV
Output Voltage: +5mV

Output - Input = 4mV = Offset

Now connect the Input to a voltage source (I used a CVP). You can use any voltage, let's say we use 5V.
Measure the output voltage and adjust it to be like this:

Output = Input + Offset
Output = 5V + 4mV = 5.004V

Check if you get the same results in all voltage areas.
You're done.

[horstronic]

I'm thinking about doing a little run of PCBs/Panels for this.
Mostly because I need some more for myself and I want to make some experiences in that area as I'm planning to dive deeper in making my own modules.
So this is kind of a non-profit thing, a set of both will cost something around 10€. Panels will be made of black PCB material.
I improved the PCB design a bit, it's gonna be panel mounted, so no more wiring.
Would there be any interest in this?

[Ginko]

I would be interested, can you get it to 4hp?

[Synthsense]

I would be interested in something like 8 PCBs, maybe more, but I also think the panel should be squashed to 4HP

Nice project, there aren't many Buffered Multiples going around Muff nowadays, I have a Polyphonic Modular project in mind that would need a good quantity of those modules

Cheers

[horstronic]

Yes, I'm trying to make it 4HP, that's my plan anyway.
I'm pretty shure it should be possible, but I need to check it out before I can say that for shure.

I'm using these jacks:
http://www.thonk.co.uk/shop/pj302m/
Has anybody used them before and can tell me how much space they need, measured from the middle to the end of the pins?



It should be something between 9.5-11mm, I guess, but it's hard to tell.

To make a 4HP panel possible, it mustn't be more than 10mm, but guess it's doable by making a thin PCB and cutting off the legs.

Edit: BTW, I'm also open to suggestions for different right angled jacks!

[roglok]

i'm still interested as well.

however, i'd prefer separate 1>3 PCBs over a single dual PCB for more flexibility. but probably that's just me. if possible, please include mounting holes (at least two) for those who fashion their own panel designs.

Give somebody an inch, and he will take a mile.

[roglok]

also, regarding your panel design: i would make it clear what is input and what is output. right now it looks like a passive mult. and also consider a normalling option. sorry i can't help with the jacks...