MAX395 examples please!

tysonlt · January 20, 2012, 10:32:54 AM

Hi everyone,

I have seen a few people mention the MAX395 and I am interested in using it for a microcontroller-based effect switcher. Currently looking at 4053's, but these max chips look promising.

Could someone please post a schematic on using these, especially in +/-5v?

PRR · January 20, 2012, 07:41:42 PM

Same as CD4053.

cloudscapes · January 20, 2012, 07:57:47 PM

I'd be too afriad of noise bleeding through. I'd use relays for that sort of thing.

tysonlt · January 20, 2012, 09:22:21 PM

Hi cloudscapes,

I have considered that, but for me ten relays would just be too expensive. I am the guy who started the "Poor Man's Actuator" thread!

When you say noise, do you mean digital noise from the control line, or some effect sound leaking even when bypassed?

tysonlt · January 20, 2012, 10:12:49 PM

Here is the schematic I am thinking for an SPST chip, minus the biasing for now:

Boxes are effects, arrows are signal in and out.

Each pair of switches is mutually exclusive. If switch A is open, signal will pass through switch B and go to the next 'choice' pair, C/D. If switch A is closed, signal will pass through the effect and on to C/D. Of course it won't propagate back because switch B will be open. I'm not sure if a diode is needed to stop signal flowing into the output jack of the effect.

Could someone please let me know if this is correct or has some basic misunderstanding! As far as I can see, this has the signal going through less silicon than with a 4053. As RG pointed out, the signal always passes through two switches per effect with 4053s, but with SPST it only passes through one switch per effect.

Later on I will post another schemo with how I think it should be biased, would really appreciate comments on that too. Gotta take the kids to the park now!

tysonlt · January 20, 2012, 10:13:20 PM

Oh and thanks Paul! You're turning into my teacher - appreciated!

PRR · January 21, 2012, 12:27:32 PM

> I'm not sure if a diode is needed to stop signal flowing into the output jack of the effect.

Diodes block one way flow, such as a battery.

Audio swings both ways. A diode won't block audio, half the wave gets through.

If you don't want audio sneaking "back" along a path, the usual tool is a buffer. While all amps/buffs "leak-back", the reverse gain is easily 1/10,000th of the forward gain so sneakage is negligible. TL072 is fine.

Your switching is wrong. Look at any "true bypass" (typically a mechanical DPDT) and emulate it with SPST.

One thought. You have controller in one box, switches in another. The switches use VERY little power, like 30uA. You could drop a pair of 9V batteries in the switch box and it would run 24/7 for years. However each opamp uses ~~2mA (2,000uA) and now you need power feed or off-switch or something.

tysonlt · January 21, 2012, 09:59:04 PM

Quote from: PRR on January 21, 2012, 12:27:32 PM
> I'm not sure if a diode is needed to stop signal flowing into the output jack of the effect.

Diodes block one way flow, such as a battery.

Audio swings both ways. A diode won't block audio, half the wave gets through.

Wow, I did not know that! I guess if we have biased the signal at 1/2v, then when it swings below that it is seen as negative voltage, so power flows the other way??

Quote
Your switching is wrong. Look at any "true bypass" (typically a mechanical DPDT) and emulate it with SPST.

I LOVE being told I'm wrong, best way to learn. Is this better?

A switches and B switches work in tandem, and A and B are mutually exclusive. This could be driven with one PIC pin if I used an inverter.

Quote
One thought. You have controller in one box, switches in another. The switches use VERY little power, like 30uA. You could drop a pair of 9V batteries in the switch box and it would run 24/7 for years. However each opamp uses ~~2mA (2,000uA) and now you need power feed or off-switch or something.

Good thought. However, I have a 9+ and a 12+ power supply on my board with plenty of milliamps left over, so I was going to use an RS232 cable between the switching box and the foot controller. This would supply power over two lines back to the controller, which would regulate it to 5v. The controller would then send a serial signal back to the latches in the switcher.

I also thought of plugging the guitar into the foot controller, and running the signal over the RS232 cable as well. Would there be any problem running signal, power, and serial data within the same cable? This would use 8 pins in total.

~arph · January 23, 2012, 11:19:25 AM

Hi,

did you look up the CMOS switching at GEO? he explains the below for CD4066's and 4053's I believe.
It's important that all inputs and outputs are at the same DC. So in my case I used a simple voltage divider to get 1/2 Vcc (just like the regular opamp Vref setup)
All inputs and outputs have 1uF caps on them and then at the IC side there is a 1meg to 1/2 Vcc. This way all ins ans outs are DC decoupled and at the same DC level at the IC. With the MAX I have no switching noise at all. But bear in mind I'm using it a low amplitude levels, not in a FX switcher.

Regards,

Arnoud

tysonlt · January 23, 2012, 08:01:06 PM

Thanks arph.

I want to switch guitar level and line level (as in, in the fx loop after the pre-amp). I have read the geo article many, many times

I'm a bit of a newbie, so I'm learning basic electron theory at the same time as getting a grip on biasing.

One burning question I have is WHY on the geofex article, it only has the 1M resistor from Vref on the chip outputs, but the chip inputs have a 1M to ground as well. On the example below, ALL lines have two resistors. Was R.G j ust using shorthand, or is there something I'm missing? I understand the two resistors, should I just put R-C-R on all lines?

~arph · January 24, 2012, 03:42:56 AM

The important bit is that all IC in and out pins must sit at the same DC level. Having the input and output at two different reference voltages as you state sounds wrong to me. As this would produce audible clicks when switching. Could you point me to that specific article?

tysonlt · January 24, 2012, 07:19:13 AM

Here is an example PRR gave me:

http://www.diystompboxes.com/smfforum/index.php?topic=95582.msg829392.msg#829392

Here is the geo article:

http://www.geofex.com/article_folders/cd4053/cd4053.htm.

I think I understand the R-C-R concept: 1/2v flows in through the first 1M resistor, charges the cap, and flows out to ground through the second resistor, until the cap is charged. The audio signal then comes and pushes the cap back the other way, causing Vref+signal to go into the chip. (is that correct?)

What I don't really understand is what just the one resistor does. Is it just pulling that line up to Vref, so there's always that much voltage on each line? If so, why the resistor to ground?

I feel a bit silly asking the same questions about biasing, but I'm getting close!

Thanks

electrosonic · January 24, 2012, 09:43:30 AM

QuoteOne burning question I have is WHY on the geofex article, it only has the 1M resistor from Vref on the chip outputs, but the chip inputs have a 1M to ground as well. On the example below, ALL lines have two resistors. Was R.G j ust using shorthand, or is there something I'm missing? I understand the two resistors, should I just put R-C-R on all lines?

Those 1M resistors at inputs and outputs are not for biasing, they are pulldown resistors to make sure the input and output are sitting at 0v DC, eliminating another potential source of switching noise.

A.

tysonlt · January 24, 2012, 08:46:26 PM

Ah, OK. I think I understand that. So if I am connecting the switch to an external effect, I should add the pulldown and cap for the loop line as well? (ie where rg has 'existing fx caps')

And if the output is going to the input of the next 4053 in the chain, I guess I only need the one biasing resistor from Vref?

News:

MAX395 examples please!