Analog switch and digitaly controled gear

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

zamproject

Well-known member
Joined
May 11, 2010
Messages
1,502
Hi all

I'm thinking about designing a 4 stereo digitally controlled GSSL switch board
My only concern for the moment is the analog switch/multiplexer

So far the only part that fit my requirement is the DG408/409
https://docs-emea.rs-online.com/webdocs/14f5/0900766b814f5ad4.pdf

...but the ON resistance is relatively high (about 40ohm for +/- 12 or 15V supply) especially for the lower attack setting,
more than 5% for the  lowest (0.3), 2.5% for 1ms, after that it's ok (below 1% resistor)
As the ON resistance is not constant VS drain voltage would you consider this as an issue,
Would you just compensate basically to have Ron + Ratt equal to "original" value ?
Or just don't care for the 5%  +/-1% (dynamic) mismatch ?

The switches logic will probably be drive by 595 SRIO and the 4 threshold and 4 gain by 8ch DAC (12 bit should be enough...)

The idea is to have all this in a 2U (with external psu) or 3 U rack, only one set of encoder/button/led with 4 unit select/link, and midi I/O for recall/dump/automation

Best
Zam
 
Hey Zam, sorry I have nothing to contribute regarding your resistance issue (other than the ol' try it out and see) , but would you please post updates as you progress?  Thank you!
 
The on resistance is relative to the load, so lightly loaded it will be fine.

Years ago I designed an electronic switched bank of channels inside a console use cheap TG CMOS switches, with unmeasurable distortion (at least on my test bench).

JR 
 
boji said:
Hey Zam, sorry I have nothing to contribute regarding your resistance issue (other than the ol' try it out and see) , but would you please post updates as you progress?  Thank you!

Sure !

My mind is more or less clear as I brainstorm since few month
But I'm at the very beginning of the drawing...open kicad project for this today  ::)

I'll probably do something that hook at already available GSSL pcb so don't know but an uncertain implementation I have in mid is to do or not an extra summing stage at VCA input with another DAC to have 8 ch VCA volume automation...

Best
Zam
 
JohnRoberts said:
The on resistance is relative to the load, so lightly loaded it will be fine.

Years ago I designed an electronic switched bank of channels inside a console use cheap TG CMOS switches, with unmeasurable distortion (at least on my test bench).

JR

I guess I don't really need to care about distortion, the switches are in the SC

The load at the attack switch time depend of the release setting, high anyway

I'll see...

Best
Zam

 
> the ON resistance is relatively high (about 40ohm ... for the lower attack setting, more than 5%...

So? Shave your 800r resistor to 760r, it is nearly the same thing.

Read the DG sheet carefully. Is "40r" a Typical or a Max? Is there any spec on variability? If it were 4 to 400, that would matter some. If it is 30-50, bah, 790 and 810 are the same for any music limiter time-constant. (Your musicians vary a LOT more than that.)
 
It is not guaranteed that an analog switch will work in an arbitrary circuit. You must consider the load or source impedance relative to the on resistance (Ron) and Ron flatness. That DG408 part has an Ron of 30-45 ohms with a +-15V supply. So imagine that you have a switch shunting signal to ground. If the source impedance is say 1K (very possible in some arbitrary circuit), that would be 1K in series with 40R to ground which only yields -28dB attenuation. Not good. The same problem occurs if the load is a little low. You can't make a switch that is completely on or completely off. Also, the on resistance is not perfectly flat (see figure "RDS(on) vs. VD and Supply" on page 6) which means you are going to get distortion. And you have gate capacitance to worry about. So you either have to design the circuit around the analog switch or find a part with really low Ron and low capacitance. ADG1414 is similar to DG408 (SPI drive instead of parallel drive) with lower Ron but still not perfect.

One thing that I have wondered is if one can make good analog switches from dual mosfets. These analog switches are effectively just P and N channel mosfets in parallel w/ one driven high and the other low. There are numerous dual mosfet packages that support 60V+, several amps of current and have an Ron of less than 1/10th of an ohm. At that Ron, flatness does not matter. For example consider DMC2450 (this part was largely picked at random but for the capacitance which seems to be quite a bit lower that other similar parts).



 
Sidechain timing, specially at fast attack times, involves rather high peak currents. Not really good for CMOS switching. I'd recommend using relays for attack/release - those functions are at relatively limited number of steps anyway..

Jakob E.
 
Hello

Thanks all for comment an suggestion !
Relay was my first idea, but cost and space consuming especially as the idea is 4 units together.
That's why I search for the analog switches...
I'll see how I can arrange resistor network with relay for possible part count improvement

Best
Zam
 
Geez. Read the plan, think about costs, use your head.

"High peak currents": the RC is fed by a TL074, not a high-current chip. And the 820r makes peak current for unlikely ~16V peaks only 20mA. Will the DG eat that? (Yes.) But actually it can't be 20mA for long because the VCA will cut way back.

If RC values are awkward, scale impedance higher and add a buffer. But HEY! This thing is already buffed with another TL084! It has infinite input impedance (fapp). Scale Attack resistors to 8.2k..2.7Meg, caps to 0.047u..0.68u, decay Rs to 1.8Meg..12Meg..33Meg. Now the Attack switch never sees <8k which is surely fine for ratings and control linearity.
 
PRR said:
Geez. Read the plan, think about costs, use your head.

"High peak currents": the RC is fed by a TL074, not a high-current chip. And the 820r makes peak current for unlikely ~16V peaks only 20mA. Will the DG eat that? (Yes.) But actually it can't be 20mA for long because the VCA will cut way back.

If RC values are awkward, scale impedance higher and add a buffer. But HEY! This thing is already buffed with another TL084! It has infinite input impedance (fapp). Scale Attack resistors to 8.2k..2.7Meg, caps to 0.047u..0.68u, decay Rs to 1.8Meg..12Meg..33Meg. Now the Attack switch never sees <8k which is surely fine for ratings and control linearity.

Geeeeeez that hurt.... but your right...

factor 10 seem high ? at this section the supply is 12V "only" so maybe factor 5 (or lower) is enough...

Best
Zam
 
Back
Top