[sdiy] Help, I'm Desperate! (Charge Injection with DG408)

Steve Lenham steve at bendentech.co.uk
Tue Dec 4 10:28:53 CET 2018


Hi David,

I feel your pain, especially if time is tight.

My thoughts:

- I do not think that there is likely to be a "reverse crosstalk" issue 
from the analogue pins back to the logic, at least not within the IC. 
There are internal buffers and level-shifters between the two.

- I doubt it is charge injection either, but you can test that by doing 
the thing you mentioned and temporarily lower the load resistance on the 
switch outputs. If it IS charge injection, lower resistance should 
reduce the click at least somewhat.

- Here is my main suggestion as to what it might be. It sounds like (we 
only have your description of the circuit to go on) you are generating 
the three-bit control word for the analogue switch from the outputs of 
several analogue comparators. How sure are you that the transition from 
one digital value to the next is clean, i.e. that all three bits change 
state at exactly the same time? If the three-bit word passes through any 
intermediate "runt" states on its way, the switch will briefly select a 
completely different input and potentially inject spikes of something 
else into the output. This is an issue in digital systems (search "race 
hazards") and is likely to be even worse if comparatively slow and 
imprecise analogue comparators are involved.

Your previous design used a DG409, which only has a two-bit address - 
the problem could theoretically still occur there, but there is less 
scope for it to do so.

- Another possibility, though IMO less likely, is simply DC offsets on 
the different inputs of your analogue switch. Most people AC-couple 
audio signals into an analogue switch, but you also need a pulldown 
resistor at the pin of the switch. Otherwise, when an input is not 
selected, leakage current in the coupling cap can allow an offset 
voltage to appear at the input of the switch. This is then discharged 
when the source gets selected, causing a glitch at the output. (This is 
less likely because you say the switch is selecting from two copies of 
the same signal).

Hope this helps a bit,

Steve L.
Benden Sound Technology


On 04/12/2018 08:17, David G Dixon wrote:
> (Sorry for so many postings...)
> 
> Actually, connecting the logic pins directly to the zeners without an
> intervening resistor should also eat the spikes, cuz that's what zeners do,
> no?  If the voltage tries to exceed the zener voltage with a spike, then the
> zeners will suck it out.  In that case, all I have to do is replace three
> resistors with wire jumpers.  To heck with it... it's 12:16 AM, but I'm
> gonna go try it right now.
> 
>   
> 
>> -----Original Message-----
>> From: Synth-diy [mailto:synth-diy-bounces at synth-diy.org] On
>> Behalf Of David G Dixon
>> Sent: Tuesday, December 04, 2018 12:04 AM
>> To: 'Ullrich Peter'; synth-diy at synth-diy.org
>> Subject: Re: [sdiy] Help, I'm Desperate! (Charge Injection with DG408)
>>
>> The other thing I thought I might try:
>>
>> I need the back-to-back zeners to create accurate threshold
>> voltages for my ADC logic circuit.  However, perhaps I should
>> drive the logic pins of the DG408s directly from the
>> comparator outputs.  The low source impedance of the opamps
>> should eat any spikes, right?  I already know that the DG4XX
>> switches happily take voltages on their logic pins up to both rails.
>>
>> Alternatively, I can put separate resistors between the opamp
>> outputs and each switch's logic pins.  This might be safer
>> for limiting the current to the logic pins.  However, I'll
>> happily sacrifice two DG408 switches to solve this problem,
>> and connecting them directly would allow me to use my current
>> layout, simply by replacing a few resistors with jumper
>> wires.  The datasheets say that the logic pins eat 10uA of
>> current, so this should be true whether I have
>> current-limiting resistors or not, right?
>>
>>   
>>
>>> -----Original Message-----
>>> From: Ullrich Peter [mailto:Peter.Ullrich at kapsch.net]
>>> Sent: Monday, December 03, 2018 11:17 PM
>>> To: David G Dixon; synth-diy at synth-diy.org
>>> Subject: Re: [sdiy] Help, I'm Desperate! (Charge Injection
>> with DG408)
>>>
>>> Hi David!
>>>
>>> There are special "slow" switchest hat alsoo have make before break
>>> characteristics so that they allow clickless/popless switching.
>>> Some information can be found here and also a list for such parts:
>>>
>>> https://www.maximintegrated.com/en/app-notes/index.mvp/id/5299
>>>    (search for "Audio Switches and the Break-Before-Make Capability")
>>>
>>> Ciao
>>> Peter
>>>
>>> http://www.ullrich.at.tt (my webpage)
>>> http://www.rc100-usb.at.tt (Roland Sampler Tool)
>>> http://synpro.heimat.eu/datadial.htm (DataDial Synth Addon)
>>>
>>>
>>>
>>> -----Ursprüngliche Nachricht-----
>>> Von: Synth-diy <synth-diy-bounces at synth-diy.org> Im Auftrag
>> von David
>>> G Dixon
>>> Gesendet: Dienstag, 4. Dezember 2018 07:22
>>> An: synth-diy at synth-diy.org
>>> Betreff: [sdiy] Help, I'm Desperate! (Charge Injection with DG408)
>>>
>>> Hello SDIY Team,
>>>
>>> This is the next sorry chapter in my saga with this
>> 9-channel scanner.
>>>
>>> You may recall that I was shocked to learn that TL074 saturates to
>>> different voltages positive and negative, and that this was
>> causing me
>>> grief with a 3-bit comparator ADC that I built to drive a couple of
>>> 8-channel analog muxes (DG408).
>>>
>>> Well, I solved that problem pretty easily: 3.3k current-limiting
>>> resistors to back-to-back 5.1V zener diodes off the comparator
>>> outputs.  I hand-selected the zener pairs to give me exactly +5.52V
>>> and -5.52V (subsequently fed to the logic inputs of the
>> DG408 through
>>> 30k resistors), and was able to size the divider resistors
>> accordingly
>>> to get exactly the threshold voltages I needed for the ADC.
>>   This all
>>> worked perfectly.
>>>
>>> HOWEVER...
>>>
>>> The scanner is creating wicked clicks when it switches the muxes.
>>> After a bunch of thinking and reading, I have come to the
>> conclusion
>>> that this is an issue of charge injection in the DG408s.
>>>
>>> Here is what I have done to try to fix the problem:
>>>
>>> 1) Decoupled the rails at the TL074 hosting the ADC
>> comparators with
>>> 100nF caps directly at the power pins (no effect).
>>>
>>> 2) Decoupled the rails at the DG408 power pins with 10uF
>> electrolytic
>>> caps (no effect).
>>>
>>> Obviously, the power rails are not the source of the clicks.
>>> This was my big clue that it was charge injection and not
>> dirty rails.
>>>
>>> 3) Added 1N4148 diodes at the DG408 logic inputs to limit
>> the negative
>>> logic voltage excursions to about -0.5V (instead of the
>> -5.5V that the
>>> zeners are providing)(no effect).  It turns out that the
>> DG4XX switch
>>> family can take the full voltage range at the logic pins,
>> all the way
>>> down to -22V.
>>>
>>> 4) Added 10nF caps off of the DG408 switch output pins, as shown in
>>> the datasheet (no effect).
>>>
>>> 5) Increased these caps to 2.2uF.  This didn't really eliminate the
>>> clicks, but it did attenuate the signals something fierce.
>>>
>>> 6) Ran the scanner output through a 6dB filter (the first
>> output of my
>>> Doc Oc filter) -- this did eliminate the clicks, but only
>> at a cutoff
>>> frequency which seriously filters the audio signals as
>> well.  Hence,
>>> adding a simple RC filter to the output isn't really a
>> viable option.
>>>
>>> 7) I tried a different pair of DG408s (no effect).
>>>
>>> The one thing I haven't tried is decreasing the downstream
>> impedance
>>> of the switch outputs.  They are going to the multiplier circuit
>>> through 100k resistors.  I could change all the resistors in that
>>> circuit to 30k, or even 10k, but something tells me that this is a
>>> fool's errand.
>>>
>>> So, I'm stumped, and about to give up.  My customer has
>> told me that
>>> the unit is useless to him with this clicking (I concur --
>> it would be
>>> useless to me too), and I basically see no alternative to
>> offering him
>>> his money back and throwing the whole mess in the trash.  I'm very
>>> surprised at the extreme level of charge injection in these DG408
>>> switches.  I didn't notice it to nearly the same extent (if at all)
>>> with the DG409 switch in my 5-channel scanner.  Earlier today, I
>>> realized that there is a dual 8-channel switch, the DG407
>> -- perhaps
>>> if I had used this instead of two DG408s, the clicking would be
>>> lesser.  However, I'm not prepared to order new chips and create a
>>> third layout for this stupid circuit.
>>>
>>> Do any of you have any insight that might help me solve
>> (or, at least,
>>> minimize) this clicking problem that I haven't already thought of?
>>>
>>> Anyone who can help me solve this problem gets a free
>> scanner PCB, or
>>> any other PCB that I make by hand.
>>>
>>> Thanks,
>>> Dave Dixon





More information about the Synth-diy mailing list