[sdiy] (Ab)use DAC as mixer?

[rsdio at audiobanshee.com]

One very promising fact is that they have 4 separate Vref inputs - not really worth the real estate if they weren’t intended to be drastically different from each other.

Another promising fact is that they quote the 100 kHz bandwidth for a Vref with a 2 Vpp amplitude. That’s a significant amplitude compared to a “DC” signal, so this must really be an MDAC (even if the term isn’t used).

On the negative side, they don’t show a graph of how bandwidth changes as the digital code is reduced from full scale, or as the Vref amplitude changes. I’ve been bitten in the past by op-amps that have a very different bandwidth for low-level versus high-level signals. The only way to know is to build something and see. You can also sign up for the Engineer-to-Engineer Forum at http://e2e.ti.com/ where they would be happy to answer all kinds of questions (in the right forum).

Note that the output can either be one or two times the reference voltage, so you might be able to exceed 1/2 Vcc for your signal levels. You should surely be able to reach 2 Vpp instead of merely 1.65 Vpp according to the specifications. If your analog mixer stage has a higher supply voltage, you might be able to get better signal-to-noise ratios by using the built-in “two times” gain. Not sure whether they might lead to clipping, though.

Brian Willoughby


On Jun 2, 2018, at 1:33 AM, sleepy_dog at gmx.de wrote:
> rsdio at audiobanshee.com wrote:
> > The thing you need to look out for is that the Vref input is probably designed for a DC signal with lots of bypass filtering
> 
> Ah! Darn, this makes too much sense ;)
> 
> Although now reading the TLV5620 datasheet, under applications, it says:
> - "Digitally Controlled Amplifiers/Attenuators"
> 
> And in operating characteristics:
> - Reference input bandwidth: 100 kHz
> 
> Sounds good?
> 
> - Steve
> 
> 
> Am 02.06.2018 um 10:14 schrieb rsdio at audiobanshee.com:
>> I had the same idea that you describe, only it was about 30 years ago. I was thinking about a digital signal, though, while using the Vref as the volume control. I assume that feeding the audio signal into the Vref would not work very well.
>> 
>> Since then I’ve learned about the MDAC, or Multiplying DAC. I haven’t found any specific part numbers (or, if I have, I’ve forgotten). An MDAC should be perfect for volume control of a signal passing through.
>> The thing you need to look out for is that the Vref input is probably designed for a DC signal with lots of bypass filtering
>> 
>> . If you send an audio bandwidth signal into the Vref pin, you’ll probably discover that it has very poor performance.
>> 
>> Brian Willoughby
>> 
>> 
>> On Jun 2, 2018, at 12:40 AM, sleepy_dog at gmx.de wrote:
>>> Hey list,
>>> 
>>> this has been floating around in my mind for some years and I just remembered it again, so maybe some of you know something about this:
>>> 
>>> Would it be feasible to "abuse" a quad DAC with 4 independant Vref inputs as a 4 channel, digitally controlled  mixer?
>>> 
>>> I.e., you feed in 4 analogue voltages into the 4 Vref inputs, and control the volumes by writing digital output values.
>>> The DAC I had in mind (TLV5620, price OK) only has 8-bit resolution and can stomach only 1/2 Vcc (1.65V I believe),
>>> and it's positive-voltage-only, but that would do away with so much effort, in comparison to other solutions I'm aware of...
>>> So If I have up to 4 audio sources which are already positive-only, *some* of which might come out of a DAC and not yet put through a reconstruction filter,
>>> I could mix them all that way, and put only one reconstruction filter for the mix?
>>> 
>>> Yeah, it's for something like waveform mixing. Not planning on a high update rate for the digital volume levels (so far), just manual knob adjustment. Although I could imagine it might be cool to modulate that somewhat, if that doesn't get too noisy.
>>> 
>>> But the layman I am, I probably don't know about some horrible side effects that may have :-D
>>> 
>>> What can you say about the feasibility of those two aspects (and then others I may have overlooked):
>>> 1) using a DAC this way in general
>>> 2) my desire to skimp on the number of reconstruction filters, filtering the mix (I'm not sure I'd be using > 1 DAC output for audio, but keep it in mind as an option, when DAC resolution is moderate)
>>> 
>>> - Steve
>> 
>