DAC output help

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: DAC output help

Post by JR. »

I'll let you do some of the work, but yes two roughly 12k resistors. One to DAC for 0-5V input and one to -12V for offset. They sum into 1k to virtual ground... which is opposite of how that schematic is drawn. Both opamps are miswired + where - should be and vice versa.

This way if 5V rail is not up first it shouldn't trash the DAC.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: DAC output help

Post by mediatechnology »

I think you're still going to have a stored charge on the 10 uF caps that will discharge through the 1K27 resistors into the outputs when the 5V collapses.

That may or may not be your "current" problem. (pun intended.)
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: DAC output help

Post by JR. »

That low z path can go away. ..

the 12k can connect directly to the DAC without the 1k and cap or two 6k with a cap in the middle.

likewise the 12 k and trim to -12v can be two 6Ks with cap in middle...


JR
Cancel the "cancel culture", do not support mob hatred.
juniorhifikit
Posts: 139
Joined: Sun Oct 04, 2009 8:37 am
Location: Oakland/Paris

Re: DAC output help

Post by juniorhifikit »

JR. wrote:That low z path can go away. ..
the 12k can connect directly to the DAC without the 1k and cap or two 6k with a cap in the middle.
likewise the 12 k and trim to -12v can be two 6Ks with cap in middle...
JR
That 1K27 resistor and 10uF cap is the the filter for the "timing" of the VCA control, so I think two 6K's with the cap in the middle is best - actually very close to the THAT data sheet. As to the total series R that figures into the VCA's scale, I'll have to work it out on paper and then trim it in place. It's never come out exactly as the formula states.

Will report back...
juniorhifikit
Posts: 139
Joined: Sun Oct 04, 2009 8:37 am
Location: Oakland/Paris

Re: DAC output help

Post by juniorhifikit »

Please excuse my ignorance, but basically I'm looking for a higher impedance path looking back to the DAC output, so that in the case of -12V powering up first (or shutting down last), current will flow somewhere else (lower impedance) and not back to the DAC output, correct?

There is no magic to -12v for the bias, I simply didn't want the regulator to have to throw away too much as heat (it's fed from the PSU's -18v leg). It's drawing very little current, so there's really very little work to be done by the regulator. At unity gain, my DAC is putting out 3.72V, so I need -3.72V of bias to get zero on both VCA control ports. I suppose I could use an LM7905 for the bias, and much less series R, and move the bias insertion point to C8 as suggested (I've yet to calculate what that series R will be) and still keep my current scale intact:
Image


The first section of the buffer opamp, which drives the negative control port, I took straight from the THAT data sheet (with a few component value changes). The voltage follower on the second stage of the opamp I thought was necessary for a low impedance, as the control port of the VCA wants to see a very low impedance:
The 2180 Series VCAs are designed to be operated with zero source impedance at pins 2 and 3, and an infinite source impedance at pin 4. (Pin 4 should be left open.) To realize all the performance designed into a 2180, keep the source impedance of the control voltage driver well under 50 W.
This is how I arrived at this version of the bi directional control port buffer, which was greatly inspired by JR's version (major thanks JR!). Am I missing something key here with regard to the reversing of the buffer opamp's inputs?

Thanks
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: DAC output help

Post by mediatechnology »

...I'm looking for a higher impedance path looking back to the DAC output, so that in the case of -12V powering up first (or shutting down last), current will flow somewhere else (lower impedance) and not back to the DAC output, correct?
Yes.

I have another idea.

Why don't you just sum (actually subtract) the required offset as a current at the summing node, pin 2 of the first op amp? If you do that, then you need to turn the second op amp into an inverter and feed it's input from the output of the first. I realize that adds two resistors and is a little less optimal from a noise perspective, but I bet if you make Rin and Rfb fairly low, you won't be able to see much difference. Since the scaled voltage is quite low, the resistors can be made lower (<<1K) without concern of running out of drive current.

Image

I realize the above has a higher noise gain in the Ec+ driver but it trims more cleanly.
You can place one passive LPF at the input. Doesn't have to be NP.
The second LPF (Cfb NP uF) can be moved to the inverter making it an integrator.
The feedback resistor values for the second inverter can be made quite low.
The Cfb around the Ec+ inverter can also be larger (nF?), but not too large; You don't want it to look like an integrator.

Changing the reference source.

Once you do that, you can make a current source as your reference that has PTAT (0 db reference) compensation.
(See: http://www.thatcorp.com/datashts/dn128.pdf)
This doesn't address the VCA driver but while you're changing things you might look at this.

Image
LM334 PTAT Bipolar Voltage Reference, Courtesy GKH THAT Corporation.

U3A and U3B are not the VCA driver but a bipolar voltage reference with a PTAT compensation built in.
This was originally done for a single-supply application. You will want to modify it to tie the 1/2 Vcc point to ground.
The top output is positive, the bottom inverts the top.
For your application you will want to use the bottom output as a negative reference supply to replace the 7912.

To eliminate the top op amp, you might want to consider connecting the LM334 input to the positive rail and have that positive, regulated, current feed the inverter.
This will give you a negative voltage reference supply using only one op amp.
Your trimmers will connect to the op amp output to provide a reference voltage that has a temperature-dependent PTAT offset which will compensate the VCA PTAT error.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: DAC output help

Post by JR. »

Yup, that looks better...

You can figure out the exact values using ohms law, and your own brain cells. I will speak in broad strokes.

Wayne.. the two relatively high impedance R's feeding a 1k to a virtual earth are not hugely different (to the DAC) from connecting directly to the opamp - input, except we gain another pole of LPF.

Re: the -current fixed offset. We are ultimately dealing with a current, so the voltage it comes from is not very important. The larger than negative voltage, the larger the R can be for a given amount of current, and more effective a simple filter. I might be tempted to use a common filter in series with all of the - offsets, but if this supply dropping R is too large, it may lead to interaction between adjustments, so separate filters will eliminate that. There does need to be a filter in this - leg too, to keep -PS noise out of the works.

Wayne: as you know there are several ways to drive, VCA control ports. Your suggestion is also good for providing a differential Ec driven by low Z. I've seen it done this way before in high performance applications. Mine uses an unusual technique where the actual noise of the inverting opamp is present on the top of the input 1k and therefore CM to both control ports. This cancellation is compromised somewhat by the divider formed with 1k feeding back into the 6k/12k input resistors. Also the Cap to ground there is a LPF that affects the CM voltage component coming back from the opamp for less accurate cancellation.

I am not smart enough to predict if my CM noise cancellation delivers much real world noise benefit or not, that will only be determined on the bench ( a very quiet bench at that). Removing the filter cap c8 might actually improve the CM cancellation of the opamp's noise, but allow in more CV path noise... That's why we build stuff and measure results, we're chasing the sound of the angel's footprints on the pin head. Using a lower noise opamp may be more practical than my cancellation trick, if this is actually an issue (probably not).

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: DAC output help

Post by mediatechnology »

Wayne.. the two relatively high impedance R's feeding a 1k to a virtual earth are not hugely different (to the DAC) from connecting directly to the opamp - input, except we gain another pole of LPF.
Not sure I understand but I think I agree.

That unlabeled R at the far right (EDIT: left) is his scaling resistor (was it 12K?) and combo LPF.
My comment WRT the noise gain was the other stage, the inverter with the 220R wrapped around it.
Isn't the noise gain of the (lower) inverter 2 vs. 1 for the follower you originally had?
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: DAC output help

Post by JR. »

mediatechnology wrote:
Wayne.. the two relatively high impedance R's feeding a 1k to a virtual earth are not hugely different (to the DAC) from connecting directly to the opamp - input, except we gain another pole of LPF.
Not sure I understand but I think I agree.
As all to often we find ourselves on different pages.... :oops:

That unlabeled R at the far right is his scaling resistor (was it 12K?) and combo LPF.
Far LEFT of your schematic? Or different schematic page.

My comment WRT the noise gain was the other stage, the inverter with the 220R wrapped around it.
Isn't the noise gain of the (lower) inverter 2 vs. 1 for the follower you originally had?
The NG of lower inverter in your schematic is indeed 2x because input 220 ohm R is connected to low Z. Gain wrt + input is 1+(220/220)=2

For my approach which I speak in terms of handy schematic nearby the noise analysis is a little more complex.

The inverter NG in the most recent juniorhifikit schematic is 1+ (1k/ (1K+3uf||6k||?k, etc). So ignoring the 3uF, the noise gain is <<2, unfortunately the 3uF pole is 50 Hz so like 2x NG for most of audio bandpass. Again ignoring the 3 UF, the voltage at the junction of the input 1k R and 6K ||- leg trim, is the opamp input noise voltage attenuated by divider 1k into <6k so <1 but more than 1/2. This voltage is added to noise voltage of unity buffer into other Ec port in phase, so CM.

I appreciate there is a lot of ignore this and that, and this is chasing vanishingly small noise sources. Perhaps easier to use a lower noise opamp that Rube Goldberg a cancellation of only one.

Perhaps a classic case of me over thinking things when I'm not the one melting solder. But it is not a lot of extra parts, just rearranging the deck chairs.

I hope this makes sense to someone besides me.

The reality is one needs to fire up a working unit and inspect what the dominant noise sources really are. And how much noise mitigation is practical ($$).

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: DAC output help

Post by mediatechnology »

That unlabeled R at the far right is his scaling resistor (was it 12K?) and combo LPF.
Yes, I meant left, not right. Duh. That dyslexia thing again. I need to stop looking at all these posts in the mirror...

Actually as I was posting that the cable techs were pulling up.

I think my point was that my suggestion, inverting the output of the inverter, had a slightly higher overall noise gain (looking at it differentially from Ec+ to Ec-) than your original using the follower paralleled at the input of the original inverter.
Post Reply