DAC output help

[juniorhifikit]

I think I suggested the same thing back in posts 4 and 6.

You did indeed. But, with my lack of EE background, I didn't get it. Now that it's been explained to me twice, I'm starting to understand Sorry Wayne!

[mediatechnology]

Well, I didn't catch the floating pin.

I'm glad I had a vague recollection of it.

[juniorhifikit]

Bought a Rigol 1052E scope, which is pretty good considering it's low cost. What I've found is that I have two issues:

1 - My 7.5V switching supply is throwing about 22mV of crap into my otherwise clean linear supply. With the switcher shut down, the +/-18V linear supply exhibits about 2.5mV of ripple. While 25mV of ripple/noise is still a very small percentage, and this isn't showing up on the subsequent 5V supplies, I would like it to be as clean as possible. I've seen input filters for switchers that resemble the output filter, with a choke, etc. Will investigate there.

2 - My +5V supply which supplies the DAC's (derived from the +18V leg) has anywhere from 25mV to over 1V of startup transient. I'm hoping to clamp this with a couple back-to-back zeners in hopes that this will not only eliminate the Vdd spikes, but also any spikes on the DATA & CLK ports, and the -12V CV bias in parallel with the DAC's outputs - which are all derived from this same +5V regulator.

Will report back...

[JR.]

I wouldn't be overly concerned about noise that gets regulated out already.

Please clarify what you mean by 1V transient? Is this 1V on top of 5V output? Where is scope ground connected? Should be at same 5V ground. If you are getting 6V at start up that seems high.

JR

[juniorhifikit]

Scope is connected to the same ground as the 5V reg. Here are some screen captures of the DC coupled scope with a 4.3V zener I had on hand. Obviously that zener is conducting all the time on a 5V supply. Maybe try a 5.1V zener or TVS diode?:

+5V supply at power-up, derived from +18V supply:



-12V bias at power-up, derived from +5V supply:


There's a negative 1V spike before the 5V supply comes up (sometimes less, sometimes more), which also results in a positive 3V spike on the negative bias supply. Still trying to understand all the different methods of protection by clamping & clipping...

[JR.]

IIRC the - bias supply going slightly positive should not be a problem. You could slow it down, also.

I am more concerned about -1V on the +5V supply. I wonder if this is coming from a half wave rectified unregulated supply that randomly comes up negative first then positive.

A diode clamp on the 5V should prevent it from going 1V negative but this may be fixed by full wave rectification in the power supply, if that is an option.

I recall years ago, an engineer working for me, came up with a trick power supply circuit to insure that the positive rail would always come up first. This was to minimize turn on transients with cheap opamps.

JR

[juniorhifikit]

IIRC the - bias supply going slightly positive should not be a problem. You could slow it down, also.

Just RC filter the output?

I am more concerned about -1V on the +5V supply. I wonder if this is coming from a half wave rectified unregulated supply that randomly comes up negative first then positive.

Full wave rectified +18V & -18V legs from center tapped transfo

A diode clamp on the 5V should prevent it from going 1V negative but this may be fixed by full wave rectification in the power supply, if that is an option.

I added 5.1V Zener and it seems to help, but the negative spike is not fully eliminated. I would expect to still see -0.7V from the diode drop, but not as much as -1.5V Is a zener too slow? Maybe a TVS diode? (is that just fancy for "fast zener"?)

The +5V digital supply, which is derived from the 7.5V switcher doesn't have a spike.

[JR.]

Are there multiple 5V supplies?

With FW rectified supply I don't see how any positive supply swings negative.

Zener clamp after the 5V regulator seems unnecessary... regulator should clamp itself at least in positive direction.

Do you see any voltage spikes with scope looking between different grounds?

JR

[juniorhifikit]

I'm out of the country mixing and will be away from my bench for a week. I'll do a bunch more testing when I return, and report back. Thanks for all of your help guys!

[juniorhifikit]

So I put a 5V zener on the output of the +5V analog supply regulator, and put the unit back into service at the studio. I then went and did the same thing on my test rig on the bench. The prototype at the studio still shows little transients now and then, but the production model on my bench seems perfect - no spikes at all. May be due to the PSU on the production model being on the same ground plane as everything else; whereas the prototype at the studio has a separate PSU and lots of daughter boards... dunno.

To answer your previous question JR, yes there are two 5V regulators - one is for the DACs and is derived from the +18V linear leg and grounds to the analog ground; while the other is for the microcontroller and is derived from the +7.5V switching regulator (which drives the LED's) and is tied to digital ground. And, there were no transients between analog and digital ground.