[time-nuts] Re: DIY Low offset Phase Noise Analyzer (Erik Kaashoek)

[Bob kb8tq]

Hi

Yes it is a pain to implement dual supplies. I ponder that issue every time 
I build one of these setups. I’ve built a lot of them …. If you are going to 
do a single supply, setting up a “virtual ground” is probably the best way
to go. Do it with a drive circuit to provide very clean 15V off of a 30V supply 
then tack everything ( including *all* the mixer grounds to that 15V supply. 

Keeping the signal undistorted before you check the beat note and use it
to drive the EFC does keep you out of various issues. You do not want to
deal with possible clipping / saturation artifacts getting into either process. 

With devices having positive side EFC, negative side EFC, and “both
sides” EFC, it’s hard to get around a dual supply in any sort of general 
purpose device. A 15V center point is not going to fit any EFC that I’ve
seen :). 

Struggling with the ground loop problem is always the big deal in any
setup. Trying to rule out / take out line noise is usually the final straw in
any series of tests. Doing that with everything at “real ground” is just a 
bit easier. 

Part of the calibration is measuring the beat note as it goes past zero. 
The ’scope gets cranked up and you look at a bit of the crossing right
at ground. Keeping the device happy while doing this is much easier if
the chassis does not need to float at 15V. 

Whatever is used as a supply turns out to be a dedicated device. The 
same ground loop / isolation stuff get in here. An old style non-switching
design is just about mandatory. Keeping switching artifacts out of things
is almost impossible. All of this makes a “build from scratch” approach
less and less crazy. Old style three terminal regulators ( so 78x18 / 79x18 )
are not as easy to find these days. They do fine if you happen to have a 
pair …. There’s really not much power used by any of this. The need for 
anything massive. 100 ma out of each side is overkill ….

As you build things up, you eventually come to the realization that a big
sheet of brass is a good idea for the ground. Tie this and that to the sheet. 
Keep everything non-essential away and likely keep it turned off. Tying
a dedicated supply to that sheet along with the amp and EFC stuff is not
at all unusual. 

Bob

> On Jul 9, 2022, at 11:11 PM, Erik Kaashoek <erik at kaashoek.com> wrote:
> 
> Hi Magnus,
> Yes, and it works very well, locking is easier as once locked it nicely stay's in lock, , even with a slow drift of either the DUT or the reference. As I could not find a bipolar capacitor the tuning potmeter has to be kept at the low side to avoid blowing the integration capacitor. Maybe a back to back series capacitor with pull down resistor is safer to use.
> Will need to update the schematic to show the small improvements.
> 
> @Bob,
> You mentioned "dual supplies with high voltage" for the first gain opamp. How much impact would dual voltage bring as its a pain to implement.
> I understand everything gets ground reference and you loose the noise of the buffer opamp but as the first gain opamp is in differential mode for its input it does not see the noise of the buffer opamp. Or am I making a mistake?
> 
> On 10-7-2022 2:02, Magnus Danielson via time-nuts wrote:
>> Have you attempted doing a PI-loop as I've suggested? 
>