Linear Power Supply and Active Current Regulation Across a Gas Discharge / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Linear Power Supply and Active Current Regulation Across a Gas Discharge

previous 1 2 3 next

Move Thread

LAN_403

Sulaiman

Sat Jan 12 2013, 11:05PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3141

my first guess would be to increase the capacitance across Rleak.

sngecko

Sun Jan 13 2013, 12:57AM

Registered Member #3447 Joined: Fri Nov 26 2010, 11:10PM
Location: North Jersey
Posts: 97

Parts of this project are spread all over 4hv, mostly because it takes me a long time to make adjustments and tests (and sometimes it blows up a bit). In the thread "Oscillations Redux," you can see some of the changes that I've made with the help of the 4hv community. It's a much better system now, but there is still the oscillation problem. Let me see if I can recap how the changes progressed and the results:

(1) The grid used to be positively biased. This led to problems in the FET, where current would (obviously) flow when Vc=0. This was corrected by setting the grid to ground. (It may sound nuts to positive bias the grid, but Proud Mary helped me uncover that this was originally done because the regulation schematic indicates a BJT which needed compensation.)
(2) Later I put the Rleak(47kOhm) and cap (.47uF) along with a TVS diode for protection.
(3) Moreover voltage protection MOV/TVS was placed on the cathode chain to protect the regulation FET from overvoltage.
(4) Lastly, I tried an integrator across the regulation op-amp that should have slowed down its responsiveness to something in the few hundred hertz range. I can't remember the values right now.

Note that none of these changes seemed to make any discernable difference to the weird symptoms as soon as the tube/laser was struck: high-pitched audible whine, a slowing of the case and boost fans when the current is turned above a milliamp or so (they quickly stop). At least during this period all of the instrumentation appears to operate properly. For more details, see here for the original circuit. The above changes have been made since.

So, when I get a chance I might try 75-100nF across R. If that doesn't work, I'm going to scope it to see where the oscillations are and what the frequency distribution is at that point. Hopefully, we can damp these suckers.

Pinky's Brain

Sun Jan 13 2013, 04:51PM

Registered Member #2901 Joined: Thu Jun 03 2010, 01:25PM
Location:
Posts: 837

Of course it might just be that their setup accidentally had enough parasitic inductance to make it all work.

I think the grid resistor works against you, it reduces the stiffness of your current source because of miller capacitance. To introduce a low pass in the regulation I'd just use a gate resistor, will low pass together with the MOSFET input capacitance.

Sulaiman

Sun Jan 13 2013, 05:00PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3141

Yes, a resistor from op-amp output to transistor gate is a good idea,
adding gate-source capacitance will make sure that there is one dominant pole in the loop, e.g. 10 kOhm and 100 nF = 16 kHz cut-off frequency.

sngecko

Sun Jan 13 2013, 07:00PM

Registered Member #3447 Joined: Fri Nov 26 2010, 11:10PM
Location: North Jersey
Posts: 97

There is a 100 Ohm gate resistor not shown in the schematic above. Sorry, that diagram was just to show the general cascode arrangement and grid resistor.

sngecko

Sun Jan 13 2013, 08:09PM

Registered Member #3447 Joined: Fri Nov 26 2010, 11:10PM
Location: North Jersey
Posts: 97

What difference might there be between (1) shunting the AC component around R with, say, a 100nF capacitor and (2) shunting the AC component to ground right at the triode cathode?

Just so you know, that cathode already has some overvoltage protection on it in the form of a MOV and TVS. I just measured its capacitance to ground at 3nF.

Pinky's Brain

Sun Jan 13 2013, 08:29PM

Registered Member #2901 Joined: Thu Jun 03 2010, 01:25PM
Location:
Posts: 837

wrote ...

What difference might there be between (1) shunting the AC component around R with, say, a 100nF capacitor and (2) shunting the AC component to ground right at the triode cathode?

Anything at the grid is going to be affected by the miller effect ... which is why generally the gate/base/grid is just plain tied to a voltage or ground in a cascode, that's it's modus operandi.

PS. I don't think it's oscillating as such, more probably arcing instead of a glow discharge.

sngecko

Sun Jan 13 2013, 08:35PM

Registered Member #3447 Joined: Fri Nov 26 2010, 11:10PM
Location: North Jersey
Posts: 97

It sure looks like a perfect glow discharge. Both the laser tubes and neon bulbs I used glowed merrily just as they should and responded appropriately when I increased and decreased the current control. The only problem was if I increased the current to above a mA or so. Then the squeal would get louder as I turned it up and the fans would get progressively slower.

Pinky's Brain

Sun Jan 13 2013, 09:44PM

Registered Member #2901 Joined: Thu Jun 03 2010, 01:25PM
Location:
Posts: 837

Do the fans drop in speed because the power supply is dipping or are they thermally controlled based on the temperature of something, if so what?

sngecko

Sun Jan 13 2013, 10:04PM

Registered Member #3447 Joined: Fri Nov 26 2010, 11:10PM
Location: North Jersey
Posts: 97

I once tried an experiment by replacing the 24V switching power supply with a battery (the one running all fans). The same odd fan behavior occurred. I concluded that the 24V switcher was not the culprit.

My current theory is that the oscillation sends weird RF signals throughout the case, increasing as the current control in increased, confusing the brushless fan controllers.

previous 1 2 3 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob