Forums :: NEW! Web Resources :: Features :: Photo Gallery :: Vintage Radio Shows :: Archives
Support This Site: Contributors :: Advertise


It is currently Oct Wed 27, 2021 5:46 am


All times are UTC [ DST ]





Post New Topic Post Reply  [ 83 posts ]  Go to page Previous  1, 2, 3, 4, 5
Author Message
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Aug Fri 06, 2021 9:49 pm 
Member
User avatar

Joined: Nov Mon 02, 2009 7:01 am
Posts: 5384
Location: Lincoln City, OR 97367
Greetings to the Forum:

Something that bb.odin said combined with a comment from my brother Jerry set off some alarm bells. bb.odin mentioned the effect that loading the other half of the dual voltage supply (the 800 V PA supply) would have on the circuit. I mentioned to Jerry that I ought to try loading the HV supply to full current and he made a comment to the effect that it wouldn't make any difference in the simulator since I was using ideal voltage sources in place of a real transformer. Jery turned out to be right about that, but I went out to the garage and measured the resistance of the high voltage winding in the actual Hallicrafters power supply. It turned out to be a bit over 26 ohms. So.... I added 13.5 ohms in series with each ideal voltage source and the whole thing fell apart. I could not maintain regulation over the required current range.

The problem was that with the supply completely unloaded, there was virtually no current drawn by the Darlington pair and thus it was necessary to size the series resistor for the zener string to limit their dissipation under that condition. When the necessary resistance was employed to satisfy this end of the load conditions, the available current through the zener dropping resistor was insufficient to supply the base of the Darlington pair under full normal load. Thus the zener current went to zero and the regulation was lost.

The problem turned out to be soluble fairly easily, once I managed to eliminate the short circuit between the headphones. :D

The value of the input capacitor in the Pi network power supply filter was carefully chosen to prevent the supply voltage from exceeding the 400 volt limit for the TIP50 pass transistor that was in the first design. This filter design was never re-thought until the regulation drop-out problem was examined in detail. It turns out that the filter exhibits very poor regulation between full and no load because of the small value of the input capacitor. It finally occurred to me that since I had long ago discarded the TIP50 and LR8N devices in favor of 800 volt devices, the voltage limitation on the filter was no longer necessary and was, in fact, highly undesirable. The first change, therefore, was to increase the input filter cap to 33uF.

While the voltage was now higher, the supply was much "stiffer", enabling me to choose a value for the zener series resistor that would allow the zeners to survive the no load condition and still regulate at the full load condition. Stability issues forced me to split the series resistor into two parts with a filter capacitor in the middle.

Limiting the transients that occurred with sudden transitions from normal load to short and back again called for some special treatment; the resistor and capacitor between the two halves of the Darlington helped a bit here.

The last and most difficult problem was that bypass capacitor values became very critical to prevent the whole thing from breaking into oscillation. I finally traced the oscillation to Q6 (the fold-back transistor). Unfortunately, the usual things one does to damp oscillations were unusable. Adding capacitance in the usual places slowed the circuit's response to the sudden short condition with the result that Q4 (the pass transistor) was seeing power excursions well over its permitted dissipation. I finally managed to kill the circuit Q enough to calm the oscillations down to a fairly minor level by splitting the resistance in series with the Darlington input and connecting the fold-back circuit (Q6 collector) to the middle.

The circuit now behaves reasonably well in all of the stress scenarios that I could put upon it. It also copes with the 800 volt supply being fully unloaded to fully loaded while being subjected to sudden shorts and removal and re-application of load. Regulation is regained fairly quickly in all cases.

As Jerry points out, a simulation's accuracy depends on how closely its parameters approximate those of the real world. I hope I haven't overlooked anything else.

Here's the final schematic. You will note that I have left in place the pair of timed switches I used to simulate the worst case transient conditions. The sudden short switch is disconnected and the sudden load / unload switch is jumpered. All I have to do to run the various tests is to break the short around one switch and / or connect the other.

Attachment:
Regulator All Discrete Darlington Vtest17.JPG
Regulator All Discrete Darlington Vtest17.JPG [ 266.38 KiB | Viewed 339 times ]


I am beginning to sound like the boy who cried "Wolf!". I hope I really have it right this time.

There is a feature in MicroCap-12 that will translate the MicroCap-12 circuit file into a Spice circuit file. I'm not sure how it would cope with the switches I use for causing the simulated failures since Spice has no such features. However, if anyone wishes to run the circuit in one of the Spice variants and try to break it, I will do the translation and provide the file. Please PM me with an E-mail address and I will send it to you. Unfortunately, I cannot post it here as the Forum software does not support such files.

Thanks again to everyone. Y'all have been very patient and helpful.

73,

_________________
Jim T.
KB6GM


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Aug Sat 07, 2021 5:35 pm 
Member
User avatar

Joined: Jun Sun 22, 2008 2:03 am
Posts: 1062
Location: Burke, VA 22015
Jthorusen wrote:
The last and most difficult problem was that bypass capacitor values became very critical to prevent the whole thing from breaking into oscillation.

That's a classical problem of an emitter follower with a parallel R-C load. This same problem could occur with an emitter follower used as an infinite impedance AM detector. If the filter capacitor is isolated from the emitter load resistor by too small a resistor (which forms an R-C lowpass filter), the transistor could break into oscillation.

Your circuit already has a current sensing resistor before the parallel RC load in the emitter circuit. That resistor provides some isolation. Since there is a current limiting transistor which forms a closed-loop, I don't know how effective that isolation resistor is.

Here's one thing you can do with the simulator to isolate the problem. Plot the input impedance as a function of frequency of the emitter follower with feedback. Check to see if negative resistance (the real part of Zin) occurs and its frequency range. To counteract this negative resistance, one way is to insert in the input circuit a large enough resistor shunted by an inductor which presents a large enough reactance at the oscillation frequency. This inductor could be one or several lossy ferrite beads in series with any resistors already in the input circuit so not to change the bias current. There are a few other techniques too such as a simple resistor in the input circuit as you have done. An inductor shunt will permit a higher resistance without affecting the dc bias.

Check this paper by Chessman and Sokal for a more detailed analysis:
Attachment:

_________________
Binh


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Aug Sat 07, 2021 11:36 pm 
Member
User avatar

Joined: Nov Mon 02, 2009 7:01 am
Posts: 5384
Location: Lincoln City, OR 97367
Greetings to bb.odin and the Forum:

Thank You for the interesting comment and the paper. I have not yet had a chance to look over the paper in detail, but your comment about using ferrite beads prompted me to look at the first page of the paper and note that the oscillation usually occurs at 50 KHz or higher in frequency.

Jerry and I made a crude approximation of the frequency of oscillation based on the simulation graphs. It is occurring at around 330 Hz or so.... too low a frequency to be quenched with a ferrite bead. The other difficulty is that the oscillation isn't occurring in the emitter follower circuit per se.... it is occurring in the fold-back control transistor Q6. Hence the mitigation (but not an entire cure) of the problem effected by adding some resistance in the control path.

Fortunately, the oscillation only occurs when the short circuit condition is present or at the ending transition of either a short or a unloaded condition. The ending transition instances damp out very quickly, so I have decided to leave well enough alone. (The oscillation when a short is actually present is of very low magnitude.)

Thank You again for your diligence in addressing what I thought would be a simple endeavor, but turned out to be quite a thorny engineering problem.

Regards,

_________________
Jim T.
KB6GM


Top
 Profile  
 
Post New Topic Post Reply  [ 83 posts ]  Go to page Previous  1, 2, 3, 4, 5

All times are UTC [ DST ]


Who is online

Users browsing this forum: No registered users and 12 guests



Search for:
Jump to:  


































Privacy Policy :: Powered by phpBB