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


It is currently Oct Wed 20, 2021 2:45 pm


All times are UTC [ DST ]





Post New Topic Post Reply  [ 83 posts ]  Go to page Previous  1, 2, 3, 4, 5  Next
Author Message
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Sun 25, 2021 5:07 pm 
Member

Joined: Aug Sat 17, 2019 8:55 pm
Posts: 101
Didn't read the whole thread - seems to be some good suggestions. I agree that the TIP50 may not be the best choice for the pass transistor. But if the issue is a current surge at startup, you might want to try simply adding a resistor as shown. Try 1K, but may need to go a bit lower.


Attachments:
Screenshot 2021-07-25 110107.jpg
Screenshot 2021-07-25 110107.jpg [ 58.08 KiB | Viewed 956 times ]
Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Sun 25, 2021 10:24 pm 
Member

Joined: Jan Thu 16, 2020 12:29 am
Posts: 1563
bb.odin wrote:
See this article by our late Leigh Bassett.
https://antiquewireless.org/wp-content/uploads/37-a_solid-state_filter_choke_or_field_coil_replacement.pdf
Attachment:
HV_Reg_modified.jpg


The design is incomplete, as the designer did not specify the thermal properties of the heat sink.

In addition this was a 100mA supply, and I would agree in that case that the TIP50 is ok with a min current gain of 30, but for a 200mA supply my view is it is better to use the Darlington I suggested which has a min current gain of 50 (also has B-E protection diode, it is reverse breakdown of that junction that likely cause the failure, not forward current as the bulk of that is sourced via the collector, so it needs a plain diode on the B-E, not a zener). Generally, it is safe to assume in a power supply design that the power transistor has a min gain of 15 or 20, regardless of the spec sheet. That way you don't get caught out with a marginal condition. In earlier times the rule of thumb assumption was 10. So one assumed the single current gain of a single BJT was 10, and a darlington was 100, even tough it could be 50 and 250 or more respectively for individual specimens.


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Mon 26, 2021 12:29 am 
Member
User avatar

Joined: Nov Mon 02, 2009 7:01 am
Posts: 5377
Location: Lincoln City, OR 97367
Greetings to ACORNVALVE and the Forum:

The supplier of the heat sink did not supply a specification for Degrees C per watt. Based on temperature measurements, the heat sink is roughly 2.5 degrees C per watt and if you add the loss across the insulating pad, it is about 3 degrees C per watt.

As far as the failure mode, upon further analysis, I think it was caused by exceeding the collector to emitter breakdown voltage. Since the regulator is feeding a load consisting only of resistance and capacitance, I don't see any possible mechanism whereby the base could become negative with regard to the emitter.

That situation WILL exist, however if I decide to bridge the pass transistor with a resistance, so I will have to add the diode. Upon close perusal of the device data sheets, a Zener across the base-emitter junction is valueless. The base voltage with respect to the emitter in the forward direction whereby the transistor is biased on cannot exceed .7 volt or so. This is the case up until the point where the maximum permissible base current for the device is exceeded.... at which point the device fails, all without ever firing the Zener.

The reverse protection diode is a good idea and will be added. I am stuck with 1N4007 diodes for now, but the next time I order, I would appreciate a recommendation for a diode with similar specifications to the 1N4007 but with faster switch times.

Thanks,

_________________
Jim T.
KB6GM


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Mon 26, 2021 12:44 am 
Member
User avatar

Joined: Nov Mon 02, 2009 7:01 am
Posts: 5377
Location: Lincoln City, OR 97367
Greetings to the Forum:

I just realized that I forgot to post the above message after proof-reading it before I went out to the garage to try the experiment. This will explain the timing of these posts.

Well... no joy for the simple solution... at least not so far. It works fine with slow ramp-up with a variac. I tried plugging it directly into the wall and the results were identical to what happened last time. The first time, it worked fine. The second time, the TIP50 shorted base-emitter. Interestingly enough, the collector was not involved, at least as far as my ohm meter is concerned. It shows open to both emitter and base, unlike the previous failure which shorted everything together.

I removed the over-voltage protection circuitry, so the output fuse did not blow. However, the 1 amp input fuse did. I am still trying to figure out the current path that caused that.... perhaps the bad transistor's collector does connect to the emitter with more voltage applied than my digital meter can supply. Possibly the LR8N shorted, although it should not have owing to its internal over-current protection. I didn't check to see... I'll do that and report back but probably not today.

Anyway, this was with a 1N4005 diode installed as a protection diode for the base-emitter junction in the TIP50. Either it does not turn on fast enough or there is some other problem.

.... Ah, well.... back to the drawing board. The parts necessary to construct the Darlington pair are on order. In the meanwhile, I see that other high voltage devices have built-in resistors across the emitter-base junction. I may try that. The TIP50's are cheap enough to be considered expendable for experimentation, although I have only three left.

Regards,

_________________
Jim T.
KB6GM


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Mon 26, 2021 2:29 am 
Member

Joined: Jan Tue 16, 2007 7:02 am
Posts: 4716
Location: Lexington, KY USA
It might be useful to re-post the circuit as it stands now.

The TIP 50 cannot take simultaneous voltage and current. If you plan series production, it might make sense to try to make it work, but for one or ten units, just spring for a bigger transistor to start with.

Even with a an oversize pass transistor, the circuit will benefit from a fold-back current limit. It should be practical to make the supply survive not only being turned on, but someone's screwdriver or probe slipping somewhere in the powered device.

The LR8 is also very limited as to output current when there is voltage across it. And you want to keep it cool; when hot, it is less robust. Try to keep the load at 1mA or so.

Darlington transistors with internal base emitter resistors have less current gain at low collector currents. The resistors hog the base terminal current. The data sheets emphasize the gain at high currents, but usually there is a curve, or some way to get an idea. Remember, the curves provided are typical, not worst case.

With present day technology, the most rugged devices at any price point appear to be IGBTs. Some are specified to dissipate full continuous power at the full collector voltage. Use of one of these would also solve the current gain problem.

The LR8 regulator data sheet does not seem to specify any output capacitor. However all the circuits shown on the sheet seem to have a 1uF capacitor. This cap can make it more difficult to survive transient overloads. Some regulator IC P/Ns are not stable without a certain amount of capacitance on the input and or output pin. It would be interesting to know more about the actual LR8 requirements. Did anyone find an app note?

Ted


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Mon 26, 2021 6:49 am 
Member
User avatar

Joined: Nov Mon 02, 2009 7:01 am
Posts: 5377
Location: Lincoln City, OR 97367
Greetings to Ted and the Forum:

Regarding the capacitor on the output: Here is a quote from the LR8N data sheet:

Quote:
To maintain stability, a bypass capacitor of 1 uF or larger and a minimum DC output current of 500 uA are required.


I am pretty sure I said this before, but can't find my post, so here it is again. I do not wish to get into any instability problems.... I am having enough trouble already. :)

As far as Ylli's suggestion of adding a resistor in series with the base of the TIP50 to slow the turn-on time, this would not be practicable without the bridging resistor I have added. With no load on the power supply, its output rises to about 420 volts. The sole load on the supply (before the latest modification) is the regulated load and all of its current goes through the pass transistor. If the pass transistor is cut off, then its emitter will be at zero volts and its collector will be at 420 volts which exceeds the maximum 400 Vceo. With the bridging resistor, this should be possible, but I am not convinced that a soft start is the solution, although I would be willing to try it.

Regarding your suggestions, Ted... I am running up against limitations in component availability. The LR8N is the only regulator that I know of that will do the job. Its low output current necessitates a pass transistor with high current gain. This is either the TIP50 or similar device. All of the devices currently available with higher Vceo ratings are very low current gain. The KSC5603 is at the top of the list with a gain of 20 depending on temperature and load; the rest of the available devices go down from there. Even the KSC5603 may be disappearing; Digikey had only 7 in stock and I bought 5 of them. So, if I can't figure out a simple way to use the TIP50, I will have to go back to plan B which is to construct my own Darlington out of separate components. This is also potentially fraught with problems as I will have twice as many devices to blow up with unexpected transient behavior.

I agree that another schematic should be supplied showing the current configuration; please find it below.

Attachment:
PS-500 Regulator Schematic end to end V3.pdf [104.15 KiB]
Downloaded 36 times


Thanks Again for all the help and patience exhibited by all.

Edit: Error in first drawing - 1N4007 protection diode from TIP50 E to C omitted... correction made.

Regards,

_________________
Jim T.
KB6GM


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Mon 26, 2021 7:05 am 
Member

Joined: Jan Thu 16, 2020 12:29 am
Posts: 1563
Jthorusen wrote:
Greetings to ACORNVALVE and the Forum:

The supplier of the heat sink did not supply a specification for Degrees C per watt. Based on temperature measurements, the heat sink is roughly 2.5 degrees C per watt and if you add the loss across the insulating pad, it is about 3 degrees C per watt.

As far as the failure mode, upon further analysis, I think it was caused by exceeding the collector to emitter breakdown voltage. Since the regulator is feeding a load consisting only of resistance and capacitance, I don't see any possible mechanism whereby the base could become negative with regard to the emitter.

Thanks,


The common mechanism of the pass transistor failing on a power cycling test, when it is an emitter follower type, relates to the charge storage on the output load. You can verify this with a simple experiment on it, if you are prepared to sacrifice a TIP50.

If the drive voltage at the base drops more than about 7 to 10V lower than the voltage at the emitter, the B-E junction zeners. Depending on the impedance in the base circuit and the charge stored at the output the effects are very destructive. To give an example, as an experiment, set the output voltage to say 200V into a resistive load. Get the same value output filter cap you have, charge it to 300V, and connect it to the regulated supply output to discharge back into the transistor's emitter and causing reverse breakdown of the B-E junction, the transistor is immediately destroyed. In essence, the same thing happens with power cycling because of the charge storage on the output filter capacitor and the base drive voltage coming and going with power cycling.

Generally there are two fixes for this problem, to prevent destruction of the pass transistor.

One is to have the diode across the B-E junction (as is seen inside the MJ10025 Darlngton). This prevents the B-E junction being reverse biased by more than about 0.8 to 1V. The method works better also when you have included some base resistance between the regulator output and the transistor base, as this limits the discharge current (from the output filter cap), via the B-E diode into the regulator IC's output, 47 to 100R is fine with the darlington. The diode backwards across the regulator output input is also helpful, to protect the regulator, if its output is forced above its input, it can also fail by the same mechanism as the transistor.

It is possible the transistor had its collector voltage exceeded too, as another mechanism of failure.

The other method is to place a diode directly in series with the pass transistor's emitter, that also prevents the pass transistor destruction if the output voltage goes (is transiently forced) higher than the transistor's base voltage as the diode becomes reversed biased.


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Mon 26, 2021 10:55 pm 
Member
User avatar

Joined: Nov Mon 02, 2009 7:01 am
Posts: 5377
Location: Lincoln City, OR 97367
Greetings to ACORNVALVE and the Forum:

I understand what you are saying, but I don't believe that it is the cause in this case. The regulator is working into a simulated load, consisting of a series string of high power wire-wound resistors totaling 1400 ohms. The 33 uF capacitor is across this string of resistors. I also have added a pair of pin jacks so that I can plug a meter in conveniently to monitor the regulated output under load. The discharge time for that 33 uF across 1400 ohms is less than a second. The only other capacitor across the output is the 1 uF cap on the regulator board and it too, sees the same 1400 ohm load. According to my Heathkit DVM which I have assigned the duty of monitoring the regulator output, the voltage is always zero before startup in my transient tests.

I have been attempting to simulate the circuit with the MicroCap 12 simulation software, but as of yet, I do not have a model for the LR8 regulator. I have been attempting to simulate the circuit using a voltage source instead of the regulator chip. The most recent simulation I have run was with a simulated pulse so that I could delay the start of the circuit after the high voltage had been applied (assuming the regulator has a turn-on delay) and I can change the rise and fall times of the pulse and its delay upon starting the simulation. I added a series resistor to the DC source in the simulation to limit its current output to 20 mA so as to try to get closer to the LR8. I do not understand what is happening there because if I read the simulation results correctly, the base current for the TIP50 exceeds 300 mA.... I don't have any idea where that is coming from. Unfortunately, the simulator is sloppy about proper use of units; and I am not familiar enough with it to see if the 300 "m" that I am seeing on the graph is 300 mA or 300/1000 of the scale factor which is already mA which would make it 300 uA.

Oh, well.... I have three TIP50 left and am willing to buy more if necessary.... they are less than a dollar a throw in quantities of 10 or more. I don't think I need to run your experiment above, though, because there simply isn't any residual charge left in the load.

My next test (after replacing the shorted TIP50 that is in there) is to physically tie the base and emitter together and repeat the transient test. If I can cycle the power a few times with out anything unusual happening, then I know the problem is with the base drive... whether too much positive or too much negative has yet to be determined.

If the TIP50 still fails under these conditions, then I know that I have to drop that device and construct the Darlington.... parts are on order for that.

BTW, if you wish any of the schematic files, simulator circuit files or board layouts (boards not yet ordered; the design is still in flux), please let me know and I will post them here or send them to you directly. That goes for anyone on the Forum.

Thanks again for sticking with me on this somewhat rocky development road.

Regards,

_________________
Jim T.
KB6GM


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Tue 27, 2021 8:15 am 
Member

Joined: Jan Thu 16, 2020 12:29 am
Posts: 1563
Jthorusen wrote:
Greetings to ACORNVALVE and the Forum:

I understand what you are saying, but I don't believe that it is the cause in this case.

Regards,


If you don't believe it, it appears that you are contradicting your description of the failure in your first post, you wrote:

"I first energized the supply with a variac.... all well and good. It ramps up and regulates without difficulty. I was able to determine that the range of line voltages one was likely to encounter would neither create too much dissipation for the heat sink system to handle nor drop out of regulation. I was just about to declare it finished when I thought I'd better try it with transients just to be sure. I plugged it directly into the wall and pulled the plug several times. No problems. I then remembered that the load it was driving had 30 uF of capacitance across its input. I soldered a 33 uF / 450 volt cap across my test load and repeated the transient test. The first time, fine. The second time completely shorted the pass transistor and took out the regulator chip along with it."

It likely failed therefore because of the charge(and energy storage ) in the 33uF capacitor, when the regulator's output cycled transient lower at the transistors base, the transistor failed as expected and took out the regulator. You could always repeat the experiment, with and without the 33uF cap and confirm your initial experiment.

In many circuits where emitter followers are pass elements for the regulator, the load currents cause collapse of the power supply output prior to the reference voltage at the base dropping significantly below the output voltage if at all. You would also not see this sort of thing in low voltage supplies very often. It is in the higher voltage range supplies like yours the event is more problematic. I am very familiar with this failure mode, I have had it happen with a 170V supply I built with a high voltage transistor with a similar configuration.

Some resistance (as mentioned) added in series with the base is helpful to protect the regulator.

Only trying to help you of course, so you don't have to take any of my advice if you don't want to, but be prepared for a higher transistor and regulator body count.


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Wed 28, 2021 1:26 am 
Member
User avatar

Joined: Nov Mon 02, 2009 7:01 am
Posts: 5377
Location: Lincoln City, OR 97367
Greetings to ACORNVALVE and the Forum:

I guess I don't understand what you are saying. In no case did the 33uF capacitor retain any charge during the transient tests. I have a DVM monitoring the voltage across the load. Since the 33uF cap is across the load, the DVM would also show any voltage present in the capacitor. The transient test was performed by plugging the supply into the wall and observing the output voltage. The first time, the voltage rose (faster than the DVM could follow, so therefore very quickly) to the desired regulated value of 270 volts. The power supply was then unplugged. Because the time constant for a capacitor of 33uF and 1400 ohms (the load resistance) is about .05 second, the 33 uF capacitance is fully discharged very quickly. The voltmeter read zero when I plugged the power supply in for the second test which resulted in the failure.

Just now, I performed another test, which consisted of shorting the base and emitter of a new TIP50 together and repeating the transient test. Because the emitter and collector of the TIP50 are bridged with a 1.5K 10 watt resistor, the supply voltage could (theoretically) not rise above the maximum permitted 400 Vceo. This appears to be the case, because I was able to repeat the transient test multiple times without the TIP50 failing. Since, with its base and emitter connected together, it is cut off, all current supplied to the load passed through the 1.5K resistor bridging the TIP50. Thus the load voltage was only 170 volts and needless to say, I did not leave the supply plugged in for more than a couple of seconds for each each test in order to limit heating of the 1.5K resistor which, without the pass transistor conducting, is operating in excess of its designed dissipation.

So, at this point, I can definitively say that the failure of the TIP50 is due to the signal injected at its base. I can also say, I think with some confidence, that that signal is not negative going because there is no charge left in the 33uF cap between transient tests.

Therefore, my conclusion is that the regulator chip is shorting immediately and applying full power supply voltage to the base of the TIP50. If the short occurs more or less instantaneously, then the maximum permissible base current of the TIP50 would be exceeded because the emitter is looking into a fully discharged 33uF cap while the base is looking at 300 plus volts.

If I'm right, the question now becomes how to protect the regulator.

Thanks again for your input and effort.

Regards,

_________________
Jim T.
KB6GM


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Wed 28, 2021 8:47 am 
Member

Joined: Jan Thu 16, 2020 12:29 am
Posts: 1563
A DVM is a useless tool for any kind of transient analysis. Some take more than a second to sample and display a reading.

The sensible move is to limit the possible current in the pathway between the regulator output and the transistor's base. With a 0.2A load and the darlington min gain 50, the max base current will be 4mA. So if you place say a 470R resistor between the regulator's output and transistor's base, its only going to drop a couple of volts. Then regardless of any calamity with the output transistor, or stored charge effects, it will limit the current that can be delivered to the regulator's output and probably prevent its failure.


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Thu 29, 2021 7:39 am 
Member
User avatar

Joined: Nov Mon 02, 2009 7:01 am
Posts: 5377
Location: Lincoln City, OR 97367
Greetings to ACORNVALVE and the Forum:

ACORNVALVE wrote:
A DVM is a useless tool for any kind of transient analysis. Some take more than a second to sample and display a reading.


Yes, of course. I am not doing transient analysis with a DVM. It is used only to measure initial conditions before the transient test... and of course, when it goes to zero or the plus rail with power applied, it indicates that the test failed.

I have been unsuccessful in modeling the circuit with computer simulation, so I have made up a simplified circuit to test the step-start concept. The load was rebuilt to be switchable between 1400 ohms and 26K which should limit the output current of the LR8N to a safe value. Of course, having just put the thing together on a breadboard, I thought I'd better sneak up on the circuit slowly with a variac to make sure it was working normally before attempting transient tests.

Interestingly enough, it would start to regulate, but would then loose all output. When I throttled the variac down, it recovered, only to fail as I approached 100 VAC of line voltage. The behavior had me puzzled for a minute because I was so fixed on an over current condition. However, the light bulb came on and I put my pinkie on the device. It was too hot to hold onto. The thing was going into thermal limit.

Obviously, I made a rather large error. I forgot the other problem waiting to bite the unwary working with solid-state devices and high voltage: device dissipation. In looking at the data sheet, the device can only dissipate about 3/4 of a watt and that is at 2.5 degrees C! Dropping from 400 volts (worst case) to 270 volts is 1.6 watts. The most probable steady state dissipation (assuming 10 mA output current and 330 volts input) is .6 watts. This is why it is able to skate by when ramped up slowly, but blows up when looking into a capacitive load. The combination of high current and high differential voltage is instantaneously fatal. Worst case: 400 volts supply voltage, zero output voltage (33 uF cap fully discharged) and 20 mA current (the internal current limiter) = 8 watts! In such a small die, I am theorizing that destruction is immediate.

To limit the device dissipation to 1/4 watt (an assumed safe value) under the worst case assumption above, the output current must be limited to 625 uAmps. The solution is a home made triple Darlington... at least at first glance. According to my calculations, with three stages of current gain (assumed to be 20) I can translate 500 microamps into 4 amperes. I think actually that in order to maintain the LR8N minimum output current, I will need only two stages with a reasonable amount of gain. There will be a lot of breadboarding and experimentation. I am going broke buying parts as the parts orders are always behind the design curve, but hopefully, I can chalk it up to a learning experience. After all, how much would a 3-unit EE course covering this subject cost me? :D

Thanks Again to all who posted..... I will keep this thread apprised of new developments.

73,

_________________
Jim T.
KB6GM


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Thu 29, 2021 8:10 am 
Member

Joined: Jan Thu 16, 2020 12:29 am
Posts: 1563
Jthorusen wrote:

I forgot the other problem waiting to bite the unwary working with solid-state devices and high voltage: device dissipation. In looking at the data sheet, the device can only dissipate about 3/4 of a watt and that is at 2.5 degrees C! Dropping from 400 volts (worst case) to 270 volts is 1.6 watts. The most probable steady state dissipation (assuming 10 mA output current and 330 volts input) is .6 watts. This is why it is able to skate by when ramped up slowly, but blows up when looking into a capacitive load. The combination of high current and high differential voltage is instantaneously fatal. Worst case: 400 volts supply voltage, zero output voltage (33 uF cap fully discharged) and 20 mA current (the internal current limiter) = 8 watts! In such a small die, I am theorizing that destruction is immediate.

73,


You must have missed what I posted on this thread on July 24:

Perhaps one other thing, make sure to put the transistor on a substantial heat sink. Although the current is low, high voltage supplies pose and equal problem with power dissipation, especially as the output voltage gets dialed down. I'm not sure how low you might adjust the output voltage down. (400-270) x 0.2A current, the power would be 26W, so ideally to keep the transistor's case temp respectable you need a decent sized heat sink, you can get these with the TO_3 holes already in them, if not easy to drill using a TO-3 mica washer as a template.

This is another reason why I suggested the MJ10025 Darlington bolted to a less than 1 dec C/watt heatsink, and gave the links to the two parts at Newark, to save you from these thermal dilemmas. Both these items were less than $10 each. I still cannot see why you couldn't use your existing heatsink (as the other was too large) and screw the MJ10025 to it , which would save you having to create darlington's from scratch and thermal management problems.


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Thu 29, 2021 4:07 pm 
Member

Joined: Jan Tue 16, 2007 7:02 am
Posts: 4716
Location: Lexington, KY USA
The regulator cannot be counted on to deliver good results when operated at or near its maximum low voltage current limit. It will do better when the load current is just slightly above the specified minimum. Note that one of the SMT packages offers quite a bit more dissipation than the TO-92.

It is not clear why the regulator is failing. It is supposed to protect itself when that is possible. Clearly the protection is not completely foolproof. This is true of the common low voltage regulators, as well. A determined effort can blow them up.

The data sheet tells us that the TIP50 will quickly fail when it sees both volts and amps at the same time. If you really really want to use this transistor, several in parallel, with a proper current limit circuit, will probably work reliably. Spreading the heat around will also make heatsinking easier.

The MJ10025 will not solve the regulator current problem, as its gain is low at 200mA Ic. It is optimized to quickly switch several amps, not to be rugged in a linear application. Best not to seek out an expensive obsolete part that will not work anyway. Those curves on the data sheet are typical values at 25C. Some units might take 20mA base current to get 200mA at the collector. Probably none will work with 1mA. The fine print by that base resistor does say 100 ohms.

http://www.solidstateinc.us/specsearch/ ... 24-ssi.pdf

The large heatsink is a very good idea. As previously noted, a hot part is not as rugged as a cool one. A lot of those datasheet numbers are for 25C. Things go south as the temperature rises. This applies to the regulator IC as well as the pass transistor.

I think the last circuit we saw posted still includes no current limiting for the pass transistor. There is no practical way to make an inexpensive transistor reliable without this. Fold-back is a useful feature to include, to reduce the current limit when there is a higher voltage across the pass transistor.

It is easy to find transistors with high voltage ratings. The regulator IC, not so much. A zener type TVS across it might be a help here. Put some series resistance between the DC source and the regulator input to avoid hammering the TVS if there is a transient. Some series resistance for the pass transistor collector might also be helpful.

Ted


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Fri 30, 2021 12:17 am 
Member

Joined: Jan Thu 16, 2020 12:29 am
Posts: 1563
Usually Lurking wrote:

The MJ10025 will not solve the regulator current problem, as its gain is low at 200mA Ic. It is optimized to quickly switch several amps, not to be rugged in a linear application. Best not to seek out an expensive obsolete part that will not work anyway. Those curves on the data sheet are typical values at 25C. Some units might take 20mA base current to get 200mA at the collector. Probably none will work with 1mA. The fine print by that base resistor does say 100 ohms.

http://www.solidstateinc.us/specsearch/ ... 24-ssi.pdf

Ted

Ted,

I disagree, the absolute worse case current gain at 200mA collector current, at low temperatures is still over 25 and it increases with junction heating, to over 50, but that is also a worst case specimen.(in this respect things don't go South with heating as you said, they improve!)The base-emitter voltage drops in the usual way with heating, which is not unhelpful.

The transistor is totally perfect for the application, the combination of existing internal diodes, saving external components, its phenomenally low thermal resistance between the junction and case a 0.7 degC/ watt, the low thermal resistance to the heatsink (0.35 deg C/watt with mica washer due to the TO-3 package). The very high voltage ratings. Plus its not even expensive at less than $10. It doesn't have to work with a 1mA drive current, the regulator can source 10mA. It is also very difficult to get near these low thermal resistances with a common flat pack epoxy device. Junction to case for that, such as the TIP50 is a whopping 3 deg C/W, 4 times poorer than the MJ10025 !

The hfe in the data sheet was obtained with the real transistor (including its 100R base resistor) which agreeably wastes about 5 to 6mA of the base current, not a transistor without its base resistor. So, if you look at the graph, with 200mA IC and a hfe of 27, then the base current is 7.4 mA, meaning about 5.5 ma for the resistor and roughly 2mA injected into the transistors actual base, which is enough to support the 200mA collector current in the second transistor and that is a worst case specimen.

I agree though, ideally that value would be higher in the range of 500R to 1K, but the 100R won't stop the part from working. It is superior in every way to the TIP50 for the task.

Obsolete, part, I concede, but the rest of the power supply could be well considered to also contain obsolete parts.And Jim is not making a commercial product, so who cares if the part is obsolete, if you can get it.

My advice would be for Jim to get one, since its a very cheap part, and at least evaluate it for the task and subject the supply to the same power cycling to determine if a failure will occur, or not.


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Fri 30, 2021 1:26 am 
Member
User avatar

Joined: Nov Mon 02, 2009 7:01 am
Posts: 5377
Location: Lincoln City, OR 97367
Greetings to ACORNVALVE, Ted and the Forum:

Thank you for your interest and analyses.

For ACORNVALVE: We seem to be talking apples and oranges. I did read your post and it is not applicable in this case. The TIP50 has a perfectly adequate heat sink; please see the discussion below. The device that is failing is in a TO-92 plastic package (see photo).

Attachment:
LR8N Physical Size.JPG
LR8N Physical Size.JPG [ 723.64 KiB | Viewed 805 times ]


This device is intended to be cooled by conduction through its leads and air surrounding the package. As far as I know, it cannot be attached to a heat sink. A TO-5, yes.... a TO-92, no.

As far as the TIP50 is concerned, it is mounted to an adequate heat sink. I have run the circuit for several hours (it performs beautifully when the input voltage is ramped up slowly) and measured heat sink and device temperatures. With about 78 degrees F ambient (about 26 C) the heat sink stabilizes at about 122 degrees F (50C). I measured the collector tab of the device with an 18" lab grade mercury thermometer by insulating the lower portion of the thermometer and allowing only the bulb tip to contact the collector tab. Heat transfer to the thermometer was aided by placing the bulb tip in a blob of heat sink compound on the collector tab. The thermometer stabilized at 55 degrees C ( 131 degrees F). Since the device is specified to operate up to 150 C, I think the matter of the pass transistor heat sink adequacy can be closed.

The problem is that there is a limit to how rapidly heat can be transferred out of the device itself. This is why all devices have specified maximum powers. For example, I could attach the TIP50 to a heat sink the size of a billboard.... but if I then proceeded to apply a transient of 10 times its rated dissipation, it would instantly fail. The transient that is being applied to the LR8N is approximately 10 times its rated dissipation. Again, no amount of heat sinking will fix that.

For Ted: Unfortunately, this is not a "clean sheet" design. I am severely constrained by a number of factors, including physical room to mount the components and again, physical room limiting the amount of board real-estate I have to work with. Also, the brute force power supply can be modified only slightly, mainly in the filter network. Therefore, the circuit has to be as simple as possible and still work.

I am open to suggestions as to how to implement current limiting in this supply using a 3-terminal regulator. I have not given it much thought because the original supply had no current limiting (other than its internal impedance, which is pretty low) and under most circumstances it is not necessary.... this supply is not a lab bench supply, its output is always securely connected to the same load.

I am going back to the drawing board with a Darlington configuration and I have found some high voltage transistors in TO-92 packages so I can squeeze them onto the board (hopefully). The maximum board size is 3 1/4" X 1 3/4". Actually, the second dimension is 2 1/4", but there is a 1/2 inch overlap where the board is attached to the end of the heat sink so only ground traces may be present there. So, I am having to be very clever to fit the components I already have onto that board. I have a Darlington squeezed onto the board now and I have some room for vertical growth (the current board is only 2.4" high, but this (and a personal loathing for surface mount components) is why I am using TO-92 devices.

I will have to look at some current limiting designs not dependent on the feature being built into the regulator IC and see what I come up with. In the meantime, suggestions for such a design are welcome.

Thanks again for all the help.

73,

_________________
Jim T.
KB6GM


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Fri 30, 2021 2:57 am 
New Member

Joined: Jan Mon 11, 2016 5:54 pm
Posts: 19
Location: Falmouth MA
Been following this one with interest. Since the design survives without the 33uf output cap; why not limit the surge current into the cap? Since this is for a Cyclone, I'm assuming this transformer also provides other voltages? You can either step start the primary, (might not be possible) or you could put a series resistor into either the plus or minus of the output electrolytic, and short this out with a relay or transistor after some desired time interval. The instantaneous dv/dt would be greatly diminished and the pass transistor would survive just fine. 200ms, 500ms? You would need to experiment. A current probe on a digital scope could capture the initial current surge. The esr of the cap, and the impedance of the power supply comes into play here. After all, the output cap is a dead short circuit initially until some voltage ramps up and it essentially becomes an open circuit; at least for DC. So you can either try to find a device big enough to handle the initial surge, which has been suggested, or just limit the startup surge.

Steve

KB1VWC

_________________
Greening the planet, one old radio at a time!


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Fri 30, 2021 6:33 am 
Member

Joined: Jan Thu 16, 2020 12:29 am
Posts: 1563
Jthorusen wrote:
Greetings to ACORNVALVE, Ted and the Forum:

Thank you for your interest and analyses.

For ACORNVALVE: We seem to be talking apples and oranges. I did read your post and it is not applicable in this case. The TIP50 has a perfectly adequate heat sink; please see the discussion below. The device that is failing is in a TO-92 plastic package (see photo).

Attachment:
LR8N Physical Size.JPG


This device is intended to be cooled by conduction through its leads and air surrounding the package. As far as I know, it cannot be attached to a heat sink. A TO-5, yes.... a TO-92, no.


The transient that is being applied to the LR8N is approximately 10 times its rated dissipation. Again, no amount of heat sinking will fix that.

73,


Can you explain how you have measured and/or calculated the transient that the LR8N would be subject to when using a Darlington pass transistor and how you have determined that this would cause the LR8N to fail ?


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Fri 30, 2021 6:39 am 
Member
User avatar

Joined: Nov Mon 02, 2009 7:01 am
Posts: 5377
Location: Lincoln City, OR 97367
Greetings to Steve and the Forum:

Thanks for your input. I will have to look into a step-start for the whole supply. I'm not sure that it will be effective, but it might be an idea. Soft-start using the traditional thermistor approach does not work well for transceivers. Since the current drain is so widely different between transmit and receive, the device is either too high a resistance for one condition or too low for the other.

Limiting the inrush current to the capacitor with a relay is not a viable option. The capacitor is located in the transceiver, not in its power supply. This approach would require extensive modification of the transceiver, which I am not willing to do... I want to preserve it in its normal state for reasons of resale value and the ability to use it with other power supplies.

Thanks to my brother Jerry, I now have a viable model for the LR8N (it is made of ideal components rather than being truly accurate), but it does allow transient modeling. I have found that a 100 mH choke in series with the input to the regulator (along with a Darlington configuration) will solve the problem. Unfortunately, there are no 100 mH chokes available in the current range that I require. They are either intended for high power solid state use (rated 1 ampere with size to match) or they are intended for use with low power circuits on PC boards (7 or 8 mA max). The latter cannot handle the current and the former cannot be fitted into the power supply due to size constraints.

So, at this time, a current limiter (fast acting) for the solid state circuit seems the only practical approach.

Again, Thanks,

_________________
Jim T.
KB6GM


Top
 Profile  
 
 Post subject: Re: High Voltage Regulator Design Issue
PostPosted: Jul Fri 30, 2021 7:07 am 
Member
User avatar

Joined: Nov Mon 02, 2009 7:01 am
Posts: 5377
Location: Lincoln City, OR 97367
Greetings to ACORNVALVE and the Forum:

There have been two different calculations, one by my brother using a simulator and one by myself using pencil and paper. I should point out that these are also apples and oranges; my calculation was NOT for a Darlington pass device, but for the LR8N driving the TIP50 directly. This analysis was intended to explain why the circuit AS BUILT was failing.

Assumptions:

1. That the LR8N current limiter is instantaneously effective.
2. That the supply voltage is 400 volts (worst case).
3. That the voltage on the emitter of the TIP50 was 0 volts (load capacitor fully discharged).

400 volts - 0 volts = 400 volts (voltage across the LR8N; diode drop represented by the emitter base junction of the TIP50 neglected).
Current through LR8N = 20 mA (internal current limit) * 400 volts = 8 watts.

Maximum permissible device dissipation of the LR8N in the TO-220 package is .74 watt (from the factory data sheet).

Conclusion: Transient power is 10 times device rated maximum.

My brother Jerry modeled a Darlington setup with MicroCap-12. He found that the instantaneous emitter current of the TIP50 with all ideal components is over 200 amps!

Of course, that neglects power supply impedance and a number of other factors. It would appear from modeling the Darlington approach that with a full Darlington or with a three stage device that the LR8N current can be reduced to microamperes without too much difficulty. The maximum permissible current with 400 volts applied and zero on its output is 625 uAmps (limiting the device dissipation to 1/4 watt for good safety margin). Unfortunately, the MINIMUM output current for the device is 500 uAmps, so we have a fairly narrow window to match. Also these calculations represent a continuous condition of a shorted output, which is not the case, although if designing for this I will achieve the current limiting that Ted wishes me to place on the circuit.

So, as I said, back to the drawing board.

Suggestions welcome.

Thanks again for the patience of everyone involved.

73,

_________________
Jim T.
KB6GM


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

All times are UTC [ DST ]


Who is online

Users browsing this forum: No registered users and 14 guests



Search for:
Jump to:  


































Privacy Policy :: Powered by phpBB