Jump to content
Electronics-Lab.com Community
Sallala

0-30 Vdc Stabilized Power Supply

Recommended Posts

Sorry, but i am unable to view the pics; Java?

You should upload your pictures to Photobucket (or similar hosting server) then provide the URL etc...
The supply does operate properly but i must admit i jumped on Audioguru concerning a current issue and he may be correct.
Yesterday, when viewing the internal of the P/S i noticed i used a 2A transformer... :o
Additionally, after reviewing the data i compiled the controlled current max was verified at @2.00A  :-[
I implemented the original design without any component substitution other than the main transformer which was reduced in secondary current to 2A.

Phew, i tend to grab/substitute components i have on hand etc...  :-\




Share this post


Link to post
Share on other sites

After drawing the diagram for the PSU I could not resist doing some tests.
I wanted to see the transient response during a step load change.

As the design went through a couple modifications to increase output, it was a good idea to look into these changes and the effects on performance.

I simulated only the voltage regulation stage (current limit circuit removed), with three different drivers. Error amp is TL081

1) BD139  (Ft 100+ MHz)
2) TIP31A  (Ft 3 MHz)
3) 2N2219 (Ft 200 MHz)

Tests were performed at 22 Volt out and a constant load current of 100mA. The current was then stepped up by pulsing a MOSFET and another load resistance. The total switched current was about 1 Amp.

Below are the results.
The blue trace is the pulse control signal to the MOSFET.
The yellow trace is the output load voltage response.
Violet trace is the error amplifier output.

We can see a possible problem with using the TIP31A as a driver. It struggles to catch up with the error amplifier output and causes overshoot and ringing. The problem seems to be that the transistor is just too slow. The phase response plot of the system confirms this as a rapid phase shift at a fairly low frequency that could cause this problem.

post-9230-14279143546833_thumb.gif

Share this post


Link to post
Share on other sites

Hi AN920,

I have to agree on your analysis concerning the biasing (quiescent point) of the TIP31A in reference to the error amplifier outlined.
A TIP31A is usually implemented as an output transistor and/or as a complement to the TIP32A etc...
A transistor with a faster response would be a better choice in this application.

Without trying to apply a large re-selection of silicon one could easily use the original components and simply drop the selected secondary current of the main transformer to avoid current issues.

When i completed the project i used the original design as stated in a prior post with very satisfactory results.

Your graph analysis is very professional and commendable!  ;)

Share this post


Link to post
Share on other sites

Hi Pawel,
The old OP07 should not  be used in this project:
1) It has a different input offset voltage adjustment curcuit requirement. Try it with RV1 removed.
2) Its max output bandwidth is only to about 2.5kHz. The TL081 is 100kHz.
3) It has diodes across its inputs.

Hi AN920,
The TIL31A is an emitter-follower without any gain in this project. It is slow but the 2N3055 output transistor is even slower so I thought the TIL31A would match the 2N3055 transistor well.
The BD139 might have a cooling problem if the supply voltage is high enough for a 30V/3A output and the 2N3055 transistors have low gain with the voltage set low and the current is high.
C6 and C9 could have their values optimized for the TIL31A driver transistor.

Hi Omni,
This project is spec'd with a 30V/3A output but it doesn't do it. Yours with a 2A max is a lot less.

Share this post


Link to post
Share on other sites

Hi Pawel_K

I was finally able to view your pics.
When viewing the NS (near-side) of the PWA (printed wiring assembly) i noticed two resistors that may be incorrect.

1. Lift one side of R7, using a VOM measure the resistance it should measure 0.47 ohms or 1/2 an ohm (not 47 ohms).
   You may have the correct value but it is always a good idea to make sure.
2. R1 should be a 1W type, when viewing your picture it appears to be a 1/2W, the latter could be the reason why it is becoming hot.

Hi Audioguru,

Yes, you are correct concerning the transformer secondary current, a 33% reduction would make a big difference...

Share this post


Link to post
Share on other sites


Hi AN920,
The TIL31A is an emitter-follower without any gain in this project. It is slow but the 2N3055 output transistor is even slower so I thought the TIL31A would match the 2N3055 transistor well.
The BD139 might have a cooling problem if the supply voltage is high enough for a 30V/3A output and the 2N3055 transistors have low gain with the voltage set low and the current is high.
C6 and C9 could have their values optimized for the TIL31A driver transistor.



Much easier said than done.  :)
The LF pole created by the slow TIP31A will cause many problems that won't be easily cured. As it stands now the design is unstable and may oscillate with certain loads. The problem is that the transistor creates a LF pole and the output capacitor another LF pole. With two poles close together you are looking at stability issues. Lowering the gain of the error amp for more bandwidth in order to get stability is an option, but will degrade , PSRR and higher frequency noise rejection. Stability is real reason for concern that needs attention.

Share this post


Link to post
Share on other sites

This is what worries me.

The plots compares the phase margins for:

1) Error amp alone with feedback network, no driver no filter cap C7
2) Error amp, with FB network and driver, no filter cap
3) Error amp, with FB, driver and filter cap
4) Error amp, with FB, driver, 2N3055 and filter cap

Phase margin with the TIP31A driver is too small. Any variation in output cap ESR will bump this into oscillation at around 20-30 kHz. Using a quality low ESR capacitor for C7 to improve ESR voltage drop during current load step will be a disaster as the bode plot shows.

post-9230-14279143547203_thumb.gif

post-9230-14279143547574_thumb.gif

Share this post


Link to post
Share on other sites

Simulator results are only as good as their component models, and frankly, most aren't great... at least the ones I've played with...MicoSim, SIMetrics, Intusoft & Workbench. While the second two are "cheap" the first two are not.


How does the TIP31A introduce a LF pole? It has a 3MHz gain bandwidth product and if the supply is running a max Iout of 3A, and the 2N3055 is at it's min beta of 5 (On-semi datasheet and that's at a Ic of 10A not 3A), the TIP31A isn't working "that hard" and your no where near it's bandwidth limit. If the high frequency zero due to the cap ESR is an issue, add a high frequency pole a decade earlier.

I'm having trouble with your bode's... the form I'm use to is more like this:

post-14697-14279143547747_thumb.png

Share this post


Link to post
Share on other sites

I agree simulators and models  are not perfect but it shows a trend of what to expect. The models will tend to be much more accurate at fairly low frequencies than at the high end, where nonlinear models would be preferred. The problem might not be obvious as most members are using the cheap grade of output  electrolytics with high ESR (1-10+ ohm for C7 ) values. The output may not oscillate but will have lousy load step performance (1A load step will drop 1-10V before the loop catches up).

About the bodes, yes I am also used to the conventional form but the software does not display gain and phase at the same time, which is silly.

The pole created by the TIP31A is the very same problem that trouble audio power amplifier design. If you have a slow driver stage you are looking at troubles. The slower output power stage does not make much additional phase difference. Doing the same simulation with any other slow (compared to the 2N2219 or BD139) transistor show similar problems. I have tried a few TIP's and also some of the higher current BD's like  241,243 and they present the same problem. We can blame the model, but all of them?

Share this post


Link to post
Share on other sites

I think the lead-compensation capacitor C6 sould have its value increased when the driver trransistor is a slow TIP31A. Now it is only 100pf which is nearly nothing.


Increasing phase lead will improve phase margin but at the same time will increase overshoot and ringing badly on a load step. Increasing the value to 330pF shows a 2V overshoot with lots of ringing. At 560pF it is at the point of oscillation with 2.5V overshoot. The oscillation problem just happen a bit higher in frequency. Increase it too much, say 1nF and we get constant oscillation at about 230 kHz, 1Vp-p at a constant full load.

post-9230-14279143547849_thumb.gif

Share this post


Link to post
Share on other sites

Since a higher value for the lead capacitor C6 causes ringing, then additionally using a higher value for the lag capacitor C9 should reduce the ringing. But then the output voltage will have a glitch when the load current is changed. 

Share this post


Link to post
Share on other sites

The BD139 well heat-sinked will be a better choice with none of these problems. One can even look at a better output transistor with better gain to remove power issues in the BD139. It would be nice if we can get something with 2A capability and response of the 139

Share this post


Link to post
Share on other sites

It is amazing that the kit maker uses a little 2N2219 that has a very low power dissipation.
With only 32V across it then it is at its max power dissipation (with a heatsink) with a current of only 56mA. Then the 2N3055 transistor needs to have a current gain of 54. The minimum gain of a 2N3055 is only 25 when it has a current of 3A. Maybe the kit maker selected high gain 2N3055 transistors.

Share this post


Link to post
Share on other sites


How does the TIP31A introduce a LF pole? It has a 3MHz gain bandwidth product and if the supply is running a max Iout of 3A, and the 2N3055 is at it's min beta of 5 (On-semi datasheet and that's at a Ic of 10A not 3A), the TIP31A isn't working "that hard" and your no where near it's bandwidth limit. If the high frequency zero due to the cap ESR is an issue, add a high frequency pole a decade earlier.


Let's analyze the TIP31A from a S-parameter view as measured on a network analyzer using the actual components . We will look at the response of it operating as a simple emitter follower alone with only resistive terminations. This is normally considered to be a very stable configuration. Surprise! :o

Good paper here for IEEE members
http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?tp=&arnumber=1082247&isnumber=23381

Looking at the first plot of a sweep between 0-10 MHz we see that this transistor is unstable with a K factor < 1 for operation above 50 kHz all the way up to 10 MHz. This transistor will oscillate without compensation. We normally aim for a K factor of at least >= 1.2
We can see a very good comparison between the measured S-param data and the model used by the simulator. We can thus assume the model is quite accurate.

Next we add a common 10uF capacitor with an ESR of 7 ohm on the output. Now we see that this setup is unstable for anything  over 70 kHz

Next we take a look at the BD139 in the same configuration. This will be stable under 1.9 MHz (big difference from 50 kHz).

Adding the same output cap we can see stability is not much affected with ample stability at 1 MHz (K = 1.9). Looking at the plot of the original 2N2219 we see a similar trend.

The original designer may have had good reason for selecting the driver he did. Maybe he was aware of the problems like I am. The importance to select a fast driver is just as important in PSU design as in audio amplifiers.

I can't make it any clearer than this.

post-9230-14279143549432_thumb.gif

post-9230-14279143549684_thumb.gif

post-9230-14279143549897_thumb.gif

Share this post


Link to post
Share on other sites

Hi aN920,
You are talking about high frequency changes in load current. Or quick changes in load current.

I think that C9 rolls off the response so that the overall gain is only 2 at 89kHz. Then C6 straightens out the phase response.

Share this post


Link to post
Share on other sites

I think that C9 rolls off the response so that the overall gain is only 2 at 89kHz. Then C6 straightens out the phase response.


That was the intention in the original design with a much better transistor. The whole picture changed. The situation will be the most problematic during load switching of high currents. It will be the worst at high output voltage settings and will get worse with lower ESR values. As-is, a 10uF cap with any ESR lower than 5 ohm will cause oscillation when switching.

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...

×
×
  • Create New...