Hey guys I'm new here this is my first post.
This isn't specifically about audio but I am very much an audio enthusiast and will no doubt be posting asking about some questions for a guitar amp I am in the process of designing.
This post though is about a power supply I have made.
I have a 57v transformer that looks like it's good for about 20 amps, I have the center tap half wave rectified to give me 30v.
I have about 500uf of capacitance (no where near enough I know I need about 30,000)
I have a lm317 variable voltage regulator I.C driving the bases of 2n3055 pass transistors. The 2n3055's are in a common collector (emitter follower) configuration with no emitter resistors or base resistors.
The output voltage remains steady and variable from 0 to about 15 volts.
Problem is as soon as I apply any load what so ever both 2n3055 transistors instantly blow and there's 30v at the output.
Could the problem be -
1. No where near enough capacitance.
2. Input voltage too high.
3. Lack of base/emitter resistors for load sharing purposes.
?
This isn't specifically about audio but I am very much an audio enthusiast and will no doubt be posting asking about some questions for a guitar amp I am in the process of designing.
This post though is about a power supply I have made.
I have a 57v transformer that looks like it's good for about 20 amps, I have the center tap half wave rectified to give me 30v.
I have about 500uf of capacitance (no where near enough I know I need about 30,000)
I have a lm317 variable voltage regulator I.C driving the bases of 2n3055 pass transistors. The 2n3055's are in a common collector (emitter follower) configuration with no emitter resistors or base resistors.
The output voltage remains steady and variable from 0 to about 15 volts.
Problem is as soon as I apply any load what so ever both 2n3055 transistors instantly blow and there's 30v at the output.
Could the problem be -
1. No where near enough capacitance.
2. Input voltage too high.
3. Lack of base/emitter resistors for load sharing purposes.
?
Welcome to diyAudio 
It always helps to see a circuit... a couple of thoughts though.
Are the transistors failing short circuit ? Just because you get 30 volts on the output doesn't automatically mean they have failed. Measure them. They will either be short collector to emitter or OK. I'm thinking that the circuit be latching up when a load is applied. That's why you need to measure them.
A 30 volt ac winding will give around 43 volts dc across the reservoir cap.
What sort of load is causing this problem to show ?
It always helps to see a circuit... a couple of thoughts though.
Are the transistors failing short circuit ? Just because you get 30 volts on the output doesn't automatically mean they have failed. Measure them. They will either be short collector to emitter or OK. I'm thinking that the circuit be latching up when a load is applied. That's why you need to measure them.
A 30 volt ac winding will give around 43 volts dc across the reservoir cap.
What sort of load is causing this problem to show ?
Here are two threads that discuss this issue about that you are describing.
Do read through this thread as this is the beginning and the good stuff comes in at about post #60 and later,
http://www.diyaudio.com/forums/chip-amps/261125-lm3886-problem.html#post4037783
Here is another guy that is having the same issues and he has a few threads on the same project but at the bottom of this page is where I explain what is happening
The Back Shed: help with circuit
http://www.thebackshed.com/forum/forum_posts.asp?TID=7676&PN=1&TPN=3
Here is one more thread here at DIYA about this type of design and I go into a bit more detail of things to consider in the design.
Again do read through the whole thing.
http://www.diyaudio.com/forums/power-supplies/264790-12-volt-30-amp-power-supply.html#post4120267
Also for the voltages you have stated you should be using a LM317HV to start with.
57v/2*1.414=40.305Vdc, The maximum input voltage for a LM317 is 40V this leaves no room for supply line voltage fluctuations.
The input voltage maximum rating for a LM317HV is 60v and would be much better suited for your application.
And as always, Too see the exact schematic you are working with, would better us to help you.
jer
Do read through this thread as this is the beginning and the good stuff comes in at about post #60 and later,
http://www.diyaudio.com/forums/chip-amps/261125-lm3886-problem.html#post4037783
Here is another guy that is having the same issues and he has a few threads on the same project but at the bottom of this page is where I explain what is happening
The Back Shed: help with circuit
http://www.thebackshed.com/forum/forum_posts.asp?TID=7676&PN=1&TPN=3
Here is one more thread here at DIYA about this type of design and I go into a bit more detail of things to consider in the design.
Again do read through the whole thing.
http://www.diyaudio.com/forums/power-supplies/264790-12-volt-30-amp-power-supply.html#post4120267
Also for the voltages you have stated you should be using a LM317HV to start with.
57v/2*1.414=40.305Vdc, The maximum input voltage for a LM317 is 40V this leaves no room for supply line voltage fluctuations.
The input voltage maximum rating for a LM317HV is 60v and would be much better suited for your application.
And as always, Too see the exact schematic you are working with, would better us to help you.
jer
Last edited:
There are way to many variables on this. 1 is that he is not stating what load is being applied, and thus I suspect where the CAP mains is only 500UF you could not possibly expect it to hold up, at much current. It could be nreaking into 60 cycle oscillation and causing the IC to deliver much ripple and such to the pass transistor. So in effect blowing it due to that it might even break into high freq oscillation. You do need to add in cap to the output side also and even add like .05 or some other stuff to prevent things from going into high freq oscillation.
Granted all this is purely a guess, but to create stability it is much better to follow a known design that to just throw together a handful or parts and hoping for the best.
Also unknown and assumed is that you put the pass transistor(s) to a heat sink.
Also unknown is the size and current that the power transformer can deliver.
Also unknown but certainly needed is either a cuircuit breaker or fuse(s) to prevent such a disaster.
Years ago I built a home stereo amp from a design, but then in the testing of it discovered quickly that the power transformer I picked to use lacked enough current ability to make it stable. So the result was that when pushing a high level of low frequency bass, it caused a lot of DC shifting. lesson Learned on that one.
The amp used Darlington outputs 2 per channel, and was fairly DC coupled at all stages. It was stable at idle, and properly biased but when pushed to even 1/3 of what the amp wanted to deliver the transformer was just not powerful enough and in the end it would just not work out. You could tell because I had a pilot light which would go dim on the peaks. However it was a good learning experience if nothing else.
Granted all this is purely a guess, but to create stability it is much better to follow a known design that to just throw together a handful or parts and hoping for the best.
Also unknown and assumed is that you put the pass transistor(s) to a heat sink.
Also unknown is the size and current that the power transformer can deliver.
Also unknown but certainly needed is either a cuircuit breaker or fuse(s) to prevent such a disaster.
Years ago I built a home stereo amp from a design, but then in the testing of it discovered quickly that the power transformer I picked to use lacked enough current ability to make it stable. So the result was that when pushing a high level of low frequency bass, it caused a lot of DC shifting. lesson Learned on that one.
The amp used Darlington outputs 2 per channel, and was fairly DC coupled at all stages. It was stable at idle, and properly biased but when pushed to even 1/3 of what the amp wanted to deliver the transformer was just not powerful enough and in the end it would just not work out. You could tell because I had a pilot light which would go dim on the peaks. However it was a good learning experience if nothing else.
Some how I missed this detail "I have the center tap half wave rectified to give me 30v".
I was assuming that it was a center tapped full wave configuration.
Even for a single diode half wave configuration your average DCrms will be .9x "IF" you are drawing some current.
But when it is just sitting at idle your circuit will be seeing the higher voltage pulses of the rectified AC waveform as ACpeak.
FWIW
jer
I was assuming that it was a center tapped full wave configuration.
Even for a single diode half wave configuration your average DCrms will be .9x "IF" you are drawing some current.
But when it is just sitting at idle your circuit will be seeing the higher voltage pulses of the rectified AC waveform as ACpeak.
FWIW
jer
Some how I missed this detail "I have the center tap half wave rectified to give me 30v".
I was assuming that it was a center tapped full wave configuration.
Even for a single diode half wave configuration your average DCrms will be .9x "IF" you are drawing some current.
But when it is just sitting at idle your circuit will be seeing the higher voltage pulses of the rectified AC waveform as ACpeak.
FWIW
jer
One "almost for sure" way to keep power transistors from popping is to implement a simple current limiter. Basically, you put a current monitoring resistor (R Sense below) into the emitter line of the transistor in question that drops about a 0.6-1 volt when the desired maximum current is reached. Hook a small like-sexed transistor (Q2) with the base and emitter connected so that the transistor is switched on by excess current. Tie the collector of the Q2 transistor to the base of the transistor being protected (Q1.
True to use the current limiter but will the LM317T as proposed counteract with the current limiting to prevent this?
I feel that the whole thing is that it needs a better regulator IC type that would sense all of this and work together to create a "smart power supply" one that is controlling the whole show, offers a preset shutdown when current reaches within 1/2 of the max rating of a 2n3055. Additionally just to have either a fuse or self resetting circuit breaker etc, of sorts would just be a wise idea no matter what.
I am not following that to rectify 47V AC is going to produce 80 volts DC, especially with any kind of load, it might go up to 60VAC without a load. Then again, simply adding a bleeder resistor would be the smart thing to do anyhow, and that also to add in capacitors to keep things somewhat stable and prevent any possible self oscillation which would lead to failure period.
Most smart supplies monitor the output of the regulator to calculate current flow between input and output of the pass transistor, and limit the voltage to the base of the pass transistor in order to prevent overloading, so that you could completely collapse the output to 0 volts and nothing will happen until the short is removed.
I feel that the whole thing is that it needs a better regulator IC type that would sense all of this and work together to create a "smart power supply" one that is controlling the whole show, offers a preset shutdown when current reaches within 1/2 of the max rating of a 2n3055. Additionally just to have either a fuse or self resetting circuit breaker etc, of sorts would just be a wise idea no matter what.
I am not following that to rectify 47V AC is going to produce 80 volts DC, especially with any kind of load, it might go up to 60VAC without a load. Then again, simply adding a bleeder resistor would be the smart thing to do anyhow, and that also to add in capacitors to keep things somewhat stable and prevent any possible self oscillation which would lead to failure period.
Most smart supplies monitor the output of the regulator to calculate current flow between input and output of the pass transistor, and limit the voltage to the base of the pass transistor in order to prevent overloading, so that you could completely collapse the output to 0 volts and nothing will happen until the short is removed.
How about to modify this to have Rbias be a Zener diode of desired output voltage? or to add a Zener diode from B of the Q1 to ground? Then add in a few .1 UF caps and some value Electroylytic caps of fairly good value on both input and output, as well as some kind of high value bleeder resistor, to help keep things stable. It just is too noisy otherwise, to make it an effective power supply, depending on the application of course.
probably exceeding max voltage for the regulator chip. calculate excess power for every load current (lots of heat loss).
link http://sound.westhost.com/project77.htm I would use uA723 based design instead of this.
Do use a full wave rectifier.
Do use a pre regulator for the chip supply and add more C here. use a large series resistor and zener bypassed with another 500uF.
Don't use more than one pass device without ballast resistors and don't expect more than 25 W loss using a big heat sink. more failures with out CL fuses.
I would plan on using this transformer with a PWM pre regulator followed by a linear design.
link http://sound.westhost.com/project77.htm I would use uA723 based design instead of this.
Do use a full wave rectifier.
Do use a pre regulator for the chip supply and add more C here. use a large series resistor and zener bypassed with another 500uF.
Don't use more than one pass device without ballast resistors and don't expect more than 25 W loss using a big heat sink. more failures with out CL fuses.
I would plan on using this transformer with a PWM pre regulator followed by a linear design.
Last edited:
two series pass 2n3055 is not enough......
what is your current at 30 volts output.....
i will probably never run a single 2n3055 with more than 1.5 amperes across c-e...
and that depends on what the 2n3055 c-e voltage will be....
All volts and no talk power supply
Thank-you for all the replies!
I'll get a schematic sorted soon.
Transistors blew as soon as less than 1 amp of current was applied at 12v.
The current rating of the transformer is unknown as I havn't tested it and part numbers had been partially rubbed off so could not find a data sheet. At a guess I'd say it's good for at least 20 amps. Tore it out of a powered 500w sub-woofer that I found on the side of the road.
The transistors are mounted to a heat sink and share a common collector. I have had a couple of people say to me that I should separate these collectors as them being connected plays a role in not allowing an equal amount of current to flow to one or the other collectors base and therefore one will loaf and drag behind the other.
Someone mentioned that the lm317 was no good as it's maximum voltage is 40v. My circuits maximum rectified voltage is 30v. The 0nly spikes I have encountered are from back emf from an inductive load but I have rectified this by placing blocking diodes at outlet of voltage regulator and also from emitter to collector to stop back emf from entering the emitters of both diodes.
I do have 10 amp fuses installed despite the datasheet for the 2n3055's stating that they can handle a maximum current of 15 amps.
Voltage level is not a problem here either I don't think as the 203055's are rated for 110v. (from memory)?
So again, blew at less than 1 amp @ 12v.
That unrectified voltage ripple caused by lack of capacitance theory sounds pretty convincing!
Thanks for your input peops
Thank-you for all the replies!
I'll get a schematic sorted soon.
Transistors blew as soon as less than 1 amp of current was applied at 12v.
The current rating of the transformer is unknown as I havn't tested it and part numbers had been partially rubbed off so could not find a data sheet. At a guess I'd say it's good for at least 20 amps. Tore it out of a powered 500w sub-woofer that I found on the side of the road.
The transistors are mounted to a heat sink and share a common collector. I have had a couple of people say to me that I should separate these collectors as them being connected plays a role in not allowing an equal amount of current to flow to one or the other collectors base and therefore one will loaf and drag behind the other.
Someone mentioned that the lm317 was no good as it's maximum voltage is 40v. My circuits maximum rectified voltage is 30v. The 0nly spikes I have encountered are from back emf from an inductive load but I have rectified this by placing blocking diodes at outlet of voltage regulator and also from emitter to collector to stop back emf from entering the emitters of both diodes.
I do have 10 amp fuses installed despite the datasheet for the 2n3055's stating that they can handle a maximum current of 15 amps.
Voltage level is not a problem here either I don't think as the 203055's are rated for 110v. (from memory)?
So again, blew at less than 1 amp @ 12v.
That unrectified voltage ripple caused by lack of capacitance theory sounds pretty convincing!
Thanks for your input peops
Have you actually measured the transistors as failing short circuit from C to E ?
I know its a strange question but just because the output shoots to max when a load is applied, and replacing the 2N3055 'fixes' that doesn't automatically mean they have blown. We just need that confirming by measurement.
Current sharing is normally done by including a low value (say 0.47ohm) in the emitter lead of each 2N3055.
12 volt output, 1 amp load... and it instantly fails to '30 volts at the output'. That shouldn't happen with genuine 2N3055's. Even if there was masses of ripple it still shoudn't be a problem.
Last question What is the load ?
I know its a strange question but just because the output shoots to max when a load is applied, and replacing the 2N3055 'fixes' that doesn't automatically mean they have blown. We just need that confirming by measurement.
Current sharing is normally done by including a low value (say 0.47ohm) in the emitter lead of each 2N3055.
12 volt output, 1 amp load... and it instantly fails to '30 volts at the output'. That shouldn't happen with genuine 2N3055's. Even if there was masses of ripple it still shoudn't be a problem.
Last question
What is the load ?
how did you measure this? considering you have DC plus large ripple voltage riding on top.
depending on who made your 317 they are limited to about 36V.
load sounds more reactive than your filtering. this could be reason for high starting loads E.g. aiding destruction of your circuit.
no max ratings are more complex.
'safe area of operation' SOA charts will tell you that.
you can NOT get max current AND high CE voltage at the same time.
limiting every 2N3055 device Ie current to 1.5A max is good practice AND with higher voltages and heat, de-rate the currents lower.
Last edited:
Thanks guys I got it sorted.
I think the problem was happening when the load was removed, the flyback voltage was frying the transistors as they only have a breakdown voltage of 60v.
I overcame this by adding diodes forward biased from emitters to collectors.
The voltage drops alot when put under load though and since I have a pretty chunky transformer would this just be to lack of capacitance? Ie - When load is increased the ripple becomes severe enough and the voltage I'm reading with my digi meter is seeing an average between the bottom & top of the DC waveform that would be severely rippled due to lack of capacitance?
I think the problem was happening when the load was removed, the flyback voltage was frying the transistors as they only have a breakdown voltage of 60v.
I overcame this by adding diodes forward biased from emitters to collectors.
The voltage drops alot when put under load though and since I have a pretty chunky transformer would this just be to lack of capacitance? Ie - When load is increased the ripple becomes severe enough and the voltage I'm reading with my digi meter is seeing an average between the bottom & top of the DC waveform that would be severely rippled due to lack of capacitance?
you mean reversed bias? sometimes called a freewheeling protection diode. how did you come about with that idea?,
hard to say if this is a good fix or not, is your load far away or inductive?. what tests have you done with and without the diode?
Also since you changed them, the replacement parts might have better SOA.
yes ripple gets larger as the load is increased, you can measure that using AC range on your DMM. but the highest peak voltage (AC + DC ) even under load is still not too far away from theory. Under transient conditions assume voltage is max at light loads initially, then current approaches the worst loading condition (usually short circuit on output) gives SOA bounds, roughly speaking.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.