50V 5A Linear Power Supply Project - Filter Cap Problems

Thread Starter

danlay

Joined Mar 2, 2016
2
I'm currently building a 50V power supply for my Dad's vintage telephone exchange equipment.

We need a 50v 5A unregulated supply to operate the system so I decided to build one.

The equipment uses a lot of large electro-magnets, coils, relays that pulse operate.
I decided that a linear PSU would be a better choice as appose to a switch mode under these quite harsh current-fluctuating circumstances.

I purchased a suitable 500VA torodial transformer (2x 115v input and 2x 50v output)
I've put the 2x 115v inputs in series as our mains here is 240v.

The AC output on each of the 2x 50v taps is over 50v, but was expecting a drop after the bridge rectifier.
I've actually added 2 separate bridge rectifiers, one for each of the outputs, effectively giving me 2x 5A supplies.

The DC output from each of the bridge rectifiers is nearly exactly 50v - so far so good.

The problem i'm having is when I add filter capacitors.. I've added a 10,000uf 100V electrolytic across the DC side of the bridge.
After doing so the output voltage jumps up to 72v.

I wondered if someone could help explain what's causing this and possibly how to remedy it?
Its still at around 72v under load of about 1A.

Dan
 

JohnInTX

Joined Jun 26, 2012
4,787
Welcome to AAC!

Sure. The 50V transformer is 50VAC RMS. When you rectify and filter it, it charges the filter caps to the peak AC voltage which is 1.414*RMS or about the 72 volts you are seeing. The solution would be to use a transformer with lower voltage or regulate the output to the 50VDC you want.
 

AnalogKid

Joined Aug 1, 2013
10,986
The equipment uses a lot of large electro-magnets, coils, relays that pulse operate.
I decided that a linear PSU would be a better choice as appose to a switch mode under these quite harsh current-fluctuating circumstances.
Not true. While there might be some linear supplies still operating out in wherever, the vast majority of telecom equipment, including old mechanical switch gear out in in wherever, runs on switching "rectifiers", 48 Vdc output switching power supplies.

Sticking with your linears, you will get more stable system performance if you do regulate the supply outputs. You will have to move some heat.

If you really want to go old school, put a large inductor between two filter caps for a pi section filter after each bridge. You will have much less ripple, to the point that solid state regulation might no longer be necessary.

ak
 

Lestraveled

Joined May 19, 2014
1,946
As John said, without a load, you are seeing the peak voltage. As you load the power supply, the output voltage will decrease. Under full load it should reach about 50 volts, depending on the quality and rating of the transformer.

I suspect that the old telephone equipment is designed to operate over a wide voltage range.
 

ScottWang

Joined Aug 23, 2012
7,397
Using a 42Vac-0V transformer and you will get the Vout = 42V*1.414-(0.9V*2) = 59.4Vdc - 1.8V = 57.6Vdc, assuming the Vf of diode is 0.9V when it flows through 5A current, but the Vf voltage could more.

And adding a TL783, 3 pcs 2N3055 to expanding the current, the current expanding circuit similar as the third circuit in creating a high current LM317 regulator, but you need to in series with an 0.2Ω/5W resistor for each E of 2N3055.

 

Thread Starter

danlay

Joined Mar 2, 2016
2
Hi All,

Thanks for your comments and help with this.

I like the idea of using the linear 2N3055s, I already have some of these.
I'm guessing that the LM317 has been substituted with the TL783 because of the voltage limitation of the LM317?
Are there any fundamental changes needed in the above circuit (besides balancing resistors) or could I just swap out the LM317 with a TL783?

The telephone equipment operates in the range of 45 to 55v max.

I was wondering if it would be simpler to return the 50V transformer I have and replace it with a 30V one?
This should then give me a voltage of around 50v when its not loaded. I could also put both 30V taps in parallel to give me 10A, so in theory it should not drop much below 50v when under load. The equipment uses less than .5A when its idle and up to around 4A when in full use. Worse case it could be 5A.

Any comments much appreciated.
Dan
 
Last edited:

ian field

Joined Oct 27, 2012
6,536
Using a 42Vac-0V transformer and you will get the Vout = 42V*1.414-(0.9V*2) = 59.4Vdc - 1.8V = 57.6Vdc, assuming the Vf of diode is 0.9V when it flows through 5A current, but the Vf voltage could more.

And adding a TL783, 3 pcs 2N3055 to expanding the current, the current expanding circuit similar as the third circuit in creating a high current LM317 regulator, but you need to in series with an 0.2Ω/5W resistor for each E of 2N3055.

That will have appalling regulation - use the method in the appnotes.
 

ian field

Joined Oct 27, 2012
6,536
Hi All,

Thanks for your comments and help with this.

I like the idea of using the linear 2N3055s, I already have some of these.
I'm guessing that the LM317 has been substituted with the TL783 because of the voltage limitation of the LM317?
Are there any fundamental changes needed in the above circuit (besides balancing resistors) or could I just swap out the LM317 with a TL783?

The telephone equipment operates in the range of 45 to 55v max.

I was wondering if it would be simpler to return the 50V transformer I have and replace it with a 30V one?
This should then give me a voltage of around 50v when its not loaded. I could also put both 30V taps in parallel to give me 10A, so in theory it should not drop much below 50v when under load. The equipment uses less than .5A when its idle and up to around 4A when in full use. Worse case it could be 5A.

Any comments much appreciated.
Dan
Most of the big component supply houses carry 48V SMPSUs designed specifically for telecoms applications - probably not cheap, but it gets the job done with no mucking about.
 

#12

Joined Nov 30, 2010
18,224
I'm guessing that the LM317 has been substituted with the TL783 because of the voltage limitation of the LM317?
There also exists a 317H which is good to 60 volts.
I was wondering if it would be simpler to return the 50V transformer I have and replace it with a 30V one?
Yes. Something in the range of 33V to 36V should work. You will need a 7 amp rating.
I could also put both 30V taps in parallel to give me 10A,
I think you're looking into a compromise here. Using a half wave rectifier will double the size of capacitor you need.
 

ScottWang

Joined Aug 23, 2012
7,397
I like the idea of using the linear 2N3055s, I already have some of these.
I'm guessing that the LM317 has been substituted with the TL783 because of the voltage limitation of the LM317?
Are there any fundamental changes needed in the above circuit or could I just swap out the LM317 with a TL783?
If your Vin-Vout<40 for LM317 and Vin-Vout<60 for LM317HV is ok, my original thought was you have a big range to be adjust, so the TL783 was suggested, but the TL783 has Voltage drop Vdo=Vi-Vo =10V, it is higher than LM317 only has 3V.

Iout_dc = Iin_ac*0.6
 

ScottWang

Joined Aug 23, 2012
7,397
That will have appalling regulation - use the method in the appnotes.
I can't find the V_drop from datasheet, and I found the Voltage drop Vdo=Vi-Vo =10V from here.
and the Vin (Max) (V) = 125V, Vout (Max) (V) = 125V,
So if the input voltage is too low, maybe don't use the TL783.
 

AnalogKid

Joined Aug 1, 2013
10,986
A couple of things.

1. Unlike the 78xx series, the 317 and 783 voltage ratings apply only to the circuit's input-to-output difference voltage. So if the regulator is configured for a 48 V output, the peak input voltage for the 317 must be below 91 V. That's a lot. Anything less, and you do not need the HV version or the 125 V TL783.

2. The 317 can make 1 A out and you need 5 A, so you need only one 2N3055 current boost transistor. And no base resistor. And no emitter resistor. If you drive the base directly from the output of the 317 as an emitter follower, the overall regulation will suffer slightly. So what? This isn't the space shuttle. It's an old phone system designed to run on hand-cranked generators and an enormous range of batteries that are miles away and connected by small gauge wires. K.I.S.S.

3. While having only one boost transistor will work, it will get hot. But if multiple boost transistors are mounted close together on a single heatsink, they are not spreading out the heat all that much. This is a tradeoff

ak
48V-Regulator-1-c.gif
 

Attachments

Last edited:

ian field

Joined Oct 27, 2012
6,536
You mean no hope is it can't use TL783 over here?
Because the Vdo = 10V is too high?
If you say so then I will agree with you.
You suggested a method that won't work.

I pointed your attention to the correct method described in the appnotes.

You won't listen and joke about it.
 

ScottWang

Joined Aug 23, 2012
7,397
You suggested a method that won't work.

I pointed your attention to the correct method described in the appnotes.

You won't listen and joke about it.
I'm not joking, I can't find the appnotes when you mentioned that, and I told you that on #11, I even can't find the Vdo in the datasheet, I always used the Vdo to do the suggest for Input voltage, if you can find the appnotes, please give me a link, thank you.
 

ian field

Joined Oct 27, 2012
6,536
I'm not joking, I can't find the appnotes when you mentioned that, and I told you that on #11, I even can't find the Vdo in the datasheet, I always used the Vdo to do the suggest for Input voltage, if you can find the appnotes, please give me a link, thank you.
The 317 is just a 3-terminal regulator with only 1.25V output.

If you really and honestly cant find a 317 appnote - use the one for the 78xx family.
 
Top