Pre-Regulator

[Overclocked]

I have a unregulated power supply from a transformer that gives out a peak voltage of 61V after the rectifier (unloaded). With a 1A load the voltage goes down to 41.5V peak. I am powering a very large LED, 32V @1A and will be using A NCL30160 to do so, but its max input voltage is 40V (lower would be better). What kind of pre-regulator should I use? A series pass transistor or a LM317HV? I dont care what happens after the 40V, its just so the chip wont fry. Adding on another SMPS in front of the LED regulator would be too costly.

To re-iterate:
Vin Min:41.5V Peak DC (1A Load)
Vin Max: 61V Peak DC (no Load)
Iout Max:1A
Voltage needed: 40V (lower would be better since I wouldnt stress the chip out).

I know there are other solutions (ie TI) but I would rather not spend $2.50 a chip.

 

[audioguru]

Your transformer is almost as bad as the cheap Chinese one I have that has an output of 18V without a load and it drops to 9V with a low current load. It gets hot even with no load because it is cheap and does not have enough turns or its core is made of rice.

 

[Overclocked]

Your transformer is almost as bad as the cheap Chinese one I have that has an output of 18V without a load and it drops to 9V with a low current load. It gets hot even with no load because it is cheap and does not have enough turns or its core is made of rice.

It was salvaged from a older power supply a long time a go. Im using it because 1) it fits the requirements and 2) Its on hand and I dont have to buy another transformer. I also really dont know the specs on it either. Im guessing by size it should be good for 1.2A..or maybe Im pulling too much.

 

[AnalogKid]

The LM317HV probably is the way to go with the least hassle. Pay attention to its capacitor and minimum load rules.

ak

 

[Diver300]

The transformer is very poor for the power you are taking. You area getting about 50% regulation, meaning that the output voltage is 50% too high with no load.

A 40 or 50 VA transformer should have around 10% regulation. Here is an example:-
https://www.farnell.com/datasheets/1508684.pdf

That data sheet also gives typical transformer sizes for the VA ratings, as you might be overloading the transformer that you have. If you are overloading it, it will get very hot on full load. Of course, that takes some time and transformers are fine with short duration overloads.

 

[spec]

I have a unregulated power supply from a transformer that gives out a peak voltage of 61V after the rectifier (unloaded). With a 1A load the voltage goes down to 41.5V peak. I am powering a very large LED, 32V @1A and will be using A NCL30160 to do so, but its max input voltage is 40V (lower would be better). What kind of pre-regulator should I use? A series pass transistor or a LM317HV? I dont care what happens after the 40V, its just so the chip wont fry. Adding on another SMPS in front of the LED regulator would be too costly.

To re-iterate:
Vin Min:41.5V Peak DC (1A Load)
Vin Max: 61V Peak DC (no Load)
Iout Max:1A
Voltage needed: 40V (lower would be better since I wouldnt stress the chip out).

I know there are other solutions (ie TI) but I would rather not spend $2.50 a chip.

Hi Overclocked,

Just thinking out loud here:

I know you said that a SWPS would be too costly but SMPS modules are available quite cheaply, around $2 US including post and packing, from Ebay for example. A step down (buck) SMPS will convert the varying voltage from the transformer into a fixed and regulated voltage for the LM317HV. Select a SMPS with an output voltage that is around 4V higher than the regulated voltage from the LM317HV.

In fact, you probably don't need an LM317HV or NCL30160: just drive the LED directly from the SMPS. An SMPS with a current limit of 1A should do the job. Most SMPSs will not take more than 30 to 40V input, so the 60V peak from the transformer would need to be sorted, but I can't see that as being a major problem., and if the LED would be connected all the time, there is not a problem anyway.

The big advantage with the SMPS approach is that it would be simple to sort out the issues and in operation it would generate little heat, unlike the linear approach where heat would be a problem, especially for the LM317HV which would be getting at least 4W with a 1A output, and would need a decent heatsink to avoid thermal shut-down.

Going back to the do-it-yourself approach, you could probably use an SCR and a zener to generate an acceptable voltage for the LM317HV, but that is just thinking out loud and would need a bit of design and development.

 

[Willen]

It gets hot even with no load because it is cheap and does not have enough turns or its core is made of rice.

Haha I never seen a rice-core transformer!

Did you mean less size core or not pure material can produce heat?

 

[Overclocked]

Hi Overclocked,

In fact, you probably don't need an LM317HV or NCL30160: just drive the LED directly from the SMPS. An SMPS with a current limit of 1A should do the job. Most SMPSs will not take more than 30 to 40V input, so the 60V peak from the transformer would need to be sorted, but I can't see that as being a major problem., and if the LED would be connected all the time, there is not a problem anyway.

In my experience those SMPS modules are really $5, use cheap components, and take a month to get here if I order from china. The NCL30160 is the SMPS for the LED (Buck converter).

Sounds like it would be easier just to get another transformer, or for the price of a new transformer, I could get a 36V Offline SMPS and be done with it. To the scrap bin with these transformers!

 

[spec]

In my experience those SMPS modules are really $5, use cheap components, and take a month to get here if I order from china. The NCL30160 is the SMPS for the LED (Buck converter).

Sounds like it would be easier just to get another transformer, or for the price of a new transformer, I could get a 36V Offline SMPS and be done with it. To the scrap bin with these transformers!

I have found the cheap SWMPs to be OK; there is not much to them.Often they are sold by stockists in different countries. I think there are some in thye UK, so delivery might not be too bad.

Wise move going for another transformar, but if you could give more details of the transformer you have, maybe something coulod be done. I should have checked what the NCL30160 was- my bad.

What arrangement do you have: bridge and resovour capacitor? what bridge/rec diodes. What value reservour cap. presumably the mains is 60hz where you are?

 

[Overclocked]

I have found the cheap SWMPs OK; there is not much to them.

Wise move going for another transformar, but if you could give more details of the transformer you have maybe something coulod be done. I should have checked what the NCL30160 was- my bad.

what arrangement do you have: bridge and resovour capacitor? what bridge/rec diodes. What value reservour cap. presumably the mains is 60hz where you are?

Thats alright . I was going to go with a full bridge rectifier with caps, which would then go to the LED Driver..But then I found out a SMPS is way cheaper. I never sat down and did the calcs for the filter caps, I was stuck on the transformer.

https://www.trcelectronics.com/View/Mean-Well/LRS-75-36.shtml

Then, unfortunately I also found a LED driver thats a all in one package, which is cheaper than doing with a custom driver. PCB + Parts will no doubt exceed $16 but then where is the fun in just grabbing a off the shelf solution?

https://www.trcelectronics.com/View/Mean-Well/LPC-60-1050.shtml

 

[audioguru]

Haha I never seen a rice-core transformer!

Did you mean less size core or not pure material can produce heat?

The windings of a transformer use their inductance to transfer power from primary to secondary and to keep the current low when there is no load. If there are not enough windings or if the iron core is too small then the primary is almost simply a piece of wire that gets hot because it has a high current in it.

 

[spec]

Hi Overclocked,

Circuit below should do your job. It stops the output from the bridge rectifier from rising above 37V. You can make the voltage what you like accoring to the forula on the schematic.

The transformer bridge and resovour caps are just included for completness and do not reflect your arrangements in that area.

When the voltage is below 37V Q1 will not conduct.

The circuit could be optimised more for your application but that would require some hands-on work to determin the voltages and currents when the LED chip and LED are operating. I suspect you can sort that out.

Be aware, though, that if the LED circuit is not connected Q1 will be dissipating 40W and will need a big heatsink. You could put a 20W 62R resistor between the collector of Q1 and 0V to reduce the dissipipation in Q? to 20W. The overal circuit dissipation would still be 40W though.

spec

 

[Overclocked]

Hi Spec, Thanks for the circuit, but why did you choose to go with a shunt regulator instead of a series pass regulator?

 

[chemelec]

Considering such a POOR Quality Transformer, That Shunt Regulator will cause the transformer to OVERHEAT Really Bad.

 

[spec]

Considering such a POOR Quality Transformer, That Shunt Regulator will cause the transformer to OVERHEAT Really Bad.

Hi chemelec,

You concerns are correct in principle, but need to be qualified. The shunt regulator will only take little more than the LED circuit or even nothing when the LED is connected and drawing 1A. It only takes current when the rectified voltage rises above 37V or a voltage defined to suit Overclocked actual circuit. That's why I said tha the circuit would need to be optimised. One way would be to put a low value resistor between the bridge rectifier and the resovoir capacitor. That would not only give the transformer an easier time by reducing the rec inrush current, but it would also lower the rectified saw tooth mean voltage, and more importantly make the supply output impedance higher which would make the job of the shunt regulator easier.

With the LED load disconnected there would be much dissipation. I mention that in my post

But all the same the priciple of your comment is good.; please see my next post and I am starting to think that I may have misunderstood Overclock's post.

spec

 

[spec]

Hi Spec, Thanks for the circuit, but why did you choose to go with a shunt regulator instead of a series pass regulator?

Afternoon Overclocked,

The reason why that circuit is a shunt rather than series regulator is because it was designed using the brain-trouble approach Appologies- I will redesign and post a new circuit.

spec

 

[spec]

Hi again Overclocked,

I am doing the detailed analysis of your requirement and really need the following information:

(1) what is the circuit of the transformer rectifier and reservoir capacitor? Is it one secondary winding and bridge. two windings and two diodes, or even half wave rectification.
(2) what bridge/rec diodes are you using? The part number would be ideal
(3) most important, what is the value of the reservoir capacitor?
(4) is the reservoir capacitor a lage low esr type or just an ordinary capacitor?
(5) are there any impedences (resistors thermistors etc) in the primary or secondary of the transformer?
(6) what is the frequencu of your mains supply 50 or 60Hz?
(7) when you say that the voltage with 1A drain is 41V5 peak, how did you measure that: with a scope or a DC voltmeter. A DC voltmeter would measure mean not peak.
(8) If you measured with a scope could you say what the peak to peak ripple voltage was at 1A drain?

Would you please post the answers to the above questions as it will help a lot

Thanks in advance

spec

 

[Overclocked]

I'll save you some trouble; as for this design I found out it would be cheaper just to use a 36V Offline SMPS . A transformer bought through digi-key or even wholesale would be on the slightly more expensive side compared to a offline SMPS. I do thank you for the help though . But I will still answer your questions, I will be using a 24VAC transformer for a similar design (its a 2A system though), but I dont have to worry about Vmax at all. Answers are in red.

Hi again Overclocked,

I am doing the detailed analysis of your requirement and really need the following information:

(1) what is the circuit of the transformer rectifier and reservoir capacitor? Is it one secondary winding and bridge. two windings and two diodes, or even half wave rectification.

Full Wave. One secondary winding, 4 diodes.

(2) what bridge/rec diodes are you using? The part number would be ideal.

Never got that fair. Part spec was the last thing on my mind. 3A, 1000V for my 24VRMS system.

(3) most important, what is the value of the reservoir capacitor?

Lets switch gears and use the 24VRMS system I mentioned above. I plan on using one 6800uF Cap @50V. Calculated it out to give out less than 3% ripple.

(4) is the reservoir capacitor a lage low esr type or just an ordinary capacitor?

Low ESR. 68mOhm

(5) are there any impedences (resistors thermistors etc) in the primary or secondary of the transformer?

No. There is a smaller onboard capacitor on the LED driver though. Its worth noting that if the LED is off, main power will also be off. The LED will be on for ~8hrs a day, 7 days a week. So basically there will be a switch controlling main power to the transformer.

(6) what is the frequencu of your mains supply 50 or 60Hz?

60Hz, 120VRMS

(7) when you say that the voltage with 1A drain is 41V5 peak, how did you measure that: with a scope or a DC voltmeter. A DC voltmeter would measure mean not peak.

I calculated peak from RMS from my voltmeter (Vrms*1.414).

(8) If you measured with a scope could you say what the peak to peak ripple voltage was at 1A drain?

Never filtered it, I was concerned with how much my unknown transformer could supply. Again, this was salvaged from something a long time ago and has been sitting in my junk box. Cost was a factor since I dont have much money these days. If I had a larger resistor, I would let it sit around to see if it heats up at all at the current I need, BUT I have to keep in mind that if I want to use it (which I probably wont), the capacitors will have to be rated for 120V, which will be quite expensive. So I'll just use a 36V Offline power supply for That LED system.

 

[spec]

Hi Overclocked,

Thks for data,

Shame you are not going to build the circuit yourself, but you are going the best way. That commercial SMPS looks good and at 16 bucks is a bargain.

By the way I have now analysed your requirements in detail including the NCL30160 data sheet- useful chip.

I now have 5 aproaches: 3 series preregulators, one using a power a PMOSFET and two using power BJTs; the originally posted shunt regulator, which inspite of the comments by chemilec and my initial panic, is fine; and a method of modifying the brige circuit to produce a peak voltage of 39V.

Why is the shunt design OK? It is Ok because:

It only takes current when the supply line is 37V and over. At 37 V input and 1A current into your high power LED the NCL30160 will be drawing 1.08 A. The impedance of the rectifying circuit is 19R5, so the 80 mA over 1A will drop the rectifier voltage to 41.5-1V56= 39V94. So, in theory, the circuit will work within design spec without any preregulation because the NCL30160 will take 40V (in practice it is bound to take more than that). But the shunt regulator is set at 37V, so it will take enough extra current to realise a further voltage drop of 39V94 - 37V= 3.94V. Thus the shunt regulator will take 3.94/19R5= 202mA.

This would be fine but could be reduced greatlly by setting the regulation voltage to 39V insted of 37V, which would still suit the NCL30160. The shunt regulator would be required to drop the voltage 39V94-39V=0V94. Thus the current that the shunt regulator would have to sink would be 0V94/19R5= 48.21 mA.

Because of the characteristics of a bridge rectifier, transformer and reservour capacitor the situation is more complex as you need to take ripple voltage into account. This means that the shunt regulator would be taking far less curent. I recon that a 4m7F capacitor would have a 2V peak to saw tooth ripple wave form which would probably be optimum.

The one drawback with the shunt regulator is that if the load were removed it would still maintain the output voltage at 37V. To do thhis it would need to drop the rectified of load voltage of 61V by 24V. From this the current drawn by the shunt regulator would be 24/19.5= 1.23A. No different to normal operation. The power transistor would be dissipating 1.23A* 37V= 45.51W: not the end of the world. A simple addition to the circuit could cut this to zero, but as you are abandoning this approach I won't take it any further.

The LM317H approach appeals but by the time you have taken the drop out voltage into consideation, thye ripple v0oltag and the thermal shut-down it seemed problamatic. But I havent done a detailed analysis so that is only my gut feeling.

The top contender, which I have just finished in hand sketch form, is the PMOSFET serial preregulator.

Nice talking to you- especially your presentation of data.

spec

 

[Overclocked]

Hi Overclocked,


The top contender, which I have just finished in hand sketch form, is the PMOSFET serial preregulator.

Nice talking to you- especially your presentation of data.

spec

I would actually like to see that, as Ive been trying to figure it out myself. Even if its a rough idea, it would still be handy to have a round.