Can anyone suggest a circuit or a ready made module please -need 52 volt regulator

Thread Starter

Lee Tracey

Joined Jun 14, 2016
6
The source is an unregulated swinging between 52VDC and 64VDC. The output needs an add-on module to regulate to 52VDC at 2 amp. Ideal would be a variable regulated output between 48VDC and 55VDC, but fixed at 52 would still do the job. If somebody has the circuit but no interest in making a working p.c.b. I would then seek an interested technician willing to make a working board for pay.
 

wayneh

Joined Sep 9, 2010
17,496
The source is an unregulated swinging between 52VDC and 64VDC. The output needs an add-on module to regulate to 52VDC at 2 amp. Ideal would be a variable regulated output between 48VDC and 55VDC, but fixed at 52 would still do the job. If somebody has the circuit but no interest in making a working p.c.b. I would then seek an interested technician willing to make a working board for pay.
Are you wanting just one for a project or a production ready PCB design? What degree of regulation is required?
 

Thread Starter

Lee Tracey

Joined Jun 14, 2016
6
Are you wanting just one for a project or a production ready PCB design? What degree of regulation is required?
The potential is 400 to 1000 but first one is needed to prove that it works. The problem is that a friend of mine has imported 400 but is committed to take 1000 Power Over Ethernet ( POE ) 8 port ( powering eight IP CCTV cameras over RG59 cable alongside the video ) units from China. The POE boards (units) are powered by a separate, stand-alone, PSU that is marked as 110-230 AC input and 52VDC output at 1.85 amps and connected via a normal DC plug. When the POE units are connected to eight IP cameras a diode on the POE board catches fire. The Chinese manufacturer has said that the diode fitted is wrong and want to send replacement diodes. However de-soldering a micro miniature machine solder flow component and then re-soldering a replacement is impossible and, in addition the diode is so close to a heavy transformer coil that it is also covered in silicone evastomer. My opinion is that there is nothing wrong with the POE board provided the correct voltage and current is supplied. I have detected 64VDC on components on the board so my opinion is that the PSU is not regulated and just runs away with the voltage and that at least one component ( the diode ) is being hit by a voltage above its safety threshold. What is that safety threshold? I do not know and obtaining any sensible answer from China is like pulling teeth with pliers and no anaesthetic. As the PSU is marked as being 52VDC output it is possible that 52 is a safe voltage but competitive devices all seem to be 48. To gain that knowledge is why I suggested a variable regulator circuit such as the TL783, but the 783 can only handle 700mA so is an illustration but not an answer. My idea of a solution therefore is a small module (SM) that sits between the output of the "52" VDC PSU and the input of the POE board and can handle at least 2 amp. If the regulator can be varied between say 70VDC down to about 40VDC, I could connect the whole system up and start at 48 and record the POE board temperatures and take the voltage up to the point where the POE board catches fire - my guess is that about 50VDC will be the safe centre voltage. But I do not know. What I do know is that at 64VDC, after a about ten hours, the POE board goes on fire. If we could obtain a circuit diagram and detailed BOM from the Chinese company then a skilled electronics engineer could probably work out with maths the correct safe voltage. My friend is in real trouble for he has sold over 300 sets at a small profit margin and now has to return the money to all the customers and take back the faulty gear. Will my idea work?? His other alternative is to seek a Chinese supplier of a replacement PSU that is regulated - but at what safe voltage? If he goes the replacement PSU route he then has the cost of the replacement PSU plus the cost of the useless PSU. My hope is that my idea will work and can be made for very little money. If only the 783 would do the job then the BOM would be less than £1 for components bought in China
 

wayneh

Joined Sep 9, 2010
17,496
Interesting. Is there any possibility of, rather than replacing the suspect diode, you could add one in parallel with it, perhaps in another location? This additional diode could even be a schottky, which means it would have a lower voltage drop and would thus take virtually all the current load off the one on the board, assuming it is not also a schottky. My point is, there's no reason to remove the old diode if you can bypass it with another one.

Your hypothesis that capping the input voltage will solve the problem sounds reasonable but it's only an hypothesis at this point. Before you pursue that strategy, you should verify that it actually works. You could use a lab power supply for testing the life of your boards with a range of highly regulated voltages from the supply.
 

Dodgydave

Joined Jun 22, 2012
11,284
It can be done with a TL431 zener and Nchannel Mosfet, make R1 5.6K, R2 39K, R3 10K pot (set to 5k) R4 2.2K,C2, 22uF, C1, C3 100uF,this will give you a voltage range of 46 to 58V...


57nMB.gif
 

Thread Starter

Lee Tracey

Joined Jun 14, 2016
6
Interesting. Is there any possibility of, rather than replacing the suspect diode, you could add one in parallel with it, perhaps in another location? This additional diode could even be a schottky, which means it would have a lower voltage drop and would thus take virtually all the current load off the one on the board, assuming it is not also a schottky. My point is, there's no reason to remove the old diode if you can bypass it with another one.

Your hypothesis that capping the input voltage will solve the problem sounds reasonable but it's only an hypothesis at this point. Before you pursue that strategy, you should verify that it actually works. You could use a lab power supply for testing the life of your boards with a range of highly regulated voltages from the supply.
Please see the image. The diode is the one under the gung. Personally I shy away from messing with the board when the fault is with the PSU. I think that my friend sent out a DO NOT USE warning after the first two caught fire so the number of damaged boards is, as we know so far, eight. So, as the old saying goes: ".. if it 'aint broke, don't fix it. In any case the cost of opening some 400 cases, removing boards, and either replacing the diodes, or finding where to solder a second diode and then soldering it; testing the boards, putting them back into the cases and then add the shipping costs both ways would be so high it would be cheaper to send a donation to the Donald Trump survival fund. For me the item to tackle is the PSU then all my friend has to do is ship out the Reg module and then both the POE board and the existing crap PSU can be used - surely by far the cheapest route to minimise the damage cost.
 

Attachments

Thread Starter

Lee Tracey

Joined Jun 14, 2016
6
It can be done with a TL431 zener and Nchannel Mosfet, make R1 5.6K, R2 39K, R3 10K pot (set to 5k) R4 2.2K,C2, 22uF, C1, C3 100uF,this will give you a voltage range of 46 to 58V...


View attachment 128579
Hi " Dodgy Dave" Many thanks for the circuit I will now see if I can get it made. A hundred years ago I would have built it myself on veroboard, but I am now 91 and after a stroke my arms/hands and fingers co-ordination is a bit awry. Can you suggest an NChannel Mosfet to select?
 

Alec_t

Joined Sep 17, 2013
14,280
Be aware that you will need a large heatsink for the MOSFET, to dissipate all those Watts of heat. With 64V in, 40V out and 2A, losses = 48W!
 

Dodgydave

Joined Jun 22, 2012
11,284
Hi " Dodgy Dave" Many thanks for the circuit I will now see if I can get it made. A hundred years ago I would have built it myself on veroboard, but I am now 91 and after a stroke my arms/hands and fingers co-ordination is a bit awry. Can you suggest an NChannel Mosfet to select?
Yes try IRF531. Rated at 20Amp 60V.
Or. NTE66 14Amp @100V.
 

ian field

Joined Oct 27, 2012
6,536
It can be done with a TL431 zener and Nchannel Mosfet, make R1 5.6K, R2 39K, R3 10K pot (set to 5k) R4 2.2K,C2, 22uF, C1, C3 100uF,this will give you a voltage range of 46 to 58V...


View attachment 128579
The TL431 tops out somewhere around 34V or so.

You can extend its voltage handling in a casc-ode arrangement. The common base (or gate) part needs a fixed bias/reference - a BJT needs the reference at or just under the 431 voltage limit, a MOSFET allows a few volts higher.
 

Thread Starter

Lee Tracey

Joined Jun 14, 2016
6
The TL431 tops out somewhere around 34V or so.

You can extend its voltage handling in a casc-ode arrangement. The common base (or gate) part needs a fixed bias/reference - a BJT needs the reference at or just under the 431 voltage limit, a MOSFET allows a few volts higher.
Hi Ian: Many thanks for the contribution. Does your comment mean that you do not support the use of the Zener TL431 because it tops out at round 34V. and I need a possible 55 or, hopefully, 52?
 

Dodgydave

Joined Jun 22, 2012
11,284
The datasheet says maximum Cathode voltage of 32V, since the circuit will work, it may require a second pnp transistor in series with the Cathode to take up the excess voltage.
 

ian field

Joined Oct 27, 2012
6,536
Hi Ian: Many thanks for the contribution. Does your comment mean that you do not support the use of the Zener TL431 because it tops out at round 34V. and I need a possible 55 or, hopefully, 52?
The TL431 has many names - but its actually a comparator with its own built in Vref.

You can extend its voltage limit by using it in a casc-ode arrangement as I stated.

You could use a second 431 to set up a stable reference within its voltage handling - but a bog standard Zener would do since it would be within the 431 nfb loop.
 

crutschow

Joined Mar 14, 2008
34,280
DD's circuit in post #5 uses an N-MOSFET source-follower output so the maximum output is always more than one threshold-voltage below the input voltage.
Below is the LTspice simulation of a similar circuit that uses a P-MOSFET so the output voltage can go essentially to the input voltage (minus only the drop due to the output current through the MOSFET ON-resistance.
Thus it has a near 0V dropout.
You can see this as the input voltage ramps from 45V to 64V, with the output tracking the input until the regulation point is reached.
This is shown for the 0%, 50%, and 100% position of pot U2.

Q1 inverts the signal from U1 and level shifts it as needed to control the P-MOSFET.
R5 and R7 limit the TL431 cathode voltage to avoid stressing it.
C2 provides compensation to roll-off the high frequency gain and avoid oscillations.

The P-MOSFET is one that I had in my model library, which in a flat-pack package.
You will want to find one with a TO-220 case so you can easily mount it on a heatsink.
The simulation shows the power dissipated by the transistor (red traces) for a 25Ω load so you can see it needs a good heatsink.
The P-MOSFET should have a minimum 100V and 10A rating.

upload_2017-6-13_16-30-56.png

Note that this circuit has only been simulated so you definitely want to bread-board it before committing to a PCB.
 

Attachments

Last edited:

crutschow

Joined Mar 14, 2008
34,280
Did you miss the word; "comparator" in their description?
Not at all.
Read the whole sentence.
They are using the op amp to "behave" as a comparator, which is not an unusual thing to do with an op amp.
But it's still an op amp, not a comparator in the usual definition of the terms as applied to ICs.
 
Top