Voltage drop.

[davelectronic]

Recently I've got to use a finished project, a linear PSU for bench top use. It's a 12 Volt linear power supply with four series pass transistors, it uses LM2940 and TIP36C PNP power transistors. The transformers are two toroidal transformers rated at 220 Volt primary windings and 16 Volt secondary windings. Both winding are wired in parallel for greater current use. I have a capacitor bank of 40000uf in four 10000uf can snap in electrolytic capacitors. The rectified voltage is 20.05 Volts before it reaches the voltage regulator. No load final output is 12.60 Volts DC, if a add a 50 watt halogen lamp, it drops the voltage to 12.35 Volts, and with two 50 watt lamps it's dropped to 11.98 Volts. I did expect something of a voltage drop, but feel this is a bit much of a voltage drop for the load on the output. The output does go through a crowbar circuit with added panel meter, all tracks and chassis wiring is 16 AWG wire for input output, and the transistors wire. I kept all chassis wiring as short as possible to avoid voltage drops, and used 1mm copper wire with solder for all high current paths. I know it's usable for its intended purpose, which is for HF radio use, and linear amplifiers. But I thought a rectified filtered voltage of just over 20 Volts DC would be enough for the LM2940 LDR linear regulator. The heatsinks for the regulator and power transistors are a pair of repurposed dell optiplex CPU heatsinks, with two 80mm fans connected to a 50°C thermal switch for thermal activation of the fans. I've checked heat at full load, its a shade over 50°C when the fans switch on, and quickly comes down once the fans are running. The image is just to give an idea on the size of PSU this is. As said I have tried every step of the way to avoid voltage drops, or minimise this affect as much as possible. Any thoughts appreciated, thank you for reading.

[SuzyC]

Show a schematic.
Your 1mm wiring for all "all high current paths" is not thick enough for 100-W loads and causing the voltage drops.

Your regulator probably is outputting the correct voltage where it is sensing the output  voltage, but this sensing point is likely some distance from the front-panel terminals, hence you see the voltage drops you posted.

A few measurements with you DVM from the output terminals to the voltage regulator PCB output point should show this voltages drop problem.

With a 100-W load,(which corresponds to an 8-amp load) and if there is ~.05-ohm wiring resistance, this would show a .4V drop. What do you measure?

[BeBuLamar]

So you detect the voltage drop at the meter? If so if it's the wiring then it's the wiring from the power supply to the meter. I would check the voltage at the other end of the wires. I don't think 1.6V drop is due to the wires alone. I think the output of your regulator drops a bit too.

[eblc1388]

The transformers are two toroidal transformers rated at 220 Volt primary windings and 16 Volt secondary windings. Both winding are wired in parallel for greater current use. I have a capacitor bank of 40000uf in four 10000uf can snap in electrolytic capacitors. The rectified voltage is 20.05 Volts before it reaches the voltage regulator.

There are discrepancies between what you said and the measurement results.

If the transformers are really 220V/16V, then when connected to the UK 240V system should give you a secondary voltage of 17.45V AC and no load peak voltage of 24.6V DC after rectification.

More likely is the transformer is only rated at 220V/12V and that would explain the actual voltage drop when loaded.

Please measure the voltage (across the main smoothing capacitor) when loaded with the lamps. If you have another meter, you can also monitor the voltage between the smoothing capacitor positive terminal and the +12V positive output, as this would tell you how much "headroom" the regulator has to perform its job.

Tell us these two voltages under loaded condition and we can tell you the reason of the voltage drop.

[davelectronic]

Thank you for your replies. I had adopted a different way of connecting the voltage regulator and transistors. I used molex connections for these, the molex plug for the transistors is rated for a maximum of 7 Amp load, the wire to these is 16 AWG. The regulator molex is rated at 4 Amps maximum load, I kept these leads under 6" long from the heatsinks to PCB. The cable to the black box (crowbar circuit) is rated for 10 Amps, it's about 10" long. The toroidal transformers measured just under 17 Volts AC when measured with a multimeter, and at the rectifiers I get 20 Volts with no load, and around 17.80 Volts DC under a two 50 watt lamp load. I think I need to make another heavy duty lead from the power supply to the crowbar circuit box. The 1mm solid copper wire has a very heavy two track run for the entire length of the copper stripboard I used. And the circuit is the 7812 and six pass transistors, I will post it, in this power supply I've used four transistors, not 6 transistors. Where ever there is a high current path I've laid heavy wire, or copper wire solder tracks.The filter capacitors are very large, and there are four of them each rated at 10000uf, as I knew 20 Volts DC was I little lower than if the transformers secondary windings where 18 volts AC, but I thought with an LDR regulator that I would have sufficient voltage for moderate heavy loads. I've connected up a CB radio that draws 1.2 Amps under load, the voltage drops from 12.60 Volts to 12.40 Volts, that's just a 4 watt radio I tried. I currently don't have a linear HF amplifier to try at the moment. I think I might look over every high current path to see if something is insufficient to carry a heavy load.

[SuzyC]

I think your problem is caused by the resistance of the 1-Amp fuse on the output of the 7812 regulator. This fuse is not needed if you have a crowbar, or else the fuse can be relocated to the 7812 input.  Keep in mind that the 7812 can instantaneously at room temp supply over an ampere and blow the fuse at high output currents.

[eblc1388]

The toroidal transformers measured just under 17 Volts AC when measured with a multimeter, and at the rectifiers I get 20 Volts with no load, and around 17.80 Volts DC under a two 50 watt lamp load.

It looks like you have a pretty good headroom for the regulator to operate(17.8V - 12.6V = 5.2V). In this case, it is likely that there are voltage drops in your leads as others has said. The GND terminal of the three-terminal regulator 7812 should be as near to the negative output terminal connection as possible to minimize any voltage drop.

If you remove the 1A fuse, you'll get exactly 12V from the 7812+series transistors instead of the current 12.6V. If 12.6V is your target voltage, you can place in series a silicone diode at the common terminal of the 7812 and 0V connection.
 

[james_s]

A fuse is completely unnecessary with a 7812 which has internal overload protection. A fuse should be used on the primary of the power transformer and ideally on the secondary before the rectifier since your power supply does not have any form of electronic overload protection on the output but you should not need any other fuses. You mentioned some connectors rated at 4A and 7A, are they carrying the full output current? As someone else mentioned, 12V at 100W is over 8 amps, that's a substantial load for a linear power supply.

[floobydust]

It's critical to have both the 7812 output and GND connecting right at the output binding posts of the PSU. This is so the IC can accurately sense output voltage without wiring voltage drops giving errors.
The 1A fuse should be on the input side, and all it's good for is limiting smoke if that IC shorts.
You should have a minimum load, at least an LED at the PSU output. I think the LM7812 spec is 5mA.

Note an LM317 is wired differently, for accurate sense overcoming wiring voltage drops, the 240R connects right to the IC's output pin, as in pic.

[davelectronic]

Yes the leads from the transistors are 16 AWG I think the molex has a pitch of 5.25mm or something like that. The regulator has 18 AWG wire and a smaller molex line socket. I have used a 1 Amp fuse in the regulator output close to it. The only other fuse is for the cooling fans at the back of the case. I get less loss directly from the terminals on the case front, but there is a drop there, more than I expected at 8 Amps load current. 12.35 Volts from those posts, and not through the crowbar circuit. Although the wiring in the crowbar circuit is substantially heavy like the PSU circuit. Yes I did add a diode just in front of the regulator, it's what has given me the extra 0.6 volts over the 12 volts from the regulator. In the past I have used two diodes in series like this. Although in a previous circuit going back some time now, I could only get the 0.7 Volts adding a second diode did nothing. So this time I just used one diode in the regulator negative pin rail. All the collectors are right at the final output of the + Volts leaving the PCB. I will try removing the fuse, but the regulator doesn't reach 1 Amp in its part of supplying the total output current from 4 X power transistors. I was hoping for at least a solid 12 Amps at 12 Volts. In the article that circuit diagram came from, the designer quoted 5 Amps from each transistor, I would be happy with 3 Amps from 4 X TIP36C transistors. Just a thought, I've used darlington transistors before and the voltage was solid at 10 Amps in an earlier PSU going back a while. Maybe that would help.

[davelectronic]

There is two led's on the front panel the power on led, the other is for fan on indication. The red led is above 5mA. The only other thing I've forgotten to mention is a high current relay, the latching red push switch is only rated for 1.5 Amps. I wanted a way to isolate the output with out turning off the mains. So a relay made sense as I means to cut the output. I didn't want a toggle switch on the front panel. But there is a second switch on the rear of the case to completely turn off power to the PSU. I doubt the relay has any bearing on the output voltage. It's almost impossible to get the IC terminals right on the front binding posts, there's about 4" of 16 AWG to the front posts, and an extra 2" for the front panel fuse. I've used an LM317 before with pass transistors, it was fine.

[james_s]

Apologies if you've mentioned it somewhere but I didn't see it. Where is the drop taking place? What is the voltage at the filter capacitors under no load and full load? What is the voltage at the output of the 7812? What is the voltage at the output of the pass transistors? The first step to diagnosing what is going on here is to figure out exactly where the drop is taking place. There is no magic involved, your power supply obeys Ohms law.

[davelectronic]

The voltage input to the LM2940 is 20 Volts, that's the rectified and smoothed voltage. The output from the LM2940 is the same at the regulator as it is at the output terminal posts. The voltage drop is when load is applied, a 50 watt 12 volt halogen lamp drops 12.60 Volts to 12.20 Volts. Two 50 watt halogen lamps drop the voltage at the terminal posts to 11.98 Volts, and 11.70 Volts at the output from the crowbar circuit. So the worst figure is from the output from the crowbar circuit. From what I can make out form above, any added distance from the regulator and transistors is going to show a voltage drop, how much seems to depend on wire / cable rating and distance. It's not until I've had time to think back that I realised this has happened before. And it was after using TIP2955 transistors, I swapped them out replaced the heatsink to a TO3 heatsink and 4 X MJ11015 transistors, the voltage drop gap closed up nicely. Only from memory, but a similar scenario as this was improved to 0.2 Volt drop with a three 12 volt halogen lamp load, it's was rock solid. I don't think the TIP36C are right for this type of circuit, or there gain is not enough under heavy loads. I'm just a hobbyist, I'm certainly no expert, but I do remember asking a question about the MJ11015, and the answer was, the voltage is less likely to sag, and if it does, it won't be nearly as much as the TIP2955 transistors. I can't be 100% sure, but I think this.would work possibly. I can only remember vaguely, it was some years ago now, but I do remember adding at least 4 X 50 watt halogen lamps, and think wow, I was expecting more volts drop, but it didn't, well 0.2 Volts from the 13.20 volts from that circuit output. I did try at every stage to limit distance from output to terminal posts, and kept tracks and wiring as heavy as was possible inside the case. It does get crowded with heavy gauge cable. But I've been using silicone cable, which is really easy to route or manipulate in tight spaces. I expect a volts drop to some extent, just not as much as 0.9 Volts with an 8 Amp load, approximately that is.

[radiolistener]

No load final output is 12.60 Volts DC, if a add a 50 watt halogen lamp, it drops the voltage to 12.35 Volts, and with two 50 watt lamps it's dropped to 11.98 Volts. I did expect something of a voltage drop, but feel this is a bit much of a voltage drop for the load on the output.

If the halogen lamp has working resistance about 3.056 Ω (12.35 V for 50 W) which leads to 12.35 Volts on it. Then two of them in parallel have resistance about 1.528 Ω which leads to 11.98 Volts on it. In such way we can estimate your PSU internal resistance, it's about 0.097 Ω. It's too much for 8 Amps load.

At a glance your wire with a cheap Chinese alligators looks very suspicious. These allgators can easily add 0.1 Ω and it will lead to voltage drop. So, I recommend to replace it with a good thick copper wires and replace weak alligators with a good crimp or soldered contacts. Also check internal wires inside PSU if their resistance are ok, then check if transformer can keep enough Voltage under full load.



 

[floobydust]

Do you have an accurate schematic? Whatever circuit you are using now, you've mention several changes.
LM2940 needs 22uF min. output capacitance for stability and given that we're boosting output current many times I wonder if an LDO is going to work with the additional gain. Have you looked at the O/P with a scope to see if it's staying stable or upsized the O/P cap?

I wasn't saying mount the IC right at the O/P binding posts but that you need dedicated wires for output and GND for the Vreg IC to them, so it is not wrongly amplifying a (wiring) voltage drop or AC hum and ripple. How you have connected together the grounds of bridge rectifier, filter capacitors, IC, binding posts etc is important. If done wrong, your Vreg will amplify ripple or do poorly at the voltage sense due to wiring voltage drops. A fuse, a relay, will have significant voltage drop as well.

[davelectronic]

I have an old Hameg scope 60 MHz I think it is, it works ok. The wire to crocodile clips, yes agreed that isn't helping after you mentioned it, I missed it. Certainly try heavier gauge wire for that. Inside every connection is  16 AMG straight to the board, and fast on tags to the rectifiers, all wires are as short as I can possibly get them, and of a gauge that should deliver the current at least 12 Amps. I think it's the toroidal transformers, (an eBay purchase) or transistors with far higher gain would work. The AC side hold up ok on an 8 Amp load, so I'm not suspecting the transformers just yet. The difference in voltage sag under load from single transistors, to using high gain Darlington transistors is significant in terms of a solid voltage under load output. I will check again incase I've over looked something.

[james_s]

The voltage input to the LM2940 is 20 Volts, that's the rectified and smoothed voltage. The output from the LM2940 is the same at the regulator as it is at the output terminal posts. The voltage drop is when load is applied, a 50 watt 12 volt halogen lamp drops 12.60 Volts to 12.20 Volts. Two 50 watt halogen lamps drop the voltage at the terminal posts to 11.98 Volts, and 11.70 Volts at the output from the crowbar circuit.

You have 20V at the input to the LM2940 while under load or without the load? I would expect it sag significantly under load, that's why you have a regulator in the first place. If that voltage is staying high enough to give the regulator sufficient headroom then you should be fine. If you want the voltage at the terminal posts to stay constant under load then you need to pick off the voltage the regulator is seeing from there. It doesn't take much resistance to create a drop of a few tenths of a volt with a load of 8 amps.

[james_s]

I have an old Hameg scope 60 MHz I think it is, it works ok. The wire to crocodile clips, yes agreed that isn't helping after you mentioned it, I missed it. Certainly try heavier gauge wire for that. Inside every connection is  16 AMG straight to the board, and fast on tags to the rectifiers, all wires are as short as I can possibly get them, and of a gauge that should deliver the current at least 12 Amps. I think it's the toroidal transformers, (an eBay purchase) or transistors with far higher gain would work. The AC side hold up ok on an 8 Amp load, so I'm not suspecting the transformers just yet. The difference in voltage sag under load from single transistors, to using high gain Darlington transistors is significant in terms of a solid voltage under load output. I will check again incase I've over looked something.

Now I'm getting confused, why would you suspect the transformer? All you have to do to rule that out is measure the DC voltage at the output of the rectifier while under load to rule this out. I don't think the problem is the transistors either, unless the regulator is driving the base with as much current as it is capable of and that still isn't enough. Your wires don't need to be super short as long as they are wired such that the regulator compensates for the drop. What do you mean by the wire "should deliver the current at least 12 amps"? The wire will carry that current without melting but it's going to have resistance. You can look up the resistance of copper wire in various gauges and calculate the resistance of the wires you're using. Then you can use Ohms law to calculate the voltage drop that will occur over a given length of that wire.

[davelectronic]

I'm familiar with wire resistance, often looking up one of the charts for wire gauge for chassis and transmission line use. No it's not the transformers I doubt, I was just voicing aloud. Bit like what's up with this bloody thing. At the filter capacitors no load output I get just over 20 Volts DC, if I load the output with two halogen lamps, that voltage drops to just over 17 Volts DC. The output directly from the regulator pins is 12.60 Volts with no load, that voltage drops to 12.20 Volts DC from the regulator pins. It's virtually the same from the front binding posts ( not the crowbar circuit meter box ) If I have 11.40 Volts DC or higher it will drive the intended load. I didn't expect an 800mV drop with  two 12 Volt 50 watt halogen lamps. I've look over every thing, there is no significant resistive path that I can find. If 6" is to far from the main PCB, then I will have to mount it directly to the board. But heatsink for the regulator becomes a difficult physical fit, well at least a suitable size heatsink for the regulator. I've looked over the data sheet for the LM2940, the best I can make out, it say 80mV drop for load regulation, I could have that wrong if I've read it wrong. Beyond that I'm at a loss to know why it's a significant voltage drop.

[SilverSolder]

Could it be that the resistance of R7 is too high at 100 ohms, so the regulator has trouble drawing enough current to reach a good operating point (with the transistors "stealing" all its current)?

Maybe try a lower resistor, give it something to work with...

[xavier60]

Also, the regulator IC might be faulty or oscillating. The bypass capacitors need to be close to it.

[davelectronic]

Here goes, I feel really stupid..
I connected up the halogen lamps again directly to the front binding posts, I had the lid off to test voltages again. Starting with 1 X 12 volt 50 watt lamp, measured from power leaving the board, and binding posts, virtually no voltage drop. Then 2 X halogen lamps, still no voltage drop to mention. Then 3 X halogen lamps, the smallest voltage drop, around 0.15 mV drop. At the same time I was measuring different areas on the circuit, I had a DC clamp meter measuring the current. It was solid at 12.46 Amps at 12.52 Volts, there is a but though. I did attempt fourth halogen lamp, but the voltage collapsed to 10.11 Volts, I don't know why that happened, the toroidal transformers are rated at 16 Amps each, and there in parallel. But the 12 Amps is a result, although I'd be interested to find out why a fourth lamp collapsed the voltage. I did remove the crocodile clips and that not so thick wire that was connecting the lamps. One up to radiolistener for calling that out, thank you. I do have to remove the multipole chassis plug and line socket to the black protection box, and wire in a heavier permanent fixed cable connection. The 100R resistor R7 might be a bit high for four X transistors, the original schematic uses 6 X transistors. But happy with what I've got, my intended load is a 30 watt radio and a 100 watt HF linear amplifier, with the radios RF power turned down to drive the linear amplifier. What I'm trying to get to eventually is a 30 Amp power supply that can deliver exactly that, at around 13.50 Volts, for a much bigger amplifier. I probably could have done that from the start, but felt I needed an interim project linear power supply first, just to get the confidence boost.

[Jwillis]

Transformer current rating isn't the only thing that can cause a sudden drop in voltage . Theirs also transformer voltage regulation and efficiency . Being toroidal transformers  are better than other constructions because the field is more consistent . But the material used  for the core and winding accuracy can have a significant effect . When the current flowing in the primary windings of the transformer is increased, the magnetic field intensity increases, but the magnetic flux density does not increase infinitely.Exceeding this limit puts the transformer into saturation . Any further current draw will drop the voltage significantly.
In many cases a transformer is rated based entirely on the calculated VA but not on actual conducted tests. This is the no load current rating .  To test a transformers regulation percentage you need to know the current at no load and at full load . Usually through a 1 load resistor.
               
                   Vs(no load) - Vs(full load
Reg% =   -----------------------------------  X 100%
                                Vs(no load)
Normally 2% to 5%  is really good  and 5%to 10% is common .  Load regulation  can be much worse the smaller the transformer gets. And even two small transformers can have a poor load regulation compared to one larger transformer.
I've experienced the same issue with a 30V 26A 800VA when nearing a current load of 15A. And voltage drops of quite a bit at 20A . Turns out I need a bigger transformer for the application . Really easy to see what is happening to the secondary with a scope but you can check with a DMM before the rectifier.

[davelectronic]

This is interesting stuff, I think the pair I bought where 160VA each, 240 Volt single primary. The secondary windings where listed as 16 Volts single secondary. But it's closer to 17 Volts AC, I've beefed up the input to the crowbar panel meter box. These transformers where really tatty winding tape, the clear plastic stuff. I re wrapped them with cotton tape, and dipped them in insulation varnish I got off RS here in the UK. I had no mounting hardware, so drill rubber sanding pads make great tough mounting rubber disks, couple of mild steel disks drilled, bingo mounted. The bottom half of the PSU has the toroidal transformers in it, everything else is in the top half of the unit. It's two cases bolted together, then drilled out holes with grommets for wiring that traverse from bottom to top half. Just using it for the radio at the moment. But I will using 100 watt amplifier, when I get round to getting one. The 12 Amps I've got I'm more than happy with.

[james_s]

Keep in mind the transformer voltage rating will be at the full rated current and at the rated line voltage. If your load is less than what the transformer is rated or if the mains voltage is higher then the output voltage will be higher too.