Troubleshooting the linear power supply

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Svart

Well-known member
Joined
Jun 4, 2004
Messages
5,134
Location
Atlanta GA USA
In our last tutorial, if you can call it that, we touched lightly on the general failures of the SMPS.

Today it's the Linear power supply's turn.

Although it seems to be much more simple than the switching power supply, a linear power supply also causes it's share of headaches.

Let us discuss the general parts of the Linear supply.

1. Input. This is usually fairly similar between linear power supplies and the SMPS as well. AC is first brought into the unit through some sort of cabling. In good supplies you will then have a fuse, breaker or other such device to limit the input current during a fault mode. From there you might have some type of filtering featuring MOVs, common mode chokes and other such goodies.

2. Current/Voltage transformation. We love our little transformers. This section is almost explains itself. The transformer uses magnetic fields to motivate electrons(or holes) to move around a coil of wire and give us a current/voltage. The primary and secondary windings can be wound at different ratios to each other to give us different voltage outputs. Easy right? There is a lot more to this but I don't want to give you a headache just yet.

3. Rectification. Again this is somewhat similar between the different types and models of power supplies. Sometimes we get full wave rectification, sometimes we get less, either way we stop the reverse currents and generally want to end up with a DC potential. There are all kinds of different rectifiers out there and some do an incredible job while some barely work without burning the house down. After the rectifiers we usually see a capacitor or even a group of different caps and resistors. Each design might be different so we will stick to the general design and forget about the details for now.

4. Voltage regulation. Now this is a really WIDE subject, much like one of my ex girlfriends but I digress. There are as many voltage regulation devices and methods as there are stars in the sky. Luckly for us they don't usually burn as hot when they catch fire though. Most linear power supplies use a linear regulator, resistive element or some kind of semiconductor pass device. Some linear regulators actually use a combination of all of these methods, but again this is just details for most people who simply want their gear to work again. Explaining how these different devices work would take me much too long. I'll let the datasheets explain instead.

5. Output filtering. Barring any other circuitry for regulation feedback or protection, the output usually has a degree of filtering to keep ripple low and to keep HF/RF noise off of the voltage rails and provide a source of instantanious current for loads that tend to create very fast current draws. This may include capacitors and resistors as well as other types of voltage regulators and other devices for protection.


Now that we have that done we can start with the short and sweet troubleshooting.

We usually need to ask certain questions that will give us clues on where to start looking for problems. Does the unit turn on at all? Did it let out smoke and smell like it burned? We don't want to be unsafe so my suggestion is that if the unit caught fire or otherwise gave the idea that it was a burning failure, don't turn it on or else parts could ignite once again.

With that said, lets get to it.

1. Without a power source your unit won't work. Check your AC cable, check your fuse and check your power switch. This is the same for all types.

2. If you have power going in to the power supply but don't have power coming out, start at the AC side and move inwards. Probe for AC on the input legs of your transformer. If it's not there then move backwards and check all the solder joints while continuing to check each node for AC. Check for a short to ground. Sometimes the MOVs and other protection circuits are meant to short to ground and blow the fuse in their fault conditions.

3 If you have AC reaching your transformer's primary, check the secondary for the correct AC voltage potential. If you see it, great move on to #4, if not then we *might* have a bad transformer. Some toroidal transformers have temperature fuses within the windings that blow when it overheats. Another possibility is that the transformer winding is internally open. Yet another possibility is that the problem is further down in the circuit and it is grounding the output of the transformer. If this is the case, your transformer will likely be hot and/or humming slightly. In any of these cases I suggest pulling the leads of the secondary out of the circuit and testing them directly. If you have voltage, the problem is elsewhere. If you don't have voltage output but you have input, your transformer may be broken.

4. If you have an output from the transformer but it grounds when you put it back in the circuit, a short is to be expected. This can come from many places that aren't very easy to troubleshoot in circuit. My next suggestion would be to disconnect the output of the power supply if you can. A lot of times we troubleshoot the powersupply only to find out that a part somewhere else on the board has shorted and pulled the PSU to ground. If you still get no output we will look at the power supply itself starting with the regulator. Next I would look at the voltage regulator. If it is a general 3 pin Vreg, pull it out of the circuit. Now you can ohm between the legs or you can ohm the board itself to see if the short to ground has gone away. Check both the input and output traces of the pcb. If the problem is gone, great, buy a new regulator but chances are that it might have started to fail by letting the full voltage on it's input come out of it's output and there may be more problems down the line. If the problem is still there we need to check the capacitors around the regulator. Electrolytic capacitors that have shorted ususally get "puffy" and leak but usually explode or fail open internally. Tantalum caps are always a terrible idea for power supplies as they always fail shorted before they explode. Ceramic caps can fail either way, poly caps rarely short and can usually take some serious juice before going bad. If there are tantalums, remove them but remember where they go and how. The band on the tantalum cap is opposite that of a diode and other parts that use bands to show the cathode, it shows the +anode side of the cap. Retest. They will usually have a slight brown discoloration on their tops if they are bad. Again, if the caps are good, move on to any other types of caps that bridge between the voltage rail and ground. Without the supply hooked up to the circuitry that it is supposed to power, you will have likely found the problem by now if the problem was in the power supply to begin with. If the short is still present you will need to likely start pulling any other parts that could have shorted the voltage rail to ground. This could include various types of diodes, resistors, transorbs and other protection devices.


If you have fixed the power supply then be proud, but be cautious, there may still be problems further down the circuit which could immediately cause it to fail once again, but hey, that is for another installment of our tutorials!

If I have forgotten something or given too little/incorrect information, please feel free to contact me or post a response here. I've been typing on this a little at a time during my day at work and I am sure I've forgotten something.
 
I always use a current limiter when powering up the first time. The lightbulb tells me if there's a short, or no current. I use a chopstick to probe the wires and power switch. If the unit I'm working on powers up, but is acting funny, I make sure it's set to 120vA/C and not 220vA/C, that's an easy fix. On older units, vintage tube amps for example, the power plug is non-polarized, so the fuse can end up on the hot, or cold side, depending on which way it is plugged in. If the fuse is on the cold side, and it blows, you may become the return path to ground, with no fuse. I fix that with a polarized plug. When checking a powered speaker, I play some rap music with good lowend, if the bass sounds weak I know it's a filter cap. I check the +/- d.c. with a multi-meter. If one side is substatially lower, I know it's lacking filter. I can verify that with the O-scope.sometimes the filtercap mounted to the circuitboard will develop a cold solder joint. I believe this is due to the joint heating up, and the excessive vibrations and sound pressure levels. I like to use silversolder to reflow those joints as it has a higher melting point, and I glue the cap onto the PCB with E6000, or plumbers goop, it dries hard unlike silicone sealer that allows the cap to wiggle.
 
Some of the open frame units 5,12,15 24/28VDC (732 IC based ones) use a voltage X2 circuit to raise the voltage for the NPN pass transistors.

This seems to be done so the base(s) can be driven higher than the emitter allowing the use of a NPN(s) and not PNP(s)

Often the pass section is a darlington so there is more that one BE drop.

If the supply has high ripple under load check/replace the X 2 cap first look for a diode 1n400something and something like a 220uf 50 VDC of bigger and higher voltage rating by the diode bridge for the big cap(s).

Check under load with a scope for looking at the ripple
 
I discovered this yesterday... my newly created 18-0-18 PSU was giving me 18-0-26 volts.

Yes... -26 volts instead of the -18 I was expecting.

Replaced the 7918 regulator 3x as I thought it was defective. Didn't fix the problem.

Turned out that these 78xx and 79xx regulators have a max input voltage, and if you go over the max input voltage, you're not guaranteed the regulated output voltage anymore.

In my case, I found out that the input voltage was -35V, which gets regulated to -26.

Weird thing is the 7818's input voltage was +35V, and it gets regulated just fine to +18.

Cause of the problems???

The toroid transformer I was using was rated 22-0-22 was not giving me 22vac... but rather 26-0-26 vac. So rectifying it, gives 36.7V which is way above the 33V max voltage input limit of 78xx and 79xx regulators.

So if you see some funky output voltage from your TO-220 regulator, check the input voltage going into the device. More than likely, you exceeded the max input value specs of the device.

Then work your way backwards and check your transformer if it's giving more voltage than it's rated for.

I checked my trafo specs, and it specifies a 26volts no load secondary voltage, and 22volts at load.
 
I'm not sure if this fits in this thread but a lot of us use the JLM power supplies for our DIY (And some the MNATS PSU)... Are they suitable for all the applications we use them for? & if not... why?

I'm currently working on an EQ circuit in which the original supply was switching & I will be using a linear supply instead - it had 470uf x 2 (per rail) on each card for filtering (I'm not reffering to the filtering on the actual power supply... these are extra caps away from the supply on the cards themselves for local filtering) - I was told the 470 x 2 (=940uf per rail) is too much for the new Linear supply :? is this true & what would be a good value to use in their place... I was told to use 47uf x 2 instead but I would like to learn why & what determins the optimum value for any given situation like this? :oops:

Also, WRT regulators (317,337's), I was also told that the 10uf caps around the regulators can cause problems & should (or would be better) being 4.7uf or even Tant's...or even leave it out???

Just wanted to bounce these things around for a broader view.
(Just for the record I'm currently using the MNATS PSU for the EQ)


Thanks,
Kevin
 
[quote author="khstudio"]I'm not sure if this fits in this thread but a lot of us use the JLM power supplies for our DIY (And some the MNATS PSU)... Are they suitable for all the applications we use them for? & if not... why?

I'm currently working on an EQ circuit in which the original supply was switching & I will be using a linear supply instead - it had 470uf x 2 (per rail) on each card for filtering (I'm not reffering to the filtering on the actual power supply... these are extra caps away from the supply on the cards themselves for local filtering) - I was told the 470 x 2 (=940uf per rail) is too much for the new Linear supply :? is this true & what would be a good value to use in their place... I was told to use 47uf x 2 instead but I would like to learn why & what determins the optimum value for any given situation like this? :oops:

Also, WRT regulators (317,337's), I was also told that the 10uf caps around the regulators can cause problems & should (or would be better) being 4.7uf or even Tant's...or even leave it out???

Just wanted to bounce these things around for a broader view.
(Just for the record I'm currently using the MNATS PSU for the EQ)


Thanks,
Kevin[/quote]

It seems like we had a discussion about this recently. I did some measurements back in the '80s with cheap 3 terminal regulators and found their output impedance rose noticeably within the audio bandwidth. This IMO is because they are based on old 741 opamp GBW technology.

I don't recall the exact value (1-2,000 uF) but found I could lower the output impedance at 20 kHz to be as low as 20 Hz by paralleling with a decent sized electrolytic.

Some suggest the importance of anti-parallel diodes across the 3 terminal regulators to insure large output caps don't discharge into the regulator backwards. I am not as convinced of the need for this in most applications but diodes are cheap. YMMV

JR
 
Back
Top