27th November
written by Todd Harrison


 I’m attempting to diagnose a problem with the motor power supply for a small metal lathe. This is part 2 so you may want to read part 1 first.

 I know the 120v AC mains are flowing through a monolith bridge rectifier and coming out as a fully rectified pulsing DC signal. This rectified DC signal is smoothed out by a very large 220v 300uF cap which is then supplying “somewhat clean” DC to the (+) terminal on the motor.  Notice I didn’t say anything about a transformer because there isn’t one.  That means the voltage to the motor is not being stepped down; just full wave rectified and smoothed out with a large cap.

READ —>:

 Even stranger I trace all the Vcc pins on the ICs back to an un-rectified 120v AC main! However, the trace does go through a big fat 1800 ohm 4watt resistor and then a large smoothing cap.  Also the IC ground pins all trace back to the rectified 0v DC trace.  For now I’m going to assume this is providing something close to the max voltage my ICs can handle per the datasheets 32v.  I will test this later but I would say it for sure it will not be a nice stable clean Vcc supply.

 Two IRF640 power transistors provide a path to ground for the motor when their gates are energized.  The gates are controlled by a UC3842 current mode pulse width modulation chip (PWM). The duty cycle of the PWM is controlled by other components which will be discussed later.

 If the PWM duty cycle is set to 0% there will be no on/off pulses to the power transistor gates hence the power transistors will never turn on hence the lathe motor gets 0V and will not run.  At the other extreme, if the PWM duty cycle is 100% the motor will get continues DC and would run at full speed. When the PWM duty cycle is somewhere between 0% and 100% the motor will receive DC power for only a portion of time relative to the duty cycle.

 Being these on/offs signals to the motor’s power supply are very fast the motor behaves as if it was being powered continually at a lower voltage relative to the PWM duty cycle.  The lower the perceived voltage the slow the motor will run.  The lathe’s motor is only rated for 90v DC continues so the UC3842 will never be allowed to go over a given duty cycle thus protecting the DC motor from being over powered. Of course the duty cycle can be dialed down to run the motor slower than its maximum speed.

 I’m now at the point where I really need to know just how the transistors and ICs are working together.  I do this by tracing out some of the circuit and mapping out a portion of a circuit diagram.  If you’re lucky like me you will have a single sided PCB board.  This means the copper traces are all on one side and the components are on the other side.  If you have a double or multilayered PCB then you might not be able to trace out much of the circuit and the board is probable not going to be serviceable anyway.

 In the below photo you see I first highlighted all the ground traces in black. 

Then in the next photo I mapped out select traces to select ICs and transistors in red.  I normally do just enough of this to get an understanding of what part of the ICs are being used and what transistors, resistors, caps ect. maybe involved.  But for this demonstration I painstakingly traced everything marking traces in red as they were added to my circuit diagram on paper. 

A complete circuit diagram is a huge help in trouble shooting a board but really I just need the circuit diagram for select pins on select ICs  to proceed with the diagnostics.  Here is a hi-res shot of the hand sketched circuit diagram.  It was way too much work but kind of fun in an obsessive compulsive way.
(click photo for hi-res)

 I will most likely take some time to redraw this diagram on several sheets of engineering paper for further reference in a cleaner layout, but truthfully that’s not needed. 

During the tracing process I did notice the below dry solder joint on a signal line to a power transistor but it electrically tested fine.  I took the photo with a couple of eye piece loupes which is why it looks so funny.  But this trick makes for super close up shots with a cheap point-n-shot camera.

 I don’t have time for anymore tonight nor do I have time to work on this project for a few days.  Most likely I will get time next week to tackle some more of this project and upload part 3.


  1. José Araújo

    Broken/dry solder joints almost always test ok, but they are not!
    Pressure from test leads will force them into contact, but with the board tuned on you can’t be sure, and won’t have any current carrying capacity.
    They’re a good indicator of where the problem origin may be. Heat stress and over-current can lead to dry solder joints. You should verify the connecting components (in this case the power transistor), and re-solder before continuing.

  2. 29/11/2010

    I do concur. I reflowed as well as added a bit more solder right after taking the photo. But I’m quite sure this is not my problem because when I was probing earlier there was no signal coming from pin 6 on the PWM chip. Even if this joint was good at this power transistor gate it wouldn’t matter. Plus, there are two power transistors in parallel here and one would have been enough to run the motor at lower speeds.

  3. Mike

    Change the 3 electrolytic caps, Not the mains filter the other 3. I service all sorts of stuff for a living and 80% of the time its just dry caps. They will measure OK on a cap tester but if you had a esr meter they would measure open.
    When testing circuits of this type use the ic’s own ground as the primary ground or neutral will give unusual readings(like 120v on a vcc leg for example).


  4. 30/11/2010

    True about caps. That is always one of my first suspicions when working with power supplies or other circuits exposed to heat and age. But in this case I don’t think it will trun out to be the problem. The supply was working great until a flake of metal from the lathe managed to get kicked into the power supply and “Pop” that was that. I’m sure I will find a bad IC or one of the small BJTs kick the bucket during the short.

    I have not had time to meter the ICs Vcc yet but when I do I will let you know what I find here. I will remember your tips.

    Any other ideas?


  5. José Araújo

    About the VCC of the ICs.
    There is a Zenner diode in there (ZD1), it is there to provide regulation. It’s pretty common to do it like that for keeping things simple and cheap.
    Verify the Zenner is ok (usualy the fail shorted). When powered up you shound have a pretty clean DC voltage for VCC of less than 35V (or else C2 whould go up in smoke).

  6. 30/11/2010

    I will for sure check. Most likely I will get some free time for this Friday night. Thanks for your input.

  7. […] he goes over the supply and identifies the main parts and tests all of the big obvious stuff. In part 2 he goes through and traces out all of the electrical connections on the board! You can see a small image of it below but check it out in full size for more details. This is […]

  8. signal7

    Two things come to mind here. A fuse that passes a continuity test may not actually be functional. Sometimes the filament in the fuse will break, but the ends will reconnect when there’s no amperage running through the fuse. The only reliable test I’ve ever found was to check the voltage across the fuse during normal operation. It should be 0V.

    Second, I’ve also found that you can’t really use a multimeter to check a cold solder joint. I’ve worked on some of the old 3/4 inch professional video tape recorders and have seen cold solder joints that test fine with the multimeter. I even had one machine where it was faster to just go over all of the joints on the board with a soldering iron than it was to spend time troubleshooting and that fixed it (but I already knew at that point the problem was a flaky connection somewhere).

    One thing the multimeter is great for is tracing a circuit board – especially when you get to two sided boards. I had a CO detector that would sound an alarm if it had a low battery which I found VERY irritating. Since this device plugs into the wall anyway, it does not *technically* need a battery at all. I’m fully comfortable with the fact that it won’t work without power, so I hacked the circuit so it would think it always had a battery. Double sided board in that unit and I don’t think I could have done it without using the multimeter to trace the circuit.

Visit Our Sponsor: MROSupply.com - An e-commerce site of Mechanical Drives & Belting