Image Map Image Map
Page 1 of 4 1234 LastLast
Results 1 to 10 of 40

Thread: Power Supply grounding

  1. #1
    Join Date
    Dec 2013
    Location
    Near Milwaukee Wisconsin
    Posts
    1,133

    Default Power Supply grounding

    I have been working on an interface between my CP/M machine and Friden SFD-V. There is some 60 cycle, AC noise on the power supply that is interfering with input data from the Friden's keyboard. I found that using a 1.0 uF capacitor between the power supply negative and earth ground, shunts the noise away. After some reading, I'm still confused as to the best grounding for power supplies. To that end, I was looking at the H724 power supply schematic of my PDP8E. I noticed that there are two different symbols regarding grounds on the DEC schematic. The normal symbol (logic ground?) with the three horizontal bars and the symbol that I equate to earth ground, looks like a pitch fork. I noticed that ground pin from the power cord plug is connected to this earth symbol, some of the primary AC circuits and the negative terminal of the DC outputs. There is also a couple of capacitors on the +5, +15, and -15VDC to the logic ground.

    In the past, I have always kept my DC power floating. I had thought that if the DC power were isolated, noise and other hazardous voltages would not or could not affect the rest of the machine. In this application, connecting to the Friden, there is 12 feet of cable between the two, including a ground connection. Apparently, these wires are acting like an antenna and impressing random signals in places I do not need them.

    I read the maintenance manual section for the H724, but there was nothing about the grounding or the difference between the ground symbols.

    Can anyone recommend a document that speaks to proper grounding and bypassing linear power supplies used in computers and other hardware.

    Thanks Mike

  2. #2
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    30,279
    Blog Entries
    20

    Default

    Linear power supplies themselves don't kick up any RFI and are very often left with the outputs floating so that they can source either + or - relative to common. They can, however, couple noise on the DC rails back into the AC line.

    Here's a guide to the symbols

    Although the stack o' lines and the triangle are often used interchangeably, the "pitchfork" almost always refers to a chassis or frame ground. Like many schematic symbols, the meanings are often abused.

    If the noise is present on the AC line, you may want to consider installing a power line filter in your equipment, such as one illustrated here. These can be very effective.

  3. #3

    Default

    The art of grounding depends on the use. Consider any current flowing through a ground path as a source of noise. If it is constant, it causes a DC offset. If it is varying it causes an AC coupling. To see why most audio equipment use a star ground connect, you have to follow the current paths. You'll see stage performers every now and then electrocuting them selves. This is because if the instruments use multiple sources of ground, there can be ground currents that come out the speakers as hum. The often cut the ground leads to get rid of the noise. When some equipment does fail with a ground fault, someone gets hurt.
    In power supplies, the ground lead of the transformer or negative lead of the bridge has just as much current noise as the positive lead. This is why a properly designed supply only has out going lead that trace back as close as possible to the filter capacitor as possible. Although, I've not seen them lately, some high quality audio equipment used to use 4 lead filter capacitors.
    So, when looking at a ground, one has to think about how much noise can be tolerated by the varying currents on the lead. This is why logic boards often have ground plain layers. Logic can tolerate some level of noise so the ground plains are designed to have minimal inductance and resistance to keep the noise below the threshold that is causes issues.
    Think this way: A wire has both an inductance and resistance. A ground plain has some less. A chassis has even less.
    You want to think about the paths that the current will be flowing through the ground. You want to keep it a small as practical.
    As for your case, if the supply you are using has high levels of AC noise, it may need filter capacitors replaced.
    Dwight

  4. #4
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    30,279
    Blog Entries
    20

    Default

    You may be able to improve the situation by installing filtering on the cable going to the Friden. Given that it's a low-speed device, you could construct a low-pass filter network that would remove most of the spikes.

    Is the Friden ground common to the frame and AC ground line? If so, you have a situation of a "ground loop", where the CP/M box is acting as a giant antenna. You may want to install bypass caps between the chassis of the CP/M machine ant the signal lines of the cable.

  5. #5
    Join Date
    Jun 2012
    Location
    UK - Worcester
    Posts
    2,416

    Default

    There are protective earths/grounds (there to stop an electrical fault from killing someone) and a signal ground lines.

    In an audio system (and anything else come to it), the protective earth conductor must NEVER (under any circumstances) be removed from the equipment.

    The signal ground (on the other hand) is always a source of amusement within a professional audio system and much time and trouble is spent trying to get rid of earth loops and the like by connecting and disconnecting the signal ground from various piece of equipment until the problem(s) go away.

    No significant current should be drawn down a signal ground line.

    Floating power supplies can also cause some interesting effects. I had a floating power supply of 10V d.c. that was sat at a potential of 50V d.c.! One side was at 40V and the other side was at 50V (but the differential was still 10V - and that was what I was reading). The high d.c. offset, however, caused problems elsewhere - and we had to take steps to remove it.

    Partly applicable to you issue - and partly ramble...

    Dave

  6. #6
    Join Date
    Dec 2013
    Location
    Near Milwaukee Wisconsin
    Posts
    1,133

    Default

    The Friden power supply is rather odd. There is no primary transformer. The 120 AC is applied directly to a bridge rectifier through a 30 ohm resistor. The frame is grounded to the utility earth ground. The series 30 ohm resistor is on the 120AC neutral. If I measure DCV between the positive/negative to frame I get +-65VDC. Measuring ACV the negative to frame is very low, less than 1 volt and 120 VAC positive to frame. The filter capacitor is just 0.1 uF. The noise that I was seeing is 60 cycle AC about 8 volts peak to peak. I'm pretty sure it is coming from the long cable connection to the Friden. My CP/M machine 8VDC supply is floating. How I resolved the problem was to add a 1.0 uF capacitor between the 8VDC negative to the frame ground of my CP/M computer, which is the same as earth ground. After thinking about this and reviewing some of my notes, I had a similar problem when I first got my Teletype ASR33. I made a 20 ma to RS232 interface for it to work with my IBM XT, CP/M and DEC PDP8E computers. I had forgotten this, but I had to add a capacitor from the negative DC supply to the frame ground to get the teletype to work. Apparently these mechanical devices with all the spinning motors and relay coils can generate some noise. When I have the PDP8 or CP/M machines to my IBM XT or Televideo monitors the bypass capacitors are not needed and do not seem to affect their use. Mike

    Dwight, I understand what you are saying about ground plains. I think my problem is coming from connecting a remote device via a cable to my computer(s). I wonder if there could be some circulating currents between computer and remote device?

  7. #7
    Join Date
    Jun 2012
    Location
    UK - Worcester
    Posts
    2,416

    Default

    >>> I wonder if there could be some circulating currents between computer and remote device?

    That's usually what happens - and is the bug-bear...

    When we had ASR-33's at work - we had them all electrically isolated from the computer. If memory serves me correctly, they were driven through an opto isolator of some description to provide the galvanic isolation.

    Dave

  8. #8
    Join Date
    Feb 2011
    Location
    Salisbury Maryland
    Posts
    768

    Default

    Wow, sounds like the terminal you are using what us old timers call a “Hot Chassis” anytime you bring in the power line to any device and don’t use a power transformer you are by default making the chassis of that device and the incoming power cords neutral the same and that’s always bad news. The neutral and the hot lead of the AC power supply are never tied to the device ground unless the device is isolated from any outside contact. Use to be a common practice in TV and radio to do this and because isolation was built in by having two separate grounds, one isolated and one hot you can do this by having only the isolated ground available to the user but if you did not know this and pulled that TV from the case and grounded the hot side of the set if the neutral was correct it was no big deal but if the AC cord was flipped in the socket or the socket wired incorrectly there was a real risk of electrocution.
    If the AC power line is wired directly to the devices ground that that is a hot chassis device and you are going to have issues no matter what you do. There are isolation transformers that will solve this and what was done to isolate hot chassis things like displays and the like was to use photoisolators or optisolators.
    The problem is anything like a DEC computer has a three wire plug, safety ground, neutral and hot and when trying to adapt anything that is just a two wire set up that’s neutral and hot you will always get cross currents between the safety ground and the neutral. The only place where the safety ground and neutral can be tied together is in the main AC distribution panel, if you have a sub feed panel or disconnect by NEC it has to have separate ground and Neutral.
    A lot of people are going to say all kind of stuff about if your plugs are wired correctly and stuff about this and that but don’t take my word for it, look for your self about the danger of dealing with hot chassis and get an isolation transformer or different I/O device, just be aware if your playing around with tying the neutral to ground and connecting different devices be extremely careful.

  9. #9
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    30,279
    Blog Entries
    20

    Default

    But, as mentioned, the frame of the Friden (Flexowriter?) is grounded to earth. So where is the hazard?

    Mike, is this a 7000 series Flexowriter?

  10. #10
    Join Date
    Dec 2013
    Location
    Near Milwaukee Wisconsin
    Posts
    1,133

    Default

    My Friden is a Flexowriter SFD-V from the late 1950's. I think the -V means it has the edge card reader attachment. Not sure what a 7000 series is. The hazard, I think is if another piece of equipment that is hot chassis-ed, is near or connected to the first, then if the neutral and hot are reversed on one machine, the frames would be at line potential. Contact could provide a shock to man and/or machine. My research has shown me that there are many different models of Friden in the 1950's. The basic writing machine is similar across the board. The various attachments make the difference. And there are different number of bits to the code used. 6 bit machines seem common. My machine is a later 8 bit machine, but uses the Friden code rather then ASCII code. Yet I have had trouble finding a schematic that is correct for my model. I have a SPD drawing from 1962 that is close, but not exact. So far I have a program that will either type a text file or reproduce text from my CP/M console. This code drives the Friden which must be more noise resistant. Lately I've been attempting to input data from the Friden and here is where the noise comes in. There are many magnets which start mechanical functions. I've taken a lot of time to open up the Friden to add snubber diodes across each magnet. This has help quite a bit, but not completely. Then there is contact bounce or contacts that stay open or kinda open. This is turning into quite the project.
    I have a 1kw isolation transformer that I can power the Friden. Doing so did not change the noise. I think my best shot is to try and suppress the noise with capacitors. Thanks for the help, Mike.

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •