Please help me bias this FET bridge

Discussion in 'The Projects Forum' started by Darkstar, Sep 5, 2011.

  1. Darkstar

    Thread Starter Senior Member

    Sep 3, 2010
    117
    1
    Thanks for the input Oldtech. I will put a 10 pF cap in parallel with Rc (later called Rs) and see what happens.
    I guess that's what you mean. I'm wondering if you looked at an older version of the circuit than I'm now using. I currently have a 0.1 uF cap between the power supply side of Rc and ground. Niether it, nor any other decoupling caps I've added, had any noticable effect.
     
    Last edited: Sep 24, 2011
  2. oldtech33709

    New Member

    Sep 24, 2011
    26
    0
    The bypass should short any transients to ground and eliminate high freq noise in the bias. The 0.1uf is good for mid freq noise.
     
    Last edited: Sep 24, 2011
  3. Darkstar

    Thread Starter Senior Member

    Sep 3, 2010
    117
    1
    Ok, I see. I had heard that 0.1 uF is sufficient and in most cases adding multiple caps of different values could cause unforseen problems. Also, I thought the 0.1 uF would work from some given frequency on up, not just for a certain frequency "range". I will make the correct changes for my next test.
    Thanks.
     
  4. Darkstar

    Thread Starter Senior Member

    Sep 3, 2010
    117
    1
    Here is the latest schematic and the graph I get from it. The colored spots show where changes have been made recently.
     
  5. Hi-Z

    Active Member

    Jul 31, 2011
    158
    17
    Firstly, why have you added the 1M resistors between chassis and the cable screens? This means that the screens will be floating to ac, so they won't be doing a very good job of screening.

    I think we ought to go back to basics before we proceed; we need to know a few things, and the best way would be to have a more direct way of monitoring antenna voltage. We could try the following:

    Disconnect the diodes on the inputs. Connect 0V directly to house ground.

    Take the jfets' drains directly to +15V (i.e. shorting across Rd and the zero pot). Connect the jfets' sources to the screened cable via 200 ohm (approx) resistors (this hopefully will prevent any chance of oscillation).

    At the indoor end, connect two 5kohm (approx) resistors to the other ends of the screened cable (so you now have a total of 5.2k in each of the fets' source circuits). Connect the other ends of the 5k resistors to -15V. (The 1.5k "tail" resistor is therefore deleted.) We now have two separate source-followers, as opposed to a differential pair.

    You can now get a good idea of the antenna dc and ac voltage by monitoring at the two 5k resistors. It would be very helpful to monitor the differential dc and ac volts over a few hours (it's a pity you can't measure both dc and ac simultaneously).

    Also, it would be very nice to know what external ground is doing (dc and ac) relative to 0V over a period: the easiest way would be to disconnect the ground jfet's drain, short its source and gate, and disconnect the associated 5k resistor from -15V. Then you are left with a 200 ohm connection to external ground.

    I hope the above makes sense!
     
  6. Darkstar

    Thread Starter Senior Member

    Sep 3, 2010
    117
    1
    I thought the 1M resistors would interfere with the oscillations. I saw a diagram which showed shield grounding for a floating sensor on the end of the cable and it had a 10M resistor between the shield and ground. I'm not sure how close this is to my arrangement, but I thought I'd try. See diagram below.

    Disconnecting the diodes is no problem. 0V is already connected to house ground.

    Connect drains to +15V is easy. Connecting sources to shields with a 200 ohm resistor at the fet will be a little tricky and require stripping some of the cable in place while crowded next to 3 other cables. If it doesn't go easily, I'll break my cable connections to the little PCB that holds the fets.

    The 5K resistors indoors are easy.

    I have 2 DVMs so I can monitor AC and DC simultaneously, but only 1 connects to the computer for continuous monitoring. The other can take simultaneous, but temporary, readings.

    There's a couple things I don't understand about the last part. 1- what's the purpose of shorting the ground fet's source and gate? 2- If you want to know what's happening between 0V and external (=house=earth ground), why not just disconnect the 0V to house gnd I made earlier and measure the DC and AC difference between the two?

    I can't say just when this will get done. Weather is threatening rain.
     
  7. Darkstar

    Thread Starter Senior Member

    Sep 3, 2010
    117
    1
    I just want to be clear about something. You refer to "external ground" as if it is different from other grounds. External ground is the house conduit ground. The ground fet gate connects to it via an outdoor electrical outlet as opposed to an indoor outlet like everything else. It is not even distant from the indoor connection, its outlet on the outside of a wall and the indoor house ground is at an outlet on the inside of the wall directly opposite to the outdoor outlet. This is the same ground all the chassies and power supply 0V ref line are connected to.
     
  8. Hi-Z

    Active Member

    Jul 31, 2011
    158
    17
    Ah, so we can expect all grounds to be at the same potential - now I see. I'd been thinking there might be a small/variable difference; well, that only makes life easier!

    As for the 200 ohm resistors, if it's awkward, you could try it without them (but unfortunately we won't know for definite if it's oscillating or not).

    Don't worry, you can omit the bit about measuring external ground relative to 0V, since 0V should be connected to house ground anyway, and that's the same as external ground.

    It would (of course) have been nice to log both dcv and acv at the same time, because then we'd have an idea about how much the interfering 60Hz was having on the dc level (perhaps both with and without your capacitive filter), but you may be able to get an idea anyway, since you have two dvms.

    Do let us know how you get on.
     
  9. Hi-Z

    Active Member

    Jul 31, 2011
    158
    17
    I was forgetting myself: the source-followers won't take kindly to a large C on their outputs, so don't try this.

    Instead, if you want to look at a filtered version of the output, just connect them to your in-amp's input filter via a couple of 2.2Mohm (approx) resistors, and set your in-amp to a gain of 1.
     
  10. Darkstar

    Thread Starter Senior Member

    Sep 3, 2010
    117
    1
    Ok, I rewired the circuit and got some measurements.
    The voltage drop on the 4.7K resistor for the antenna fet were stable.

    The voltage drops for the ground fet were not steady. I got graphs with and without using the decoupling caps. I measured oscillation of 45 KHz to 65 KHz without them and only 60 Hz with them. I measured about 2.22VAC @ 60Hz.

    The graphs are attached. Here are some measurements taken on the circuit:

    Monitoring Antenna FET with Decoupling Caps:


    Vcc = +15.07 V
    Vee = – 15.07 V
    Rs1 = 4680 Ω (-15V to shield, antenna side)
    Rs2 = 4680 Ω (-15V to shield, ground side)
    VdRs1 = 17.04 V across 4.7K for ant FET
    VdRs2 = 12.20 V across 4.7K for gnd FET
    Rstotal1 = 4680 Ω (-15V to ant shield)+ 217.2 Ω (ant shield to source) = 4897.2 Ω
    Rstotal2 = 4680 Ω (-15V to gnd shield)+ 217.3 Ω (gnd shield to source) = 4897.3 Ω
    Is1 = 17.04 V / 4680 Ω = 3.64 mA calc
    Is2 = 12.20 V / 4680 Ω = 2.61 mA calc

    I hope this is clear.
     
  11. Hi-Z

    Active Member

    Jul 31, 2011
    158
    17
    Hi again,

    Can you include a circuit diagram for what you've now got? I think we need to be clear about what's being measured here. Also, what (where) are the decoupling capacitors you refer to?

    Just to recap, what we should have are two fets, each with their drains connected to +15V, their sources connected to individual 200 ohm resistors, and the other sides of these resistors connected via the screened cables to 5k resistors. The other side of the 5k resistors should connect to -15V. The screens of the cables should be connected to 0V, and 0V should be connected to house ground. You should be measuring voltages at the junctions of the 200 ohm and 5k resistors.
     
  12. Darkstar

    Thread Starter Senior Member

    Sep 3, 2010
    117
    1
    Here's the current schematic. When you mentioned monitoring the 5K resistors, I thought you meant to monitor the voltage drop across them, which is what I did.

    The decoupling caps are the same ones that have been in the circuit.

    The 0V line from the power supply is connected to house ground now within the power supply. I did not understand that you wanted 0V connected to the shields. Where would I connect 0V to the shield, in between the 200 and 5K resistors?
    I think I see where I misinterpreted. Where you said to connect the 200 and 5K resistors to the "screened cable", I thought you meant to connect them to the "screen" and use it to deliver current to the sources. With the original conductor still connected, I had 30V across the FETs. I figured I fried them. So, I disconnected the original conductor. It was all I could see to do with the way I had wired the screens.

    It won't be a problem to correct this.
     
  13. Hi-Z

    Active Member

    Jul 31, 2011
    158
    17
    Yes, the 200 ohm resistors should be connected to the 5k's via the inners, with the screens connected to 0V. This fet configuration shouldn't be loaded with any capacitors at all, so don't connect to the filter ("decoupling") capacitors. Ultimately, we can make use of the filter when we engage the in-amp - but only via 2M2 resistors.

    If the above results in apparent "oscillation" then we'll tackle that as it arises - but it may not happen.

    I really should get my act together in terms of posting images - it would save a lot of misunderstanding!
     
  14. Darkstar

    Thread Starter Senior Member

    Sep 3, 2010
    117
    1
    Ground FET:

    I measure 2.37 VAC output at 60 Hz.
    When measuring frequency, I only get 0.0, so if there is oscillation, it must be out of range of the DVM.

    Current through ground FET (4.7K resistor) = 22.8V/4680 Ω = 4.87 mA, 111.08 mW

    There are small pos and neg jumps on the graph (larger than the bkg) which were due to me turning TV and room light on or off.

    See monitoring ground fet graph.

    Antenna FET:

    DC output is a steady 16.30 V from the start.

    AC V is only 0.3 mV. No freq can be measured.

    Measuring specifically for frequency gives no reading (0 Hz).


    Current through the antenna FET (4.7K resistor) = 16.3V/4680 Ω = 3.48 mV, 56.77 mW

    This is not affected by anything being turned on or off in the room.

    See antenna fet monitoring graph.

    Graph "working" is the schematic.
     
  15. Hi-Z

    Active Member

    Jul 31, 2011
    158
    17
    Well, the antenna fet seems to be behaving itself. The ground fet is quite alarming, really, because it seems to have a fairly large dc offset and also a significant ac reading.

    A quick check would be to short the antenna fets's gate to the ground fet's gate, and just check that it results in similar dc and near-identical ac readings. Even better would be to measure the external ground directly with the dvm (relative to 0V/house-ground) - though this would mean disconnection of the ground fet. Thinking on it, I reckon this is going to be an essential test to do, because it will indicate actual dc offset, plus whether or not the ac is caused by the fet (i.e. oscillating), or is actually present on the external ground.

    As for the frequency reading of 0.0, I don't know what we can deduce from this.
     
  16. Darkstar

    Thread Starter Senior Member

    Sep 3, 2010
    117
    1
    Ok, I have a lot of data here.

    The 0Hz frequency measurement mentioned above may mean there is nothing being detected since the default display with no input is 000.0 Hz. Or, it may mean that there is a frequency there, but it is beyond the frequency/sensitivity range of the meter. See attachment.

    This first test completely bypasses the outside connection to ground and makes a more direct connection of the ground FET gate to power supply 0V. See schematic 1. The detector chassis is also connected to 0V, and the power supply 0V is connected to house ground inside the power supply chassis. See schematic 2.


    -----------------


    I Checked antenna FET output first and it is perfectly stable at 16.4V.

    The connection from 0V inside on the power supply and the inside end of the shielded cable connected to the ground FET gate outside is just plain, unshielded wire. See schematics 1 and 2.

    The detector chassis has always had its own house ground. For these tests, the detector chassis had 2 grounds: First, the connection to the house ground. Second, direct connection to the power supply 0V line.


    Measuring AC V on the ground FET output, I get 8 to 9 mA. Freq = 0-24 Hz.
    Measuring frequency specifically, I get 0 Hz.
    There was only a brief, small 30mV blip when the room lights were turned on and off. (A little later there was another blip in the output, TV turned on and off had no effect.
    Measuring DC V, it was perfectly stable at 16.4V.


    -------------

    Monitoring Ground FET(detector chassis house gnd connection has been removed) (Dark green earth on schematic 1.)


    When monitoring ground FET DC V output, it began about 16.5V and gradually climbed like yesterday.
    When monitoring AC V, I got about 2.3 V and 60Hz.
    When monitoring frequency, I got 0 Hz.
    When DVM is disconnected for a few seconds to switch from DC V to Frequency and back, the DC V output again begins at about 15 and rises fast at first then much more slowly. Current to the FETs or from the power supply has not been interrupted. If the long curve is from warming of the FET during use, it should stay warm and the graph should not start at 15 again.

    --------------

    I did a test where I disconnected 0V from house ground in the power supply. Then I monitored the ground FET line using the house ground and 0V supply virtual ground separately (I know this isn't correct but I wanted to see the difference.) See schematic 3.


    First I monitored ground FET DC V output with the detector chassis and ground FET gate connected only to the power supply 0V. Nothing is connected to house ground but the power supply chassis. The power supply circuitry is not connected to the chassis.



    For ground FET line DCV, I got a nearly identical curve as before with 0V connected to house ground.
    For ACV, I got 2.13V @ 60 Hz
    For frequency measurement alone, I got 0Hz.

    ---------

    Next, I remembered I have a 12VDC cooling fan powered across the rectifier bridge and having some EMI filter caps going between bridge pos and neg to the chassis at house ground. I disconnected the caps from the chassis. See schematic 3.



    The DCV for ground FET line started out briefly at around 12V and rose quickly 17V, then slowly rose to 22.5V within 10 minutes.
    For ACV I got about 1.06 V @ 60 Hz
    For freq I got 0 Hz

    -----------

    Next, I disconnected the detector chassis and ground FET gate from 0V and connected them to house ground. Fan filter caps still disconnected.
    0V line is not used anywhere now. See schematic 4.



    DCV for ground FET line began at 19V and soon began climbing to 23V in 20 minutes.
    ACV was 2.34V @ 60 Hz.
    Freq measurement was 0 Hz.
    When I switched back to DCV, the output again started about 18V and began climbing again.

    ---------------

    To compare the external house ground with the 0V from the power supply, I disconnected the permanent connection between the two made recently within the power supply and then monitored the floating 0V virtual ground to the house ground with the DVM. Power supply is not plugged into detector so readings are not influenced by operation of detector. See schematic 5.



    For DCV, house gnd is neg, power supply 0V line is pos. Graph shows from about 20 to 47 mV difference, wildly varying.
    ACV measures 1.75V @ 60 Hz.
    Frequency measurement varies wildly between 12 and 24 KHz. Turning fluorescent room light or nearby TV on and off makes no difference.


    I'm sure this is confusing, take your time. It's late and this is as clear as I can get at this time so if you have questions, please ask.
     
    Last edited: Oct 2, 2011
  17. Darkstar

    Thread Starter Senior Member

    Sep 3, 2010
    117
    1
    New news.
    I was sitting here looking at the graph of DC voltage difference between the power supply house-grounded chassis and the floating 0V reference, wondering if the polarity would change. As I watched, the graph jumped from -20mV to +55mV then the noisy graph stayed in the +70mV range. Here is a blow up of the transition region.
     
  18. Hi-Z

    Active Member

    Jul 31, 2011
    158
    17
    Sorry to be taking so long to get back to you, but, you're right, I am a bit confused! A lot of the measuements seem to be with a floating 0V, which just won't work with this circuitry.

    A couple of posts ago, it seemed as though you had a lot of ac and dc volts on external ground relative to house-ground (to which 0V was tied). I think we need to investigate this as first priority. I have to say, I found it quite surprising that the antenna seemed to be totally quiet relative to house-ground, whereas external ground was so "busy".

    Measuring external ground relative to house ground ought to be fairly easy to organise - no fets, just a dvm.

    Let me know how you get on!
     
  19. Darkstar

    Thread Starter Senior Member

    Sep 3, 2010
    117
    1
    I've measured this in the past, just to be sure kind of thing, but results were as expected.....0.0V. Since external ground and house ground are the same conduit there is no difference. See diagram.

    I know the floating ground of my last measurement is not the right thing to do, but I thought it might be showing something going on with the power supply, otherwise why is the 0V ref line changing. I don't know what to make of the reversals in polarity. I believe the last measurement taken was the only one with a floating 0V ref line. All other circuit modifications were grounded inside the power supply or out.

    I tried to cover a lot of ground with all the measurements, too much I guess.
     
    Last edited: Oct 2, 2011
  20. Hi-Z

    Active Member

    Jul 31, 2011
    158
    17
    Well, that's interesting. In post #114 you measured over 22V (corresponding to +7V relative to 0V) dc and over 2V ac at the output of the ground fet, while the antenna fet output was sitting quietly near ground. If this was truly the case, then the ground fet must have been oscillating violently (that's the only conclusion I can draw); however, it would make complete sense if the results for the two fets were the other way round.

    Can you confirm that the ground fet really was the ground fet for this test? Possibly you could repeat the measurements (which have 0V tied to house-ground)?
     
Loading...