Results 1 to 17 of 17

Thread: MOSFET power amplifier repair -biasing issues

  1. #1
    Member knight2001dts's Avatar
    Join Date
    Jan 2008
    Posts
    317
    Thanks
    44
    Thanked 17 Times in 15 Posts
    Rep Power
    146
    Reputation
    137

    Default MOSFET power amplifier repair -biasing issues

    Hi!

    A good mate of mine has a MOSFET power amp he is currently repairing.

    He has re-capped the amplifier modules and is trying to set the biasing (should be measured as 100mv across a biasing resistor when cold and 300Mv when warm according to the MFR)

    Once he has set the bias correctly, it remains stable for a short time and then begins to fluctuate wildly - it seems to drift way off as soon as things start to warm up and then the extra biasing starts to warm things up further etc....

    I have suggested that he try checking or replacing the zenners and small signal diodes and transistors, but is there anything else that anyone here has seen as a common cause of this kind of behaviour? Again, all of the electrolytic caps have been replaced, but there are a number of tantalum/poly caps too....

    The rails remain stable and the power supply is a large transformer not a SMPS.

    The amp in question is an ashly ftx2001 SERIES 3 .... i can get hold of the schematic if you are able to assist or suggest something else to check.

    Thanks for any suggestions or advice!

    CHEERS!



Look Here ->
  • #2
    Premium Member
    hinekadon's Avatar
    Join Date
    Jan 2017
    Location
    still above ground level
    Posts
    598
    Thanks
    1,019
    Thanked 577 Times in 223 Posts
    Rep Power
    247
    Reputation
    10246

    Default

    ok method wrong for checking , first put ampmeter into supply rail ,short input to allow no noise . turn off bias to each fet then indivually lift each one to same level . All mosfets are different and require different bias , as per mfr specs do you know what i mean ? cheers don
    . Please treat people as you would like to be treated yourself.

  • The Following User Says Thank You to hinekadon For This Useful Post:

    knight2001dts (08-12-18)

  • #3
    Senior Member
    Reschs's Avatar
    Join Date
    Aug 2010
    Location
    Outside a few schooners
    Posts
    2,464
    Thanks
    487
    Thanked 1,351 Times in 692 Posts
    Rep Power
    530
    Reputation
    17318

    Default

    Check Q17 the VBE Multiplier. 2N4125. Make shure it is in good thermal contact with the heatsink.

  • The Following User Says Thank You to Reschs For This Useful Post:

    knight2001dts (08-12-18)

  • #4
    Senior Member
    nomeat's Avatar
    Join Date
    Jan 2008
    Location
    Commonly found in a pantry or the bottom of a fridge, searching for grains, fermented or distilled
    Posts
    4,193
    Thanks
    1,157
    Thanked 2,086 Times in 1,267 Posts
    Rep Power
    976
    Reputation
    35218

    Default

    Quote Originally Posted by knight2001dts View Post
    but there are a number of tantalum/poly caps too....


    CHEERS!
    That deadly word in red could be your problem.
    Ditch them all and replace with ceramic or poly if possible. You can get small ceramic up to 10F, just watch the voltage rating.
    Last edited by nomeat; 05-12-18 at 01:02 PM.
    $$$...Welcome to Australia where corruption is legalized...$$$

    The only group that does not have a lobby is the middle income taxpayer.

  • The Following 2 Users Say Thank You to nomeat For This Useful Post:

    hinekadon (05-12-18),knight2001dts (08-12-18)

  • #5
    Senior Member
    Reschs's Avatar
    Join Date
    Aug 2010
    Location
    Outside a few schooners
    Posts
    2,464
    Thanks
    487
    Thanked 1,351 Times in 692 Posts
    Rep Power
    530
    Reputation
    17318

    Default

    Quote (should be measured as 100mv across a biasing resistor when cold and 300Mv when warm according to the MFR)

    1.5mv cold and 3.0mv warm according to circuit. The values you quoted would easily explain your problem.
    What was the original fault ?

  • The Following User Says Thank You to Reschs For This Useful Post:

    knight2001dts (08-12-18)

  • #6
    Premium Member
    hinekadon's Avatar
    Join Date
    Jan 2017
    Location
    still above ground level
    Posts
    598
    Thanks
    1,019
    Thanked 577 Times in 223 Posts
    Rep Power
    247
    Reputation
    10246

    Default

    oops sorry i misunderstood thought he had replaced mosfets reschs seems to be on the case but as with all mosfets the bias is set by current not by voltage but to read current they take a shortcut and give you a voltage across a resistor and as a result the bias changes with the supply to the source .In your case i would short the input out and see what gets hot over time shorting the input is critical to getting correct levels . Nomeat is correct those sneaky tants can sneak up on you so replace them definitely poly is a good choice if its running to any extent it is unlikely to be a semiconductor / but dry temp paste can cause problems with thermal runaway as resches suggested regards don
    . Please treat people as you would like to be treated yourself.

  • The Following User Says Thank You to hinekadon For This Useful Post:

    knight2001dts (08-12-18)

  • #7
    Member knight2001dts's Avatar
    Join Date
    Jan 2008
    Posts
    317
    Thanks
    44
    Thanked 17 Times in 15 Posts
    Rep Power
    146
    Reputation
    137

    Default

    Hi!

    Thanks to everyone for your input and Ill start by correcting my figures (was going from memory) - 1.5mV and 3.0mV are the correct biasing voltages taken across a resistor at the biasing pins which represent 100mA and 300mA respectively.....

    I was wondering about those pesky tantalum caps myself as have had issues with a few in the past - will get him to change those out, and an awesome tip re:Q17

    Will investigate further!

    He has a good amp as well, so Im able to swap parts out to try theory pretty easily....hes just lazy!!!!!

    Hinekadon- I would love some more info on your method of biasing please?

    To turn off each mosfet would be done by desoldering the gate pins from all of the power mosfets and leaving only one soldered onto the pcb? Or did I get that wrong?

    This is a common gate amp and I assume that the mosfets are matched so that they all use the same bias current?

    Learned this stuff 25yrs ago and quite rusty on my theory - please be kind lol


  • #8
    Member knight2001dts's Avatar
    Join Date
    Jan 2008
    Posts
    317
    Thanks
    44
    Thanked 17 Times in 15 Posts
    Rep Power
    146
    Reputation
    137

    Default

    Will also get him to try shorting the input and monitor - great suggestion!!

  • #9
    Senior Member
    Reschs's Avatar
    Join Date
    Aug 2010
    Location
    Outside a few schooners
    Posts
    2,464
    Thanks
    487
    Thanked 1,351 Times in 692 Posts
    Rep Power
    530
    Reputation
    17318

    Default

    Looking at the circuit, I do not see anywhere a Tantalum would be used in the main part of the amplifier.
    There are two types.
    1 Bead. Yes a problem. Because no one reads the specs on how they should be used.
    2 Solid. No problems.

  • The Following User Says Thank You to Reschs For This Useful Post:

    knight2001dts (08-12-18)

  • #10
    Member knight2001dts's Avatar
    Join Date
    Jan 2008
    Posts
    317
    Thanks
    44
    Thanked 17 Times in 15 Posts
    Rep Power
    146
    Reputation
    137

    Default

    Thanks Reschs!

    Looks like I got some bad intel there re:tantalum caps...

    Hopefully one less thing it might be.

    Im going to try and grab the amp from him so I can have a look and not have to do the Chinese whispers thing :P

    Cheers!!

  • #11
    Senior Member
    Reschs's Avatar
    Join Date
    Aug 2010
    Location
    Outside a few schooners
    Posts
    2,464
    Thanks
    487
    Thanked 1,351 Times in 692 Posts
    Rep Power
    530
    Reputation
    17318

    Default

    The 10N16 and 10P16 FETS behave the same as normal transistors with a positive temperature coefficient and are even more subject to thermal runaway.
    The Original Hitachi Mosfets, used in amplifiers, had a negative temperature coefficient and were self controlling.
    Correct biasing is crucial. The bias needs to be adjusted from zero after no less than 10 mins and final after no less than 30 mins.
    Not saying you do not have another issue.

  • The Following 2 Users Say Thank You to Reschs For This Useful Post:

    knight2001dts (08-12-18),tristen (07-12-18)

  • #12
    Premium Member
    hinekadon's Avatar
    Join Date
    Jan 2017
    Location
    still above ground level
    Posts
    598
    Thanks
    1,019
    Thanked 577 Times in 223 Posts
    Rep Power
    247
    Reputation
    10246

    Default yep

    thats basically correct , each added fet is the same added bias to the total current draw short the input means no signal so no amp current only the quiescent current of the operating circuit when you start with no fets connected thats the quiescent current join one fet add 100ma , two fets 200ma total added.and so on if you come across massive jump in current thats where the problem is . is this clear enough for you , cos im ratshit at explaining my methods regards don
    . Please treat people as you would like to be treated yourself.

  • The Following User Says Thank You to hinekadon For This Useful Post:

    knight2001dts (08-12-18)

  • #13
    Senior Member
    Reschs's Avatar
    Join Date
    Aug 2010
    Location
    Outside a few schooners
    Posts
    2,464
    Thanks
    487
    Thanked 1,351 Times in 692 Posts
    Rep Power
    530
    Reputation
    17318

    Default

    Quote Originally Posted by hinekadon View Post
    thats basically correct , each added fet is the same added bias to the total current draw short the input means no signal so no amp current only the quiescent current of the operating circuit when you start with no fets connected thats the quiescent current join one fet add 100ma , two fets 200ma total added.and so on if you come across massive jump in current thats where the problem is . is this clear enough for you , cos im ratshit at explaining my methods regards don
    100ma per pair of FETS is for the old HITACHI FETS only. It was necessary to achieve the negative temperature coefficient.
    The FETS here only require 10 to 20ma per pair.

  • The Following 2 Users Say Thank You to Reschs For This Useful Post:

    knight2001dts (08-12-18),tristen (07-12-18)

  • #14
    Member knight2001dts's Avatar
    Join Date
    Jan 2008
    Posts
    317
    Thanks
    44
    Thanked 17 Times in 15 Posts
    Rep Power
    146
    Reputation
    137

    Default

    Hi and thanks again for the info....

    So in an amp such as the earlier FET2000 series Ashlys that uses 4 pairs of the Hitachi devices (2SJ50 and 2SK135), the bias current would be approx 400mA?

    If the FTX series amps have 4 pairs of the 10n16 and 10p16s - how come they need 150-300mA of bias current if they only need 10-20mA per pair? (That works out at only 40-80mA)

    Cheers!

  • #15
    Premium Member
    TVguy's Avatar
    Join Date
    May 2009
    Location
    Peripatetic
    Posts
    953
    Thanks
    369
    Thanked 786 Times in 393 Posts
    Rep Power
    362
    Reputation
    10728

    Default

    Fault finding on any DC stabilised amp can be a hideously tedious process. The fact that the issue is thermally induced can only point to an active device rather than any passives.
    Don't rule out the input stage as it does determine the overall output and offset bias. The image below may help to lift the curtain (so to speak) on circuit topology.
    Unless you want to try the hairdryer and freezer technique, this is a check by replacement task.

    "Any sufficiently advanced technology is indistinguishable from magic."

    - Arthur C. Clarke

  • The Following 4 Users Say Thank You to TVguy For This Useful Post:

    hinekadon (10-12-18),LeroyPatrol (09-12-18),nomeat (09-12-18),tristen (10-12-18)

  • #16
    Senior Member
    nomeat's Avatar
    Join Date
    Jan 2008
    Location
    Commonly found in a pantry or the bottom of a fridge, searching for grains, fermented or distilled
    Posts
    4,193
    Thanks
    1,157
    Thanked 2,086 Times in 1,267 Posts
    Rep Power
    976
    Reputation
    35218

    Default

    Quote Originally Posted by TVguy View Post
    Unless you want to try the hairdryer and freezer technique, this is a check by replacement task.
    LOL, since I discovered the use of freeze spray for car Aircons and the cost of the bloody stuff these days I have abandoned that practice.
    Today I just reduce my soldering tip to a bit less than 200˚C and place it on the individual components.
    That way I know exactly which one does what thermally, especially when everything is so closely spaced.

    Looking at that circuit the zeners D22 D23 and D25 D26 might be the first components I would place my tip on.

    DAYOR warning... yes under power, so skilled and not shaky(shorting) placing please.
    Last edited by nomeat; 09-12-18 at 10:47 PM.
    $$$...Welcome to Australia where corruption is legalized...$$$

    The only group that does not have a lobby is the middle income taxpayer.

  • The Following 3 Users Say Thank You to nomeat For This Useful Post:

    hinekadon (10-12-18),tristen (10-12-18),TVguy (10-12-18)

  • #17
    Senior Member
    Reschs's Avatar
    Join Date
    Aug 2010
    Location
    Outside a few schooners
    Posts
    2,464
    Thanks
    487
    Thanked 1,351 Times in 692 Posts
    Rep Power
    530
    Reputation
    17318

    Default

    Quote Originally Posted by knight2001dts View Post
    Hi and thanks again for the info....

    So in an amp such as the earlier FET2000 series Ashlys that uses 4 pairs of the Hitachi devices (2SJ50 and 2SK135), the bias current would be approx 400mA?

    If the FTX series amps have 4 pairs of the 10n16 and 10p16s - how come they need 150-300mA of bias current if they only need 10-20mA per pair? (That works out at only 40-80mA)

    Cheers!
    The 100mA per pair for the Hitachi's was to put them in negative temperature co-efficient.
    The newer FET's do not have this feature and behave exactly like transistors and are prone to thermal runaway.
    A lot of designers were not up on this and created problems for themselves.
    In this case the designer has taken this on board and provided Q17 as a thermal sensing VBE Multiplier to compensate. Hence the importance of good thermal contact with the heatsink.
    The choice of such a high quiescent current could either be a mistake or a serious overkill to achieve low distortion.

  • The Following User Says Thank You to Reschs For This Useful Post:

    hinekadon (10-12-18)

  • 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
    •