Announcement

Collapse
No announcement yet.

Help with vertical mosfet power amp improvements

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Help with vertical mosfet power amp improvements

    Hi
    I have this power amp that is on a board complete with power supply. Looks Italian.
    It works, but I have substantial ''crossover sounding'' distortion if a play a 200 to 450 HZ sine wave into it.
    Bias is set to 40mA monitored with amp meter in the positive rail.
    On the scope I can just see the distortion. The same amount I would leave for any other amp, but for some reason on this amp I can still hear it in the previous mentioned freq range.
    If I then increase bias until I hear no more distortion, amp meter shows quiescent current of 150mA. It heats up a bit more in idle.
    I then let the unit cool down, and when switched back on, bang, the Mosfets are taken out along with the mains fuse. Obviously it doesn't like the higher bias.
    I have attached a schematic, and would like to improve on the design.
    Any ideas would be great.
    Also note that on the schematic I have written that the biasing transistor is not mounted on the heatsink.
    Attached Files

  • #2
    I think that you should provide some more info:
    how do you know that it looks Italian?
    does it have any name?
    is there a manufacture year, or at least date on capacitors?
    were the output transistors replaced previously?
    how do you know that it was designed for vertical MOSFETs? Maybe it was designed for lateral MOSFETs and someone replaced the transistors.

    For me the higher bias looks much better that the lower one. I wonder why the ballast resistors are missing. This, and the fact that the bias transistor is not on the heatsink somehow suggests that the amp was designed for lateral MOSFETs. Answers to the questions above (or even a photo of the power amp board) could greatly help. For starter, I would add ballast resistors. Please note that if you are measuring power supply current, you don't know whether quiescent current of one transistor is 5 mA and the other 145 mA, or they are both 75 mA. The correct way of measuring this is with ballast resistors.

    Mark

    Comment


    • #3
      Mark.
      OK to come clean. It is from a powered speaker, still being manufactured from a reputable "name" company in Italy. I have been warned by the local agents here in South Africa not to disclose any schematics with the mention of their name or logo. That is why I hand drew the schematic.
      I have the original schematic, and I can assure you all the components are correct.

      Comment


      • #4
        Well, I know why diydidi thinks it's Italian

        Not long ago somebody (was it him?) mentioned repairing Montarbo amplifiers; powered speakers to be more precise.

        And yes, they do use Vertical Mosfets, do not mount the sensing transistor in the heatsink but on the PCB, a few inches away (completely defeating its purpose) and despise ballast resistors.

        Yet those amps and cabinets sell well and usually last years, only explanation is that they do crossover-distort ... there's no other way.

        My Montarbo shematics are in another computer, I'll post one when I get there.

        2 thoughts come to mind:

        1) if 200/450 Hz distortion is barely seen but quite annoying, maybe there's a cabinet crossover or speaker component problem which increases its audibility; if cabinet response is very poor (cabinets have 10dB or worse peaks and notches all over the place) in that range, harmonics become much more prominent to the ear.
        I have already lost time chasing such ghosts.

        2) not so much for distortion but for safety, epoxy that transistor (with suitably heatshrinked wires) to the heatsink, close to one or in the middle of power transistors, add some kind of bracket or clamp over it so it's surrounded by hot metal (even if iron).
        Add ballast resistors; .33 or .47 ohms suggested.

        Check that the Zobel network is fine.

        Ok, I found the schematic for a Montarbo 8"+driver 120W RMS powered speaker.

        Notice the lack of ballast resistors and the MosFets mounted away from the main board (which carries the bias transistor), joined by flat ribbon cable.

        Click image for larger version

Name:	Montarbo01.jpg
Views:	1
Size:	138.4 KB
ID:	837073

        Click image for larger version

Name:	Montarbo02.jpg
Views:	1
Size:	289.5 KB
ID:	837074

        As of
        how do you know that it was designed for vertical MOSFETs?
        the biasing string can be adjusted between 5 and 10 "diode drops" which means Verticals.
        Of course, fully agree with the need of ballast resistors.

        FWIW this is some Montarbo produce, which sells very well in Europe, and which surprisingly sounds GOOD (maybe using lots of RCF speaker components is a reason):


        More proof than investing in better speakers is often more useful than endlessly tweaking minor electronics details.
        Last edited by J M Fahey; 03-10-2015, 04:25 PM.
        Juan Manuel Fahey

        Comment


        • #5
          Compare the schematic to Ampeg SVT3Pro (which does not have such problems).
          As Juan mentioned, add 0.47 Ohm ballast resistors, connect thermally the bias transistor to the heatsink, set the quiescent current to the lower value. Measure the voltage drop on ballast resistors and see whether they are balanced. Heat up the transistors with hear-dryer and see how the currents change. Make sure that the diodes on MOSFET gates are OK. Switch the amp off and on to see whether it fails.

          It seems to me that the most important issue (and their mistake) is luck of thermal contact between the bias transistor and the output MOSFETs.
          It looks like they were trying to save some money.

          Mark

          Comment


          • #6
            JM I just had a laugh. Very sneaky indeed.
            Ok I can see that adding ballast resistors and Thermally coupling the bias transistor to the heatsink might help with reliability.
            But what about the distortion I'm hearing??
            I have tried running the amp through other speaker cabinets, and its still there.
            Just out of interest, would lateral fets need a different biasing arrangement and voltage than verticals?
            Also if I use the 0,47 ballasts, will I need to bias for 25mV as suggested in the SVT3/4 pro schematics? Thats about 53mA per FET.

            Comment


            • #7
              Originally posted by diydidi View Post
              Also if I use the 0,47 ballasts, will I need to bias for 25mV as suggested in the SVT3/4 pro schematics? Thats about 53mA per FET.
              Originally posted by MarkusBass View Post
              set the quiescent current to the lower value.
              This is what you are going to find out. I suggested to start with the lower value of the quiescent current - 25 mA per transistor and then increase it if you need it (and the amp won't fail).

              Mark

              Comment


              • #8
                Originally posted by diydidi View Post
                JM I just had a laugh. Very sneaky indeed.
                Ok I can see that adding ballast resistors and Thermally coupling the bias transistor to the heatsink might help with reliability.
                But what about the distortion I'm hearing??
                I have tried running the amp through other speaker cabinets, and its still there.
                Just out of interest, would lateral fets need a different biasing arrangement and voltage than verticals?
                Also if I use the 0,47 ballasts, will I need to bias for 25mV as suggested in the SVT3/4 pro schematics? Thats about 53mA per FET.
                Ok, I have no signed agreements with anybody, and feel free to comment on the Italian Football Team (a.k.a. "I Azzurri") match strategy, so here's their formation for the next 2 matches against Argentina (which of course we'll win 5:0 just because they do not thermally compensate nor use buffers to protect the goalie):

                Just in case, incomplete and unlabelled

                Click image for larger version

Name:	Azzurri01.jpg
Views:	1
Size:	102.2 KB
ID:	837076

                Click image for larger version

Name:	Azzurri02.jpg
Views:	1
Size:	46.0 KB
ID:	837077

                Almost forgot: thanks Elektrotanya

                Sorry: almost forgot: if you can't lower distortion, first scope amp with a real speaker load, maybe an inductive load will show some unstability which resistive ones do not.

                Then scratch and cut tracks close to power gates and add individual 1k resistors in series with each, I mentioned multi MHz instabilities, so high that my scope screen would display them as solid bars, not resolving separate waves.

                And anyway you don't want to *see* them, you want them OFF !!!!

                Absolute worst case, accept that distortion.

                100% of customers will complain and insist you re-repair a dead amp for free, 1% will make an issue with residual distortion which can be heard only at bedroom levels,so choosing between dangerous silky smooth and slightly harsh safe and stable is a no brainer in my book.

                You are supposed to bring back to life a dead amp, not reengineer design sins comitted thousands miles away and for which you are not responsible.
                Juan Manuel Fahey

                Comment


                • #9
                  OK, so I have taken advice from all of you. Thank you all very much!!
                  I'm in the proses of adding 0,47ohm Source resistors, and want to attach the two Vbe multipliers to the heatsink. I say two, because this is a bi-amp system. My question is, where do I mount them? I was thinking of using TO-126 packages for ease of mounting.
                  Do they have to be mounted close to their relevant FET's? ie. the Vbe multiplier for the tweeter, close to the tweeter Fets, and the other close to the woofer Fets? ;or doesn't it matter?
                  Also, can i mount them on top of one of the output Fets using the same mounting bolt?
                  I will however need to link these transistors to the preamp board with 5cm wires. Anything I need to lookout for when doing this?
                  I have included a photo of the output/heatsink board.
                  Attached Files

                  Comment


                  • #10
                    The unit now works great!!!
                    No distortion, and I can see Vbe multiplier is working great. I mounted them directly on top of the output devices.
                    I have one more question. The Fets for the tweeter aren't paralleled, as it only uses one irfp9240, and one irfp240 for the tweeter amp. Do they also need ballast resistors?

                    Comment


                    • #11
                      Yes, because ballast resistors are used not only for current sharing but mainly for thermal stability.

                      5cm long wires are no problem, just heatshrink each leg properly and apply some hot glue , silicone sealant, or contact cement to avoid vibration.

                      And ... congratulations.

                      If possible post a couple more pictures, I'm interested in the heatsink, its fins, and how is it mounted on the backplate.

                      Also how the backplate is mounted and the cabinet cutout for it.

                      Is the backplate flat or has a lip all around?
                      Juan Manuel Fahey

                      Comment


                      • #12
                        I will post a couple of pictures as soon as I'm done testing etc.
                        Thanks JM for all the help..

                        Comment


                        • #13
                          Thanks
                          Juan Manuel Fahey

                          Comment


                          • #14
                            Originally posted by diydidi View Post
                            I will post a couple of pictures as soon as I'm done testing etc.
                            Thanks JM for all the help..
                            Have you finished yet?
                            Juan Manuel Fahey

                            Comment


                            • #15
                              Hi JM. Completely forgot. This issue has been resolved. I have Modded almost a hundred of these units until now. Not one comeback. The info on the mod has been sent and reviewed by Montarbo HQ. lets see if they’ll implement it?

                              Comment

                              Working...
                              X