2 Newbie Questions: matching MOSFETs, and humming power supplies

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
hey now,

i'm attempting to build the SOZ amplifier as a test project using parts i have lying around, before actually building the real deal with massive heatsinks. the aim of this project is to get ~2W from a pair of monoblocks into 2 altec 755a's mounted in an open baffle (so efficiency is not a consideration, but NOT blowing up the speakers IS!)

here's the 2 questions, i apologize in advance for their complete stupidity, if applicable

1- i set up the circuit to match my MOSFETS's - IRF540 as described in the PASSDIY.com how-to section. Used a 24VDC power supply and a 50ohm 10W resistor. the problem i have is that after plugging in the MOSFETS's the Vgs starts out as ~3.9V, but falls down to ~3.7V, presumably as the transistor heats up, in approximately 1 minute. what am i doing wrong? or how am i supposed to match these?

2- i plan to use 4 regulated power supplies (lab retirees from the university) that each source 24VDC-8A, (running +/- 24VDC rails for the 2 monoblocks). but unfortunately, found that they hum! i.e. mechanically vibrate (electrically rock stable - measured AC is < 0.1mV). is this just because these have old transformers with shaky laminations? is there something i can do to fix them up, i would hard pressed to build something like these on my own! my backup plan was to use 4 12-volt car batteries and setup a relay to put them on trickle charging when amp is switched off. is this a feasible idea?
Let me give my opinion on your questions:

The change of Vgs is normal. You should wait untill it is stabilized, that means: untill the Mosfet/heatsink reaches its final temperature. Depending on the size of the heatsink it can take up to half an hour (very large heatsink). You'll see Vgs get stabile at this point.

By adding a copper strap surrounding the entire transformer at bobbin height you decrease the radiated magnetic field by 20dB or so. Make sure the copper surrounding is shorted. Mechanical vibrations are more difficult to overcome, I don't have a good solution for that.
A battery powered amp is a good thing, in fact you'll get rid of any noise and ripple. So I recommend the last, but you have to spend some serious money.

On the FET matching issue, I think as much of an answer as you would ever expect is already here, but you'll have to use the search function. Search for 'match' or 'matching' in he Pass Labs forum, and you'll get lots of info.

On the other, I think it's the transformer laminations, and I don't know what to suggest. Maybe try pouring epoxy over the thing. I fix everything with epoxy.;)
Shellac is the traditional answer for "glueing" transformers back together. Easy to work with, cheap, and dries in about 15 minutes. Epoxy will soften at high temperatures; shellac won't.
We're assuming that the power supplies are in good electrical health, yes? No shorted windings, not overloading them, etc.

Keep in mind that mains supplies are often very dirty. Transformers will hum if DC is present on the line; and it often is. Try running the powersupplies off an isolation transformer and see if they quiet down. Should this be the case, you will need to do some work on the AC.

Thanks for all the advice!

thanks to all the experienced folk for their advice. i will do the MOSFET matching with properly installed heatsinks for longish periods of time to ensure that the voltage stabilizes.

also, regarding the PS mechanical hum, i will try all the aforementioned options (dipping laminations in shellac, adding copper sheilds). i did already try an isolation transformer ahead of the power supplies, and the hum did reduce. unfortunately the only isolation transformer i have is only 500VA rated, and itself hums quite a bit!

I too had confusion with the MOSFET matching. When I did the checking of IRFP240s for my Aleph60, I wired it according to the diagram in A40 doc. I used binder clips to attach the MOSFET to a heatsink and placed 3 100R 5Watts resistor wired in parallel. I applied 17V into the circuit. The resistors do heated up so I put them on an additional heatsink. I kept the MOSFETs for 5 to 7 minutes to allow it to heat up and stabilize. I got reading of 3.81V to 4.10V for 30 devices. During that time, the voltage registered to say 3.90, later after 2-3 minutes goes to about 3.88V. It was all same scenario for all MOSFETs, that is, it goes 0.02V lower. But to my surprise, when I touched the heatsink, I felt almost zero increase in temperature. Everyone says it must heat up. Have I done the matching correctly? Is my voltage of 17V not enough? (the doc says 15V only, some says 24V)
Anyone has a recommedations?:(
If my power supply has a maximum output of 12V, what resistor value and power rating should I use in a mosfet testing circuit? Do I need to mount the mosfets on a heatsink or can I safely test them without one? How long must I wait for the temperature to be right for measurement? What would be the answers to these questions if the supply was set to 10V?
If you are just going to test your amp why bother with mosfet matching?

just power it up and measure the DC at the speaker terminal. If its too big (>.1V)) then try another FET. It's the easier way.

SOZ has only 2 fet's /ch ! :D

I made my 1st SOZ without any testing and got DC offset of 0.040V.
no, no, no

I didn't said you have a SOZ.

I did say that you are going to make a test amp( or even real SOZ amp) then don't match the fet's.

just mount everything.

If DC offset is to big then replace MOSFET's. After everything is on.

I have build a 5W SOZ and it works fine without any matching(am I'm lucky?)
Okay, I'm still alive but I would like the behavior of some mosfets explained. I used the circuit described in the A75 article. My power supply outputs 9.85-10.16V and with a 36ohm (measured) power resistor in series to the drain, that should put me between 165-175 mA through the device. I am testing IRFP150's and some don't drop their volt reading. My reading range is 3.621-8.694V. Not a single device was stable before 4 minutes in the test circuit. None of the devices ever got too hot to hold firmly between the fingers but were very warm. I had a small heatsink handy and noticed that some of the stable mosfets would raise their reading when placed on the heatsink and some stayed put. Others would continually dance within a 180mV range and some required a full hour to wake up. My meter has a 2 Mohm/volt sensitivity and is running on a fresh set of batteries. My leads have 0.20 ohms of resistance. I get the feeling that this matching procedure isn't being performed properly so where is the problem?

Probably you madesome mistake.

The mosfet dissipate alot of heat, so medium size heatsink is required.

is your power supply big enough for the fet's powers?

I still say: don't match anything.

did you bought all the fets from the same location at the same time?

the mosfets are from the same series and are Harris-Intersil parts. I read that HarryHaller has used these same parts and likes them alot. Why would you assume that they get really hot if there is only 166mA going through them? I could understand when they are driven in a circuit where they are biased to 3A but in this circuit they feel about the right temperature when and if they wake up.

HarryHaller, I know smart guys like you probably don't give a damn about simple stuff like this but since you are familiar with this part, maybe you can clue me in on what's wrong with my matching test. If you're reading HarryHaller, an explanation of these haywire mosfets would be appreciated.
If anyone has any experience getting these IRFP150's to match please enlight me with your experience. Do I need apply more voltage to wake these up? Heatsink or no heatsink? How long should it take for them to settle? The A75 article tells us to aim for 170mA but this ain't cuttin it so should I be thinking in watts instead of amps. Should I use a 12ohm resistor with my 10V power supply?
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.