Symmetrical amplifier conundrum

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

I am trying to get the first channel of my symmetrical amplifier (designed by Anthony Holton) running, but I'm having some trouble setting up the output stage. I've been waiting for an e-mail response from him for a few days but I'm quite eager to fix this problem so I figured I would ask the forum.

The power supply works fine, voltage is a little high at +/- 74-76V, and the voltages on the board all checked out before installing the output stage.

I first plugged the amplifier in, DC offset was a few millivolts, and set the bias as usual, then realized that some of the output Mosfets were getting no bias (there was no voltage across their source resistors). After a few minutes there was a small spark and two of the output stage Mosfets blew (I unplugged it immediately). The rest of those along with Q10, the biasing transistor, checked out OK with multimeter tests. It turns out a few connections were loose since the pins on the Mosfets did not reach fully through the PCB (I'm not sure if this is an error in the PCB's mounting system or if I made a silly mistake). Nothing was burnt, no smoke or anything, just the gate was shorted to drain on those two fets.

Anyway I got two more Mosfets that were matched close enough, soldered small pieces of wire to extend all the pins to ensure good connections, and double checked everything before powering up again. So when I plug it in, I see DC offset varying from 1 to 1.2 volts, and there is no voltage across any of the source resistors in the output stage. Where did the bias go?

My questions would be: what parts of the circuit should I check to verify that the rest of the amplifier has not been damaged? And what would cause a DC offset and lack of bias? I am not familiar enough with the amplifier's design to troubleshoot this myself.

Thanks very much for any help, I really appreciate it.

Michael
 
Oy vey

Hi again,

The above post can be mostly ignored because it turned out that I had switched the base and collector leads of Q10 (wired off-board) after I had taken it out to check it. What a stupid mistake... I fixed this. All the other transistors on the board are working as well. I had some solderpads come off the board in my, but this was easily fixed with a few jumper wires.

Of course however, there is no free lunch.

After looking over the connections yet again, I turned on the amplifier and found 3mv DC offset. I set the bias to 18mv across a source resistor and kept adjusting it because it was decreasing over time (a few minutes).

Then there was a big pop, so loud it scared me. I think the red clipping indicator LED lit momentarily as well, but I'm not sure. Of course I shut the thing down (actually the fuses did for me) and checked everything. There was a little burning smell around the Mosfets, and no visual damage. I removed one of them from the board and it tested OK, thankfully. (In the circuit they all produce the same readings so I think they're probably all intact). Q10 looks like it's still working too. I am quite afraid to turn the amplifier on again and check it in operation (since that could endanger expensive components and my safety), until I have some idea of what is going on.

I've considered power-supply over voltage as a problem... I forget the exact value when I was running the amplifier, but now (30 minutes later) it is +/- 76.8V at no load. This is with a 50V transformer (which is recommended for the circuit), so I would think such a discrepancy would be acceptable.

I had read about N-channel amplifiers going up in flames because of not mounting the thermal sensing device on the same heatsink on the output stage, so I had piggybacked it on Q11 (one of the N-channel Mosfets) with silver thermal paste and held it down with superglue on the edges. So I don't think that's the problem either.

What happened? This time I really thought I had it right. Sorry to bother you guys... and even some moral support would be greatly appreciated. Thanks again.

An externally hosted image should be here but it was not working when we last tested it.
 
Not easy to know what happend.

Seems you have found nothing more than a fuse, that has burned.
How much is 18mV - in current?
Is there any other safety in the circuit
beside the fuse?

With +-75 volt
even not so much current, can be a lot of Power.
P=UxI

Have you any other details to tell us, now?

/halo - wants to know - what happend?
 
Halojoy,

The 18 millivolts is across a .22 ohm source resistor, and there are four Mosfets per output bank (N and P channel), setting the bias current at about 330mA. There is no protection anywhere in the circuit except a 5 amp slow blow fuse on each voltage rail.

Another thing I thought of was that maybe because of imperfect matching, the bias on some of the Mosfets was way high. The amplifier uses a 20 turn pot for biasing... it took like 15 turns to get any measurable voltage, and then the next turn took it past 18mv. Possibly the bias on the other Mosfets was very high because of this sensitivity to adjustment... well, I don't know. The resistors are 5% tolerance and the Mosfets were matched within 0.03v if I remember correctly.

The sound reminded me of when I was carrying my other power supply around and the wires touched for a moment... a snap or pop leaving a warm smell. I don't remember seeing a spark or anything. I think I will take out the rest of the Mosfets and make sure they are all in working order, because it would be really something if this failure didn't damage anything.

Well, at this point I may as well just check all the components again, probe around with the circuit off and wait for some more ideas to pop up. I guess I'm just a bit paranoid in not wanting to fire this one up again (or the other channel which is ready and waiting) until I know what the deal is.

And thanks for posting, maybe helpful people like you can help me get somewhere.

Michael
 
my amp is actualy rinning at 77v and works just fine. getts a little warm but heh oh well. here is some pic's. did you put them resisters on the bottem or the top. : O ) if not then that is your problem.
 

Attachments

  • dscf0002.jpg
    dscf0002.jpg
    65 KB · Views: 915
Looks very nice Jason, I see that is the N-channel amplifier that seems to be more popular than the one I have? How does it sound?

I have Xicon 0.22 ohm 5W/5% wirewound resistors soldered to the bottom of the board. And I hope heat is not a problem, since the heatsink is at worst 0.28 C/W. I have two channels of Zen V4 at 1.3 amps (~140 watts dissipated all the time) mounted horizontally on one of those and it's at a safe temperature.
 
Jason seems to be repeatedly get confused between Anthony Holton's N-channel amp and other designs, in this case his symmetrical amp design wich can be made with L-type mosfets as well as IRF devices.

Kilentra, how closely have you matched the IRF Mosfet banks and what methods did you use for matching. I would suggest that you check all components, particularly SS transistors, driver/Current source transistors and also check for leaking capacitors. The trouble seems to be at a stage behind the output stage for you to be facing new problems, each time you switch on the amp. If in doubt replace parts that you don't have equipment to test - eg., capacitors. Small signal and other transistors should be checked out of the circuit with an hFe tester and check for leakage.

When you have replaced all parts, use another transformer with a lower voltage or a variac and use only one pair of output devices. It would also be wise to replace insulation kits, ground the heatsink, add 10n and 100n supply decoupling caps, remove source resistors and parallel 1 or 2 watt CFR resistors to obtain the required low value resistance and twist +- rails leads. This will eliminate Hf oscillations if any. When all is well, add the paralleled output devices and up the voltage rungs.

Hope it helps.
 
Sam,

On this board the N-channel devices were matched to between 3.94-3.96v and the P-channel between 3.88-3.90v. This was done by simply hooking the Mosfet up with an 18v power supply and 120 ohm resistor in the matching circuit given by Holton, and measuring the voltage across the pins. He does state they should be matched to +/- 100mv, which is a pretty wide tolerance.

Thanks for the suggestions, I will check all the components this afternoon. I don't have another transformer readily available, but I think I can temporarily wire this one's primaries in series to halve the voltage. I can't say I have an hFe tester or oscilloscope either - just analog and digital multimeters. Don't have enough non-inductive power resistors on hand to try that modification, either. I'm afraid I am poorly equipped.

So I will definitely try adding power supply coupling caps and grounding the heatsink... along with checking over everything again and removing all but one pair of Mosfets. I'm low on spare parts but I can borrow from my other channel for the time being.

Is it safe to run this amplifier at full supply voltage using just one pair of output stage Mosfets? Heck, I should rather be asking if it is safe for a person like me to be running this amplifier at all. :)

Anyway, thanks a lot guys. I'll see how my detective work goes today.
 
Kilentra:

I'm going to build same amplifier as you and will buy the components next Monday, except transformers, supply capacitors and output transistors. I will test the amplifier without expensive mosfets with laboratory supply in the school. If it's working properly then I'll buy rest of the components.

Have you read the Anthony's construction notes for the symmetrical amp? On page four there is "Pre-flight test". It says solder 10 ohm resistor across zener diode ZD3. So you can test the differential and voltage amplifiction stages without MOSFETs.

Oscilloscope would be nice but if you haven't got access to one testing have to be done with multimeter. You can get sine wave test signal from CD player by playing test CD. Measure the input signal voltage. Then you multiply this by 32 and should get approximately same value as output voltage reading.
 
Naula,

Good to see someone else is taking on this project. Hopefully with good testing equipment and more experience you will not have the same problems as I did.

I have read the construction manual and did the pre-flight test... all the voltages were right on the mark. It was only when I tried to bias the output stage that I started having problems. The first problems were due to my stupid mis-wiring and now I have damaged some things because of that. Just be careful with your own amp... I am a good example of what not to do.

So anyway, I removed the output stage and performed the pre-flight test again. This was the result (correct values in parentheses):
R3: (1.6V) 0.60V
R5: (1.6V) 0V
R15: (1.0V) 0.06V
R12: (0.5V) 0V
R13: (0.5V) 0V
R8: (14.6V) 14.61V
ZD1: (15V) 15.14V
R11 (offset): (0-0.1V) 51.5V... wow.

I lost another P-channel Mosfet and two N-channel Mosfets from the output stage. Let's hope I don't destroy any more.

Judging from these results it appears as if the input stage current source is OK, but everything else is screwed up. It looks like I will be replacing all of the transistors. Does anyone know of a U.S. source for these small signal devices that does not charge high shipping and handling?

What I'll do is pre-flight test the other channel... re-drill the PCB's mounting holes so the Mosfet pins are long enough... and be extra careful. It would be nice to get one channel running to boost my confidence. Thanks again for the help so far.
 
Good Luck Klientra!

I am sure you will get it right in the end.

It only needs one little wrong connection
somewhere in the circuit
and nothing will work alright, at the output.

Thinking of all the mass of connections and components
there is in a Power Amp,
it is a wonder they will work at all.

/halo
 
Hi!

Im in the starting fase of building the symmetrical amp, I have the Heatsink/transformer/PCB and all the MOS-FETs. The problem I have is regarding some parts value....

So I thought Id ask you what you use.... (Anthony has not answered on my e-mail)

* C3 - schematic = 100nF, part list = 10nF, what is right?

* Do you know a good replacement for ZD3 and ZD4 or would any simular
work?

* What is the rating of C12? - ,47 !? Is that 0,47uF?

* Is R42 1Mega ohm?

Hope anyone can help me as these questions are the last barrier...
Thanks!
 
Yay, someone else is trying it too

Hi richeros,

For C3 I am using 100nf. I did not receive a parts list. In fact I made the "mistake" of ordering high-voltage polypropylene caps (notice how on my board there are some large red/brown caps instead of the blue rectangular MKT caps normally used). I am not sure if this will improve the sound, but it did cost a little more and was a bit of trouble fitting some of them into the PCB.

ZD3 and ZD4 are specified as 1N4737, I assume any old 7.5V zener diode will work. C12 I assume is 0.47uf. At least that's what I used. I also used R42 as 1Mohm. I just used all the values from the schematic in the online construction manual.

I recommend you order some extra small signal transistors (MJE340s, BC546s, etc), because if anything goes wrong you'll need some. They only cost maybe 50 cents each. Also see if you can get some extra Mosfets so you can match them more closely, and so you can replace some when they get destroyed. :)

I'll have to ask again if anyone knows where I can get a few of those transistors in the U.S. without the usual handling costs. Why can't these places just put them in an envelope and mail it?

Anyway, I used a drill to extend the Mosfet mounting holes closer to the pins. This made them close enough so I could solder them directly into the board and they would still fit. The second channel passed the pre-flight test and is just waiting to be fired up... I'm not at home now but I will post a picture soon.

I have one last question. Long after I unplugged the second channel the power supply wires touched somewhere and the clipping indicator LED lit briefly. I think the supply was down to just a few volts, and there are diodes on the rails onboard to prevent switching polarity, but is there any chance this damaged something on the board? I would rather not have a similar "conundrum" with the second channel! (And I cannot easily do the pre-flight test again, for the Mosfets are all soldered and firmly mounted.) Thanks again guys.
 
2nd channel worked

Hi again,

I have completed the second (first one working) channel of the amplifier and it works, which to me is quite a relief. However, I connected the output to the large solderpad on the PCB next to ground (which I assumed was the right place for the - output) and after a few minutes of music at moderate levels, R9 (10 ohm resistor to ground; I used 0.6W version) started smoking. It's quite burnt as you can see, it used to be light blue.

Duh! Well, of course if you connect 10 ohms between the speaker and ground you will pass a lot of current through the 10 ohm resistor. This is the kind of brain-fart I'm trying to avoid. :)

An externally hosted image should be here but it was not working when we last tested it.


The lesson here: if you are building the symmetrical amplifier, don't solder your output wire into the spot on the PCB where you might think it should go - connect it directly to ground.

So I replaced the resistor, reconnected the output and it works again. But who knows, this could have damaged something else. The replacement parts for the first channel are on the way. Again, thanks for the help.
 
Hi, I have one last question. :)

How am I supposed to mount Q10, the Vbe multiplier device, on top of the output stage Mosfets?

I used a dab of silver thermal paste in the middle and applied superglue to the edges to hold it down, which I assumed would create a good enough thermal coupling. However, the tops of the Mosfet cases are noticeably warmer than the top of the BD139 casing... they are around 45C I think (fairly hot) whereas Q10 is barely warm, which I assume cannot be a good thing for the stability of the amp. I am wondering how others who built the N-channel amp mounted the similar device from that project?

I read the thread where Arne K's amp went up in smoke from improper thermal stability and I've been reluctant to turn up the volume on this one because of those concerns.

Is the thermal coupling I have adequate, and if not, what do I do with this thing?

An externally hosted image should be here but it was not working when we last tested it.
 
As you can see the mounting hole on the BD139 is far smaller than that of the IRFP240 and I would be afraid of damaging the die by drilling to enlarge it. Electrical isolation is not a problem with the plastic case of the Mosfet.

Arne, I have an overvoltage (as high as 77V) power supply and difficult speakers as well. Degenerative resistor power should not be a problem... at 500 watts output into 4 ohms there would be 11 amps RMS current, spaced over eight 0.22 ohm resistors so each would have to dissipate less than one watt.

BTW, how does it sound?

Did Mr. Holton ever tell you just how closely coupled Q10 should be? Maybe I could rip it off the Mosfet, put in longer wires and bolt it to the aluminum bracket there? (Actually, that would probably be an improvement over what I have now). I am having trouble reaching him by e-mail.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.