Mosfet amp problem, Slone amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
This is design 11.8 in the high power amp book. Two sets of output fets 250watt 4 ohms.

I have built two of these as in two channels and one works beautifully and I thought the other one did too, but it doesnt.

The problem is that when I increase the volume (this is amp is driving a bass driver via and active xover) there is a horrible sound produced from the speaker. Using sine waves at 50hz this becomes noticeable at 6.1V, this is driving and XLS 10.

Now I know its the amp and not anything else because I have done the usual swapping of channels etc to make sure.

This never used to actually be a problem when I was driving 850146 because they didnt need EQ so subsequently needed less power from the amp. The problem was still there but I actually thought it was the speaker, and it only was a problem when it went VERY loud so as you can imagine, I was pushing the driver a bit so I expected it to make a few odd noises.

But now I have the XLS and EQ they draw quite a bit more power then the 146's did.

The output stage is not clipping I dont think because I can easily turn it up past this point and the amp isnt running out of power, but the sound does get worse.

I have replaced the pre drivers, whenever Ive had problems its always been those but to no avail. I have also checked that all the resistors are of the same value from amp 1, which works, to amp 2 which half does.

The next thing to do would be swap the fets over from one amp to the next and see if that does anything but this is quite a large thing to do to the amps and reasonably demanding of the circuit boards. I have also increased the values of the resistors that do the SOAR protection thingy to see if it was that, it wasnt.

I wanted to ask if anyone had any idea what it could be before I go crazy with the soldering iron. Before the problem sets in the amp works beautifully and sounds great too.

Thanks in advance Matt
 
A likely cause is tripping the protection circuit. The amplifier parts may be working fine, but if the protection circuit is not, it may trip prematurely, causing a terrible sawing sound on high power peaks. I would not expose a tweeter to this amp until you get it working. Do testing on a dummy load and observe the behavior on a scope.

You might try temporarily disabling the protection circuit by removing the diodes D5 and D6 to localize the problem area. If that helps, then you know it is in the protection circuit itself. Perhaps one of the transistors or diodes is leaky. Increase the time constant caps (C15, C16) from 0.047 to 0.22 mF to see if that affects it.

On an unrelated note, I think the compensation caps at 5 pF are seriously too small on this design, and would suggest replacing them with at least 22pF and possibly higher value parts, times 4 (C11,12,13,14). Also, the 0.33 resistors on the drains are silly, they belong on the sources to encourage current sharing. All the outputs could be sensed through a common drain resistor of high wattage, say 0.1 ohm 10W.

Let us know how this resolves.

I confess to having grave doubts about the stability of this output stage, CFP with L-Mosfets. With V-Mosfets I get cross-conduction of Mosfets a lot with this configuration, and I am struggling to make a stable amp. The problem seems to be faster turn on than turn off, resulting in cross conduction which is fatal instantly. When it works it sounds marvelous, but I have blown out many parts. I am regarding it as a challenge to make reliable, but very frustrated right now.
 
This may not be germaine to your case, but in the last few weeks I have stumbled on something that triggers oscilations that I never previously thought much about. - Solder joints that look good but actually are not. It's a long story I won't recount here but it turns out that it is possible to have problems when the component lead heats faster than the pad. When touching the solder to the joint the lead is hot enough to cause it to flow and the pad is hot enough so a nice meniscus shape forms but not to form a good bond. Later if you apply a little heat and some force the bond breaks neatly at the pad while the bond to the lead holds. Low or lower temp type solder seems to be involved in this. Modifying the my technique so that the solder is touched only to the pad until it starts to flow seems to be the cure. At least two boards that were re-worked this way suddenly worked fine. Reverting to standard solder may have helped as well.

To repeat: the "bad" joints were visually indistinguishable from "good" ones so we are not talking about grossly bad solder joints produced by obviously bad techniques. In less sensative circuits these joints would most likely have been completely acceptable or maybe I should say functional.

My first thought was that the thin boundary at the pad acted like a dialectric and was thus a source of small but unwanted capacitance. Entirely by coincidence, I was looking through one of JL Hood's books where he mentions that bad joints can behave like diodes. In either case the same end result. A CFB (BJT or FET) output is said to be fussy in any case but I imagine any slight disturbance such as cap-like or diode-like behaviour in a solder joint or two could set it off. The offending joint may not even need to be in the on one output section components.

The fact that you have one amp board working perfectly suggests that something in the physical construction that is not obvious to the eye should be considered as a possible culprit. At least it is a possability to be checked out and eliminated.
 
Ok cheers for all your responses, Ill try changing some of the things recommended.

With regards to the protection circuit I have increased the values of the SOAR resistors would this not accomplish a similar thing?? It didnt do anything when I did. Either way I've got plenty more things to try.

Cheers once again Matt
 
OK i went in for the kill 1st of all and removed D5 and D6 this removed the problem. SO as it seems to be the protection circuitry im now increasing the cap values. This had no effect. It didnt improve anything whatsoever, if I could say it did anything then it made it worse.

I have tried replacing the diodes too and that might have improved the situation a little but not by much.

As I now know that the problem is specifically in the protection circuitary is there anything else that could be triggering false activation. Diode swaps, cap alteration, increasing the res values to allow for a bigger SOAR, these all contribute nothing to solving the problem. What about replacing Q9 & Q10?
 
You wrote: .. removed D5 and D6 this removed the problem.

Aha, it's nice when your long distance diagnostic guesses are correct. I would replace the two transistors in the protection circuit because one is probably leaky and they are too cheap to worry about which one.

Then as another poster suggested, re-heat all the solder joints to make sure there is a good bond. I had a bad joint on a 100 ohm base emitter resistor which fooled me for a while, the amp worked but I kept losing the output transistor. By chance I noticed it, reheated it and it was fine. It's pretty easy to have a bad joint when you're in a rush, but oh so frustrating to find.

And finally, a toast to JLH for his wisdom and good grace. He's gone now, but I was always impressed by the balanced judgement he showed in his writings and by his gentlemanly behavior. I hope we can have that tenor of discourse here on the board.
 
As I thought about this further, it could be tricky. The problem could be in the protection circuit or the output stage. I'd replace all the transistors and diodes in the protection circuit for starters. Cost is nil, and it's acting like a component in the protection circuit is marginal. In my opinion, it's not worth localizing to which one, you just want to get it fixed.

If that does not cure it, The protection circuit could be triggering early because one output transistor is current hogging or an emitter resistor is high in value, or the resistor divider feeding the protection circuit has the wrong values. (The circuit seems a little hair-triggered, try changing R24 and R25 to higher values, say 470 ohm.) Or the output circuit could be breaking into momentary oscillation and triggering the protection. On his latest designs, Slone has put a 10 pf cap from gate to drain on all the outputs, which tells me he has noticed some high-frequency troubles there. Personally, I prefer to use ferrites on the gate leads and good layout.

Let us know how this turns out.
 
Under further scrutany I have discovered that it is the N channel side of the amp that is the problem.

This I found out by removing the diodes one at a time in the protection circuit. With the P channel left as it is there is no problem, I can get literally window rattling bass 20hz both channels, the room and window rattles without amp or speaker producing any untoward noise. Putting back in the N channel diode and removing the P the problem is present again. This simplifies things sumwhat as it narrows down where the problem could be.

The N channels in both amplifiers run colder then the P channels is this normal amp behaviour? in my other amps 11.4 design both NPN and PNP run at the same temperature.

I have replaced the diodes and trannies in the protection circuit. I am assuming that is diode 5 and Q8, this does nothing.

I had some problem with oscillations in one of my 11.4 amps but this was a broken track and when this happened the amps got VERY HOT. Now this time it doesnt, even with the protection off, the other P channel still gets hotter. I was thinking that when music plays if the oscillations are only very quick then any temperature increase because of this will be negligable as its only transient. BUT if I am running a sine wave because if its duty cycle there is no point where there is low amplitude so to speak. Therefore if the amp was oscillating if a sine wave was played thru it and it did oscil then you would expect a good heat increase. If you remove the protection and run powerful sinewaves there is no additional heat produced.

I also dont think that increasing component values is the key to sorting this out as the other channel works flawlessly. This to me spells disaster, if the other channel is fine and no protection jumps in then I would assume the protection circuitry is fine for the task and that there is an alternative problem causing this.

I am continuing to look into this and will report back if I find anything else.

Once again cheers Matt
 
If anything the N-channel side should run cooler, due to lower Rds(on). You probably have some high frequency oscillations sucking power on the N-channel side. As I mentioned, Slone is putting 10pF caps from gate to drain on his latest amps, I presume to quench oscillations. The other traditional fix is to increase the gate resistors to 1K (from 470). Try the caps or resistor changes and let us know how it goes.
 
I changed the res values from 460- 940, I didnt have any 1ks so i just added another 460 in series.

This seems to have solved it although I cant be too sure yet as I cant do any really loud tests due to parents in bed.

However it seems the SOAR is actually tripping quite early anyway, although this is LOUD, the XLS are not running out of excursion lol at this level not by anymeans they are still perfectly linear.

Loud as in the mid amp clips before this level so I thinks im OK.

Once again cheers for solving my mosfet mystery! Another thing learnt then.
 
The protection circuit protects the output transistors, and does not necessarily protect your load. It is set to trip rather early in this amp, at about 2 amps per transistor or a little more when the load is shorted to ground. These transistors will handle up to 7A each with low voltages across them, so this is rather conservative. Your speakers probably have low impedance at low frequencies and are sucking lots of power, bottoming out the protect limit.

This sort of protection circuit works well to protect against shorts from output to ground, which are the most common fault by far (shorted speaker cable).

But -- this protection circuit will cause the amp to clip immediately, as the negative feedback loop senses that the output is low. The result is the hash which fries tweeters. So this is good for the amp, but bad for speakers. Which one is more valuable to you?

Personally, I delete protection circuits and fuse the outputs. My amp may die, but hopefully not my speakers. My speakers are much more valuable than my amps. But this is a choice you get to make since you are building the amp.
 
Well as the only thing the amps are going to be driving are XLS10 directly they will have a fairly easy time so it would probably make sense to up the protection a bit, id rather keep it there mind you to protect from short circuit just incase.

So Ill increase the values a bit to give me some more head room just incase its needed. The speakers can take it and if the amp can give it then why not set it up that way.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.