Vi limiter design

[J M Fahey]

Not a critic, everybody has his own preferences, just thinking aloud:

* why is C5 connected to positive rail instead of ground?
I fear it will introduce ripple into the amp.
In any case not sure what is it purpose.

* C1 is too large, it will introduce a loooonnnngggg turn on thump until it charge and voltages stabilize.
It *might* be related to your turn-on problem.

* U12/U5 are passing 36mA idle, which is the peak current available to drive U5/U14, more than enough for that task.
So Q5/Q9 are quite unnecesary.

* simplify VI limiter to a plain I limiter which is enough anyway because you have 3 power transitor in parallel.
If you remove R20/11/12/19 the simple I limiter will still allow 2.4A peak per transistor, so 7.2A total.
More than enough for 8 ohm loads and just enough for a 4 ohm one, not bad.

[teemuk]

Worth to point out in general is that most VI limiters only protect for a limited time. Despite limiter doing its thing the output devices will dissipate a lot of power during short circuit and will eventually fail. Voltage amplifier stage driving the output devices may also be subjected to excessive dissipation.

Effective protections in addition...
- Limit VAS current, i.e. another VI limiter
- Mute the input signal

[J M Fahey]

Agree.
Personally I use simple current limiting protection (such as I suggested above) which is good enough for a bad cable/broken plug (somebody stepped on cable)/frayed cable where some "hairs" brush against ground, etc.
Musician notices lack of sound or sparking sound, etc. , stops playing and checks.

BUT as said above by Teemu, IF you get a solid short and feed signal for enough time, it will eventually die.

Peavey/Acoustic in the old days used simple current limiting, a lot of power devices (6 or 8 for 200W) and added a thermal switch to heatsink, so after a minute or two at most , mains was switched off until cold.

Now I have some "no clue" users, such as grade school teachers, small Churches, etc. , where output may be shorted, music fed into amp, NO sound output (REAL short, including miswiring, not fully plugged cables, etc.) or a zillion speakers in parallel so amp "works" but, say, into 1 ohm load (or less) and amp burns yet they don´t notice it.
Not within , say, 5 minutes at least.

Fuses blow but *after* the fact, when amp burns.

In those cases, I add a short protection based not on transistors but a small SCR, TO92 case, which triggers a small relay, cutting speakers for good.

Replacing bad cable is not enough, you must turn amp OFF and wait until caps are **fully** discharged so SCR self turns off.

Curiously, Musicians HATE that protection , so I only add it in fopr the "non Musician" clients mentioned.

[Manowar1985]

Blew up the ops again.Could the low quality of the output transistors be the cause.

NOTE

The transistors are not bad .They are new but very cheap and I assume they are low quality.Maybe they blew up because of oscillations .I also run an AC sweep(with global feedback) at the simulation and have seen that all the frequencies close to having phase shift -180° were at a gain below unity(0.5v/v of less)Could they be prone to breakdown by themselves.

Thank you.

[Jazz P Bass]

Hmm.
Took notice that you are using a 220uf output coupling capacitor on a dual polarity power rail.
Why not connect it directly to the speaker?
And where is the zobel network?
And why are you using 'switching' transistors ?

[Manowar1985]

I used the capacitor only for testing.If something fails and I get high dc voltage at the output and then burn the ops and the speaker too.
If I use a resistive 8 ohm load do I really need to have zobel network and an inductor,because I read somewhere that those are needed only if we connect a real loudspeaker.
Those transistors are very cheap and are only for testing.If the circuit gets stabilized(no oscillations or no blow up) then I will replace them with other ones like : 2sc3281-3sa1302 or 2sc5200-2sa1943 or mj15003-mj15004.That is because I don't trust my design yet,so I used very cheap bjts.
So I really need help I have blown maybe 100 bjts trying to find the right design.I am sure that I am not exceeding the soa


So do you think that with different transistors the ops will not blow up with the same design

[Jazz P Bass]

In my mind the only ways to 'blow' the opt's is to:
1: have both sides turn on at the same time.
2: severe oscillation drives them into self destruction.

[Manowar1985]

So.My amp works and thank you for your support.Unfortunately it has considerable crossover distortion at larger signal and/or higher frequencies. That's why it is biased at 0.7 while it should be at 2.8v(simple darlington output).When I increase the bias and current begins to flow it is not stable.It keeps rising.Maybe its thermal runaway. I have placed emitter resistors at 0.27 ohms.Any comments on this would be appreciated.

[Dave H]

When I increase the bias and current begins to flow it is not stable.It keeps rising.Maybe its thermal runaway. I have placed emitter resistors at 0.27 ohms.Any comments on this would be appreciated.

Is the bias transistor in good thermal contact with the power transistor heatsink?

[Manowar1985]

yes it is but the think it doesn't heat up at all because it is running at class b mode.Whenever I try to increase the bias once it is close to 2.8 volt it starts contacting instantly and it can't be controlled.It going to thermal runaway once it starts contacting even a little.I should say that I had placed emitter resistors of 0.27 ohm and now I have 1 ohm and in both cases It is doing the same thing.Do you think that it might be the resistor between the emitters of the drivers or maybe the current of vas stage.
Thank you

[Dave H]

yes it is but the think it doesn't heat up at all because it is running at class b mode.Whenever I try to increase the bias once it is close to 2.8 volt it starts contacting instantly and it can't be controlled.It going to thermal runaway once it starts contacting even a little

It may be bursting into oscillation as it starts to conduct. I'd scope the output with it in class B mode then increase the bias voltage until it starts to conduct. Does it burst into oscillation? Monitor the voltage across the output emitter resistors to check the idle current. Is it stable when set to a low value 10mA say?

[Manowar1985]

I just found out something new.

Sometimes in class B there is some oscillation back to the input.The oscillation is so big that you can't even see the input waveform.The input was at peak 100mV when it was clean.Of course the osxcillation was transfered to the output too.That is without the zobel network.With the zobel network the oscillation in the input stayed the same and in the output was improved by a big factor.In both cases the transistors where heating up when the oscillations occur but I had the time to switch of the amplifier(all that in Class B).When I removed the resistor between the emitters of the drivers all that oscillation disappeared .Maybe it is the wrong value of this resistor..Now I think that this resistor maybe the cause for blowing ups the ops when I try to bias it into class AB.I will try to bias without it and let you know. Thank you.

[nevetslab]

In looking at the selection of semiconductors from front end thru the output stage, I see the use of the BD911/BD912 in so many locations. 3MHz ft parts.....slow, as has been stated previously. You might consider changing to 2SA1943/2SC5200 for PNP/NPN outputs, being 30MHz parts, 150W 230V rated parts. Drivers, voltage gain stage xstrs....might try MJE15032/MJ15033 parts (TO-220 parts), 250V Vce, 5A Ic rated, 30MHz parts, so you've got speed from front end thru the output stage.

Looks like National Instruments Multisim in use...not sure what version. I use Multisim 11....though I haven't checked to see if the spice models for these parts suggested are resident in the database.

I've never used a large power xstr for a bias xstr....there's very little power flowing thru there, and the Vce across the bias part is always low.....I'd use a smaller part there....could be a TO126 body, or even a TO-92 body, either clamped or slipped into a hole in the heatsink, loaded with thermal compound.

[Jazz P Bass]

Just a thought.
I had a Luxman R1120 come through and I noticed on power on the output came up really goofy.
There is a turn on delay on the speaker relay and without that delay the outputs would have cooked with a load connected.

[nevetslab]

You can also use the 2N5401 PNP for U7/U11 slow start/constant current source in the front end instead of the BD912. TO92 parts are typically used there in most designs.