Can oscillation destroy output transistors?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
My test amp is running happilly with a pair of MJ15003/4 when I decided to try out a pair of 2SC5200/1943 on it and see how it sounds. This amp has no enclosure (it's a test amp)and when I connected the new transistors I used wires to connect to the pcb. After following the test procedures it worked, but only for about 5 seconds. :bawling: A resistor came on fire too. :cannotbe:

When I was trying to analize what went wrong I noticed that the wires I used to connect the transistor to the pcb came too close to the shielded wire connecting the input to the potentiometer (volume control). Can this cause oscillation? I have a DMM connected to the V+ and V- reading 126VDC (my supply rails are +/-63VDC) when I turned the pot I noticed it went down to 112VDC:confused: There was no input program at the time of testing. Load was an 8 ohm speaker which was also destroyed:mad:

Can this be caused by oscillation? By putting the transistor wires close to the pot means putting the output close to the input right? Please help me, any suggestions will help.

BTW, the transistors which were destroyed were Toshiba 2sc5200/2SA1943. Labels were laser printed and cannot be removed by fingers nor acetone, hopefully originals, bought it from a reputable shop here in my country.

Jojo
 
Unfortunately it's possible. Try to use short wires to connect the output devices, or use small (1-2ohms) resistors series with the base of the output devices. The series resistor is useful when You can't use short wires, or connect paralell more devices.

Sajti
 
AX tech editor
Joined 2002
Paid Member
oscillations

Hi Jojo,

Yes oscillation can be the cause. But you didn't say whether you adjusted the bias current when you changed the xsistors. If not, possibly the bias current was way too high, which would explain the sagging of the supply voltage, dying of the xsistors and the demise of the speaker.

Whether the closeness of the wires gives the oscillations, if that's what it was, I doubt it (how close is too close). Is this an inverting or non-inverting amp? Is there a big difference in Ft between the old and new xsistors?

Jan Didden
 
Hi Jan

Jan,

I did adjust the idle to 25ma. Too close in my case was side by side (stupid me, output transistor wires + input wires side by side:scratch: ). Ft of the MJ15003/4 according to datasheet was 2MHz while the new 2SC5200 was 30MHz.

Many might think that substituting the MJs with the 2SC/A is not wise due to the dissipation but what I had in mind was just to test it first then parallel more 2SC/A so that there won't be any problem with Pd.

Back to your question, what does Ft do to an amps response/behaviour?

Jojo
 
Jojo,

I think by now you have the answer.

Oscillation caused cross conduction, which destroyed your outputs through overcurrent. It's a bitter lesson, but a very significant one. Since we can't observe this phenomenon directly, we note only the effects and must become detectives instead!

Step 1: Examine your lag compensation capacitor very carefully. This cap, typically 100pF, lies across the collector base of the voltage amplifier, which is normally the next transistor in the chain after the input diff pair. As you raise the rail voltage, the amplifier is rapidly more stressed, and prone to oscillation. Correspondingly, you usually need to increase this capacitor to prevent oscillation, which occurs because the amp still has considerable gain at the point where the negative FB turns to positive FB (the pole). You should not use too high a lag comp cap, but you must have it to avoid oscillation. The art of amp design lies in manipulating circuit topologies and dimensions so that the size of this cap can be minimized.

Step 2: Insert a 100R base stopper on each driver base, and a 10R base stopper on each output transistor base. These resistors inhibit the rapid migration of charge to and from the bases of these devices, thus dampening oscillation whilst permitting the flow of audio signals. Excellent insurance.

Step 3: Ensure there is a 0.1uF capacitor and 10R 1W resistor in series strapped across the output from the amplifier. This is a Zobel network. By modifying phase angles (between current and voltage) at high frequencies, this allows the amp to drive real world loads, that is loudspeakers, which are highly reactive.

Step 4: Insert a 4uH inductor/10R 1W resistor in parallel into the output of the amp. Do NOT include this in the feedback loop. This prevents the entry of RF from the unshielded leads, and could prevent oscillation being triggered by a noisy RF environment. This is usually the last precaution you need to consider.

Step 5: Keep all input and feedback circuitry well away from inputs and rails, and make sure all feedback tracks and wires are short - preferably no more than 5mm.

All of these steps are complementary; the idea is to prevent the amp oscillating in any operational environment. Once you have achieved that, you are well on the way to making an amplifier that sounds good, because a surprising influence on sonics is short term bursts of oscillation, which basically rob resolution.

Hope this helps,

Cheers,

Hugh
 
mrfeedback,

that's a very nice trick, thanks for sharing.

now with the new transistors...

just bought them awhile ago, got them setup, made wires as short as possible... it works! :bigeyes:

the new 2sc5200/1943 works great, they sound good too. i am still in the break-in process and i keep monitoring my supply rails, idle current, and hope that what happened last night won't happen again :mad:

thanks to everyone!

jojod818
 
jojo,
it is possible to have oscillation problems in an amp specially with a common emmiter type output stage, if you are fammiliar with the "kinetic" amps sold locally, these amps are notoriuosly prone to such oscillations...so that whenever i am asked to repair one, i politely turn down unless i can change the output stage...Ft is the gain bandwith rating of a power transistor, it is when the small signal gain of the transistor reaches unity at a maxed frequency, it is common for power transistors to have ft's of 2-4mhz while small signal transistors have 100-up mhz....i use 10ohm 1 watt resistors as fuses when testing amps, you know by now how fast you can loose your power transistors with a slight mistake such as a reversed bias diodes.....cheers...
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.