I've replaced all the electrolytics on both channels. 2263-2267 but i have kept the old ones and they where measureing fine but they where 20+ years old. And fault had already occured before i replaced those.
Here is the receipt of what i bought (i'm from greece):
Are you talking about the 2 zeners and 2 x 1N4148 after the 2 x 100ohms resistors? On the repaired channel i did replace them and that channel did hold up.
I had no reason to change anything other than the caps on the good channel (which is now fried).
How would "i see" if there is an oscillation present? Let's say i borrow an oscilliscope and feed 1khz on the input. How would the output waveform look like if i had an oscillation?
Also note that the resistors 3286 and 3285 + the rest of the feedback circuit from the output emitters back to the differential amp have all been checked on both channels (just in case but it still blew)
Look at post #7. Do as I explained there already...
Start with resistive load. Monitor the output, sweep the timebase and look for oscillations. Sometimes they'll show on top of (superimposed) sinewave; sometimes you'll see large spikes. Change to squarewave, (REDUCE the amplification first!!!), and look for overshoot oscilations.
Go back to sinewave. Increase the gain (look at post #7)
If no oscillations, add a 0.047uF cap in parallel to the resistive load. Repeat.
Now include the speaker cables; place the load at the end of the speaker cable. Repeat the test. If there's spike in instability, the speaker cables might be too capacitive.
If you see oscillations, trace them back until you find the culprit. Use 10:1 probe setting. You may see oscillations everywhere; however, only the culprit transistor/diode will cause a large spike in oscillations at the output, once you poke it with a probe.
Start with resistive load. Monitor the output, sweep the timebase and look for oscillations. Sometimes they'll show on top of (superimposed) sinewave; sometimes you'll see large spikes. Change to squarewave, (REDUCE the amplification first!!!), and look for overshoot oscilations.
Go back to sinewave. Increase the gain (look at post #7)
If no oscillations, add a 0.047uF cap in parallel to the resistive load. Repeat.
Now include the speaker cables; place the load at the end of the speaker cable. Repeat the test. If there's spike in instability, the speaker cables might be too capacitive.
If you see oscillations, trace them back until you find the culprit. Use 10:1 probe setting. You may see oscillations everywhere; however, only the culprit transistor/diode will cause a large spike in oscillations at the output, once you poke it with a probe.
Thank you i will try to find (or buy) an oscilliscope and see what's going on there. after i replace all the faulty components.
Your previous post was really helpfull at this point. I did not take it into consideration preiously because i did not have an oscilliscope in oreder to test.
But now i have no choice if i want to get this thing repaired (and learn something more in the proccess)
Yes, get a solid oscilloscope and learn how to use it. Play with cheap amplifiers, try different loads... get a tone generator. I think these days you can get a software application that will generate all 3 types of waveforms and all the required frequencies. I still use an old-fashion tone generator, though. Then you can get a variac... and you're done. Navigating the maze blindfolded would become a thing of the past.
Bias is set with no load attached as any DC offset can skew the bias result. Having no load removes that possibility.
The current may be quite dependent on supply voltage and so should always be set on full mains (no bulb). Thermal stability is very important and the bias current should be checked when the amp is properly warmed. What you do not want is any tendency for thermal runaway where the bias current just keeps creeping up.
Can you explain this in a bit more detail waht is thermal runway? Can bias keep increasing while it gets hotter?
Also i need you to explain to me how the bias is measured in this specific case because it troubles me. Let me explain:
- The manual states that i should check the bias on the end to end pins and not the middle pin like this:
-On the schematic though it seems that this resistor connects the emitters of the npns and pnps. So how is the measurement being produced won't one voltage negate the other since they are opposite?
Wouldn't it be preferable to measure B-E voltage on each Output transistor directly and set the bias accordingly?
Here is the part of the schematic that confuses me:
From what i can tell the middle pin goes towards the output and the end to end pins go to the emitters? Maybe i was measuring wrong?
Bias can increase with heat. If the bias generator (the circuit around the transistor with the bias preset) generates a fixed bias voltage then the bias current in the output transistors will rise as temperature increases. This is because the current gain of any transistor will increase with temperature. As more current flows it gets hotter still and that continues sometimes destructively. Also the base/emitter is also varies with temperature and reduces as the transistor gets hotter. That also means more current flows.
The two 0.18 ohm resistors are in series and the speaker output is from the centre pin. We have no speaker attached for setting the bias.
The manufacturer will recommend a bias current (lets say 80 milliamps) and that means the voltage across each 0.18 ohm resistor should be 0.080*0.18=0.014 or 14 millivolts per resistor. So they would say set the voltage to 28 millivolts as measured end to end across both series resistors. You get the same result measuring across just one resistor and setting the voltage to 14 millivolts.
As those are low values you can get better accuracy setting for 28mv across both rather than 14mv across just one.
It doesn't happen like that. Firstly the voltage measured from ground to the resistor doesn't matter in relation to the current through the resistor (although the offset voltage should be near zero in a working amp).
If the amp had a DC offset of +10 volts (and we assume no load connected) then a bias current of 80 milliamps means that the centre pin of the resistor is at +10 volts. The upper end going to the NPN will be at 10.014 volts and the end going to the PNP will be at 9.986 volts. The voltage across the two resistors is 10.014-9.986 which is 28 millivolts and 28 millivolts across the two 0.18 ohm resistors is 0.028/0.36 which is 80 milliamps.
That would never work
The voltage is not an absolute but varies with transistor to transistor and with temperature. It also varies between different manufacturing processes. When we say the B/E voltage is usually around 0.6 volts that is true only up to a point, it is not an absolute that can ever be used to set the operating point of a circuit.The bias generator (transistor and preset) senses the temperature and as it heats up it conducts more and that pulls the generated bias voltage down reducing the current in the output stage. It is called a Vbe multiplier.
https://en.wikipedia.org/wiki/Rubbe...e-to-emitter voltage (V BE) of the transistor.
Thank you for these valuable information i understand it now. From what i can tell this is the main bias circuit:
At some point during previous repairs i found that i could not set bias it was always zero. That is not the case anymore.
But i'm interested in understanding bias better because at that time i had left the bias as it was and it still blew the output transistors even though i was getting a zero reading at the output resistor.
What is usually the cause of zero bias? My understanding is that all transistors up to the final outputs should have a b-e voltage as per the schematic and if one in the chain is missing you have no "drive" to the output transistors leading to a zero reading. But then again how could it have blown since the current was in theory zero since the output transistors where not biased?
Could it be severe oscillation as Extreme_Boky suggests?
I will check with an oscilliscope of course. As soon as i get paid i will buy this one which is good (and relatively cheap) for beginners like me:
https://www.skroutz.gr/mp/144173164...rafos-pagou-UTD2102CEX-2-kanalion-100MHz.html
Thanks in advance for taking the time writing and explaining all these concepts. If you have any good resources i can read to learn more please feel free to share any links!
At some point during previous repairs i found that i could not set bias it was always zero. That is not the case anymore.
But i'm interested in understanding bias better because at that time i had left the bias as it was and it still blew the output transistors even though i was getting a zero reading at the output resistor.
What is usually the cause of zero bias? My understanding is that all transistors up to the final outputs should have a b-e voltage as per the schematic and if one in the chain is missing you have no "drive" to the output transistors leading to a zero reading. But then again how could it have blown since the current was in theory zero since the output transistors where not biased?
Could it be severe oscillation as Extreme_Boky suggests?
I will check with an oscilliscope of course. As soon as i get paid i will buy this one which is good (and relatively cheap) for beginners like me:
https://www.skroutz.gr/mp/144173164...rafos-pagou-UTD2102CEX-2-kanalion-100MHz.html
Thanks in advance for taking the time writing and explaining all these concepts. If you have any good resources i can read to learn more please feel free to share any links!
If you look at the circuit you will see you have the output transistors (forget the parallel pairs, they make no difference to this), you have the driver stage feeding the output transistors and you have the pre drivers feeding the main drivers.
That means you have six base/emitter junctions to forward bias in order to turn them all on and get current flowing in the output transistors. Working to 0.6 volts B/E voltage for a silicon transistor means we need around 6 * 0.6 which is 3.6 volts. That is the voltage that must be developed across between the base's of the NPN and PNP pre drivers. In other words the voltage the bias generator has to provide.
It is important to realise the 0.6 volts is not an absolute. Some older transistors may turn on at closer to 0.48 volts and some newer ones closer to 0.65 or 0.7 volts. Its all down to the manufacturing processes and physical properties of the transistor.
So we can never set bias by simply setting an assumed bias voltage at the generator or setting the B/E volts to some supposed value. It just won't work doing that. You must monitor the final current flowing in the output transistors to know it is correct and measuring voltage across a fixed low value resistor is the easiest way (the 0.18 ohm resistors).
The 0.18 ohm resistor also provide a massive degree of thermal stability.
Look at the NPN side. If the current in the output transistor tends to increase the emitter voltage also increases and provided the base is held at a constant voltage that means the applied B/E voltage effectively decreases slightly and that reduces current flow. If those resistors were shorted out the bias would become very critical and always be tending to increase destructively.
That means you have six base/emitter junctions to forward bias in order to turn them all on and get current flowing in the output transistors. Working to 0.6 volts B/E voltage for a silicon transistor means we need around 6 * 0.6 which is 3.6 volts. That is the voltage that must be developed across between the base's of the NPN and PNP pre drivers. In other words the voltage the bias generator has to provide.
It is important to realise the 0.6 volts is not an absolute. Some older transistors may turn on at closer to 0.48 volts and some newer ones closer to 0.65 or 0.7 volts. Its all down to the manufacturing processes and physical properties of the transistor.
So we can never set bias by simply setting an assumed bias voltage at the generator or setting the B/E volts to some supposed value. It just won't work doing that. You must monitor the final current flowing in the output transistors to know it is correct and measuring voltage across a fixed low value resistor is the easiest way (the 0.18 ohm resistors).
The 0.18 ohm resistor also provide a massive degree of thermal stability.
Look at the NPN side. If the current in the output transistor tends to increase the emitter voltage also increases and provided the base is held at a constant voltage that means the applied B/E voltage effectively decreases slightly and that reduces current flow. If those resistors were shorted out the bias would become very critical and always be tending to increase destructively.
Either a definite fault like an open junction or insufficient voltage developed across the bias generator.
Oscillation usually causes an increase in current but a meter may or may not show that accurately. If the oscillation is high frequency the conduction of the transistors (NPN and PNP) can overlap and leave both on momentarily on each cycle and that causes heat generation usually... and distortion.
Zero bias breaks the feedback loop authority and thus tends to lead to high frequency oscillation in some circuits. So its probably best to roughly set bias with current-limiting measures in place, as both under-bias and over-bias can lead to high dissipation - and never set the bias with a speaker attached in case something does oscillate and risks frying the tweeter. But people can and do set bias by starting at zero and cranking it up slowly whilst observing the current (well, voltage across the resistors), without any current limiting between the PSU and amp.
I will do that while on the dim bulb to make sure that the bias circuit functions properly and that i can actually measure bias and then reset it to minimum and set it without the dim bulb tester while the amp is warm enough. A important note here is that the output transistors where cool and not hot when all the failures occured.
I'm writing this reply as a repair log until the final output transistor replacements and the oscilliscope arrives.
I think it will make an excellent case study for any other beginners like me and maybe one of you can correct me if i'am wrong:
In the beginning when i first got the amp:
- I got the amplifier from a friend and it was producing a lot of distortion. Turned out it had broken solder joints in alsmost all driver transistors.
- I resoldered the joints and tested the amp. It was working fine! (little did i know).
- I did not check bias and dc offset when i first repaired the solder joints (big mistake)
- My guess is that my friend might have opened the amp and tinkered with the bias trim pots in an attempt to fix it himself without telling me. Because i saw that the main power switch had been bridged with a wire by someone, it was not staying on you had to hold the switch, which means someone had tinkered with this thing previously! I replaced the switch with a new one later on.
So what happened so far? As far as the feedback loop is concerned and the damage on both channels:
Right (Previously repaired and repeatedly getting fried but now good channel):
When the fault first occured i had extensive damage.
Testing components from the output and going backwards (check attached schematic for reference):
- Output transistors and their base resistors 10ohm resistors fried
- The 150ohm resistor connecting those 10ohm resistors at the bases got fried (it blew a hole in the board!!!!)
- The pair of the transistors before the 150ohm resistor were shorted as well (2sc4793, 2sa1837 pair)
- The 2x 47ohm resistors at the bases of 2sc4793, 2sa1837 pair where burned also
- One of the bf422 - 423 pair transistors was sorted and the damage stopped there
- Both 2x0.18ohm output resistors where completely open
- As far as the feedback loop of the right channel is concerned the 27k feedback loop resistor was also open.
As far as the right channel is concerned all those components have been replaced repeatedly....
Why? Because i had not measured bias and had left it as it was assuming it was the factory setting!!! And to make things worse after the 2nd time it had failed i tinkered with the dc offset and bias until i "saw" the voltages the schematic states at the bases of each stage IGNORING the bias measurement procedure the manufacturer states!! (Man i screwed this up BIG TIME!!)
So it kept frying the output! I set the bias properly and it works fine now. (with 30mv dc offset can't get it to go any lower after repeated repairs)
Left (Previously good channel):
This one presented fault a few days ago and the damage was not as extensive as the other channel. This one needs investigating further:
- The 2xR18 ohm output resistors where measuring 2x 0.33ohms (measured out of circuit) i replaced those with the correct value (2x0.18)
- The 27k feedback loop resistor was reading about 40k (out of circuit) which means it had degraded // should be 27k - i replaced this one as well
- The output transistors got fried/ shorted along with one of the 10ohm resistors at their bases (i replaced all 4 just in case) - still waiting for the replacements for the output transistors
- The bias trim port was set to about 20ohm of the 0-100ohm range (that seems it was a bit high about 80% towards maximum // that's what made me think that someone had tinkered with it). I have reset it and I had measured 15mv across the output resistors on this one but i think for some reason the bias kept kreeping up until it destroyed the channel - will keep investigating- and replace the pots with multi-turn sealed ones
- All the other components leading back to the 2 differential amplifiers in the input where measuring fine.
So the damage was not extensive on this one. But still i need to investigate further.
So that's it basically. A mixture of previous tinkering by someone else (my guess i haven't validated that 100%), my ignorance about proper biasing and the lack of proper tools (like an oscilliscope) got me to where i am now.
Attachments
Last edited:
When measuring low value resistors you need to also account for the meter lead resistance which for low priced DVM's is going to be around the 0.2 ohm region.
The replacement resistors did meausre 0.18 ohms. My multumeter is this one: https://www.skroutz.gr/s/17055176/Q1-True-RMS-PSifiako-Polymetro.html
To be honest that didn't cross my mind i will check again with only the leads to see if there is any compensation for the multimeter leads. When reading such low values i should be more carefull. Yet another thing you taught me today! Thanks!Fair enough
The final transistors finally came and i installed them along with a few resistors.
I checked all the transistors before the outputs down to the 2 transistors before the bias chip and everything checks out fine.
I did not go further back to the 2 differential amplifiers. Should I? All the voltages there measure fine.
The Bias on both channels is always zero and i cannot set it set no matter where i turn the trimmers even if i let it sit for 30 minutes. (I'm on the dimbulb).
Note / Question:
When i had an input signal in (1khz sine wave from my computer to the rca inputs, used this website: https://onlinetonegenerator.com/):
I did measure dc bias on the 2 ends of the output resistors and it reached at about 20mv when the volume knob was half volume.
But i know this is not an accurate reading and it has nothing to do with idle current. What the hell is this though?
While i had the speakers connected and the amp was on the dim bulb i put a 1khz sine wave in and cranked the volume to 50% then i used a portable osciliscope (that's all i could get for now) to look a the wave form for any irregularities such as oscilation, clipping etc. here are a few photos of the oscilliscope:
So no zero crossing distortion that i can see and no oscillation. But i do see a small difference between the positive and negative side on the oscilliscope.
So what do i do now? I cannot set bias. how should i procceed?
I checked all the transistors before the outputs down to the 2 transistors before the bias chip and everything checks out fine.
I did not go further back to the 2 differential amplifiers. Should I? All the voltages there measure fine.
The Bias on both channels is always zero and i cannot set it set no matter where i turn the trimmers even if i let it sit for 30 minutes. (I'm on the dimbulb).
- I put 8ohm speakers on both channels and it sounds fine no audible distortion even though i'm on the dimbulb and the bias reads zero.
- For the 30 minute test on 50% volume the output transistors were cool to the touch, the predriver transistors where quite hot (that was always the case since i got the amplifier).
Note / Question:
When i had an input signal in (1khz sine wave from my computer to the rca inputs, used this website: https://onlinetonegenerator.com/):
I did measure dc bias on the 2 ends of the output resistors and it reached at about 20mv when the volume knob was half volume.
But i know this is not an accurate reading and it has nothing to do with idle current. What the hell is this though?
While i had the speakers connected and the amp was on the dim bulb i put a 1khz sine wave in and cranked the volume to 50% then i used a portable osciliscope (that's all i could get for now) to look a the wave form for any irregularities such as oscilation, clipping etc. here are a few photos of the oscilliscope:
So no zero crossing distortion that i can see and no oscillation. But i do see a small difference between the positive and negative side on the oscilliscope.
So what do i do now? I cannot set bias. how should i procceed?
Most output faults only take out components up to the driver stages. There is no need to check anything else without good reason to do so.
The bias measurement is only valid with no signal and no load attached as it is a static DC measurement and setting.
If you turn the volume up with signal applied then each 0.18 ohm sees the load current delivered to the speaker and this is in the form of a halve wave signal in each resistor. The actual current in the resistors could be several amps given enough signal and the voltage across them several hundred millivolts. The DVM doesn't stand a chance in showing what that current may be.
Add to that the bias voltages as shown in the schematic assume that you’re not limiting the supply by using a dim lamp. The dim lamp is used as a quick go, no-go test, once the lamp dims down, after the big ecaps charge up, in seconds, the test is over. Next test is to apply full AC to be able to read proper voltages and set bias.
Make sure all the bias pots are set to the lowest value, ccw for these initial tests.
Make sure all the bias pots are set to the lowest value, ccw for these initial tests.
Last edited:
