Hi,
Sorry I couldn't answer earlier. Regarding DC, no there wasn't any and when measuring the bias current I short circuit the input and place a 6.8 resistor at the output.
I will implement the idea with the light bulb. Question: can the quiescent current be reliably set using the lamp in series?
Sorry I couldn't answer earlier. Regarding DC, no there wasn't any and when measuring the bias current I short circuit the input and place a 6.8 resistor at the output.
I will implement the idea with the light bulb. Question: can the quiescent current be reliably set using the lamp in series?
You said that you tried increasing the bias into class AB. I presume by adjusting R11.
Looking at the photo of the PCB, T4, the Vbe multiplier transistor is not mounted on the heatsink. This means:
- There is no thermal feedback from the output stage temperature into the Vbe multipler.
- As the output stage heats up, the Vbe multipler will not track (well not closely) the temperature of the output stage
- This amplifier is not made to be biased into Class AB.
As a starting point, I would suggest:
- You leave well alone, and leave the amplifier biased in class B
- I suspect that if the oscillations went away with compensation on T8, you have learned something very important about the problem you have. Don't ignore that.
If you really want to get into modifying this (and it is and always will be a PA amplifier so you might ask why...)
- You will need to thermally attach T4 to one of the output devices
- You will want to be very careful how you set things!!!
- You can swap out the to92 device for a TO126 (eg BD139/140) - and bolt this to the heatsink if you need.
- You would also need to to a bunch of testing, notionally:
- Set the bias current
- Allow the amp to run with no load, watching the current for drift. Watch out, when they tak off they go fast sometimes.
- Once stable, run the amp into a 4 ohm load to get the output stage warm.
- Remove the signa and check the bias current, it will have shifted, check it is not going crazy
- If all is good, repeat and get the output stage HOT (i.e. you can only just touch it) - check again
If the above gives you trouble, and without shifting T4 it will, I would call it a day and leave this as a Class B amplifier.
As a personal comment: I would be tempted to recognise this as a PA amp, crank the current back, and let it be!.
So: There is a very good reason that the amplifier went into runaway - and these things do take time sometimes, so the 10 minute delay does not surprise me.
At least 2N2955/3055 devices are cheap!
Oh, by the way, I would think long and hard about dropping in fancy fast op amps.
Looking at the photo of the PCB, T4, the Vbe multiplier transistor is not mounted on the heatsink. This means:
- There is no thermal feedback from the output stage temperature into the Vbe multipler.
- As the output stage heats up, the Vbe multipler will not track (well not closely) the temperature of the output stage
- This amplifier is not made to be biased into Class AB.
As a starting point, I would suggest:
- You leave well alone, and leave the amplifier biased in class B
- I suspect that if the oscillations went away with compensation on T8, you have learned something very important about the problem you have. Don't ignore that.
If you really want to get into modifying this (and it is and always will be a PA amplifier so you might ask why...)
- You will need to thermally attach T4 to one of the output devices
- You will want to be very careful how you set things!!!
- You can swap out the to92 device for a TO126 (eg BD139/140) - and bolt this to the heatsink if you need.
- You would also need to to a bunch of testing, notionally:
- Set the bias current
- Allow the amp to run with no load, watching the current for drift. Watch out, when they tak off they go fast sometimes.
- Once stable, run the amp into a 4 ohm load to get the output stage warm.
- Remove the signa and check the bias current, it will have shifted, check it is not going crazy
- If all is good, repeat and get the output stage HOT (i.e. you can only just touch it) - check again
If the above gives you trouble, and without shifting T4 it will, I would call it a day and leave this as a Class B amplifier.
As a personal comment: I would be tempted to recognise this as a PA amp, crank the current back, and let it be!.
So: There is a very good reason that the amplifier went into runaway - and these things do take time sometimes, so the 10 minute delay does not surprise me.
At least 2N2955/3055 devices are cheap!
Oh, by the way, I would think long and hard about dropping in fancy fast op amps.
googlyone's advice is good.
Using a faster opamp will probably make things worse. That said, there is no reason you can't use a faster/better opamp and then add the appropriate compensation externally by placing an additional capacitor across R4.
Using a faster opamp will probably make things worse. That said, there is no reason you can't use a faster/better opamp and then add the appropriate compensation externally by placing an additional capacitor across R4.
Btw i simulated the circuit in the PDF and it is theoretically stable at 20mA. 80degrees phase margin.
The only reason that R21 would burn is that you had a thermal runaway condition in your biasing which caused the output transistors to fail short circuit. T7 may also fail if the output transistor fail, so best to check that it is still ok.
Putting leads on T4 and mounting it on the heatsink would be a good start.
The only reason that R21 would burn is that you had a thermal runaway condition in your biasing which caused the output transistors to fail short circuit. T7 may also fail if the output transistor fail, so best to check that it is still ok.
Putting leads on T4 and mounting it on the heatsink would be a good start.
Last edited:
Hi guys and thank you again for your advices.
@googlyone:
- T4 was mounted on a heatsink together with the output transistors. I had it removed from the heatsink once I started to have oscillations and bias runaway to reduce the variables. I will remount it, I use a clamp for that I mount on the heatsink fins.
- I guess that the guy who was selling those kits (in the 80's in former Yugoslavia it was very hard to obtain electronic parts) wanted to avoid complains from customers and decided to place 1.5K for both R11 and R12 and virtually leave it in class B. With 1.5K, there is almost zero bias current on the output. BUT, he does claim on page 3 that this is an AB class "Hi-Fi" amplifier and instructs how to lower or increase bias on page 4. Page 1 states 10-20 mA of bias
- The whole PCB layout is impractical, forcing the use of output transistors far from it.
- I will eventually leave it in class B if I'm unsuccesful in my last attempt. However, I managed to set the bias on the left channel and I keep it drifting +- 30% without T4 being mounted on the heatsink. It sounds decent, too, bass is very controlled and I don't have complains for the mids and highs.
- I'm not tempted to use faster opamps, my main objective is stability. I learned from your comments that a faster opamp may increase oscillation susceptibility, so I'm not even considering that.
@TMM
- It's been about 22 years since I last used PSPice. I downloaded somethng called qucs for Mac and managed to calculate DC biases. With R11=1K, I get 23 mA of bias current on output transistor. What program do you suggest with easy learning curve?
- How can I stabilize the bias circuit from thermal runaway? Maybe use the topology with two transistors like on the picture?
@googlyone:
- T4 was mounted on a heatsink together with the output transistors. I had it removed from the heatsink once I started to have oscillations and bias runaway to reduce the variables. I will remount it, I use a clamp for that I mount on the heatsink fins.
- I guess that the guy who was selling those kits (in the 80's in former Yugoslavia it was very hard to obtain electronic parts) wanted to avoid complains from customers and decided to place 1.5K for both R11 and R12 and virtually leave it in class B. With 1.5K, there is almost zero bias current on the output. BUT, he does claim on page 3 that this is an AB class "Hi-Fi" amplifier and instructs how to lower or increase bias on page 4. Page 1 states 10-20 mA of bias
- The whole PCB layout is impractical, forcing the use of output transistors far from it.
- I will eventually leave it in class B if I'm unsuccesful in my last attempt. However, I managed to set the bias on the left channel and I keep it drifting +- 30% without T4 being mounted on the heatsink. It sounds decent, too, bass is very controlled and I don't have complains for the mids and highs.
- I'm not tempted to use faster opamps, my main objective is stability. I learned from your comments that a faster opamp may increase oscillation susceptibility, so I'm not even considering that.
@TMM
- It's been about 22 years since I last used PSPice. I downloaded somethng called qucs for Mac and managed to calculate DC biases. With R11=1K, I get 23 mA of bias current on output transistor. What program do you suggest with easy learning curve?
- How can I stabilize the bias circuit from thermal runaway? Maybe use the topology with two transistors like on the picture?
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
The lamp technique should be used for testing purposes it will save a lot of fried transistors in case of malfunction. You may adjust the bias although the amplifier is operating at a lower mains voltage. You will need to readjust the bias after the lamp would be removed.
Most amplifiers include a coil in series with the output and an RC filter which helps to keep oscillations down.
As regards the Vbe multiplier transistor it should be mounted directly on the heatsink. It gives a good thermal tracking of the output stage and adjusts the bias accordingly. Hm...is not always the case of course.... I have built lately a stereo amplifier delivering 300W per channel and I have used 8 BDY56 per channel which I mounted on a large heatsink and used forced air cooling. The cables connecting the driver board and the output transistors are 20cm long. Checking carefully the output with an oscilloscope I could not trace any signs of oscillation. I use it the last 2 months and has no problems at all.
By the way I agree with Googlyone but remember that class B amplifiers suffer from Xover distortion.
There are 2 bundles of 3 wires for each channel, I twisted the wires in each bundle and then run it in parallel with the other one.
Back to work. My heatsinks are screwed to the chassis, which is at ground potential. Is thsi OK or should I take messures to isolate it?
Okay, I've used a similar circuit designed in 1966 for 40xxx RCA transistors (slower than snot) and the design didn't make it through 1971 when the transistors got faster. The dynaco "TIP mod" included
1. putting 51 pf from b to c of both T7 and T8
2. putting 1000 ohms 2 watt to ground from speaker out to PS negative
3. putting a "zobel" between junction of R22 R23 and the speaker, inside the chassis. This is about 11 turns wire around a AA battery or china marker, in parallel a 1 ohm high watt resistor. This keeps RF (like from your CELL PHONE. DIGITAL TV TUNER, PC MAINBOARD, DSL MODEM from getting in the amp and causing oscillation. These things all became popular AFTER the amp went through checkout.
In addition, I put 10 ohm 1 watt carbon comp resistors between the drivers T7 T8 and the output transistors T9 T10. While I did this for fusing purposes to keep flames from burning too many parts when the OT's went short, somebody have said "base stop resistors" are handy to prevent oscillations in too fast output transistors located on remote heat sinks. Oddly enough, the value I picked was about what they recommended, only I think they said 4.7 ohms was enough.
I had a bias control circuit go out of control due to a sense transistor mismatch, and run my amp at 250 ma idle current, for about a year, and no damage occured. I did have fans on the output heat sinks, since the ones in my amp are so patently inadequate to product the 60 w/ch advertised. I was aiming for idle current of 30-40 ma.
I was using NTE60 OT's, which are even faster than modern ST2n3055, and are maybe equivalent to MJ802 or something.
I did not twist up the leads from PCB to the output transistor heat sink.
In my later experiments with a more conventional sense circuit, I've been putting an 8 turn inductor on the sense circuit back from the device on the heat sink, to the bias transistors. These I salvage from PCAT power supplies, and I hope will keep RF from getting in the long wires and into the driver circuit.
BTW, if you want to experiment more, keep the digital devices listed in bold out of the experiment room. that goes with digital scopes IMHO. Having compliance with FCC regs doesn't mean a digital device won't drive an amp crazy, IMHO, especially with the case off and no zobel on the output. In addition parallel c2 with a .1 uf ceramic cap, there is a **** of a lot more digital noise floating around now than in 198*. Also put 33 pf across R4, to cut the gain on the op amp at RF.
If you just want the amp to sound good and work fast, put in a LM3886 IC. It should work fine with this power supply. See the datasheet which you can download from datasheetcatalog.com with example circuits. I paid $7 each for a triple of them a couple of years ago.
Good luck and have fun.
1. putting 51 pf from b to c of both T7 and T8
2. putting 1000 ohms 2 watt to ground from speaker out to PS negative
3. putting a "zobel" between junction of R22 R23 and the speaker, inside the chassis. This is about 11 turns wire around a AA battery or china marker, in parallel a 1 ohm high watt resistor. This keeps RF (like from your CELL PHONE. DIGITAL TV TUNER, PC MAINBOARD, DSL MODEM from getting in the amp and causing oscillation. These things all became popular AFTER the amp went through checkout.
In addition, I put 10 ohm 1 watt carbon comp resistors between the drivers T7 T8 and the output transistors T9 T10. While I did this for fusing purposes to keep flames from burning too many parts when the OT's went short, somebody have said "base stop resistors" are handy to prevent oscillations in too fast output transistors located on remote heat sinks. Oddly enough, the value I picked was about what they recommended, only I think they said 4.7 ohms was enough.
I had a bias control circuit go out of control due to a sense transistor mismatch, and run my amp at 250 ma idle current, for about a year, and no damage occured. I did have fans on the output heat sinks, since the ones in my amp are so patently inadequate to product the 60 w/ch advertised. I was aiming for idle current of 30-40 ma.
I was using NTE60 OT's, which are even faster than modern ST2n3055, and are maybe equivalent to MJ802 or something.
I did not twist up the leads from PCB to the output transistor heat sink.
In my later experiments with a more conventional sense circuit, I've been putting an 8 turn inductor on the sense circuit back from the device on the heat sink, to the bias transistors. These I salvage from PCAT power supplies, and I hope will keep RF from getting in the long wires and into the driver circuit.
BTW, if you want to experiment more, keep the digital devices listed in bold out of the experiment room. that goes with digital scopes IMHO. Having compliance with FCC regs doesn't mean a digital device won't drive an amp crazy, IMHO, especially with the case off and no zobel on the output. In addition parallel c2 with a .1 uf ceramic cap, there is a **** of a lot more digital noise floating around now than in 198*. Also put 33 pf across R4, to cut the gain on the op amp at RF.
If you just want the amp to sound good and work fast, put in a LM3886 IC. It should work fine with this power supply. See the datasheet which you can download from datasheetcatalog.com with example circuits. I paid $7 each for a triple of them a couple of years ago.
Good luck and have fun.
No need to isolate the heatsink. Make sure though that the C-B-E terminals of the output transistors are well insulated form the heatsink. You should use an insulating sleeve over the B-E terminals which will not allow an accidental contact of the terminal to the hole's wall...Check the mica and the plastic washers of the collector screws....replace them if they are damaged. Check the insulation with an Ohm meter...it should be infinite.
Last edited:
Was setting up an amp yesterday and encountered the same condition (couldn't bias it, as soon as i got any significant bias it would spring into oscillation) which made me think of this thread.
The problem in the end? Output zobel filter wasn't connected properly. I see now that the schematic doesn't include an output zobel filter, try adding one - 10ohm in series with 100nF from output to ground. It would also be wise to put an inductor in series with the speaker output terminal. 10 turns around a 1/2" diameter should suffice.
The problem in the end? Output zobel filter wasn't connected properly. I see now that the schematic doesn't include an output zobel filter, try adding one - 10ohm in series with 100nF from output to ground. It would also be wise to put an inductor in series with the speaker output terminal. 10 turns around a 1/2" diameter should suffice.
...Oh I get by with a little help from my friends...
Hello dear friends,
These verses came to my mind once the reason for blowing output transistors is found. I replaced 4x2N3055 with only 2xMJ15004, T3, T5, T6 as well and mounted T4 on the heatsink. I also removed the BC compensation capacitor. I used the series bulb powering method and I can tell you it saved me from couple of more blowouts.
There were no oscillation, but I was randomly getting DC on output and in my penultimate attempt T3 had a collector emitter irreversible blowout.
Finally, T3 was replaced and there was no DC and the output transistors were having about 50mA of bias.
My friend offered help and came to have four eyes and helped me with his experience. One more try and DC is back again. PCB triple checked, no loose contacts no blown components. He analyzed the circuit and concluded that the OP could be latching to Vcc because it is supplied with near (or as we measured, a little over) maximum allowed voltage. He said the amp is designed too much in the border zone of some of the components. He proposed to try other OPs. We first tried the LF351 (which should almost be identical to TL081) and the amp was working happily and there was no trace of oscillations. To prove the assumption, we put a new TL081 and again the output was showing DC.
He next proposed to try with NE5534 which has higher maximum voltage and again there was no DC on the output. Hooray! It was already too late and I decided to continue the observations the next day.
The next day revealed that with NE5534 it is enough to measure DC on output to be able to cause slight oscillations. The "slight" oscillation immediately heat the output transistors. Also, even without measurable oscillations NE5534 is so hot that I can barely stand ten seconds touching it.
Still, since NE5534 has higher supply voltage, I would prefer to use it instead of LF351 and after the weekend, I will try adding the Zobel network and compensation capacitor across R4 and see if the oscillations and heating is gone.
If all is fine, I'm connecting a speaker and observing the 2xMJ15003 heating (on paper they should be able to substitute twice the number of 2N3055)
I will follow-up. I know this is mostly all known to the experienced diy-ers, but the information about TL081 latching could be added value.
Hello dear friends,
These verses came to my mind once the reason for blowing output transistors is found. I replaced 4x2N3055 with only 2xMJ15004, T3, T5, T6 as well and mounted T4 on the heatsink. I also removed the BC compensation capacitor. I used the series bulb powering method and I can tell you it saved me from couple of more blowouts.
There were no oscillation, but I was randomly getting DC on output and in my penultimate attempt T3 had a collector emitter irreversible blowout.
Finally, T3 was replaced and there was no DC and the output transistors were having about 50mA of bias.
My friend offered help and came to have four eyes and helped me with his experience. One more try and DC is back again. PCB triple checked, no loose contacts no blown components. He analyzed the circuit and concluded that the OP could be latching to Vcc because it is supplied with near (or as we measured, a little over) maximum allowed voltage. He said the amp is designed too much in the border zone of some of the components. He proposed to try other OPs. We first tried the LF351 (which should almost be identical to TL081) and the amp was working happily and there was no trace of oscillations. To prove the assumption, we put a new TL081 and again the output was showing DC.
He next proposed to try with NE5534 which has higher maximum voltage and again there was no DC on the output. Hooray! It was already too late and I decided to continue the observations the next day.
The next day revealed that with NE5534 it is enough to measure DC on output to be able to cause slight oscillations. The "slight" oscillation immediately heat the output transistors. Also, even without measurable oscillations NE5534 is so hot that I can barely stand ten seconds touching it.
Still, since NE5534 has higher supply voltage, I would prefer to use it instead of LF351 and after the weekend, I will try adding the Zobel network and compensation capacitor across R4 and see if the oscillations and heating is gone.
If all is fine, I'm connecting a speaker and observing the 2xMJ15003 heating (on paper they should be able to substitute twice the number of 2N3055)
I will follow-up. I know this is mostly all known to the experienced diy-ers, but the information about TL081 latching could be added value.
Hi sharbatgula,
You twisted the leads to the outputs! Don't do that.
Guess what I am going to suggest you do with the lead dress? Make it like it was before you started. How the wires are run in an amplifier can be critical, it may not take much to upset a marginal design. When you twisted the leads, you increased the coupling between the wires. That can cause oscillation very easily. You may have other problems, but start with this. It is probable that you have got by this because so much had been replaced.
Do you have an oscilloscope? You need to check for oscillation with that. Also check near the peaks on a sine wave signal. You may get bursts of oscillation there.
Okay everyone, if you work in any equipment, make sure the wires are placed in the same position that you found them in. Don't get creative.
-Chris
You twisted the leads to the outputs! Don't do that.
Guess what I am going to suggest you do with the lead dress? Make it like it was before you started. How the wires are run in an amplifier can be critical, it may not take much to upset a marginal design. When you twisted the leads, you increased the coupling between the wires. That can cause oscillation very easily. You may have other problems, but start with this. It is probable that you have got by this because so much had been replaced.
Do you have an oscilloscope? You need to check for oscillation with that. Also check near the peaks on a sine wave signal. You may get bursts of oscillation there.
Okay everyone, if you work in any equipment, make sure the wires are placed in the same position that you found them in. Don't get creative.
-Chris
OK, I admit I got carried away with creativity and the urge to pack everything neatly and visually appealing. But, I'm also confused with the twisting. I was peeking some amps in magazines and spotted plenty of twisted wires. So, at the moment I have twisted wires feeding the supply (about 12cm/5" long), twisted wires on the output tranistors (about 10cm/4") and twisted wires to the Vbe multiplier (about 10 cm/4").
Wires to the output terminals (speaker) are not twisted.
Do you suggest to untwist them all? Admittedly, the amp's previous wiring was without twisting but all wires were longer and it looked like a forest of wires in every direction.
Wires to the output terminals (speaker) are not twisted.
Do you suggest to untwist them all? Admittedly, the amp's previous wiring was without twisting but all wires were longer and it looked like a forest of wires in every direction.
An amplifier should be stable regardless of the way the interconnecting wiring is set up..(except the input and the star ground point)
I have build the amplifier below a year ago and checked it carefully for any signs of oscillations....I found absolutely nothing at all.
I think the amp is finally behaving stable. I left the faster NE5534 in and added in steps: Zobel parallel network, removed C3, added unity gain compensation capacitor between pins 5 and 8 and finally a compensation capacitor. The results of the steps were:
1. Mr. Zobel made the amp even more instable. It was self oscillating this time
2. Removed C3 (from OP out to ground). Can't read oscillations on the voltmeter, but after a while the series bulb glows indicating high current. No DC on output.
3. Added unity gain compensation capacitor of 10pF (between pins 5 and 8). This completely stabilized the amp. No signs of oscillation
4. For the sake of safety added a 33pF compensation from output to input. That should limit the bandwidth to about 100kHz according to the simulation.
Dilema: leave it with NE5534 plus compensations or put back LF351 without any compensation?
1. Mr. Zobel made the amp even more instable. It was self oscillating this time
2. Removed C3 (from OP out to ground). Can't read oscillations on the voltmeter, but after a while the series bulb glows indicating high current. No DC on output.
3. Added unity gain compensation capacitor of 10pF (between pins 5 and 8). This completely stabilized the amp. No signs of oscillation
4. For the sake of safety added a 33pF compensation from output to input. That should limit the bandwidth to about 100kHz according to the simulation.
Dilema: leave it with NE5534 plus compensations or put back LF351 without any compensation?
As I said previously, I'd put a hf gain killing disk capacitor (33 pf) around the feedback resistor of the op amp, whether old or new. There is waay more RF noise around these days compared to when this was designed. I didn't know what the knee point was, I don't do simulations on a Linux op system computer, but the sound was fine on top octave grand piano source on my op amp disco disco mixer. A very hard sound to reproduce.
Hi John,
You are recommending a bad practice to another person for reasons only known to you. Try not to be so cavalier about things like lead dress.
-Chris
Why do you think that? You have just inserted random coupling into a gain device without even investigating possible problems. And you expect a commodity designed product to have all the fine engineering that people pay extra for? I'll put forward this concept. Not all designers are competent. My 35 + years in service have illustrated that quite clearly. I have seen more examples of twisted leads creating instability than I can recall.
Okay, so how exactly have you checked your amplifier for signs of unwanted behavior? Have you used an oscilloscope? Without basic instruments, you can't say anything about whether your amplifier is stable or not. A DVM type meter doesn't count.
You are recommending a bad practice to another person for reasons only known to you. Try not to be so cavalier about things like lead dress.
-Chris
Hi sharbatgula,
Well, the same goes for your testing. An oscilloscope (or spectrum/network analyzer) is required to know for sure if the amp is stable or not. Previous stages can be oscillating like crazy without RF being present on the output of the amp. As I mentioned earlier, you also have to check for HF bursts with dynamic signals or a sine wave as well. Then also with actual speakers connected.
Your compensation cap between pins 5 and 8 made all the difference. You may have to experiment with the value as you have greater gain than unity. The LF351 has internal compensation built in for unity gain.
The zobel network is an R-C network, just to make certain we are talking about the same thing. Sometimes you need an inductor to isolate the output stage from capacitance. Those are the coils you may have seen in other equipment in series with the output signal. A raw capacitance from the output to ground is extremely odd. Normally the resistor is added to damp the resonance. You have to try the zobel with the capacitor removed from circuit, you can't / shouldn't use both at the same time. I would say that a zobel is better than a straight capacitor for stability.
4. What do you mean you added a 33 pF cap between input and output? That normally goes across the feedback resistor and has a lower value normally.
Right now you are working completely in the dark without an oscilloscope. There are some basic instruments you need to work on these things, a good DVM and oscilloscope are two main instruments you can't do without. You can get a used 'scope (a good one) for less than $200. The new ones are cheaper than ever before. Likewise a good voltmeter (HP 3478A (5.5 digit)- say) is also not expensive. I picked up an HP 974A for less than $100 CDN, and that is a fantastic meter (4.5 digit, 100 KHz hand held meter). If you can score an HP 34401A, you'll have one of the best meters you can use for any service work. You actually do need a half decent meter, and you would never know this until you have one. They are available. Every single hobby there is has some cost associated with it. But these things drastically improve your joy in the hobby.
-Chris
Well, the same goes for your testing. An oscilloscope (or spectrum/network analyzer) is required to know for sure if the amp is stable or not. Previous stages can be oscillating like crazy without RF being present on the output of the amp. As I mentioned earlier, you also have to check for HF bursts with dynamic signals or a sine wave as well. Then also with actual speakers connected.
Your compensation cap between pins 5 and 8 made all the difference. You may have to experiment with the value as you have greater gain than unity. The LF351 has internal compensation built in for unity gain.
The zobel network is an R-C network, just to make certain we are talking about the same thing. Sometimes you need an inductor to isolate the output stage from capacitance. Those are the coils you may have seen in other equipment in series with the output signal. A raw capacitance from the output to ground is extremely odd. Normally the resistor is added to damp the resonance. You have to try the zobel with the capacitor removed from circuit, you can't / shouldn't use both at the same time. I would say that a zobel is better than a straight capacitor for stability.
4. What do you mean you added a 33 pF cap between input and output? That normally goes across the feedback resistor and has a lower value normally.
Right now you are working completely in the dark without an oscilloscope. There are some basic instruments you need to work on these things, a good DVM and oscilloscope are two main instruments you can't do without. You can get a used 'scope (a good one) for less than $200. The new ones are cheaper than ever before. Likewise a good voltmeter (HP 3478A (5.5 digit)- say) is also not expensive. I picked up an HP 974A for less than $100 CDN, and that is a fantastic meter (4.5 digit, 100 KHz hand held meter). If you can score an HP 34401A, you'll have one of the best meters you can use for any service work. You actually do need a half decent meter, and you would never know this until you have one. They are available. Every single hobby there is has some cost associated with it. But these things drastically improve your joy in the hobby.
-Chris
- Status
- Not open for further replies.