Hi Chris,
I'm using a 10 Ohm resistor in series with 0.1uF and that combination goes from output to ground. I don't understand the Zobel with removed capacitor. do you mean to keep only the inductor?
Ironically, I sold my old and extremely bulky 2x100MHz Russian oscilloscope just after I thought I fixed the amp and before the first blowout. I bought it long time ago but rarely used it. I'm planning to buy a USB based digital oscilloscope with the money. Right now I'm "blind" without it.
I'm using a 10 Ohm resistor in series with 0.1uF and that combination goes from output to ground. I don't understand the Zobel with removed capacitor. do you mean to keep only the inductor?
I mean from pin 6 to pin 2. I selected the value by simulating the circuit. The -3db on output is at about 500kHz. I believe that in reality it is much lower.
Hi Chris !
I also have 35 years experience in servicing and building of amplifiers. According to my experience if an amplifier oscillates the problem is not due to the twisted cables connecting the driver board to the output stage but rather to bad circuit and probably printed circuit design .
I don't think I recommend a bad practice to another person.....I just told my opinion and posted the picture of an amplifier with twisted cables in the output as an example. The fact is that the amplifier is working. It's up to anyone's will to follow my example.
Yes I have an oscilloscope.... two actually ...A Trio and an HP1740A which has a useful bandwidth up to 150Mhz . I use voltmeters just to measure the DC volts value. I also use audio oscillators to check the response of amplifiers.
I wonder though...could you explain how the twisted cables affect the operation of the amplifier. Added capacitance between the cables ?
Not the capacitance. It's the inductance that causes problems. An EF's (or CPF's) stability depends partially on the collector (power supply side) being at AC ground at all frequencies. That supply impedance goes too high and stability is degraded. These things won't show up in a simulation - unless you model it. PCB mounting everything without all those long twisted wires just more easily facilitates putting the proper bypass cap(s) where they are needed.
Twisting the wires (as opposed to not twisting) is generally a good thing if the pairs contain equal and opposite currents. This prevents radiation into other parts of the circuit. This issue is not a stability problem per se, but affects noise and distortion.
Twisting the wires (as opposed to not twisting) is generally a good thing if the pairs contain equal and opposite currents. This prevents radiation into other parts of the circuit. This issue is not a stability problem per se, but affects noise and distortion.
You can check for oscillation with an analog VOM with a 20 VAC and 2 VAC scales. Normally you put a .47 uf 600 v cap or something like that in series with one lead to keep DC voltage from reading on the AC scale.
If you have an AC voltage at a point, you can change the cap to 390 pf or 220 pf or something. If you still have the AC voltage, it is ultrasonic. Music won't go through such a small cap. If you hear not much music and have a big loud AC voltage, that makes one suspect the oscillation and start changing caps. I use clip leads on the caps, makes this a fast test.
Worked on my Simpson 260-6XLPM meter, the 5000 ohms/volt at AC one.
Another sign of oscillation: use rock music from a radio to exercise the amp while you are looking at AC. The beats of the drums will show up on the pointer of an analog meter. No beats, you probably have an oscillation instead of music.
Last edited:
This is exactly how I check for oscillations, analogue VOM and capacitor in series.
Hi John,
I'm really happy your amp isn't acting up. Many would misbehave with that configuration. If not outright oscillation, bursts on or near peaks. You have compensated for that with yours, however the designers of the amp in question have not. wg_ski is correct in that the inductance is the most problematic, and this is actually distributed L and C. A braid would have been the better way to dress the leads if you had to go that route, but most just run the wires in a loose grouping. If the output stage had gain, you would not find it easy to make is stable without using excessive C-B capacitance.
I think the main idea is to dress the leads as you have found them when working on a manufactured amplifier. Not unless the lead dress was a problem to begin with.
I just rebuilt an HP 1722A, but would love to restore a 1740A as well. That old scope works fine now, but it had a tangle with TIM. Technician Induced Malfunction for those unfamiliar with the term. My mainstay scopes are a Tek 2465B and an Agilent 54642D. Both lovely scopes. I have a Philips PM 3070 for back up and a Gould OS 255 for working on tube stuff, or when a scope has to go visiting. Over the years one tends to accumulate things.
-Chris
I'm really happy your amp isn't acting up. Many would misbehave with that configuration. If not outright oscillation, bursts on or near peaks. You have compensated for that with yours, however the designers of the amp in question have not. wg_ski is correct in that the inductance is the most problematic, and this is actually distributed L and C. A braid would have been the better way to dress the leads if you had to go that route, but most just run the wires in a loose grouping. If the output stage had gain, you would not find it easy to make is stable without using excessive C-B capacitance.
I think the main idea is to dress the leads as you have found them when working on a manufactured amplifier. Not unless the lead dress was a problem to begin with.
I just rebuilt an HP 1722A, but would love to restore a 1740A as well. That old scope works fine now, but it had a tangle with TIM. Technician Induced Malfunction for those unfamiliar with the term. My mainstay scopes are a Tek 2465B and an Agilent 54642D. Both lovely scopes. I have a Philips PM 3070 for back up and a Gould OS 255 for working on tube stuff, or when a scope has to go visiting. Over the years one tends to accumulate things.
-Chris
Detecting RF
Using a radio doesn't always work, but I imagine the AM band might be the best place to look. If you are looking for RF, make an RF probe. You need a crystal diode (Ge) and something around 0.01 uF with a VOM or (better) VTVM. Of course an oscilloscope is the best for finding things like this.
New Scopes. What out as many have a maximum input voltage of 50 VDC + AC pk. That's with a 10X probe! They are easy to damage. If you are fixed on something like this, the Rigol is probably your best low cost bet. Look also at Keysight and the 1000 and 2000 series. (yes, they plug into the wall). The USB scopes do tend to be a pain to use since its a lot easier to simply twist the knobs instead of clicking on things.
I have had some scopes for eval, and my USB scope died for no reason as far as I can tell. It died on power-up, worked fine the day before. But, if you ask me, an analog scope is the better bet for the work most of us do. Digital scopes are as good, but as good tends to run $20 K and up. That means I will probably always have an analog scope on the bench. One thing to really think about is aliasing. That can be a problem with low end digital scopes.
-Chris
Using a radio doesn't always work, but I imagine the AM band might be the best place to look. If you are looking for RF, make an RF probe. You need a crystal diode (Ge) and something around 0.01 uF with a VOM or (better) VTVM. Of course an oscilloscope is the best for finding things like this.
New Scopes. What out as many have a maximum input voltage of 50 VDC + AC pk. That's with a 10X probe! They are easy to damage. If you are fixed on something like this, the Rigol is probably your best low cost bet. Look also at Keysight and the 1000 and 2000 series. (yes, they plug into the wall). The USB scopes do tend to be a pain to use since its a lot easier to simply twist the knobs instead of clicking on things.
I have had some scopes for eval, and my USB scope died for no reason as far as I can tell. It died on power-up, worked fine the day before. But, if you ask me, an analog scope is the better bet for the work most of us do. Digital scopes are as good, but as good tends to run $20 K and up. That means I will probably always have an analog scope on the bench. One thing to really think about is aliasing. That can be a problem with low end digital scopes.
-Chris
Hi Chris!
You are right and I agree with you that the layout of cables may present an oscillation problem. Apparently some designs suffer.
Bias instability
Hi again,
After many hours of tweaking and testing, I can say I have an amp that works stable and produces decent sound. I finally settled around NE5534 due to higher supply voltage and generally faster turn-on which manifested as significantly reduced thumping when turning-on.
The schematics of the moded amp is attached.
The sound is also very detailed and open and when idle I can barely hear any noise.
One think that bothers me though is the bias current. I set it at around 45 mA after about 20 minutes of "warming-up". It's quite cold actually and it takes a lot of time for the Vbe multiplier to pick-up the heat from the sink and start lowering. So I have about 40% overshoot and after a while it's 45 mA.
If I play music loudly though, the output transistors warm-up normally and then when I measure the bias it's about 40% lower than what I set.
I finally set the idle current to 45 mA when hot, but that means about 70mA when cold.
I don't like this wide swinging of the bias. Do you have suggestions how to stabilize it? I read somewhere that increasing the Vbe multiplier resistors can increase bias stability.
Hi again,
After many hours of tweaking and testing, I can say I have an amp that works stable and produces decent sound. I finally settled around NE5534 due to higher supply voltage and generally faster turn-on which manifested as significantly reduced thumping when turning-on.
The schematics of the moded amp is attached.
The sound is also very detailed and open and when idle I can barely hear any noise.
One think that bothers me though is the bias current. I set it at around 45 mA after about 20 minutes of "warming-up". It's quite cold actually and it takes a lot of time for the Vbe multiplier to pick-up the heat from the sink and start lowering. So I have about 40% overshoot and after a while it's 45 mA.
If I play music loudly though, the output transistors warm-up normally and then when I measure the bias it's about 40% lower than what I set.
I finally set the idle current to 45 mA when hot, but that means about 70mA when cold.
I don't like this wide swinging of the bias. Do you have suggestions how to stabilize it? I read somewhere that increasing the Vbe multiplier resistors can increase bias stability.
A temperature feedback loop can undershoot or over shoot if there is nothing there to adjust the gain of the loop. Its because of the thermal mass of the heatsink.
Some people glue the Vbe multiplier transistor directly to an output transistor for faster feedback.
Some people glue the Vbe multiplier transistor directly to an output transistor for faster feedback.
Heat conduction through metal also is a lot faster than through plastic, so using a very short and fat trace for the T4-T8 connection may be worth a shot, too.
- Status
- Not open for further replies.