The inductance of a speaker coil is too high to get much current through it at RF. And if there is much RF the typical Zobel circuit will light off really quickly (As R . Marsh knows).
Having built a few of these lousy oscillators (unstable amps) my experience suggests its very hard to hide the instabilities if you make a modest effort to look for them. A .1 uF cap is usually a one stop test, just make sure the leads are short. And the oscillation relates to the differential gain phase issues as the change of phase at different drive levels is very closely related to what gets the system oscillating.
Having built a few of these lousy oscillators (unstable amps) my experience suggests its very hard to hide the instabilities if you make a modest effort to look for them. A .1 uF cap is usually a one stop test, just make sure the leads are short. And the oscillation relates to the differential gain phase issues as the change of phase at different drive levels is very closely related to what gets the system oscillating.
Filler comments -
Comment -- if anyone remembers analog TV and roof top antennas --- one easily got RFI from bad cable lead-in connection outdoors due to contact corrosion. The problem would appear on other TV channels from the harmonics producted at the bad contact... appearing as interference. [Find info from ARRL handbook on this common problem.]
The above article describes the distortion from a hysterisis view.... I talked about it from a non-linear Z perspective... same results. -RNM
Comment -- if anyone remembers analog TV and roof top antennas --- one easily got RFI from bad cable lead-in connection outdoors due to contact corrosion. The problem would appear on other TV channels from the harmonics producted at the bad contact... appearing as interference. [Find info from ARRL handbook on this common problem.]
The above article describes the distortion from a hysterisis view.... I talked about it from a non-linear Z perspective... same results. -RNM
Last edited:
Smoke testing of Zobal networks -
Did I hear my name? Once apon a time - Incoming inspection on new amps were done by techs who ran the power amp at full drive at freq well above audio.... the zobel network was being smoke tested.
There are similar issues of phase change with complex loads and high power which will create PIM? ??
Did I hear my name? Once apon a time - Incoming inspection on new amps were done by techs who ran the power amp at full drive at freq well above audio.... the zobel network was being smoke tested.
There are similar issues of phase change with complex loads and high power which will create PIM? ??
Mike, once again regarding transient spikes at the output during light switch turn-on. As a cure, try:
- everything in class II.
- no ground loops. Especially protective earth loops must be avoided. This is assured when components in class II.
- shielding continuity must be fulfilled, everywhere and in every component. Audiophile signal link cables where shield is connected only at one end are not allowed if you fight with EMI/RFI.
- no signal transformers outside amplifier cases.
- properly shielded circuits with high HF PSRR.
In case you keep these recommendations, you would not need to use additional mains filters.
- everything in class II.
- no ground loops. Especially protective earth loops must be avoided. This is assured when components in class II.
- shielding continuity must be fulfilled, everywhere and in every component. Audiophile signal link cables where shield is connected only at one end are not allowed if you fight with EMI/RFI.
- no signal transformers outside amplifier cases.
- properly shielded circuits with high HF PSRR.
In case you keep these recommendations, you would not need to use additional mains filters.
You won't hear the oscillations, but if the circuit is misbehaving in that fashion then the chances are that it can't be functioning as intended in the audio region; again, especially as far as the FB mechanism is concerned. It seems as if I'll have to play with LTspice, deliberately simulating non-linear behaviour in the output, to see what signal frequencies the FB area of the circuit is experiencing ...
Frank
It's interesting that some of the more awkward elements affecting correct functioning of audio systems is now being discussed; IME, every single one of these has to be knocked over, got under control - you know that you've passed the barrier of "sufficiency" of effort when the sound snaps together, and it's just, subjectively, "right". No instrument is required to measure anything, your hearing gives you the thumbs up ....
Frank
Frank
Thanks for this George. I remember this well. Like much of the Great Guru's writing, all wannabe gurus are directed to study on their knees 5x a day. When you become a guru in your old age, you will still come back to these and find new insights.
This is was advertised as a 7 part series but the last part was never written.
Part 7 was to have been further analysis of 2 pole compensation. Self shows the advantage of this in Fig 6.19 of 'Audio Power Amplifier Design Handbook' 4th ed.
Here are perhaps GG Baxandall's last thoughts on Power Amps. The Baxandall Papers 20 yrs on. I wish I wasn't beach bum and could afford to buy magazines.
I prefer to call their TMC, the Cherry capacitor. Prof Edward Cherry, Monash U, Oz, was the first to present a detailed analysis. The 'pure' version of Cherry in #2060 is key to its SOTA performance with a stone age circuit. GG Baxandall & Self propose a bastardized version which doesn't have all it's advantages.
_______________________
May I suggest you get a switched capacitance box like in school science labs.
You may find a few nF more of a problem than 0u1 or larger loads but the box lets you check this out quickly.
You need to check at different signal levels, with & without a resistive load in conjunction with the capacitance box. And with the amp cold and warmed up.
Then repeat the test with a big guitar or PA speaker instead of capacitance at frequencies between 100Hz & 1kHz.
If you want SPICE to show what you get in real life with this, you need to have real life inductance/resistance in the leads to your PSU, decoupling & earthing system too.
Try this with all the Golden Pinnae amps you can lay your hand on.
________________________
JC, we're still waiting with bated breathe for your Hirata test results on Blowtorch. If the results were not as you expected and you now think the test is rubbish, that's OK but us faithful would like to know.
I'm sure Blowtorch will do much better on the Quan test.
Last edited:
Yes Mr.Marsh, I remember.
The article is based on calculating intermodulation products (f1+f2, f1-f2) with varying delta freq and delta amplitude between two main frequencies fed in cored inductors .
The authors beg for more research on the topic with more frequencies applied simultaneously (I haven’t found any yet) as they consider their two frequency experiments which confirm theory, falling behind real life signals.
Thus they don’t go into generalized conclusions, apart from one. That is, such a distortion is a function of signal current squared.
George
Connected to the output device, of course ? I had not made measurements, but all my system is connected that way. Dead silent, no HF.
Of course, i use symmetrical cables with some coper, not thin ones.
Preamp and amp are flat up to > 2Mhz.
I filter the amp inputs after that with a low pass network at the input. (~200Khz)
11 devices are connected: Turntable, CD player, Minidisk, DAT, K7, Tape recorder, Tuner, Digital active filter, preamp, computer, amp.
I use ferrite on all connectors near outputs.
[edit] Computer use optical connections, no ground connection.
Last edited:
Only a partial help. Any shield discontinuity is a big problem for HF EMI suppression.
I'm just saying that you have not posed a problem in enough detail so there is someting to discuss, fairly anecdotal observations are not enough.
As a counter example I once built a Borbley FET follower a forgot to bypass it, it oscillated at 60MHz but still sounded fine. Even John has stated that simple circuits (especially those using high capacitance JFET's) often need better power supplies.
I see no arguments against circuit complexity, after all is a cascode complexity? A Siklai pair (quite popular here ) is well known to have stability concerns, a simple emitter follower can oscillate with a nasty load (speaker?), the list goes on.
Like everybody here, i have often experienced unwanted oscillation problems with a little help from Mr. Murphy. Even at high level.
As long as it don't drive the audio signal to clip, most of the time, it do not change the sound texture a lot, with correct audio circuits, on my experience. Well i can feel them, sometimes, by a strange feeling, something white in my brain i can't explain, discomfort.
If the sound is greatly changed i thing it is because some circuit tend to IM, not fast enough to follow the signal, while the hf is not enough filtered ?
As each configuration/circuit is different, it is hard to generalize any experience; but i don't believe in magic like some part in a loop witch 'tend to oscillate'. i believe we know all with a square wave on a fast scope.
If there is some oscillation at very high HF, you will see-it on the clipping level in the audio signal. Right ?
And if it is low level, you don't have to allow 60Mz signals to enter the first stage of your power amp. Right ?
Scott, I hope you would not mind if I post a measurement of 250MHz oscillations in a BJT preamp ...
Attachments
- Status
- Not open for further replies.