Discrete audio amplifier project PT14.1 oscillation fixed and first power test

[@1pcfred]

Congratulations on finding a solution to your problem.

[@killcar5nbike2]

I do enjoy the trial and error of R&D bug finding. Good work John.

[@raymondheath7668]

Wow, been following along and you have done a lot of good troubleshooting and offered up some reasonable good explanations

[@bennyjorgensen]

Thanks for your videos on this discrete amplifier. I have created a board with two amplifiers and one of them oscillated and using the ideas from the video helped. The two amplifiers are by the way, one for the tweeter (current feedback - 12w/8ohm) and one for the midbas (normal voltage feedback - 40w/4ohm) and then the electronic crossover and baffel correction.

[@anindyamitra5091]

I am very satisfied with the results, keep it up!

[@iblesbosuok]

Great video. So hard to explain ultrasonics oscillation due to negative feedback stability. Explaining RF oscillation even much harder. Oscillation issues make our brain so hot. Fortunately your meow cools it down back to nice-warm state. Cheers from Indonesia

[@rrangana11]

Well done... Very much enjoying watching how this project is progressing.

[@Radfordperson]

The classic Quad 303 power amp can run into oscillation problems if modern high frequency output transistors are used, In fact the original hometaxial devices are difficult to find.

[@nowt1002]

Well done on curing the oscillation. I've been designing a similar amplifier using similar speed transistors. I've not built the prototype yet so this will be very helpful when I come to build and test it. Thanks

[@bennyjorgensen]

Thanks a lot for this video. I'm constructing a small active loudspeaker and in that i'm using two amplifiers (Tweeter: 14W/8 ohm + Bas:50W/4 ohm) and i'm using 2SD2390 and 2SB1560. I get heavy oscilation when the idle current is set to more than 1mA (target is 300mA) and i totally forgot base stoppers. Thanks for your video. I now hope that I can solve my problem, as you did. Thanks.

[@joohop]

Good Work Earthling

[@MarkTillotson]

Its pretty common to put 100 ohm stoppers on the _driver_ bases, which greatly reduces the ability of the output section to work as a common-base circuit at VHF frequencies, which I think is the oscillation mode usually seen.

[@bixy9347]

Great stuff, very educational 👍

[@musicstevecom]

My Phase linear amp 400II I get a Perfect SQ wave but when the speaker relay kicks in, The speaker output going into a dummy load and a Parallel .1uf and 5 ohm resistor to ground and I get a terrible Sq Wave, I took the dummy load off (120watt ) and added a Non-inducted 8 ohm cross over resistor (only 10 watts for testing) and it helped a lot but not perfect. 1. or Going to add another capacitor to the + Bus and - Buss (+80vdc and -80vdc) currently it is .31uf to ground sense 2. Stopper resistor help? 100 ohm? I have a predriver on the control board that goes to a bootstrap output driver and a 40pf cap I added to the predive Base to collector?

[@johnsenchakinternetsecurit8935]

time @ 19:50 Snickers number one fan !!!!!!!!!!!!!!!!!!!!!!!!!!!!!

[@T2D.SteveArcs]

come on John cat brains run at 61.7Mhz everybody know that.... well done on debug happy days

[@ubergeeknz]

So @johnaudiotech in terms of the circuit... was the final piece of the puzzle some 102 (that seems small) caps from the output transistor rails (collectors) to ground? I'd be really interested to see what happens if you leave those in place but remove the stoppers

[@czmik]

You know it's bad when your oscillation is oscillating lol

[@1pcfred]

Their oscillations were modulating.

[@terencekaye9948]

Nice Work John,, Way to go. :)

[@Moonbrony]

Is your bench power supply a linear or switch mode one? If it's switch mode, then the high frequency noise could be coming from that. In which case you wouldn't need the base resistors or the extra decoupling caps. Maybe build up the proper linear power supply and test it either way?

[@johnyang799]

It's oscillation not the noise from power supply.

[@ferrumignis]

It's not noise, it's the emitter follower output stage oscillating. This is a well known phenomenon with emitter followers, all it takes is a bit of inductive reactance in the collector circuit and you form a resonant circuit with the intrinsic base collector capacitance. A resistor in the base connection adds damping to this resonant circuit. As John says, a good pcb layout with a ground plane and local decoupling at the collectors will fix this.

[@johnyang799]

Plus, you can see the brand and model of the power supply. Mind spending 30 seconds searching up on Google?

[@pliedtka]

Did you do SPICE on let say LT Spice. You can add stopper resistors on drivers and also on power trannies. Sometimes stopper parallel with small caps or ferrite inductors may help, some Harman amps used the ferrites on bases. Also depending on your bias the transistors might start to oscillate. Usually Darlington triple is much harder to stabilise with fast transistors. I put 100n, 1uF parallel with resistor load - if it doesn't oscillate, it OK.

[@jazbell7]

That problem is called squegging, reference: en.wikipedia.org/wiki/Squegging

[@JohnAudioTech]

Thanks James. I learned something new.

[@chrisdado]

Every amplifier circuit secretly wants to be an oscillator. An RC snubber might tame the oscillation but as you say, it defeats the purpose of fast output transistors. My advice is to bite the bullet and experiment with trace layout on DIY etched PCB's. Also check if that 40MHz is due to ringing (points finger at ceramic caps and their location). If you want max' performance from your design, snubbers are not the answer.

[@boonedockjourneyman7979]

Fantastic. If you use a classic signal tracer can you localize the oscillation? In the last video you indicate that the oscillation was everywhere in the circuit. I am curious how that works. It may be both obvious and simple to you but it's not to many of us. I can see the global feedback but I also see a feedback mechanism through the rails. Audio circuits are not all that clear to digital people.

[@JohnAudioTech]

The feedback will carry the signal back to the input so that it can appear at places in the amplifier where it wasn't created. It shows up on the rails to some extent as well.

[@montynorth3009]

Did you try connecting the caps between collector and base of the O/P transistors as I would be interested to know if this would have worked? Also,will the series resistors on the bases reduce the O/P impedance and therefore damping factor of the amp?

[@JohnAudioTech]

I believe that it would have helped because when I added the 100pf caps across the collector to base of the drivers it had no effect. It had to be the output transistors oscillating. Supply bypassing was the biggest problem with this amp.

[@jeffm2787]

My first thought was that the tabs are capacitively coupling via the mica washers and the heat sink.

[@n.shiina8798]

nice try, John!. I honestly didn't think of decoupling caps on my design. It might be my solution too!. how did you determine the base stopper value?. I know the value for BJT is around 2.2~10Ohms but I always use 3.9Ohms for silly reason (i was thinking lower value should be better for speed)

[@JohnAudioTech]

Without getting into deep explanation, the R stopper value should be kept as low as possible to take care of the oscillation as its presence in the circuit can have an adverse effect on performance. It is important to take care of other issues first such as supply bypass and layout before resorting to stopper resistors.

[@n.shiina8798]

@@JohnAudioTech thank you for the reply!. it is my first PCB layout and indeed, i didn't place supply bypass near the output stages but instead, directly on the main supply caps (0.47uF film caps. i'm silly enough to put them on main caps instead of PCB)

[@1959Berre]

If I remember well Xraytonyb discussed similar problems of oscillation in one of his latest video's when he was refurbishing a Carver amp featuring high speed output transistors.

[@DAVIDGREGORYKERR]

What about a 50hz notch filter on the input and what about a band-stop filter on feedback loop.

[@johnnytheangel1]

good work m8, but getting of track a little I have an amp with leaking caps 50v 6800uf 30mm x 50mm, I have two on hand but they are only 30mm x 35mm in size would they be ok to use or do I have to use the correct size caps,

[@JohnAudioTech]

I would say you are okay to use them. Newer caps tend to be smaller in size. The caps must be rated to handle the ripple current. For audio amplifier use, I doubt there would be an issue.

[@johnnytheangel1]

thanks m8, looking forward to yr next vid

[@johnnytheangel1]

well that didn't work out, noticed the smaller caps were only 85C instead of 105 C, looks like its an ebay job, suppose 80v10000uf 85C for audio would be out of the question (got a doz of those lying around)

[@nigelwright7557]

I have designed amps for about 30 years. Oscillation is either Vas capacitor too small or outputs need base stopper.

[@38911bytefree]

Vas cap is the miller cap from collector to base on the predrivers. I have a vintage amp and it is pretty much factory, I cant solve the oscillation. I got better grounding all over and somehow improved but I have a 4Mhz osc, that goes up when I turn the volume up (no signal). And once the amp want to do its job with signal, the oscillation is all over the test tone, sounding extremely harsh ..... It is a tipical TIP41-42 with RCA 01C16/17 predrivers in TO-39. It is an split power supply design, no oput cap. Rail are pretty contmited with oscillation ... but is more a a consequence of the amp itself. Where I should start looking. I got really lost scoping. Noise is everywhere ... due to feedback. Any trick I can use ?. Thanks

[@fredfabris7187]

I think snickers left the room at the moment you connected the capacitors supporting the brain wave theory

[@twotone3070]

Job done, almost. :)

[@Kenshindegozaru]

Hi! Amazing project. I have a question: is a good idea to ground the heatsink if it will have just one IC? Im making a circuit for the LM3875.

[@JohnAudioTech]

I would use an isolating washer and screw so that you can ground the heat sink.

[@Kenshindegozaru]

@@JohnAudioTech I'm not sure, but I think the tab is connected to ground. In that case, isolation is necesary anyway? My concern is the heat. I have seen that, screwed directly to the heatsink, improves heat dissipation.

[@JohnAudioTech]

@@Kenshindegozaru The tab is connected to pin 3 which is the negative supply voltage when the IC is used in its normal dual supply configuration. The mica insulator and insulating shouldered washer for the screw is good enough provided the heat sink is of proper size.

[@Kenshindegozaru]

@@JohnAudioTech thank you very much!

[@jimhough6233]

Yeah we knew you'd track it down! I know you will fix the oscillation problem too. However, is it true that a 30KHz oscillation on an audio amp will kill the outputs or other parts?

[@deerdave]

Hi John ...just wondered if you have thought of using ferrite beads to eliminate the high frequency oscillation ?

[@musicstevecom]

Phase Linear did it w the original slow Output Transistors But seem not to work for the New Faster Transistor Good Luck

[@MrDunk66]

John Those LM1875 boards have no pots for bias adjustment. When two are paired up for a stereo amp, is additional circuitry required?

[@n.shiina8798]

LM1875 is a single channel amplifier. using two of them for stereo is just like making a dual mono module, nothing else required. if you want to parallel them, using 2 per channel, you'll need a small value resistor to each output legs before the output got paralleled for current balancing. if you want to bridge them, you'll need to invert the signal on one of the chip.

[@MrDunk66]

Just want to use one per channel in a DIY stereo amp. Thanks

[@n.shiina8798]

@@MrDunk66 Just hook them straight up, then. you'll be fine. good luck!

[@JohnAudioTech]

S. Kojina adequately answered your other question. As for the bias pot, bias is internally set within the IC. There nothing to adjust so no reason to worry about it.

[@MrDunk66]

Thanks, I am fixated on the LM1875 at the moment (building an amp) and every time I see a a similar package I think it is a chip amp. Yours is made of discrete components - - I need to wake up 🤪

[@NotMarkKnopfler]

Well done! When can I buy a kit? 😆

[@aljivenalejo3904]

nah theres a schematic for that online

[@NotMarkKnopfler]

@@aljivenalejo3904 be nice to have a PCB though

[@aljivenalejo3904]

seriusly?? man if i want to tell you i gave you some pcb will you trade me some cash??

[@SuburbanDon]

Great news! Cancel my order cancellation !!!

[@JohnAudioTech]

As I find more problems with this design, you will have to cancel your canceled cancellation!

[@aljivenalejo3904]

so meaning to say the preboosted?? or the main output stage??

[@vicentecortez8636]

Great Job??.... Wil be waiting for the final outcome!! Hey Jhon, could you help me!. I'M looking for a diagram for a PerformancePteknique ICBM-1200.0. I bougt this broken amplifier but ir came whith no out put transistors! 😖

[@steveswan5714]

Again great work and very interesting John 👍 say hello to Snickers for me he is like me wee cat Thomas I had many years ago he was tiny for a tom but sadly he got poisoned by some moron 😢can't wait for the next instalment 😬

[@T2D.SteveArcs]

what does transition frequency mean please

[@HillsWorkbench]

Kind of the upper limit for the transistor, frequency where gain has fallen down to unity.

[@T2D.SteveArcs]

@@HillsWorkbench thanks for that

[@ruikazane5123]

Pulsing waveform means two frequencies playing at the same time, like me using 40Hz and 30Hz together for a wave bass sample. When you added the resistors, you killed one of those frequencies. Add the capacitors and you dampened everything to millivolts! But I was thinking that your power supply may be producing the oscillation due to the fact they are switching supply, but that is quite impossible if you ask me because it is a perfect sinusoidal wave. And proper bypassing is always the solution. What if you place the power input nearer the power transistors, and kill the supply instability from there!

[@JohnAudioTech]

The supply I use is a linear type. But you are right, bypassing the supply near the outputs cleared up the problem.

[@ruikazane5123]

JohnAudioTech Be careful of setting the parts as close as you may be able to. A star-point ground works better for improved ground impedance. The close loop part of the amplifier must always be isolated from others, learned that the hard way. If your final design would employ a double-layer board, make sure the audio traces stay near to ground, and not any other power potential. Just saying. 02ルイ

[@JohnAudioTech]

@@ruikazane5123 Yes, Each ground lead has its own connection to the ground around the bypass capacitors and each block of the amp has its own supply leads off the bypass bank. It turns out I just had too long of supply leads running out to the output stage.

[@ruikazane5123]

JohnAudioTech Sounds about right. It can also help seperating the preamlifier stage's power supply connections and the power stage connections as well. And you may then try connecting the two together. Inductance itself is created with frequency, causing oscillations which are typical in Class B/AB amplifiers, and hazardous on pure class A topology amplifiers...I killed my amp because of that!

[@diedraak1]

How about limiting the amp bandwidth by adding a low pF cap across the feedback resistor.? You don't need anything above 50KHz. -3dB point will still be above 20khz.

[@kennmossman8701]

output transistor tabs???? do you mean the Collector?

[@johnyang799]

Wtf is wrong with you? Your son died?

[@joohop]

Best Amplifier Ever Designed John ?

[@anindyamitra5091]

I wonder, what would happen if the 220Ω speedup resistor was reduced to 33Ω.

[@JohnAudioTech]

That would reduce the current gain of the output stage as it shunts more current from the base of the output transistor.

[@anindyamitra5091]

@@JohnAudioTech , got it! Thank you!

[@jonathanstott3153]

Hmm - those oscillating pulses look just like a self quenching super-regenerative radio receiver.

[@kennmossman8701]

little transistors? do you mean the small signal transistors? So Why are they hot???

[@johnyang799]

package size

[@kennmossman8701]

@@johnyang799 Which is?

[@ferrumignis]

@@kennmossman8701 Use your own eyes. Apparently you think you know everything so you should be able to work it out.

[@kennmossman8701]

@@ferrumignis Hahahaha Great answer - it shows how ignorant you are!

[@kennmossman8701]

That you don;t know is obvious. Just a bunch or imbeciles sitting around in a circle jerking each other off saying how clever you are.

[@anindyamitra5091]

I want you to conduct a last experiment before you wrap up everything and proceed to make the PCB. The experiment is to run the amplifier to by 100Hz sine wave tone into clipping(I mean hard clipping, so that the 100Hz sine wave appears like a square wave) on ±36VDC(O-L voltage, yeah current limit will sag the voltage which is ok) with a 100mA current limit and while connecting a 4Ω load at its output, and observing the output on the oscilloscope. Please do the above experiment, it is simple, it won't destroy the amplifier, neither it will take much time but it might lead to a conclusion, and I want to know that! I hope that you will do the experiment. Thank you in advance for your time!

[@JohnAudioTech]

I can use square waves for the fast rising edge if that's what you need to see. Do you mean that you want me to set the current limit on my supply to 100ma? What are are you checking for?

[@anindyamitra5091]

@@JohnAudioTech , I want to check for the effect on the amplifier due to hard clipping. Or, I must clear things up a little bit and say that, I want to see the effect on the amplifier due to highly unequal supply rails.

[@JohnAudioTech]

I will be performing square wave tests (step response) soon. I can adjust the input signal to drive the amplifier into clipping to see if there is any stability issues there. If the supply rails have unequal voltage. The output will clip on the rail with the lower voltage first. This should not cause output offset voltage problems.

[@anindyamitra5091]

@@JohnAudioTech , excellent test! And you can further reduce one of the rail voltages, like you can say, for example, to supply the amplifier with +ve rail having +36V or more and -ve rail having -9V or less, at 100mA current limit. And see there, if you can simulate any stability problems or not.

[@recenzo8800]

47p//22k FB

[@mrjohhhnnnyyy5797]

I was thinking the same, that would slow down the whole amp. High frequency gain rolloff

[@johnyang799]

That won't do any good.

[@antigen4]

those caps are only filtering out the signal. I would rather find the SOURCE of the oscillation and fix it there.

[@kennmossman8701]

How does one 'slow' transistors? Very odd turn of phrase. Do you actually mean reduce the bandwidth???

[@HillsWorkbench]

A small capacitor, C to B, or stopper resistors as John described.

[@kennmossman8701]

@@HillsWorkbench stopper resistors? wtf did you get all this freaking bent terminology

[@n.shiina8798]

it's actually a simplified terminology. it's easier to understand for non-engineer peoples i guess?.

[@n.shiina8798]

@Hill's Workbench stopper resistor is actually base resistor. what i know, this resistor acts as damper for the previous stage that might be overloaded and causing instability. i don't know the technical explaination, though.

[@kennmossman8701]

@@n.shiina8798 so a brain is a 'thinker', fingers are wriggly things, a zener diode is a voltage capper? bloody ridiculous

[@jimhough6233]

30MHz!

[@kennmossman8701]

78 Watts? .........and that is RMS?

[@HillsWorkbench]

Watts RMS, JAT's favorite term! :) (He did a rant on that one...)

[@johnyang799]

78 Watts with sinusoidal signal. This is the proper term.

[@kennmossman8701]

@@johnyang799 There is peak, average, RMS, etc. To say only Watts without qualification is BULL-SH^T.

[@johnyang799]

@@kennmossman8701 Lol. Vrms is shown clearly on the scope and the load is 4ohm So what? Can't you calculate for yourself?

[@johnyang799]

@@kennmossman8701 There is no RMS for power. Average it is if you wish. But square wave and dc can also shown as average power so it's simply useless. Another way to effectively imply the condition is to use 1% thd.

[@khuif]

This looks like am radio modulation

[@ubergeeknz]

It is exactly that. A high frequency waveform modulated by a lower frequency.