L12/2 Version 5 Amplifier. First impressions.

[@MichaelBeeny]

THANKYOU TO EVERYONE WITH SUGGESTIONS ON HOW TO FIX THE PROBLEM. Once we have the new circuit diagram worked out, (still waiting for the kit of parts) fault finding should be relatively easy. The real purpose of THIS video was just to look and show what YOU would get if you purchased the amplifier. It does look like this is NOT just an upgrade but a very different L12 indeed!! We shall see!

[@cobar5342]

This is truly interesting. By the way.is 'LJM' a person or organisation?

[@MichaelBeeny]

@@cobar5342 He is a person. I have no idea what, if anything is behind him.

[@x5-acousticguitarstuff.2]

When I was 14 years old, I made my first "Integrated Amplifier" using Kits from Dick Smith and Jaycar Electronics. 1: ETI 480 Power Amplifier 2: Preamplifier with Tone Controls 3: ETI Power Supply Kit +/- 39Volts. 4: Large Instrument Case. 5: Power Transformer 6: Mains Lead and Power Switch 7: Black Letter Transfer Kit. I am 60 Years old and I still love to Study Electronics almost everyday.

[@MichaelBeeny]

Only 60! still a baby lol

[@x5-acousticguitarstuff.2]

@@MichaelBeeny 61 next week. Yesterday I lifted a 460 Pound 900cc Road Bike that fell in a Hole next to the Road. A strong Mind creates a Strong Body.

[@Poult100]

I think I remember that kit! I'm 64

[@Poult100]

Did the preamp use an op amp lm380, if I remember rightly. Noisy thing!

[@duskobikic8180]

Good to see you in regular business Michael. I wish you lots of luck and good health.

[@MichaelBeeny]

Thank you, most appreciated.

[@trevorjohnson2315]

Hey Mr. Beeny, it’s good to see you back! I got interested in making a biamp’ed system for a couple sets of rotating speakers for guitar when you had just announced you were “signing off” KZbin. So fun to find myself in the middle of all the investigation and development of these affordable little amps - with the guy who got me started, nonetheless! Cheers from Ohio, USA!

[@ats89117]

Not just Heathkit. Just about everything including equipment such as test instruments from HP and Tektronics came with a schematic inside the case fifty years ago...

[@MichaelBeeny]

Very true, sadly those days of good customer service are mostly long gone. A distant memory. Much of my early test gear was HealthKit, all built from kits and always worked first time.

[@ClausDreisig]

Hi Michael. It look very much like paracitic oscillation and not loop instabillity. Have seen this many times in amplifier design. Also in my own builds, I would use a small pickup coil (five winding hooked up to the scope probe) and search on the board were there problems arise. My guess is around the driver or output transistors. Sometimes the issue can be fixed by a small base resistor on the driver or output transistor. Sometimes a capacitor in series with a small resistor mounted from the base to the collector on the culprit can cure it. Claus from Denmark

[@klaushollinetz7494]

This effect will probably go away with a small capacitor (around 100p) between the base and the collector of the VAS transistor, if missing. The other L12/2 amp boards I have, do in fact incorporate such a capacitor. Schematic is available at Calvins Audio Pages.

[@ronschauer839]

Yes, that was one of my first thoughts as well. When I receive the kits I will look into it.

[@MichaelBeeny]

@@ronschauer839 Sadly not the problem, changing the value changes the shape but does not stop it.

[@ryanchappell5962]

I agree, if you buy a kit, the diagram should be available.

[@MichaelBeeny]

....and it will be soon.

[@MichaelBeeny]

Too expensive to landfill, I WILL get it going sooner or later.

[@ryanchappell5962]

How bout that! You just saved some people from buying an unstable amplifier! Well done sir.

[@cyrilmarchand237]

Hello Mister Beeny. I am glad to see you in good shape. I await your analysis of the cards with as much impatience as for an episode of Star Wars 😅 See you soon

[@MichaelBeeny]

Wow Cyril, not sure I can compete with Star Wars! but a very kind thought however.

[@brianconyers6655]

Hi! Michael, It's good to see you back again. I was very disappointed when you announced that you were packing it in due to poor health. It's great to see you looking well and cheerful once again. Kindest regards, Brian

[@MichaelBeeny]

Thank you, Brian, for your kind words. Clearly, I'm not cured but I have good days and not so good weeks! I do what I can, when I can. Good to see you here Brian.

[@brianconyers6655]

Dear Michael, Season's greetings to you and the family. I'm looking forward to next episode of the L12-2 amplifier, I have bought the kit but I won't commence with constructing it until I get some positive info from you. Happy and healthy 2025 to you and all your followers. Kindest regards, Brian.

[@stevenellerbruch6023]

Anxiously waiting for your follow up video

[@MichaelBeeny]

Working on it right now, won't be long.

[@claudiusdan6165]

Hello, Nice videos, thanks! Regarding the issues you pointed out: - Oscillation: based on the V4 schematic the supply decoupling capacitances are only 100u elcos. You use rather thin and long wires so, in my opinion, you should add some at least 1000u elcos and 100n ceramics. This will make true the power supply AC short circuit assumption. - Setting the power stage idle current may be independent of power supply voltage depending on the way the driver stage is biased. If the driver uses constant current the output actually copies it in a multiplied version. In V4 the driver is biased from VBE/R current generator that is supply voltage independent. It also has the advantage that the current decreases with temperature. - I didn’t make any calculations nor any simulation but maybe 15V supply is too low… I’m eagerly waiting for the updated schematic to check my assumptions. Thanks again, Claus

[@Creeptonic]

It so nice to see you kind Sir during this season of Advent ! Sending you wishes for your good health .

[@MichaelBeeny]

Thank YOU, much appreciated.

[@ronschauer839]

Stay tuned... 😁

[@thegrowl8874]

shout out from Philippines!!! advance merry Christmas sir mike...

[@MichaelBeeny]

I hope you in the Philippines and my viewers from all over the world also have a good Christmas and everyone LIKES my video. That's all I need for Christmas!

[@stuartwadsley]

Any further updates/fixes for this?

[@MichaelBeeny]

@@stuartwadsley Working on it right now, be patent, we have to get it right. We are making progress.

[@fredcailloux2526]

Hello Mr Beeny! You look healthy! I'm an early buyer and purchase a kit few days after his announcement by LJM on DIY forum, for now I just test the transistor they supply using a PeaK DCA 75 Pro curve tracer and I'm not happy with the result at all. Before I attempt to replace all of them with better one, I will wait for you and your friend Ron conclusion. Let's hope you both save the day "again" but my confidence in some LJM resellers is pretty low for now.

[@MichaelBeeny]

We will do our best for everyone. At present we are still waiting for delivery of the kit version.

[@peterlarkin762]

First thing I'd try is some 100nf caps decoupling both power rails on the board. And it looks like there aren't any miller compensation caps on the outputs. Not always needed but maybe in this case.

[@ronschauer839]

I always add ceramic decoupling caps across power rail caps as a general practice. Unfortunately this looks like something else entirely, but we will see...

[@Obsfucation]

Happy Holidays to you, Michael. I do have to respectfully disagree about the bias adjustment being set at the factory being not ideal for various rail voltages or just in general. This is not an issue for two reasons and can be good for one reason: First, the quiescent current set in a typical Lin topology amplifier is based on the voltage set across the VBE multiplier to provide the series base to emitter voltages required to turn on the output devices (or the drivers if it's a Sziklai output). This is unaffected by the rail voltages and the bias current is only defined by the above. Second, the basic operation of a BJT makes it largely unaffected by the rails since the output characteristics show a pretty constant current output. Again, the voltage across the CE doesn't change the current set by the BE voltage. The good reason is that the factory (hopefully and probably) has the bias set reasonably well and that allows a DIY person to avoid that adjustment being required to get good results, i.e. low distortion. Most folks don't have the test equipment and/or the experience to set the bias for the sweet spot of low distortion but, reasonably low output idle current to minimize dissipation. I believe that LJM made the correct design choice.

[@mfr58]

I agree, although it is nice to have the option of setting the bias yourself.

[@lohikarhu734]

Gosh, not being clued in about "audiophile" stuff, I have no idea what an L12 is .. I have L3/4/5 back surgery, but L12, new to me! I dropped back to this video after a quick look at the latest.

[@cobar5342]

Great to see Michael. I have just finished the version 4 (after seeing your video) with known name capacitors and 1% resistors and about to complete the construction. It will be interesting to see how V5 turns out. From across the Tasman

[@MichaelBeeny]

Me also, I really want this to be good, preferably better than the original version. If it's not better, what's the point?

[@MichaelBeeny]

So far you have the best version, the version 5 is a nightmare, but we are working on it still.

[@duskobikic8180]

I have identified exactly the same problem of oscillation at the positive side of the waveform in L20SE. LJM didn't bother addressing the problem. He simply switched to Chinglish, pretending the problem doesn't exist. All LJM cares is to keep his product sales up and to keep his customers in dark. All my LJM products ended up in the bin.

[@ronschauer839]

I suspect that you are right about LJM and have observed what you referred to (denial) on the DIYAudio forum. I made a number of changes to the L12/2v4 boards I have used in my stereo for a couple of years and do like them. However, as originally supplied they were, frankly, not all that good. One of the changes was to add Mike's (and Calvin Audio's) QI adjustment, which is a MUST-do change. However I have also reduced the overall gain (it was FAR too high), changed most of the components at the input, and so forth. This is the primary reason I buy kits rather than finished boards, and always trace out the schematic from the bare board even in the odd occasion when a "representative" one is supplied. As Mike says, with these sorts of kits, you are pretty much on your own.

[@laurynaslauris9677]

Hi. Mr. Beeny. You should ad ceramic capasitors 100nf across the power rails as close as posible to power transistors. That should solve your problem. Good luck!

[@MichaelBeeny]

I have tried this, made zero difference. We have the circuit now and will publish it very soon. Still have the problems. I will publish it very soon on KZbin.

[@aidysas]

Very informative video 👍👍👍👍👍

[@calvinaudio8141]

Ji, nice to see You up an running again. 👍 Take care and merry christmas, Calvin

[@MichaelBeeny]

Thank you, Calvin, good to see you here also.

[@calvinaudio8141]

@@MichaelBeeny + Hi, the layout of the amp has obviously changed considerably. The common cooling area for the biasing generator and (pre-?)drivers , which are so typical for Sziklai-output stages, have seemingly be split up in two independent parts. I wonder if the bias generator has been changed to a 2-transistor design also, where each one of them is thermally attached to a corresponding driver transistor? Could you already test for Bias stability?

[@MichaelBeeny]

@@calvinaudio8141 Hi Calvin, The follow video should be issued in the next week or two. A few surprises are coming. I for one cannot wait to see it. lol For example, when is a FET not a FET??

[@Shanachie707]

It might be set at a Voltage +/-50V as in the data sheet power output is shown at this setting

[@beamer.electronics]

Interesting, it could be: Supply decoupling, Miller effect issues and/or circuit layout! All the best.

[@larrygaines7462]

My tpa3255s monoblocks @ 50vdc into 8 ohm . Running 2years continues to perform great

[@Poult100]

Interesting fault 🤔 Looks like ringing. It reminds me of a power amp a friend and I built which, we realised was also a transmitter, when we turned a MW radio on in another room and heard the music we were testing our amplifier with! We reckoned that the speaker cables were ringing and it was cured by putting an inductance in series with the output. It was a low value which didn't affect the audio frequency range, but detuned the cables. Worth trying to see if it cures your instability. If it does, I'd try changing the upper output transistor first, seeing as it is the positive half wave which is affected on your amp. Good luck!

[@MichaelBeeny]

No, the case here, the circuit has so far only seen a resistive dummy load. It's instability within the amplifiers itself, not dependent on inductance or load. Watch this space.

[@chrisharper2658]

I've got a home brew Hafler like amplifier I built up around 1990 and have maintained and upgraded it over the years. At one time I had the same oscillation that would rear it's ugly head in the form of sibilance at moderately high levels. I believe I had to add capacitance somewhere around the output FETs but have since changed out those driver PCBs with a different design. Interesting that your finding it. I'm not familiar with the L12-2 design but I'm curious now.

[@chazmo77]

I assembled Heathkit and Dynaco. I loved those!

[@MichaelBeeny]

Long lost days for sure.

[@mfr58]

I suggest you check the output inductor to make sure it's not a ferrite core version that they quite often use on these Chinese modules.

[@MichaelBeeny]

This amplifier out of the box, does not have any inductors on the output of any kind. I make no comment at this stage!

[@bloothedog4443]

Good to see you!

[@ianhaylock7409]

As you have the circuit diagram, you could always enter it into Microcap. That will enable you to easily change components and analyse the difference they make without having to touch a soldering iron.

[@MichaelBeeny]

We don't yet have the circuit. As explained in the video, we are waiting for the kit version to arrive. It's easier to reverse engineer a blank PCB than one already assembled.

[@ianhaylock7409]

@@MichaelBeeny LOL Shows how well I listen these days.

[@MichaelBeeny]

@@ianhaylock7409 I never even thought for one moment that could be true Ian. lol

[@ronschauer839]

I prefer LTSpice for simulations, and traditional CAD for formal schematics. Unfortunately we do not have the schematic quite yet, but stay tuned...

[@larrygaines7462]

I remember kits from layfayette, heath , hafler and learned a lot. Then tv and speaker technology.

[@Чемы]

Это очень интересно! Надеемся на хороший результат!

[@Thewatchinglad]

Thank you.

[@kenielsen]

What happens if 100nf is added to each rail close to the outputs? As others have posted, I would start probing the outputs and working backwards. Good video

[@MichaelBeeny]

There are no direct big electrolytics across the supply, this will be explained in the next video, coming soon.

[@sc0or]

Greetings. I wonder aren't buffer capacitors to small for its power? An idle current stabilization transistor is also under a question: why doesn't it touch a heatsink?

[@MichaelBeeny]

We will know the answer as soon as we are able to reverse engineer the circuit. However, on face value the decoupling caps do look a little small.

[@silvano2010gomes]

Zobel Boucherot network filter is missing from the output

[@MichaelBeeny]

Yes, it is missing. I will add one later. Not the cause of this problem as I am only feeding into a resistive load. No inductance at all.

[@reverend11-dmeow89]

Thank you for visual confirmation the two 30V 10A power supplies I chose hook positive on one to negative on the other as Ground for my +-15V dual supply analog audio projects. Does Negative on the 3.3V & 5V supplies hook to the same or kept separate for mixed signal circuits for Audio with MIDI Digital power supplies? I ask as before I built my own with 78xx & 79xx supplies. And now there's more I need to understand Digital vs. Analog on PCBs I need to learn. Thank You, Sensai I am all cornfussed re. Equipment Chassis, Circuit, Analog, Digitalx, and Line Grounds. ARRRRRRGH

[@MichaelBeeny]

The extra supplies tend not to be able to link to other voltages. They all differ from one and another.

[@reverend11-dmeow89]

@MichaelBeeny Thank You, Master, it is Ground that concerns me, rules for Breadboard vs. PCB mostly, realizing Audio vs. Data Ground in Mixed Signals circuits is a different sort from Mains vs. Chassis vs. Circuit (any I missed) Grounding. I learned a harsh lesson repairing a Tapco mixer that had a very creatively backwards bent DIP IC upside down underneath the PCB , which my infinite wisdom said, "Hey, let's Really get this grounded," and proceeded to hook Chassis Ground Directly to the plus/minus 15 Volt Circuit Ground. Now this 40 year break in between then and now sidetracked away, so it's super important I get the Whole Holistic Grounding perspective before I design ANY PCB, ever. Tapco now Mackie was nice enough to provide schematic for it. Even when they told me it had to have been bastardized from parts Only accessible from an employee there, since the chassis paint job never hit the street. The second the soon to be ex-husband of my sister's band plugged it into way more power amp rhan I had, which was instantly depowered before the Chassis hum blew some speakers. Yeah, he was the Bass player with wandering pants.

[@ClausDreisig]

Just to wrap up. If you have the schematic, it would be a good idea to simulate the open loop gain and phase margin. This is fairly easy done in LTSpice. You can send me the schematic and I can do ii for you. Think there was a comment on the miller compensation that was missing?

[@MichaelBeeny]

Next video out soon complete with full circuit and a few surprises. Sadly, not good ones.

[@ConorHanley]

Oh dear, presumably they tested the new 'improved' L12/2 before building to sell. You'd hope/think so but that's rather an issue if its not jut your pair being faulty though both having the same problem rather discounts anything but faulty by design or very poor manufacturing quality control. Good thing we have you to inform the world ⛔

[@reverend11-dmeow89]

Heathkit was a huge step up for me with their Intersil-based multimeter kit and Analog & Digital Electronics (separate packages) training kits. Two weeks hernia repair time-off gave me TMS9900 college training package and HP41C with electronics plugin was blast-off. Then damned DOS then Windows support & test in Micro$oft jobs took over to pay bills. It has taken over a decade to undo the damage from the latter. Now I have to fight over scam parts. Do you have any tips for testing transistor types and their friends for actual functionality compared to alleged datasheets that are passed off as theirs, what are the important parameters for matching pairs used in these sorts of circuits? At 69, i feel 17 years old all over again coming home to my intermediate-level electronics workbench.

[@reverend11-dmeow89]

I couldn't or didnt catch, Linear or Switched-Mode power supplies for this testing? My brain got really intrigued when the upper half shows the issue while not in the lower. Also, what is actual upper peak V vs. lower peak V? I can't read the scope on my phone. ;-]

[@MichaelBeeny]

My test power supplies are SMPS but will eventually be powered with a linear supply.

[@reverend11-dmeow89]

@@MichaelBeeny ah, a Clue, perhaps, with no backing to say yay or nay at the moment. It was the totally unrelated title of the first post about the same board in my search for reasons why that effect would occur, sir. I am wondering if 0V isn't really balanced, causing the positive V pushing something into oscillations I can see the more power being output, it is oscillating somewhere, but I don't see its actual schematic, the whole point. ;-) Title unrelated to most of its content.

[@Douglas_Blake]

@@MichaelBeeny I was thinking the same thing .... that positive half cycle "fuzz" looks a lot like current limiting from an SMPS. Simple test ... swap power supplies and see if the problem moves to the negative half cycle.

[@MichaelBeeny]

@@Douglas_Blake Those power units can supply 10 amps. If they go into current limiting a red LED comes on and the voltage collapses. With a 45 volt supply, I was only running 15 watts into 8 ohms. A million miles from 10amps, and of course it was a single amplifier only being driven. I can understand why you would consider this point, however. I will double check tomorrow. Nothing can be ruled out .

[@Douglas_Blake]

@@MichaelBeeny I agree... it's kind of out there. But it wouldn't be the first time I've forgotten to release the current limiting on my power supply and spent half of my lunch hour puzzling it out.

[@jameswarren1831]

How much is shipping from the Diyaudio store in the states? It might be nice to have some guaranteed stuff to build.

[@cobar5342]

There is no need for DIYAudio in this case. Shipping from Aliexpress is very reasonable

[@MichaelBeeny]

@@cobar5342 Post free (well, included) in some cases.

[@jameswarren1831]

@ The point is getting a quality product that works, free shipping doesn’t repair the defect.

[@MichaelBeeny]

@@jameswarren1831 Very true James, both Ron and I are working on a solution right now. I hope it will not be long before we have a solid solution. The boards we were waiting for have arrived today, that's a good start!

[@electrovoltmce]

but is there any product that is able to output 100 w / 8 ohm completely clean?

[@MichaelBeeny]

Any amplifier that is unstable has been poorly designed. It's as simple as that. Time will tell if it's PCB layout or incorrect choice of components.

[@Buzz_Purr]

Looking at the schematics of L12-2 V.4, it is probably prone to the same problem as V.5 Q17 and Q19 are meant as a diode clamp. But Q19 never turns on, because of the saturation voltage of Q13. When Q13 goes into saturation, Q12 turns off completely. Ideally, the collector voltage of Q13 should not go (far) below its base voltage. ======= Another problem: The base of Q3 is not decoupled, causing a common mode signal in the input stage. The result is a positive feedback loop when the output of the amplifier is high. This problem will be much worse when Q12 turns off, and Q4 goes into saturation! If that's the cause of the problem, the oscillation should be measurable on the base of Q3. To fix the clamp, an extra diode is probably needed to lift up the voltage of Q19.

[@MichaelBeeny]

This version has little in common to the V4. We will be publishing the circuit for V5 very soon on this channel.

[@Buzz_Purr]

@@MichaelBeeny Thank you very much, I'm looking forward to it.

[@ronschauer839]

For what it may be worth, I have now received, constructed and tested this version 5 of "L12" board. And I have provided Mike with an accurate schematic of it including an annotated component side view (with component numbers). There are a few surprises in the schematic, which Mike will share, I am sure. **Unfortunately** (or perhaps fortunately, depending upon how you wish to look at it) I have been able to replicate *ALL* of Mike's results independently. The problems are quite real. This version of the L12 series is simply a bad one, IMHO. Changing the CDOM capacitor makes the parasitic oscillations worse regardless of whether its value is increased or decreased... It merely moves the "fuzz" around in the waveform a little. Adding ceramic bypass capacitors in various places made no difference whatsoever. There are NO supply rail capacitors present at the V+/V- inputs by the way, despite appearances. The four 100uF capacitors and additional transistors that are fitted on the board are for LJM's "active" filtering. They merely replace a pair of 68 ohm resistors on the ver4 and older boards (why???). I have discovered that bypassing the transistors completely (jumpering them out) does in fact make things much more stable. However more testing is necessary before any victories can be claimed. Raising or lowering the QI setting doesn't make enough of a difference to the instabilities to even mention, other than that the "recommended" 30ma is much too low. I had to raise it to 43ma to completely eliminate crossover distortion artifacts. However, as Mike discovered a day or so later, the QI adjustment is *highly* unstable..! I have an idea or two as to how to tame it down, but in the end this may not matter because of all the other problems. My take on it- *In its "as-shipped" form it is a complete waste of money and time.* However we will keep working with them to see what happens. There may yet be hope.

[@jorgemellooliveira9611]

Ótimo vídeo esta certo OK

[@thegrowl8874]

dont throw it sir if you dont like send it to me here in Philippines

[@glennlove461]

start looking for oscillation at rails and then each device.. easy fix

[@MichaelBeeny]

Easy to say Glenn, not quite as easy to fix other people work. You can spend many hours looking for instability and end up with a poor PCB layout. We will be publishing the circuit in the next few days. We shall see.

[@glennlove461]

@@MichaelBeeny You found it will be watching...

[@larrygaines7462]

Fans and cool operation costs the least of all your gadgets. No brainer.

[@mfr58]

Having purchased a lot of stuff from China, I'm afraid their business mind and practice is very different from the likes of you and me. They really don't care about anything except getting your money. Of course there are some exceptions, but this is the general case in my long experience.

[@MichaelBeeny]

Sadly, in MANY cases you are right.

[@etiennecahurel4584]

traduction ia insupportable

[@MichaelBeeny]

Sorry, I have no control of any translation. This is auto generated by KZbin.

[@andymouse]

@@MichaelBeeny Tolly wibble screnvo ? LOL ! heyop dribble furdemstil 20 mW @1 swiktoggle !

[@MichaelBeeny]

@@andymouse Did I say that?? I think I needed a coffee.

[@andymouse]

@@MichaelBeeny :)

[@Robert-ps8fj]

Good Day Mr. Beeny and I am a DIY audio amplifier lover from the Philippines and one of your subscriber,,Sir Is a linear power supply a good power supply for a diy digital amp?.

[@MichaelBeeny]

SMPS are much better in regulation but for reliability I will always use a standard transformer type SP. For many people fixing a SMPS is a nonstarter. Most people could probably fix a standard power supply, little to fail in the first place. Welcome to my channel.