L12-2 CFP Output amp 120W*2 8R

Does anyone know which version of L12-2 this is? It shows darlington ouputs, but their part names are not....

L12-2_V2_skjema.jpg
 
I wonder how many faulty boards have been sold so far. I bet they are continued to be sold, not withdrawn as they should be. So, do all these people have to go and purchased more transistors? What about people that purchased completed boards. They might well have enough knowledge to wire them up but would not be happy making changes to the board.
Changing these transistors will not correct the very poor PCB layout. Remove the track to no ware! Unstable bias and way to hot thermally linked bias circuit.
Overhigh gain and FETs that are not FETs.
Add a QI pot to the V4.2 you have a much better solution.
I'm asking myself do I spend more money and time replacing transistors, re testing. I think it's time to move on.
A total of 1837 problematic transistors were sold C 4793 has the trademark SEN.
There is no problem with JJAG's trademark.
Approximately less than 10 faulty units were sold in total. And I have already resent it to the customer for replacement.
The L12-2 VER5 that I sell now undergoes three tests. Oscilloscope testing. Connect the speaker for trial listening. I have probably tested many sets. I haven't found any problems with the new ones.
 
So, as I planned, I replaced the last A1837 transistor on the PCB with the Toshiba A1837. The result is the same oscillation as previously. I can add pictures of PCB, Scope, and transistors. But I don't think I'ts necessary after Michael Beeny showed absolutely the same results.
There is no problem with JJAG's trademark.
Nice hardware, working with the selected brand of semiconductors only. Some amps require matched transistors or massive filter capacitors, but an amp that needs transistors of a special brand is something unique. I see such for the first time.
I ordered Wolverine PCB, the amp that simply works well.
Another question is, how with an unstable amp, LJM get a THD plot below 0.001% ...
 
The resistors on Vers 4 and Version 5 have 0,1 Ohm original. Dont know aubout Vers. 4.2 ,but ithing they have the same value.

Peter
Thank you, Peter.... I find the various VERSIONS of LJM something that makes buying them a difficult decision. Some versions are sometimes, in effect, a different amplifier and other times just one or two components different and easy to upgrade earlier versions. As you say, there is now a version 4.2, so how do I know what one I will get when I buy one? AliExpress sellers still show version 1 photos on their ads, so you have a chance of getting any one of the 6 variants
 
Last edited:
There is no problem with JJAG's trademark.
There is a problem: Google can't find either JJAG A1837 C4793 transistors or JJAG COMPANY at all!
It looks like those transistors were made in some garage or the basement near "LJM Laboratories"
One of the most important features of DIY devices is serviceability; while we have a circuit, drawings, and access to components, we can repair our device in the case of a fault. Here we don't have an official circuit, and the components used in that amp are not obsolete only, but impossible to obtain!
It is possible to afford hard-to-find components if the performance of the circuit is outstanding and unique, but why do we need it just for another average-quality device ( actually V5 amp with its stability is well below average quality)? There are SYMSYM, Honey Badger, GB150D, AKASA75, Ultimate Fidelity, and other DIY amps, that build from easily available components.
What is a glue-sealed transistor array on a DIY board? It's it is a trademark of greed, a sign of the decency absence.
 
@dakku, first, we are all friends here - there is no need to refer to anyone with "Sir". Though, I understand that as a habit of a courteous gentleman.

Correct, except connection to the input onboard "GND" should not be connected. Your signal input ground is already connected to the board ground. You can verify this with multimeter continuity check.

There is no need for duplicate connection of the input "GND" through an alternative route. (Occam's razor)
Below is a wiring diagram from the internet by Scott Campbell published in his great article entittled "A Complete Guide to Design and Build a Hi-Fi LM3886 Amplifier".
Go and read it - there is plenty of useful information in this article even though it is about another amplifier configuration. Basics of wiring are the same.


View attachment 1410060
Hi Berlusconi. I am a little puzzled that you suggest to dakku that he "duplicating" the inputs grounding connection? (POST1,024) I can only see ONE gnd connection the to the star earth on his diagram. The amp wiring diagram you show, (above) does however do it the very same way as I would do it, IE: with BOTH input grounds remaining separate all the way to the star-earth. Many thanks for explaining things.
 
  • Like
Reactions: Berlusconi
B
I can only see ONE gnd
Hi sonicles
Thanks for pointing to this: you're right. My bad.
L12-2 has GND so close to the input terminal that I've thought his purple lines (GND) have been connected to the input ground. That is the most common cause of hum. Input GND on the board is useful just when testing individual boards but when the both boards are connected, the input GND should be disconnected. It is impossible to overemphasize the need to use thck wire and the shortest path from GND to the star.
Thanks again @sonicles... and appologies to @dakku. His last schematics was correct.

EDIT: In fact Campbells schematics has double GND connection to the star. It should be single.
EDIT: When I think (contemplate) this issue I use less common concept "return path" instead of "ground" and even more confusing "earth".
 
Last edited:
  • Like
Reactions: sonicles
Thanks for replying to me Berlusconi. However, you say above: "EDIT: In fact Campbell's schematics has double GND connection to the star. It should be single.". I have to differ and say, in my opinion, Campbell's method is correct. I think dakku should also have used 2x individual wires from the input grounds to the Star Earth, (he shows just the one GREEN wire on one channel). Some builders short the incoming signal ground at the input sockets and have one wire back to the main earth point. Having two wires makes certain there are no earth loops before the signal is grounded, and it will result in less hum or other signal being induced into the signal input. HOWEVER, it is often true that both channels have their earth side (but not always) joined together inside the preamplifier, so maybe one should try both a single earth (as dakku shows) or use 2x separate earth wires as I usually prefer. (as does Cambell). One method should result in less noise pickup
 
  • Like
Reactions: Berlusconi
it is often true that that in both channels, the earth/ground side will, very likely, (but not always) be joined inside the preamplifier, so maybe one should try both a single earth (as dakku shows) or use 2x separate earth wires as I usually prefer. (as does Cambell). One method should result in less noise pickup
Indeed. This line of reasoning correctly involves the need to analyse the system as a whole, including the wiring of the signal cables. However, my reasoning is as follows: If we connect both GNDs to the star we end up with two conductors of different impedance hence creating unnecessary difference of potentials, however this difference low is. Let me again refer to the good old Occam's Razzor. (Ockham: pluralitas non est ponenda sine necessitate, “plurality should not be posited without necessity.”)
But you're absolutely right: different paths and cable connections should be investigated because you never know in advance, in the real world, which is the best, without empirical investigation. When unsure I measure distortion for different configurations using FFT analysis. Human hearing simply isn't reliable enough for this purpose.
Thanks for sharing your thoughts with me, and all of us.
 
  • Like
Reactions: sonicles
Thank you, Peter.... I find the various VERSIONS of LJM something that makes buying them a difficult decision. Some versions are sometimes, in effect, a different amplifier and other times just one or two components different and easy to upgrade earlier versions. As you say, there is now a version 4.2, so how do I know what one I will get when I buy one? AliExpress sellers still show version 1 photos on their ads, so you have a chance of getting any one of the 6 variants
I measured L12-2 repeatedly today. The reality is that I haven't found any unstable places. I just don't want to argue with some people about these things.
Because I found that 100% of people are discussing opinions. It's all meaningless. Actually, the unstable situation discovered by the friend above. Just because of his SMPS issue.
Just replace it with a circular transformer. Add two 4700UF capacitors. L12-2 will not oscillate.
Including some friends who are passionate about studying L12-2. Please provide me with your revision suggestions. I also tried. Actually, there won't be any changes.
More precisely, L12-2 VER5 is more stable than VER4.2. But many people told me that 4.2 is stable. VER5 is unstable. Actually, it's just that he used SMPS which caused high-frequency oscillation.
I repeatedly measure. Including the use of speakers. I didn't find any problems. Its signal-to-noise ratio. Distortion. The noise is excellent.
But I don't want to discuss these issues too much. After all, some people are too subjective. They always believe that they are right.
Additionally, I accidentally discovered during testing that a 10V1000UF feedback capacitor was used. Replacing with 25V470UF would be better.
It prevents the speaker from making a loud pop sound when turned on.
 
  • Thank You
Reactions: sonicles
I have 2x37VAC 400W "circular transformer" and 6x22000uF filter…
I measured another 20 sets of L12-2 VER5 today.
I mailed it to the merchant quite early. I thought there was a problem. So let the merchant mail it to me. Replace the parts.
In fact, I have repeatedly measured. No issues were found.
There is no problem with the transistor brand either.This has wasted a lot of my time.
There is only one circuit board. Because I tightened the screw too tightly. The 2SA1837 plastic was crushed. I just need to replace it.
The only component I replaced was 10V1000UF. Replace with 25V470UF. This way, when starting up. The charging time of the capacitor is reduced.
It will not produce a loud explosion sound on the speaker.
And I compared L12-2 VER4.2. The performance of VER5. Noise. Everything is better.
So I have decided not to make any further changes to L12-2 VER5. Upgrade.
 
Just for the record LJM.... I was only asking about the different versions that make me ask what version is better. I am not critical or found anything wrong with them myself. I was only reading what other people have said. I am a big fan of your L20 V9.2 amplifier. It is extremely transparent and have no trouble with them. I always recommend them to anyone. Thanks LJM for making them 🙂
 
So Version 5 with the Toshiba transistors. Look like that with 66000uF per rail the amp feels better than with SMPS. And here a two questions: what about PSRR, and/or who can ensure that either short-time stability issue will not cause an amplifier failure? Then the bias stability: Last time I set the bias to 4-5mV, but now after tests, the bias was 15mV. 300% deviation?! WTF! I Allowed the amp to cool, then adjusted the bias to 5mV, and then I heated the amp again near 45 degrees C. Bias being 7.5mV.
While there are no visible artefacts, the performance of the amp, due to the "strange" behaviour of the bias, is still questionable. So there is time for the Cosmos ADC and THD IMD Noise plots. Stay tuned to that channel...

... P.S. When I finished writing all of the above, I tested bias again at 3.6mV. Stability is here!
33.5V.jpg
34V_clip.jpg
34V_clip 2.jpg
 
Last edited:
  • Thank You
Reactions: Berlusconi